[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
snip

I don't see where RDH4 says the noise depends on frequency, or that the
EINR was calculated based on measurements at RF.

It doesn't AFAIK. However, the title of the Harris paper that derived
the 2.5/gm formula makes it clear he is talking about high frequencies.
Chech out the reference at the end of that chapter in RDH4.
There's a reference to Harris, page 946, item B27, but no reference to
RF.

You are quite right. In RDH4 is calls Harris' paper 'Fluctuations in
vacuum tube amplifiers and input systems'

However, the Burgess paper I have gives it its full title which is:

'Fluctuations in Space Charge Limited Currents at Moderately High
Frequencies, Part V - Fluctuations in Vacuum Tube Amplifiers and Input
Systems'

Clearly the RDH4 reference is to just one of a series of articles but
they all seem to be concerned with 'moderately high frequencies'. Indeed
Burgess talks exclusively about rf circuits.

On page 937, there's a table attributable to Harris with formulas for
noise and the table has 4 sections, triode amps, pentode amps, triode
mixers, pentode mixer, multigrid mixer.

The only simple formula for noise resistance is the first one for triode
noise resistance, 2.5/gm., and there's no mention of RF specificness, so
I think that 2.5/gm is the official triode EINR formula regardless of F.

I agree its official but I am not sure it is correct at af.

Well, Johnson noise in resistors is is a fact of life at all F.


Yes, but we are talking about shot noise which is a different beast
entirely. It is just convenient to model it as an equivalent inout
Johnson noise resistance.

If you just connect various values at the input of an amp you can listen
to it.

And when they say the input of a triode consists of an equivalent input
noise resistance between grid and cathode, then that's the model, and I
don't think there is any special consideration with RF. As bandwidth
extends up past the audio band, so does the noise and the noise voltage
at a rate of the square root of increase in BW. So thus having a bw =
2Mhz instead of only 20kHz, means noise should become 10 times greater.
Going to 20MHz would raise noise by 3.16 more times. Then if you have a
tuned circuit at say 1MHz, and BW = 18kHz as for a broadcast
AM radio station, then the reduction of BW is 18,000 / 1,000,000, or
about 1/50, so expect noise to be about 1/7 of when you have bw = 1Mhz.


Agreed. I would not argue against Johnson noise, I am just saying I am
not convinced by the 2.5/gm formula.


All the others look difficult to work out, but none have reactive
components in the formula.

If you use a 6BE6 as a mixer for two inputs to make one output then
it'll be noiser than if you have two triodes with a common anode load.

All that's important is that triodes are quieter than all other types of
tubes and that the higher the gm the better, and then all one needs to
consider is to have Ia high, and Ea low right at the input of a phono
stage, and hope it works out.

I agree it is important that triodes are quieter than all other types.
What is also important is how quiet they ahould be expected to be. If
you don't know this how can you possibly select tubes for low noise?

You don't have to know all the theory; just general concepts a bit. Then
you find out what works by experimentation. Its the hands on work that
keeps the studio recordists happy with their gear.

That's fine as long as the general concepts are accurate. So far I don't
fond 2.5/gm to be any where near reality.


Usually it does for MM. And when Denon invented an MC with less than 20
ohms Rout, and with very low Vout, a transformer with 1:10 ratio was the
normal way to up the Vo and yet keep noise lower than with MM. The ZR of
100 meant that 20 ohms became only 2,000 ohms from grid to 0V of a tube
input stage. If EINR of 2,000 ohms gives 2uV of noise, then 20 ohms
makes about 0.2uV, and when transformed up this noise becomes 2uV, and
when added to the series 2uV of the tube you get 2.8uV. The cart output
of 0.3mV has become 3mV, and snr = -60dB, and thus just OK considering
vinyl noise will be higher.

Now if you have a 2sk369 j-fet with 0.14uV input noise ( typical ) and
0.4mV of rated MC cart Vo, then snr = -69dB, and there's less LF noise
so less grumble in the signal.

It was j-fets that allowed a whole new generation of bugs to be
developed for spying because it was far easier to conceal a little
battery and a fet than have to wire up a triode. AND the fet picked up
more clean sound without the noise of itself ruining the time spent by
CIA operatives in vans outside the russian embassy.


The rules about noise are valid right down to dc once you have said that
the noise is due to a resistance.

So where you have bandwidth of dc to 20kHz, and say you have an R giving
2uV of noise, then if R is reduced by 10 then noise reduces by 1/3.16 or
about a third. And if the bandwidth is effectively reduced from 20khz to
50 as it is with an RIAA filter then noise is reduced the square root of
the amount of bandwidth reduction, ie, sq.rt ( 50 / 20,000 ) = 1 / 20,
or -26dB.

Where did that factor of 50 come from. According to Morgan Jones it is
nearer 1.
50 is the new bandwidth after RIAA is added to the circuit because the
response is dc to -3db at 50Hz.
Sorry that makes no sense to me. How does the bandwidth suddenly become
reduced to 50Hz?

A phono preamplifier is in fact a bandpass filter. Noise enters the amp
at the front end with wide bw maybe in excess of 50kHz.
But the RIAA filter has its first pole at 50Hz, -3dB point and
attenuation after that is first order for awhile, a bit of a flatter
curve around 1kHz, then back to first order at -6dB per octave again.

I suggest you set up a 6CG7 triode for test with RL about 100k and
bypassed Rk and about 3mA of Ia. You can measure the anode signal as we
have discussed to test noise. We can also place a 1M resistance from
grid to 0V to vastly increase the noise at the grid over having the grid
connected directly to 0V for tube noise testing. Now suppose the anode
load R plus Ra = 10k. If we place cap of say 0.318uF from anode to 0V,
will will see a dramatic reduction in noise.

Now with Rg = 1M at the the triode input, the noise bw of the 1M is
high, but then the R has some self C and then you have the miller C
between anode and grid of maybe 50pF so as F rises the noise measured at
the anode gets less because of the miller effect which is a negative
feedback effect. So don't expect much noise at the 6CG7 at above 10Mhz.
The capacitances beome low impedances which shunt the noise across
higher impedances involved.

But what we would find is that noise voltages above 50Hz would be
attenuated at 6dB/octave with 0.318uF from anode to 0V.

The RIAA filter acts very similarly to having just a singe 0.318uF from
anode to 0V on a 6CG7 set up as described.

The bandwidth of signal is measured between poles in the response. Now
the 6CG7 tested this way has BW from dc to 50Hz. But if we add a CR
coupling as we might use on a second amp stage then there may be a LF
pole at say 20Hz. So signal BW becomes 20Hz to 50Hz, or only 30Hz.
If the initial signal bw at the front end of the amp was dc to 20kHz,
then the bw reduction factor becomes 30/20,000 = 0.0015 and the noise
reduction factor = 0.038 = 28.2dB, a bit more than what we worked out
for bw = dc to 50Hz. The noise below 20Hz is missing.


Morgan Jones disagrees with you. He says the effective noise bandwidth
is 188Hz not 50Hz and also that you have to subtract the extra 19dB or
so gain you have to add to bring the response back at 1KHz which means
the overall noise improvement is only just over a couple of dB. He
cites the Texas Instruments Bi-FET manual IIRC as his source which I am
trying to find on the net now.



Knowing about noise and its bandwidth and the value of capacitors to
reduce it means that if we have a noisy signal from some device under
test and we want to see the hum voltage in the blur of noise on the CRO,
than all we need do is place a shunt C of the right value to shunt all
the HF junk above say 150Hz, and we can see the hum clearly.

The RIAA filter cleans up the signal from the phono cartridge which
would otherwise sound like a rock being dragged down a canyon were it
not for the boosting of mid-treble 40dB above the 10Hz signal level
during recording and the filtering in the playback preamp. Along with
reducing noise in the recording, the RIAA also cuts the distortions of
the first am stage a little because higher harmonics of fundementals get
attenuated more than the fundementals.

Our ancestors were clever, doncha think?

BW was 20kHz before RIAA.
Therefore BW reduction factor = 50Hz / 20,000Hz = 1/400. The noise
reduction factor is the square root of bandwidth reduction = sq.rt of
1/400 = 1/20 = -26dB.

Now in vinyl recording they boost the hell out of treble above the low
bass reference, +40dB at 22 kHz compared to 10Hz.
Depends what you are talking about. If you are talking about the
amplitude of the recorded groove then it is nearly constant with
frequency - there's just a small boost in the mid frequency range.
That's why ceramic cartridges need very little equalisation.

The reason a MM or MC cartridge gives a rising level with increasing
frequency is because its output is proportional to rate of change of flux.

Well OK, but they boost the signal sent to the cutting head for LPs.


Not as a rule, only as they approach the inner grooves and the linear
velocity reduces.

Say you just recorded with a flat signal. You'd not be able to fit bass
grooves on the record very well. High F treble signals would be
seriously affected by competing levels of noise.


Vinyl IS generally recorded with a near flat amplitude. Frequency
dependent boost on record is need because of the head characteristics
not to combat noise. Head cutters can cut bass into a record with no
problem. Its the playback carts that have the problem staying in the
groove because their compliance is frequency dependent.

On playback, noise noise would be intolerable.
So the RIAA filter flattens the music response and disposes of most of
the noise.
I am still not clear on this. I need to check what Jones says.

The disc recording process is a bit of a compromise.
The ears of most people don't notice LF noise as much as
noise above say 300Hz. Now because tubes make a fair amount of LF noise,
the theoretical -26dB noise reduction doesn't get achieved.
There will still be some noise. But on good vinyl its often tape hiss or
venue noise which is the highest noise of the recording. In FM radio HF
is boosted or "emphasised" and after detection the signal is
de-emphasized. Exit much noise.

That's different and is done deliberately to reduce noise. RIAA is not
for that reason.

But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.

No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency. A magnetic cart has a
rising output with frequency (just like a take head) because its output
is proportional to rate of change of flux. The top cut is necessary to
compensate for this.


Cheers

Ian

In fact if we use 1kHz as the reference, then the bass is cut, and
re-boosted i the playback amp and the noise in bass regions isn't much
reduced by RIAA; 100Hz signals are attenuated about -12dB compared to
1kHz. Noise at around 100Hz in the grooves and input amp isn't much
reduced but is amplified with the LF signal where most ppl agree the
extra bit of noise due to the vinyl and electronics does not matter due
to our ears being less sensitive to a given spl of noise as F goes low.
The noise in the bass music is low because its a small part of the whole
20kHz F range. But after 50Hz the signal recoverd from the record groove
has an increasing amplitude at about +6dB per octave. The snr of the
recovered signal gets better as F goes higher and then C of the RIAA
filter just flattens this signal and reduces noise enormously and
preserves a good snr. And we need all the noise attenuation we can get
because our ears become very sensitive to higher F.

I'm speaking of a concept. The exact numbers don't matter. Just use an
amp with low enough noise and keep other sources of noise low say by
cleaning records well and you'll enjoy a vinyl.

If your'e gonna build an MC amp its worth taking a little extra trouble
compared to the lack of trouble taken by many bottom end budget disc
replay systems.

Patrick Turner.

[Patrick Turner]

Ian Bell wrote:

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
snip

I don't see where RDH4 says the noise depends on frequency, or that the
EINR was calculated based on measurements at RF.

It doesn't AFAIK. However, the title of the Harris paper that derived
the 2.5/gm formula makes it clear he is talking about high frequencies.
Chech out the reference at the end of that chapter in RDH4.
There's a reference to Harris, page 946, item B27, but no reference to
RF.

You are quite right. In RDH4 is calls Harris' paper 'Fluctuations in
vacuum tube amplifiers and input systems'

However, the Burgess paper I have gives it its full title which is:

'Fluctuations in Space Charge Limited Currents at Moderately High
Frequencies, Part V - Fluctuations in Vacuum Tube Amplifiers and Input
Systems'

Clearly the RDH4 reference is to just one of a series of articles but
they all seem to be concerned with 'moderately high frequencies'. Indeed
Burgess talks exclusively about rf circuits.

On page 937, there's a table attributable to Harris with formulas for
noise and the table has 4 sections, triode amps, pentode amps, triode
mixers, pentode mixer, multigrid mixer.

The only simple formula for noise resistance is the first one for triode
noise resistance, 2.5/gm., and there's no mention of RF specificness, so
I think that 2.5/gm is the official triode EINR formula regardless of F.

I agree its official but I am not sure it is correct at af.

Well, Johnson noise in resistors is is a fact of life at all F.


Yes, but we are talking about shot noise which is a different beast
entirely. It is just convenient to model it as an equivalent inout
Johnson noise resistance.

Somebody with far more insight than I have decided an imaginery
resistance between grid and cathode modelled tube noise OK.

I'm wondering that if I look further at this noise issue, will I make
more silent amps?

If you just connect various values at the input of an amp you can listen
to it.

And when they say the input of a triode consists of an equivalent input
noise resistance between grid and cathode, then that's the model, and I
don't think there is any special consideration with RF. As bandwidth
extends up past the audio band, so does the noise and the noise voltage
at a rate of the square root of increase in BW. So thus having a bw =
2Mhz instead of only 20kHz, means noise should become 10 times greater.
Going to 20MHz would raise noise by 3.16 more times. Then if you have a
tuned circuit at say 1MHz, and BW = 18kHz as for a broadcast
AM radio station, then the reduction of BW is 18,000 / 1,000,000, or
about 1/50, so expect noise to be about 1/7 of when you have bw = 1Mhz.


Agreed. I would not argue against Johnson noise, I am just saying I am
not convinced by the 2.5/gm formula.

I'm not much convinced either. But when Rin is lower, there is less
noise. Even with bjts, paralleling a lot of them with very low base
current leads to more silent bjts. Less R.


All the others look difficult to work out, but none have reactive
components in the formula.

If you use a 6BE6 as a mixer for two inputs to make one output then
it'll be noiser than if you have two triodes with a common anode load.

All that's important is that triodes are quieter than all other types of
tubes and that the higher the gm the better, and then all one needs to
consider is to have Ia high, and Ea low right at the input of a phono
stage, and hope it works out.

I agree it is important that triodes are quieter than all other types.
What is also important is how quiet they ahould be expected to be. If
you don't know this how can you possibly select tubes for low noise?

You don't have to know all the theory; just general concepts a bit. Then
you find out what works by experimentation. Its the hands on work that
keeps the studio recordists happy with their gear.

That's fine as long as the general concepts are accurate. So far I don't
fond 2.5/gm to be any where near reality.

So what is a better solution?


Usually it does for MM. And when Denon invented an MC with less than 20
ohms Rout, and with very low Vout, a transformer with 1:10 ratio was the
normal way to up the Vo and yet keep noise lower than with MM. The ZR of
100 meant that 20 ohms became only 2,000 ohms from grid to 0V of a tube
input stage. If EINR of 2,000 ohms gives 2uV of noise, then 20 ohms
makes about 0.2uV, and when transformed up this noise becomes 2uV, and
when added to the series 2uV of the tube you get 2.8uV. The cart output
of 0.3mV has become 3mV, and snr = -60dB, and thus just OK considering
vinyl noise will be higher.

Now if you have a 2sk369 j-fet with 0.14uV input noise ( typical ) and
0.4mV of rated MC cart Vo, then snr = -69dB, and there's less LF noise
so less grumble in the signal.

It was j-fets that allowed a whole new generation of bugs to be
developed for spying because it was far easier to conceal a little
battery and a fet than have to wire up a triode. AND the fet picked up
more clean sound without the noise of itself ruining the time spent by
CIA operatives in vans outside the russian embassy.


The rules about noise are valid right down to dc once you have said that
the noise is due to a resistance.

So where you have bandwidth of dc to 20kHz, and say you have an R giving
2uV of noise, then if R is reduced by 10 then noise reduces by 1/3.16 or
about a third. And if the bandwidth is effectively reduced from 20khz to
50 as it is with an RIAA filter then noise is reduced the square root of
the amount of bandwidth reduction, ie, sq.rt ( 50 / 20,000 ) = 1 / 20,
or -26dB.

Where did that factor of 50 come from. According to Morgan Jones it is
nearer 1.
50 is the new bandwidth after RIAA is added to the circuit because the
response is dc to -3db at 50Hz.
Sorry that makes no sense to me. How does the bandwidth suddenly become
reduced to 50Hz?

A phono preamplifier is in fact a bandpass filter. Noise enters the amp
at the front end with wide bw maybe in excess of 50kHz.
But the RIAA filter has its first pole at 50Hz, -3dB point and
attenuation after that is first order for awhile, a bit of a flatter
curve around 1kHz, then back to first order at -6dB per octave again.

I suggest you set up a 6CG7 triode for test with RL about 100k and
bypassed Rk and about 3mA of Ia. You can measure the anode signal as we
have discussed to test noise. We can also place a 1M resistance from
grid to 0V to vastly increase the noise at the grid over having the grid
connected directly to 0V for tube noise testing. Now suppose the anode
load R plus Ra = 10k. If we place cap of say 0.318uF from anode to 0V,
will will see a dramatic reduction in noise.

Now with Rg = 1M at the the triode input, the noise bw of the 1M is
high, but then the R has some self C and then you have the miller C
between anode and grid of maybe 50pF so as F rises the noise measured at
the anode gets less because of the miller effect which is a negative
feedback effect. So don't expect much noise at the 6CG7 at above 10Mhz.
The capacitances beome low impedances which shunt the noise across
higher impedances involved.

But what we would find is that noise voltages above 50Hz would be
attenuated at 6dB/octave with 0.318uF from anode to 0V.

The RIAA filter acts very similarly to having just a singe 0.318uF from
anode to 0V on a 6CG7 set up as described.

The bandwidth of signal is measured between poles in the response. Now
the 6CG7 tested this way has BW from dc to 50Hz. But if we add a CR
coupling as we might use on a second amp stage then there may be a LF
pole at say 20Hz. So signal BW becomes 20Hz to 50Hz, or only 30Hz.
If the initial signal bw at the front end of the amp was dc to 20kHz,
then the bw reduction factor becomes 30/20,000 = 0.0015 and the noise
reduction factor = 0.038 = 28.2dB, a bit more than what we worked out
for bw = dc to 50Hz. The noise below 20Hz is missing.


Morgan Jones disagrees with you. He says the effective noise bandwidth
is 188Hz not 50Hz and also that you have to subtract the extra 19dB or
so gain you have to add to bring the response back at 1KHz which means
the overall noise improvement is only just over a couple of dB. He
cites the Texas Instruments Bi-FET manual IIRC as his source which I am
trying to find on the net now.

OK, I thought signal BW was defined by the -3dB points.

But yes, if 1khz is the reference, and you have a low level bass signal
then snr at bass does not improve at bass.

But consider a microphone amp. Response is flat, and noise at input is
all amplified by a uniform gain.
So if there was 2mV of signal and 2uV of noise at the input, 20kHz bw,
snr = -60dB, and this will also be at the output, unless the amp
contributes considerable other noise in the path.

The same amp when used with an RIAA filter in the signal path has the
same noise below 50Hz, but the bass signals are very low, say 0.2mV at
low bass compared to mid F based around 1kHz at 2mV and so bass signal
snr is worse, but not a huge fraction of total. 2mV at 1kHz enters the
amp and noise and 1khz are both attenuated equally. Then the treble is
boosted so that 20mV appears at 22kHz so the noise above 1kHz begins to
be much lessened while the signal is flattened.

Just how quiet things sound depends on choice of low noise parts at the
amp input.



Knowing about noise and its bandwidth and the value of capacitors to
reduce it means that if we have a noisy signal from some device under
test and we want to see the hum voltage in the blur of noise on the CRO,
than all we need do is place a shunt C of the right value to shunt all
the HF junk above say 150Hz, and we can see the hum clearly.

The RIAA filter cleans up the signal from the phono cartridge which
would otherwise sound like a rock being dragged down a canyon were it
not for the boosting of mid-treble 40dB above the 10Hz signal level
during recording and the filtering in the playback preamp. Along with
reducing noise in the recording, the RIAA also cuts the distortions of
the first am stage a little because higher harmonics of fundementals get
attenuated more than the fundementals.

Our ancestors were clever, doncha think?

BW was 20kHz before RIAA.
Therefore BW reduction factor = 50Hz / 20,000Hz = 1/400. The noise
reduction factor is the square root of bandwidth reduction = sq.rt of
1/400 = 1/20 = -26dB.

Now in vinyl recording they boost the hell out of treble above the low
bass reference, +40dB at 22 kHz compared to 10Hz.
Depends what you are talking about. If you are talking about the
amplitude of the recorded groove then it is nearly constant with
frequency - there's just a small boost in the mid frequency range.
That's why ceramic cartridges need very little equalisation.

The reason a MM or MC cartridge gives a rising level with increasing
frequency is because its output is proportional to rate of change of flux.

Well OK, but they boost the signal sent to the cutting head for LPs.


Not as a rule, only as they approach the inner grooves and the linear
velocity reduces.

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.



Say you just recorded with a flat signal. You'd not be able to fit bass
grooves on the record very well. High F treble signals would be
seriously affected by competing levels of noise.


Vinyl IS generally recorded with a near flat amplitude. Frequency
dependent boost on record is need because of the head characteristics
not to combat noise. Head cutters can cut bass into a record with no
problem. Its the playback carts that have the problem staying in the
groove because their compliance is frequency dependent.

I've always thought there was considerable bass cut and treble boost in
recording.

But I have not recorded anything or cut a record. All I know is that
records sound fine with RIAA, and lousy without it.

On playback, noise noise would be intolerable.
So the RIAA filter flattens the music response and disposes of most of
the noise.
I am still not clear on this. I need to check what Jones says.

The disc recording process is a bit of a compromise.
The ears of most people don't notice LF noise as much as
noise above say 300Hz. Now because tubes make a fair amount of LF noise,
the theoretical -26dB noise reduction doesn't get achieved.
There will still be some noise. But on good vinyl its often tape hiss or
venue noise which is the highest noise of the recording. In FM radio HF
is boosted or "emphasised" and after detection the signal is
de-emphasized. Exit much noise.

That's different and is done deliberately to reduce noise. RIAA is not
for that reason.

But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.

No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency. A magnetic cart has a
rising output with frequency (just like a take head) because its output
is proportional to rate of change of flux. The top cut is necessary to
compensate for this.

Well, I'll have to think more about that but it won't alter how I make
amplifiers much.

Patrick Turner.

Cheers

Ian

In fact if we use 1kHz as the reference, then the bass is cut, and
re-boosted i the playback amp and the noise in bass regions isn't much
reduced by RIAA; 100Hz signals are attenuated about -12dB compared to
1kHz. Noise at around 100Hz in the grooves and input amp isn't much
reduced but is amplified with the LF signal where most ppl agree the
extra bit of noise due to the vinyl and electronics does not matter due
to our ears being less sensitive to a given spl of noise as F goes low.
The noise in the bass music is low because its a small part of the whole
20kHz F range. But after 50Hz the signal recoverd from the record groove
has an increasing amplitude at about +6dB per octave. The snr of the
recovered signal gets better as F goes higher and then C of the RIAA
filter just flattens this signal and reduces noise enormously and
preserves a good snr. And we need all the noise attenuation we can get
because our ears become very sensitive to higher F.

I'm speaking of a concept. The exact numbers don't matter. Just use an
amp with low enough noise and keep other sources of noise low say by
cleaning records well and you'll enjoy a vinyl.

If your'e gonna build an MC amp its worth taking a little extra trouble
compared to the lack of trouble taken by many bottom end budget disc
replay systems.

Patrick Turner.

[John Byrns]

In article ,
Ian Bell wrote:

Patrick Turner wrote:

But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.

No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.

That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
snip

I don't see where RDH4 says the noise depends on frequency, or that the
EINR was calculated based on measurements at RF.

It doesn't AFAIK. However, the title of the Harris paper that derived
the 2.5/gm formula makes it clear he is talking about high frequencies.
Chech out the reference at the end of that chapter in RDH4.
There's a reference to Harris, page 946, item B27, but no reference to
RF.

You are quite right. In RDH4 is calls Harris' paper 'Fluctuations in
vacuum tube amplifiers and input systems'

However, the Burgess paper I have gives it its full title which is:

'Fluctuations in Space Charge Limited Currents at Moderately High
Frequencies, Part V - Fluctuations in Vacuum Tube Amplifiers and Input
Systems'

Clearly the RDH4 reference is to just one of a series of articles but
they all seem to be concerned with 'moderately high frequencies'. Indeed
Burgess talks exclusively about rf circuits.

On page 937, there's a table attributable to Harris with formulas for
noise and the table has 4 sections, triode amps, pentode amps, triode
mixers, pentode mixer, multigrid mixer.

The only simple formula for noise resistance is the first one for triode
noise resistance, 2.5/gm., and there's no mention of RF specificness, so
I think that 2.5/gm is the official triode EINR formula regardless of F.

I agree its official but I am not sure it is correct at af.
Well, Johnson noise in resistors is is a fact of life at all F.

Yes, but we are talking about shot noise which is a different beast
entirely. It is just convenient to model it as an equivalent inout
Johnson noise resistance.

Somebody with far more insight than I have decided an imaginery
resistance between grid and cathode modelled tube noise OK.

I'm wondering that if I look further at this noise issue, will I make
more silent amps?

Depends. If you simply believe the 2.5/gm mantra without question then I
guess you probably won't.


If you just connect various values at the input of an amp you can listen
to it.

And when they say the input of a triode consists of an equivalent input
noise resistance between grid and cathode, then that's the model, and I
don't think there is any special consideration with RF. As bandwidth
extends up past the audio band, so does the noise and the noise voltage
at a rate of the square root of increase in BW. So thus having a bw =
2Mhz instead of only 20kHz, means noise should become 10 times greater.
Going to 20MHz would raise noise by 3.16 more times. Then if you have a
tuned circuit at say 1MHz, and BW = 18kHz as for a broadcast
AM radio station, then the reduction of BW is 18,000 / 1,000,000, or
about 1/50, so expect noise to be about 1/7 of when you have bw = 1Mhz.

Agreed. I would not argue against Johnson noise, I am just saying I am
not convinced by the 2.5/gm formula.

I'm not much convinced either. But when Rin is lower, there is less
noise. Even with bjts, paralleling a lot of them with very low base
current leads to more silent bjts. Less R.


I can understand paralleling because the signal is coherent so it
increases by 6dB when you parallel two devices but the noise, being
separate uncorrelated noise sources, add their powers so the noise
increses only by 3dB. Result S/N improves by 3dB. It is exactly the same
with track width in magnetic rtape recorders.


All the others look difficult to work out, but none have reactive
components in the formula.

If you use a 6BE6 as a mixer for two inputs to make one output then
it'll be noiser than if you have two triodes with a common anode load.

All that's important is that triodes are quieter than all other types of
tubes and that the higher the gm the better, and then all one needs to
consider is to have Ia high, and Ea low right at the input of a phono
stage, and hope it works out.

I agree it is important that triodes are quieter than all other types.
What is also important is how quiet they ahould be expected to be. If
you don't know this how can you possibly select tubes for low noise?
You don't have to know all the theory; just general concepts a bit. Then
you find out what works by experimentation. Its the hands on work that
keeps the studio recordists happy with their gear.
That's fine as long as the general concepts are accurate. So far I don't
fond 2.5/gm to be any where near reality.

So what is a better solution?


Now that is a really good question. Someone needs to do some work first
to verify that there is some relation between gm and EIN and secondly to
quantify it. Actually that's rubbish. All we need to know is what are
the achievable EIN values for popular audio tubes. So some experiments
are in order. I was planning to make the 40dB amp for measuring 6CG7
noise for my mic pres anyway so I guess I'll be the one to make a start.


Usually it does for MM. And when Denon invented an MC with less than 20
ohms Rout, and with very low Vout, a transformer with 1:10 ratio was the
normal way to up the Vo and yet keep noise lower than with MM. The ZR of
100 meant that 20 ohms became only 2,000 ohms from grid to 0V of a tube
input stage. If EINR of 2,000 ohms gives 2uV of noise, then 20 ohms
makes about 0.2uV, and when transformed up this noise becomes 2uV, and
when added to the series 2uV of the tube you get 2.8uV. The cart output
of 0.3mV has become 3mV, and snr = -60dB, and thus just OK considering
vinyl noise will be higher.

Now if you have a 2sk369 j-fet with 0.14uV input noise ( typical ) and
0.4mV of rated MC cart Vo, then snr = -69dB, and there's less LF noise
so less grumble in the signal.

It was j-fets that allowed a whole new generation of bugs to be
developed for spying because it was far easier to conceal a little
battery and a fet than have to wire up a triode. AND the fet picked up
more clean sound without the noise of itself ruining the time spent by
CIA operatives in vans outside the russian embassy.


The rules about noise are valid right down to dc once you have said that
the noise is due to a resistance.

So where you have bandwidth of dc to 20kHz, and say you have an R giving
2uV of noise, then if R is reduced by 10 then noise reduces by 1/3.16 or
about a third. And if the bandwidth is effectively reduced from 20khz to
50 as it is with an RIAA filter then noise is reduced the square root of
the amount of bandwidth reduction, ie, sq.rt ( 50 / 20,000 ) = 1 / 20,
or -26dB.

Where did that factor of 50 come from. According to Morgan Jones it is
nearer 1.
50 is the new bandwidth after RIAA is added to the circuit because the
response is dc to -3db at 50Hz.
Sorry that makes no sense to me. How does the bandwidth suddenly become
reduced to 50Hz?
A phono preamplifier is in fact a bandpass filter. Noise enters the amp
at the front end with wide bw maybe in excess of 50kHz.
But the RIAA filter has its first pole at 50Hz, -3dB point and
attenuation after that is first order for awhile, a bit of a flatter
curve around 1kHz, then back to first order at -6dB per octave again.

I suggest you set up a 6CG7 triode for test with RL about 100k and
bypassed Rk and about 3mA of Ia. You can measure the anode signal as we
have discussed to test noise. We can also place a 1M resistance from
grid to 0V to vastly increase the noise at the grid over having the grid
connected directly to 0V for tube noise testing. Now suppose the anode
load R plus Ra = 10k. If we place cap of say 0.318uF from anode to 0V,
will will see a dramatic reduction in noise.

Now with Rg = 1M at the the triode input, the noise bw of the 1M is
high, but then the R has some self C and then you have the miller C
between anode and grid of maybe 50pF so as F rises the noise measured at
the anode gets less because of the miller effect which is a negative
feedback effect. So don't expect much noise at the 6CG7 at above 10Mhz.
The capacitances beome low impedances which shunt the noise across
higher impedances involved.

But what we would find is that noise voltages above 50Hz would be
attenuated at 6dB/octave with 0.318uF from anode to 0V.

The RIAA filter acts very similarly to having just a singe 0.318uF from
anode to 0V on a 6CG7 set up as described.

The bandwidth of signal is measured between poles in the response. Now
the 6CG7 tested this way has BW from dc to 50Hz. But if we add a CR
coupling as we might use on a second amp stage then there may be a LF
pole at say 20Hz. So signal BW becomes 20Hz to 50Hz, or only 30Hz.
If the initial signal bw at the front end of the amp was dc to 20kHz,
then the bw reduction factor becomes 30/20,000 = 0.0015 and the noise
reduction factor = 0.038 = 28.2dB, a bit more than what we worked out
for bw = dc to 50Hz. The noise below 20Hz is missing.

Morgan Jones disagrees with you. He says the effective noise bandwidth
is 188Hz not 50Hz and also that you have to subtract the extra 19dB or
so gain you have to add to bring the response back at 1KHz which means
the overall noise improvement is only just over a couple of dB. He
cites the Texas Instruments Bi-FET manual IIRC as his source which I am
trying to find on the net now.

OK, I thought signal BW was defined by the -3dB points.

But yes, if 1khz is the reference, and you have a low level bass signal
then snr at bass does not improve at bass.

But consider a microphone amp. Response is flat, and noise at input is
all amplified by a uniform gain.
So if there was 2mV of signal and 2uV of noise at the input, 20kHz bw,
snr = -60dB, and this will also be at the output, unless the amp
contributes considerable other noise in the path.


Agreed.


The same amp when used with an RIAA filter in the signal path has the
same noise below 50Hz, but the bass signals are very low, say 0.2mV at
low bass compared to mid F based around 1kHz at 2mV and so bass signal
snr is worse, but not a huge fraction of total. 2mV at 1kHz enters the
amp and noise and 1khz are both attenuated equally. Then the treble is
boosted so that 20mV appears at 22kHz so the noise above 1kHz begins to
be much lessened while the signal is flattened.

Just how quiet things sound depends on choice of low noise parts at the
amp input.


Knowing about noise and its bandwidth and the value of capacitors to
reduce it means that if we have a noisy signal from some device under
test and we want to see the hum voltage in the blur of noise on the CRO,
than all we need do is place a shunt C of the right value to shunt all
the HF junk above say 150Hz, and we can see the hum clearly.

The RIAA filter cleans up the signal from the phono cartridge which
would otherwise sound like a rock being dragged down a canyon were it
not for the boosting of mid-treble 40dB above the 10Hz signal level
during recording and the filtering in the playback preamp. Along with
reducing noise in the recording, the RIAA also cuts the distortions of
the first am stage a little because higher harmonics of fundementals get
attenuated more than the fundementals.

Our ancestors were clever, doncha think?

BW was 20kHz before RIAA.
Therefore BW reduction factor = 50Hz / 20,000Hz = 1/400. The noise
reduction factor is the square root of bandwidth reduction = sq.rt of
1/400 = 1/20 = -26dB.

Now in vinyl recording they boost the hell out of treble above the low
bass reference, +40dB at 22 kHz compared to 10Hz.
Depends what you are talking about. If you are talking about the
amplitude of the recorded groove then it is nearly constant with
frequency - there's just a small boost in the mid frequency range.
That's why ceramic cartridges need very little equalisation.

The reason a MM or MC cartridge gives a rising level with increasing
frequency is because its output is proportional to rate of change of flux.
Well OK, but they boost the signal sent to the cutting head for LPs.

Not as a rule, only as they approach the inner grooves and the linear
velocity reduces.

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.


I'll check him out.



Say you just recorded with a flat signal. You'd not be able to fit bass
grooves on the record very well. High F treble signals would be
seriously affected by competing levels of noise.

Vinyl IS generally recorded with a near flat amplitude. Frequency
dependent boost on record is need because of the head characteristics
not to combat noise. Head cutters can cut bass into a record with no
problem. Its the playback carts that have the problem staying in the
groove because their compliance is frequency dependent.

I've always thought there was considerable bass cut and treble boost in
recording.


This is a common misconception.


But I have not recorded anything or cut a record. All I know is that
records sound fine with RIAA, and lousy without it.

That is true, but only for a magnetic cartridge of course.


On playback, noise noise would be intolerable.
So the RIAA filter flattens the music response and disposes of most of
the noise.
I am still not clear on this. I need to check what Jones says.
The disc recording process is a bit of a compromise.
The ears of most people don't notice LF noise as much as
noise above say 300Hz. Now because tubes make a fair amount of LF noise,
the theoretical -26dB noise reduction doesn't get achieved.
There will still be some noise. But on good vinyl its often tape hiss or
venue noise which is the highest noise of the recording. In FM radio HF
is boosted or "emphasised" and after detection the signal is
de-emphasized. Exit much noise.
That's different and is done deliberately to reduce noise. RIAA is not
for that reason.
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency. A magnetic cart has a
rising output with frequency (just like a take head) because its output
is proportional to rate of change of flux. The top cut is necessary to
compensate for this.

Well, I'll have to think more about that but it won't alter how I make
amplifiers much.


No, please don't. From what I have seen your amps are pretty good and
RIAA is the right thing to use for magnetic cartridges.


Cheers

Ian
Patrick Turner.
Cheers

Ian
In fact if we use 1kHz as the reference, then the bass is cut, and
re-boosted i the playback amp and the noise in bass regions isn't much
reduced by RIAA; 100Hz signals are attenuated about -12dB compared to
1kHz. Noise at around 100Hz in the grooves and input amp isn't much
reduced but is amplified with the LF signal where most ppl agree the
extra bit of noise due to the vinyl and electronics does not matter due
to our ears being less sensitive to a given spl of noise as F goes low.
The noise in the bass music is low because its a small part of the whole
20kHz F range. But after 50Hz the signal recoverd from the record groove
has an increasing amplitude at about +6dB per octave. The snr of the
recovered signal gets better as F goes higher and then C of the RIAA
filter just flattens this signal and reduces noise enormously and
preserves a good snr. And we need all the noise attenuation we can get
because our ears become very sensitive to higher F.

I'm speaking of a concept. The exact numbers don't matter. Just use an
amp with low enough noise and keep other sources of noise low say by
cleaning records well and you'll enjoy a vinyl.

If your'e gonna build an MC amp its worth taking a little extra trouble
compared to the lack of trouble taken by many bottom end budget disc
replay systems.

Patrick Turner.

[John Byrns]

In article ,
Patrick Turner wrote:

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.

Sounds like he is talking about the ultrasonic time constant used in the
recording equalization network for velocity type cutter heads. We have
talked about this time constant here before, I think it is called the
4th time constant, or is it the 5th?

I don't know about instability issues, but it is obviously impractical
to keep the treble boost going up to indefinitely high frequencies, how
high would we go, 100 MHz? How much extra gain would that require? Out
of practical necessity we must include an ultra sonic pole in the
equalizer for a velocity cutter head.

This same problem exists with the preemphasis/deemphasis used in FM
broadcasting, only nobody worries about it because people are not as
picky about their FM reception as they are about vinyl playback.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.

That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".



Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Cheers

Ian

[Patrick Turner]

Ian Bell wrote:

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
snip

snip,

Somebody with far more insight than I have decided an imaginery
resistance between grid and cathode modelled tube noise OK.

I'm wondering that if I look further at this noise issue, will I make
more silent amps?

Depends. If you simply believe the 2.5/gm mantra without question then I
guess you probably won't.

I am a questioning man. I am a PITA for the extremist theoretician,
because I ask "And what do we get in practice?"

So, OK, to me I had a choice of persuing low noise with careful
selection of high gm triodes all paralleled up and then I thought I
could use one j-fet costing $1.0 instead. That iddy biddy j-fet, the
2SK369 will produce 1% THD per volt of output with most loads so that
with 5V out you'd get 5% THD. Its a quite terrible result compared to a
we set up 6CG7. BUT, the THD declines towards 0.0% at 0.0Vo, so that if
the Vo was 0.01V, THD should have declined to 0.01%, which is quite low.

And so I use a j-fet at my phono amp inputs. With MM, the j-fet has a
lot of local current FB from its larger Rs to keep its gain low and the
worst case is with the possible 50mV input at 22kHz. If the fet gain is
10, you have 0.5Vo, and the first harmonic is at 44khz and inaudible.
But in fact high levels of treble in music don't much occur.

With MC, input at 10Hz is perhaps 0.03mV, at 1khz, perhaps 0.3mV, and at
HF perhaps 3.0mV. So THD is low even with less local current FB and more
fet gain. When I test a phono preamp I apply a level sig gene signal
through a recording RIAA filter.
A fet can produce much higher output than required and still have low
THD.

If one wants less THD then one can use a balanced input with two j-fets
because most j-fets produce similar distortion dominated by 2H like a
triode.

snip,

So what is a better solution?


Now that is a really good question. Someone needs to do some work first
to verify that there is some relation between gm and EIN and secondly to
quantify it. Actually that's rubbish. All we need to know is what are
the achievable EIN values for popular audio tubes. So some experiments
are in order. I was planning to make the 40dB amp for measuring 6CG7
noise for my mic pres anyway so I guess I'll be the one to make a start.

Getting less than 1uV noise at the input of any tube is problematical,
especially with 6CG7, 12AT7, 6DJ8, 12AX7, 12AY7, 12AU7 and many other
pentodes in triode mode.

But feel welcome to try things.

Maybe consider some higher gm tubes like 6C45pi or 417 or E80F/6BX6,6EH7
in triode and at various Ia and see what happens.


snip

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.


I'll check him out.

He was i think the first to use a high gm j-fet to drive a triode in
cascode.

Its very simple and it works.

snip,


Well, I'll have to think more about that but it won't alter how I make
amplifiers much.


No, please don't. From what I have seen your amps are pretty good and
RIAA is the right thing to use for magnetic cartridges.

I can't be too wrong then?

Patrick Turner.

[Patrick Turner]

John Byrns wrote:

In article ,
Patrick Turner wrote:

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.

Sounds like he is talking about the ultrasonic time constant used in the
recording equalization network for velocity type cutter heads. We have
talked about this time constant here before, I think it is called the
4th time constant, or is it the 5th?

some records have low bass removed, 3,180uS, then you have standard
318us, then 75us, then ultrasonic at say 7.5uS, which isn't very
critical.

I don't know about instability issues, but it is obviously impractical
to keep the treble boost going up to indefinitely high frequencies, how
high would we go, 100 MHz? How much extra gain would that require? Out
of practical necessity we must include an ultra sonic pole in the
equalizer for a velocity cutter head.

The HF cutting head amp boost is done by using NFB. Boost cannot be
infinite, and there is a level plateau in response above 50kHz.
So, in the playback amp we level our correction to suit.

Allen Wright talks about it in his Tube Amp Cookbook, and give anecdotal
evidence that such fine tuning of RIAA preamp networks makes the vinyl
experience optimal.

When vinyl was king for so many, there were agonising groans from the
hi-fi cognescenti if a manufacturer's RIAA R&C filter was even +/-
0.25dB inaccurate.

But some hi-fi sets I saw made in the 1960s just had one R&C filter with
a -3dB pole at 50Hz and that was that, and so you got slightly more bass
relative to treble, and it compensated for poor bass response elsewhere
in the amps and everyone thought the rock and roll sounded marvellous.
And jazz. And even Motzart. My mum used to use a Kriesler system like
this. I still have it, much modified though.

There was **so much** processing and equalisation done in studios that
what RIAA does is only ever "about right" according to what the studio
guy thought sounded right to get sales.



This same problem exists with the preemphasis/deemphasis used in FM
broadcasting, only nobody worries about it because people are not as
picky about their FM reception as they are about vinyl playback.

Some ppl are very picky. But fact is without the treble
emphasis/de-emphasis, noise in FM would be worse.

And with a boosted HF, the waves are easier to decode when F gets higher
when trying to synthesize the L and R channels which are derived from a
sub carrier with a rather low 38kHz frequency. FM would have been better
if the sub carrier was at 100kHz and the pilot tone was at say 30kHz,
but the boffins on 1955 forsaw multiplexing and the possibilities of
other sub-carriers F at 67kHz and 96khz beckoned.

So the 1961 Zenith-GE agreed FM system was lowest common denominator.
But it sounds a lot better than AM radio. Live FM broadcasting when it
happens is still very very good to listen to. All the gear is in a
little suitacse type of box and the signal goes to satelites and its all
far more complex than live broadcasts in 1963, but it still sounds good.
My mum used to listen to live concerts on AM from Sydney Town Hall in
the 1960s, but the damn Kreisler gram-receiver was a bunch of crap with
AM audio BW of only 3kHz. We still enjoyed Motzart drifting around the
house.
Nobody I knew in 1960 ever had an AM receiver capable of a flat 10kHz if
it was being transmitted on AM. And it was transmitted then if it didn't
interfere with other stations, so AM could be almost as good as FM, if
you had the gear, but nobody did. Nobody had tweeters capable of a flat
response to 20kHz either.

When FM came to Oz in around 1972, the local electronics industry was in
its death throes as tariffs on cheap asian mades were being reduced.
FM sets were horribly complex like TV sets and so the only economic way
of providing receievrs for FM sets was to allow the cheap new chip
filled radio sets to flood the country. Integrated circuit chips
revolutionised the information and enetertainment industry. AM took a
real back seat and its bandwidth in most SS sets was a paltry 1.5kHz if
you were lucky and with poor selectivity. So SS gear really killed much
AM listening. But we still have AM stations and with a good set they
sound well, and there is any official emphasis except that some stations
boost bass and boost treble and then compress and de-ess and use MP3
music files and it is then OK on the poor cheap hi-fi sets and ghetto
blasters the lower orders use.

Only a few FM stations and AM station are worth listening to.



Patrick Turner.



--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Patrick Turner]

Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.

That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".


Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

The 75uS time constant in an RIAA network give a preamp response which
means the HF above 1kHz is down -3dB at 2,112Hz. above this F the rate
of attenuation becomes close to -6dB per octave so that if a 22kHz
signal enters the preamp it is about 20dB below the 1khz reference level
at the preamp output.

To get a flat response with vinyl and a magnetic cart, the signal from
the record contains a boosted HF so that a 22khz signal is 20dB above
the 1 khz reference level.

The +/-12dB points in the process are around the 10khz point, see the
official RIAA attenuation charts.

Patrick Turner.

Cheers

Ian

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
snip

snip,

Somebody with far more insight than I have decided an imaginery
resistance between grid and cathode modelled tube noise OK.

I'm wondering that if I look further at this noise issue, will I make
more silent amps?
Depends. If you simply believe the 2.5/gm mantra without question then I
guess you probably won't.

I am a questioning man. I am a PITA for the extremist theoretician,
because I ask "And what do we get in practice?"


That's what I say too, and my answer so far is 'way more than 2.5/gm'.


Cheers

Ian


So, OK, to me I had a choice of persuing low noise with careful
selection of high gm triodes all paralleled up and then I thought I
could use one j-fet costing $1.0 instead. That iddy biddy j-fet, the
2SK369 will produce 1% THD per volt of output with most loads so that
with 5V out you'd get 5% THD. Its a quite terrible result compared to a
we set up 6CG7. BUT, the THD declines towards 0.0% at 0.0Vo, so that if
the Vo was 0.01V, THD should have declined to 0.01%, which is quite low.

And so I use a j-fet at my phono amp inputs. With MM, the j-fet has a
lot of local current FB from its larger Rs to keep its gain low and the
worst case is with the possible 50mV input at 22kHz. If the fet gain is
10, you have 0.5Vo, and the first harmonic is at 44khz and inaudible.
But in fact high levels of treble in music don't much occur.

With MC, input at 10Hz is perhaps 0.03mV, at 1khz, perhaps 0.3mV, and at
HF perhaps 3.0mV. So THD is low even with less local current FB and more
fet gain. When I test a phono preamp I apply a level sig gene signal
through a recording RIAA filter.
A fet can produce much higher output than required and still have low
THD.

If one wants less THD then one can use a balanced input with two j-fets
because most j-fets produce similar distortion dominated by 2H like a
triode.

snip,

So what is a better solution?

Now that is a really good question. Someone needs to do some work first
to verify that there is some relation between gm and EIN and secondly to
quantify it. Actually that's rubbish. All we need to know is what are
the achievable EIN values for popular audio tubes. So some experiments
are in order. I was planning to make the 40dB amp for measuring 6CG7
noise for my mic pres anyway so I guess I'll be the one to make a start.

Getting less than 1uV noise at the input of any tube is problematical,
especially with 6CG7, 12AT7, 6DJ8, 12AX7, 12AY7, 12AU7 and many other
pentodes in triode mode.

But feel welcome to try things.

Maybe consider some higher gm tubes like 6C45pi or 417 or E80F/6BX6,6EH7
in triode and at various Ia and see what happens.


snip

Allen Wright says a lot about how treble is boosted and bass is cut
acording to the RIAA recording curve which is the opposite to the
playback curve. He says its important to have just the right RIAA
playback filter to allow for the fact the treble isn't boosted
infinitely, but the only up to maybe 50kHz, or else they have
instability troubles with the cutting head amp and its loops of negative
feedback to ensure distortion is minimized.

I'll check him out.

He was i think the first to use a high gm j-fet to drive a triode in
cascode.

Its very simple and it works.

snip,

Well, I'll have to think more about that but it won't alter how I make
amplifiers much.

No, please don't. From what I have seen your amps are pretty good and
RIAA is the right thing to use for magnetic cartridges.

I can't be too wrong then?

Patrick Turner.

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".

Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

The 75uS time constant in an RIAA network give a preamp response which
means the HF above 1kHz is down -3dB at 2,112Hz. above this F the rate
of attenuation becomes close to -6dB per octave so that if a 22kHz
signal enters the preamp it is about 20dB below the 1khz reference level
at the preamp output.

To get a flat response with vinyl and a magnetic cart, the signal from
the record contains a boosted HF so that a 22khz signal is 20dB above
the 1 khz reference level.


Nope. The reason the HF signal from a mag cartridge is boosted is not
because it is boosted on the record (it isn't) but because (as I have
said several times now) of the magnetic cartridge's rising output with
frequency (for a constant amplitude input) due to its output voltage
being proportional to rate of change of flux i.e. in this case frequency.


Cheers

Ian


The +/-12dB points in the process are around the 10khz point, see the
official RIAA attenuation charts.

Patrick Turner.
Cheers

Ian

[John Byrns]

In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.

That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".

Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
Patrick Turner wrote:

Some ppl are very picky. But fact is without the treble
emphasis/de-emphasis, noise in FM would be worse.

And with a boosted HF, the waves are easier to decode when F gets higher
when trying to synthesize the L and R channels which are derived from a
sub carrier with a rather low 38kHz frequency. FM would have been better
if the sub carrier was at 100kHz and the pilot tone was at say 30kHz,
but the boffins on 1955 forsaw multiplexing and the possibilities of
other sub-carriers F at 67kHz and 96khz beckoned.

Hi Patrick,

Another fact is that if a 100 kHz sub carrier had been used for FM
stereo as you suggest, the noise would have been much worse than it is
with the 38 kHz sub carrier. FM stereo is already something on the
order of approximately 22 dB noisier than monophonic FM. Using a 100
kHz stereo sub carrier frequency would make the FM stereo noise level 8
or 9 dB worse than it is with the 38 kHz sub carrier due to the effect
of FM's triangular noise spectrum. As a result you want the stereo sub
carrier to be as low in frequency as possible, 100 kHz is a bad choice
for this reason. FM's triangular noise spectrum is the tradeoff old
E.H. Armstrong made to achieve wideband FM's noise reduction in the
audio range below 15 kHz.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Patrick Turner]

Ian Bell wrote:

Patrick Turner wrote:

Ian Bell wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".

Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

The 75uS time constant in an RIAA network give a preamp response which
means the HF above 1kHz is down -3dB at 2,112Hz. above this F the rate
of attenuation becomes close to -6dB per octave so that if a 22kHz
signal enters the preamp it is about 20dB below the 1khz reference level
at the preamp output.

To get a flat response with vinyl and a magnetic cart, the signal from
the record contains a boosted HF so that a 22khz signal is 20dB above
the 1 khz reference level.


Nope. The reason the HF signal from a mag cartridge is boosted is not
because it is boosted on the record (it isn't) but because (as I have
said several times now) of the magnetic cartridge's rising output with
frequency (for a constant amplitude input) due to its output voltage
being proportional to rate of change of flux i.e. in this case frequency.

I thought the signal sent to the cutting head amp was passed through a
recording RIAA filter which cuts bass and boosts treble.
Thus the signal from the record has low bass and high treble and the
preamp applies a filter doing the opposite of what a recording filter
does and the music response is thus flat at the preamp output.

See RDH4, page 767 (v), Equalisation of the cutter.
They say equalisation is applied to the cutter to obtain the desired
recording characteristic.
There were 12 different characteristics before RIAA became the standard
for LP, but some were already close to RIAA.

If you do a Google on 'RIAA equalization recording' there are lots of
hits.

eg, see http://www.phonopreamps.com/faq.html

"""What is RIAA equalization?

Because of limitations in the LP recording process, an equalization
curve must be applied to the music or other sonic content prior to it
being cut onto vinyl, so as to reduce backround noise and sibilance.
Removing this equalization affect (called the RIAA curve) and restoring
the music's original frequency response curve during playback is an
important part of the phono preamp's job and differentiates it from
other preamps used for microphones and musical instruments, which
provide gain but no other modification of the original sound quality.
Proper RIAA re-equalization during playback is a must in faithfully
producing the original musical content without coloration or distortion.
(top)""""

Patrick Turner.

[Patrick Turner]

John Byrns wrote:

In article ,
Patrick Turner wrote:

Some ppl are very picky. But fact is without the treble
emphasis/de-emphasis, noise in FM would be worse.

And with a boosted HF, the waves are easier to decode when F gets higher
when trying to synthesize the L and R channels which are derived from a
sub carrier with a rather low 38kHz frequency. FM would have been better
if the sub carrier was at 100kHz and the pilot tone was at say 30kHz,
but the boffins on 1955 forsaw multiplexing and the possibilities of
other sub-carriers F at 67kHz and 96khz beckoned.

Hi Patrick,

Another fact is that if a 100 kHz sub carrier had been used for FM
stereo as you suggest, the noise would have been much worse than it is
with the 38 kHz sub carrier.

Well the increase in bandwidth needed to have the subC at 100kHz would
increase noise.


FM stereo is already something on the
order of approximately 22 dB noisier than monophonic FM.

Hmm, I find that difficult to believe going by what i hear from better
FM tuners.

Using a 100
kHz stereo sub carrier frequency would make the FM stereo noise level 8
or 9 dB worse than it is with the 38 kHz sub carrier due to the effect
of FM's triangular noise spectrum. As a result you want the stereo sub
carrier to be as low in frequency as possible, 100 kHz is a bad choice
for this reason.

I was guessing, but 50kHz would give less noise.

FM's triangular noise spectrum is the trade off old
E.H. Armstrong made to achieve wideband FM's noise reduction in the
audio range below 15 kHz.

Did EHA consider stereo?

Patrick Turner.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.



Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.

Cheers

Ian

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".

Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
The 75uS time constant in an RIAA network give a preamp response which
means the HF above 1kHz is down -3dB at 2,112Hz. above this F the rate
of attenuation becomes close to -6dB per octave so that if a 22kHz
signal enters the preamp it is about 20dB below the 1khz reference level
at the preamp output.

To get a flat response with vinyl and a magnetic cart, the signal from
the record contains a boosted HF so that a 22khz signal is 20dB above
the 1 khz reference level.

Nope. The reason the HF signal from a mag cartridge is boosted is not
because it is boosted on the record (it isn't) but because (as I have
said several times now) of the magnetic cartridge's rising output with
frequency (for a constant amplitude input) due to its output voltage
being proportional to rate of change of flux i.e. in this case frequency.

I thought the signal sent to the cutting head amp was passed through a
recording RIAA filter which cuts bass and boosts treble.

It is, but because the recording head is also velocity operated, if you
had no equalisation, the amplitude recorded on the disc would fall
continuously at 6dB/octave. For constant amplitude on disk what you
really need is a rising EQ at 6dB/octave. Cutting the bass effectively
makes the frequencies below the bass turnover frequency constant
amplitude. Boosting the treble effectively makes the frequencies above
the treble turnover frequency constant amplitude.

Thus the signal from the record has low bass and high treble and the
preamp applies a filter doing the opposite of what a recording filter
does and the music response is thus flat at the preamp output.


If you think in terms of recorded velocity then what you say is 100%
correct. I find that hard to do and prefer to think in terms of amplitude.

See RDH4, page 767 (v), Equalisation of the cutter.
They say equalisation is applied to the cutter to obtain the desired
recording characteristic.

I am not disputing that EQ is needed.

There were 12 different characteristics before RIAA became the standard
for LP, but some were already close to RIAA.

If you do a Google on 'RIAA equalization recording' there are lots of
hits.

eg, see http://www.phonopreamps.com/faq.html

"""What is RIAA equalization?

Because of limitations in the LP recording process, an equalization
curve must be applied to the music or other sonic content prior to it
being cut onto vinyl, so as to reduce backround noise and sibilance.

Yes, I have seen many such references. Unfortunately most are just plain
wrong or at the very least blithely ignore the difference between
velocity and amplitude.

Removing this equalization affect (called the RIAA curve) and restoring
the music's original frequency response curve during playback is an
important part of the phono preamp's job and differentiates it from
other preamps used for microphones and musical instruments, which
provide gain but no other modification of the original sound quality.
Proper RIAA re-equalization during playback is a must in faithfully
producing the original musical content without coloration or distortion.
(top)""""


No argument there except it only applies when you use a velocity
sensitive transducer like a magnetic cartridge.


Cheers

ian
Patrick Turner.

[John Byrns]

In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.

Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.

Precisely!

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.

Precisely!



It has always puzzled me why they did not attempt to get close to
constant amplitude from the start. It's the velocity operated head
cutters that gave them the excessive bass excursions that caused them so
much problem in the very early days. They were bright enough to know
what the head response was so why not correct for it? I guess there has
to be some practical factor that makes this harder than you would
expect. I was going to say it is no harder than driving a mgnetic tape
record head but I suspect the powers involved in a disc cutting head are
a lot higher.

Cheers

ian

[Patrick Turner]

John Byrns wrote:

In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.

Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.

Precisely!

Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

Patrick Turner wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.
Precisely!

Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.

OK, Here goes.


The (amplitude) response of the cutter head falls by 6dB/octave. The
inverse RIAA curve, which is used on record starts, with a rising curve
from 50Hz to 500Hz at 6dB/octave. The combination of these two means a
constant amplitude gets recorded on the disc between these two frequencies.

From 500Hz to 2.12Khz the RIAA curve is flat, so because of the head
response, the amplitude recorded on the disc falls by 6dB/octave for
about 2 octaves.

From 2KHz upwards the RIAA record curve again slopes upward at
6dB/octave which, when combined with the head response of -6dB/octave
means the recorded amplitude is constant versus frequency.


On playback the mag cartridge has a response which rise at 6dB/octave.

The RIAA playback curve falls at 6dB/octave from 50Hz to 500Hz. So the
constant amplitude on disc plus the rising characteristic of the mag
cart means the pre-amp input signal rises at 6dB/octave between these
frequencies and is compensated for by the falling characteristic of the
RIAA replay curve.

Between 500Hz and 2.12Khz the signal from the disc falls at 6dB/octave
which combined with the 6dB/octave rising characteristic of the mag cart
means the signal at the pre-amp input is constant between these two
frequencies and the flat part of the RIAA replay curve between these
frequencies leaves the input signal unchanged.


From 2.12KHz upwards the constant amplitude from the disc combined with
the 6dB/octave rising characteristic of the cart means the signal at the
pre-amp input rises at 6dB/octave which is compensated for by the
6dB/octave falling characteristic of the RIAA replay curve.


Hope that helps.

Cheers

Ian
--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
Patrick Turner wrote:

John Byrns wrote:

In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the
record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on
disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates
the
frequencies above approximately 2 kHz by more than 12 dB, not my idea
of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals
on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?

Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference
on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the
more
commonly used constant velocity reference. To change the reference
from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency.
When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and
a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts
rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122
Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.

Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.

Precisely!

Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Because I am talking about the RIAA "standard", not "real world
hardware". "Real world hardware" then attempts to meet the requirements
of the standard as closely as possible within the bounds of the relevant
economic and technical limitations.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[John Byrns]

In article ,
Ian Bell wrote:

It has always puzzled me why they did not attempt to get close to
constant amplitude from the start.

It's not obvious to me why they would want to do that? Presumably there
are a lot of factors that go into choosing the best equalization curve,
including among others the spectral distribution of the energy in the
source material, and the capabilities of the recording medium vs.
frequency. Perhaps the high frequency amplitude shelving during
recording is because of limitations on the velocity that can be recorded
at high frequencies.

It's the velocity operated head
cutters that gave them the excessive bass excursions that caused them so
much problem in the very early days. They were bright enough to know
what the head response was so why not correct for it? I guess there has
to be some practical factor that makes this harder than you would
expect.

IIRC early electrical recordings were constant amplitude below the
turnover frequency, which I think was somewhere in the 250 Hz to 400 Hz
range, and constant velocity above the turnover frequency. This
equalization was mechanically implemented by the primary resonance of
the cutter head at the turnover frequency, so that the cutter was mass
controlled above the turnover frequency, and compliance controlled below
the turnover frequency. At least that is an approximation of the
explanation I have heard. I'm not sure about the earlier acoustic
recorders, but I think the equalization was similar. Eventually they
found that they could add some electrical boost to the constant velocity
high frequency region to allow reducing the high frequency noise during
playback with a corresponding high frequency rolloff.

I was going to say it is no harder than driving a mgnetic tape
record head but I suspect the powers involved in a disc cutting head are
a lot higher.

Yes, the power involved is more like the power required to drive a
loudspeaker than it is like the power needed to drive a magnetic tape
recording head.

--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Patrick Turner]

Ian Bell wrote:

Patrick Turner wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.
Precisely!

Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.

OK, Here goes.

The (amplitude) response of the cutter head falls by 6dB/octave.

OK, so as F rises, there is less amplitude in the grooves, no?.

So the bass grooves are the biggest, so if you cut bass and boost treble
somewhere near 6dB/octave, then the groves of any F will all become
constant amplitude.

The
inverse RIAA curve, which is used on record starts, with a rising curve
from 50Hz to 500Hz at 6dB/octave. The combination of these two means a
constant amplitude gets recorded on the disc between these two frequencies.
From 500Hz to 2.12Khz the RIAA curve is flat, so because of the head
response, the amplitude recorded on the disc falls by 6dB/octave for
about 2 octaves.

But the RIAA has about 5dB difference in eq between 500 and 2,112Hz.....

There is NO flat part on the RIAA recording or playback curves between
say 20Hz and 22kHz.

From 2KHz upwards the RIAA record curve again slopes upward at
6dB/octave which, when combined with the head response of -6dB/octave
means the recorded amplitude is constant versus frequency.

So the slight change inthe rate of RIAA attenuation/boost between about
500Hz and 2,112Hz means that there is a change in amplitude between F
ranges below 500 and above 2,112?


On playback the mag cartridge has a response which rise at 6dB/octave.

The RIAA playback curve falls at 6dB/octave from 50Hz to 500Hz. So the
constant amplitude on disc plus the rising characteristic of the mag
cart means the pre-amp input signal rises at 6dB/octave between these
frequencies and is compensated for by the falling characteristic of the
RIAA replay curve.

Between 500Hz and 2.12Khz the signal from the disc falls at 6dB/octave
which combined with the 6dB/octave rising characteristic of the mag cart
means the signal at the pre-amp input is constant between these two
frequencies and the flat part of the RIAA replay curve between these
frequencies leaves the input signal unchanged.

But preamp input isn't constant between the two F.....

From 2.12KHz upwards the constant amplitude from the disc combined with
the 6dB/octave rising characteristic of the cart means the signal at the
pre-amp input rises at 6dB/octave which is compensated for by the
6dB/octave falling characteristic of the RIAA replay curve.

Hope that helps.

I hear what you are saying, and basically the faster the cutter has to
move side to side to make a groove then the less it travels.

And the faster the cartridge moves side to side to track gooves, the
higher the amount of voltage output and its all a fairly linear system.

And to avoid having high bass groove swings which would take up too much
room on the record they cut the bass signal to the cutting head and to
avoid treble grooves having too small an amplitude they boost the treble
signal to the cutter. The groove amplitude is thus more or less flat
across the whole range.
But the cart then puts out little bass and a huge amount of treble
because its response is linear with F and so the RIAA playback curve
reduces treble and boost bass in reference to 1kHz. Disc Noise is
reduced at HF above 1kHz at least, and amplifier noise somewhat more
methinks.

Patrick Turner.





Cheers

Ian
--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.
Precisely!
Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.
OK, Here goes.

The (amplitude) response of the cutter head falls by 6dB/octave.

OK, so as F rises, there is less amplitude in the grooves, no?.


Yup.

So the bass grooves are the biggest, so if you cut bass and boost treble
somewhere near 6dB/octave, then the groves of any F will all become
constant amplitude.


Yup


The
inverse RIAA curve, which is used on record starts, with a rising curve
from 50Hz to 500Hz at 6dB/octave. The combination of these two means a
constant amplitude gets recorded on the disc between these two frequencies.
From 500Hz to 2.12Khz the RIAA curve is flat, so because of the head
response, the amplitude recorded on the disc falls by 6dB/octave for
about 2 octaves.

But the RIAA has about 5dB difference in eq between 500 and 2,112Hz.....


Yes, I was using the Bode straight line approximations - in practice the
real things are curves. In practice 500Hz and 2.12KHz are the 3dB point
so if you add 2*3dB to the 5dB you see, you get 11dB. However, there is
still 12dB amplitude difference on disc between frequencies removed from
the 3dB points.

There is NO flat part on the RIAA recording or playback curves between
say 20Hz and 22kHz.

In practice no but again I was using Bode approximations - see above.

From 2KHz upwards the RIAA record curve again slopes upward at
6dB/octave which, when combined with the head response of -6dB/octave
means the recorded amplitude is constant versus frequency.

So the slight change inthe rate of RIAA attenuation/boost between about
500Hz and 2,112Hz means that there is a change in amplitude between F
ranges below 500 and above 2,112?


Yes



On playback the mag cartridge has a response which rise at 6dB/octave.

The RIAA playback curve falls at 6dB/octave from 50Hz to 500Hz. So the
constant amplitude on disc plus the rising characteristic of the mag
cart means the pre-amp input signal rises at 6dB/octave between these
frequencies and is compensated for by the falling characteristic of the
RIAA replay curve.

Between 500Hz and 2.12Khz the signal from the disc falls at 6dB/octave
which combined with the 6dB/octave rising characteristic of the mag cart
means the signal at the pre-amp input is constant between these two
frequencies and the flat part of the RIAA replay curve between these
frequencies leaves the input signal unchanged.

But preamp input isn't constant between the two F.....

No, but see above.

From 2.12KHz upwards the constant amplitude from the disc combined with
the 6dB/octave rising characteristic of the cart means the signal at the
pre-amp input rises at 6dB/octave which is compensated for by the
6dB/octave falling characteristic of the RIAA replay curve.

Hope that helps.

I hear what you are saying, and basically the faster the cutter has to
move side to side to make a groove then the less it travels.

And the faster the cartridge moves side to side to track gooves, the
higher the amount of voltage output and its all a fairly linear system.

And to avoid having high bass groove swings which would take up too much
room on the record they cut the bass signal to the cutting head and to
avoid treble grooves having too small an amplitude they boost the treble
signal to the cutter. The groove amplitude is thus more or less flat
across the whole range.


That's is basically it.

But the cart then puts out little bass and a huge amount of treble
because its response is linear with F and so the RIAA playback curve
reduces treble and boost bass in reference to 1kHz.


That's right

Disc Noise is
reduced at HF above 1kHz at least, and amplifier noise somewhat more
methinks.



Amplifier noise certainly, disc noise I am not so sure about.

Cheers

Ian

Patrick Turner.




Cheers

Ian
--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Patrick Turner]

Ian Bell wrote:

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.
Precisely!
Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.
OK, Here goes.

The (amplitude) response of the cutter head falls by 6dB/octave.

OK, so as F rises, there is less amplitude in the grooves, no?.


Yup.

So the bass grooves are the biggest, so if you cut bass and boost treble
somewhere near 6dB/octave, then the groves of any F will all become
constant amplitude.


Yup

The
inverse RIAA curve, which is used on record starts, with a rising curve
from 50Hz to 500Hz at 6dB/octave. The combination of these two means a
constant amplitude gets recorded on the disc between these two frequencies.
From 500Hz to 2.12Khz the RIAA curve is flat, so because of the head
response, the amplitude recorded on the disc falls by 6dB/octave for
about 2 octaves.

But the RIAA has about 5dB difference in eq between 500 and 2,112Hz.....


Yes, I was using the Bode straight line approximations - in practice the
real things are curves. In practice 500Hz and 2.12KHz are the 3dB point
so if you add 2*3dB to the 5dB you see, you get 11dB. However, there is
still 12dB amplitude difference on disc between frequencies removed from
the 3dB points.

There is NO flat part on the RIAA recording or playback curves between
say 20Hz and 22kHz.

In practice no but again I was using Bode approximations - see above.

From 2KHz upwards the RIAA record curve again slopes upward at
6dB/octave which, when combined with the head response of -6dB/octave
means the recorded amplitude is constant versus frequency.

So the slight change inthe rate of RIAA attenuation/boost between about
500Hz and 2,112Hz means that there is a change in amplitude between F
ranges below 500 and above 2,112?

Yes



On playback the mag cartridge has a response which rise at 6dB/octave.

The RIAA playback curve falls at 6dB/octave from 50Hz to 500Hz. So the
constant amplitude on disc plus the rising characteristic of the mag
cart means the pre-amp input signal rises at 6dB/octave between these
frequencies and is compensated for by the falling characteristic of the
RIAA replay curve.

Between 500Hz and 2.12Khz the signal from the disc falls at 6dB/octave
which combined with the 6dB/octave rising characteristic of the mag cart
means the signal at the pre-amp input is constant between these two
frequencies and the flat part of the RIAA replay curve between these
frequencies leaves the input signal unchanged.

But preamp input isn't constant between the two F.....

No, but see above.

From 2.12KHz upwards the constant amplitude from the disc combined with
the 6dB/octave rising characteristic of the cart means the signal at the
pre-amp input rises at 6dB/octave which is compensated for by the
6dB/octave falling characteristic of the RIAA replay curve.

Hope that helps.

I hear what you are saying, and basically the faster the cutter has to
move side to side to make a groove then the less it travels.

And the faster the cartridge moves side to side to track gooves, the
higher the amount of voltage output and its all a fairly linear system.

And to avoid having high bass groove swings which would take up too much
room on the record they cut the bass signal to the cutting head and to
avoid treble grooves having too small an amplitude they boost the treble
signal to the cutter. The groove amplitude is thus more or less flat
across the whole range.

That's is basically it.

But the cart then puts out little bass and a huge amount of treble
because its response is linear with F and so the RIAA playback curve
reduces treble and boost bass in reference to 1kHz.

That's right

Disc Noise is
reduced at HF above 1kHz at least, and amplifier noise somewhat more
methinks.


Amplifier noise certainly, disc noise I am not so sure about.

Definately disc noise above 1 kHz is much reduced by the boosting of HF.
Just think if the amplitude of HF were to dwindle down with no RIAA
recording HF emphasis. The HF disc noise would still be high and signal
low and snr lousy. Someone may say that the higher HF amplitude means
more noise is generated but afaik, most experts say the disc noise is
much less when its done the way we know.

The luck with vinyl was that it was so much quieter than shellac
records.

And many people still take vinyl so seriously.

Patrick Turner.


Cheers

Ian

Patrick Turner.




Cheers

Ian
--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/

[Ian Bell[_2_]]

Patrick Turner wrote:

Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

John Byrns wrote:
In article ,
Ian Bell wrote:

Patrick Turner wrote:
But RIAA reduces noise. It allows bass signals to "fit" onto the record
to give use 20 minutes a side from LP.
No it does not. As I said, in amplitude terms, what is recorded on disc
is nearly constant amplitude versus frequency.
That depends on how you define "nearly constant amplitude versus
frequency"? During the recording process the RIAA curve attenuates the
frequencies above approximately 2 kHz by more than 12 dB, not my idea of
"nearly constant".
Mine neither. You are saying that above 2KHz the amplitude of signals on
the disc is attenuated by 12dB or more. That makes no sense to me. Do
you have a reference for this?
Hi Ian,

I don't have a reference handy, although this issue was discussed in
another group a few years back and someone actually found a reference on
the web.

Let me explain, as you said "in amplitude terms, what is recorded on
disc is nearly constant amplitude versus frequency." The operative bit
here is that your stated reference is "constant amplitude", not the more
commonly used constant velocity reference. To change the reference from
constant velocity to constant amplitude we must apply a correction
factor to the "RIAA" recording curve that rolls off across the entire
audio band at a rate of 6 dB per octave with increasing frequency. When
the conversion is made from a constant velocity to a constant amplitude
reference we are left with a recording curve with a pole at 500 Hz and a
zero at 2,122 Hz, ignoring the 50 Hz and ultra sonic time constants.
What this says is that on an amplitude basis the response starts rolling
off at 500 Hz at a rate of 6 dB per octave, and then flattens out again
at 2,122 Hz. Between 500 and 2,122 Hz the attenuation gradually
increases to approximately 12 dB above 2,122 Hz. In other words, the
response recorded on the LP is shelved down approx. 12 dB above 2,122 Hz
relative to the response below 500 Hz, on the basis of recorded groove
amplitude basis.
Got you. It was clear to me there was a kink in the amplitude response
due to the separation of the two mid time constants between which the
response would fall at 6dB/octave. I had not realised they were two
octaves apart. So to summarise, the recorded amplitude response is flat
from 50Hz to 500Hz, then rolls off at 6dB per octave to 2.12KHz where it
is about 12dB down and thereafter is flat out to 20KHz.
Precisely!
Without being able to view relevant diagrams and response curves of real
world hardware, I am precisely confused why John is so precisely in
agreement.

Patrick Turner.
OK, Here goes.

The (amplitude) response of the cutter head falls by 6dB/octave.
OK, so as F rises, there is less amplitude in the grooves, no?.

Yup.

So the bass grooves are the biggest, so if you cut bass and boost treble
somewhere near 6dB/octave, then the groves of any F will all become
constant amplitude.

Yup

The
inverse RIAA curve, which is used on record starts, with a rising curve
from 50Hz to 500Hz at 6dB/octave. The combination of these two means a
constant amplitude gets recorded on the disc between these two frequencies.
From 500Hz to 2.12Khz the RIAA curve is flat, so because of the head
response, the amplitude recorded on the disc falls by 6dB/octave for
about 2 octaves.
But the RIAA has about 5dB difference in eq between 500 and 2,112Hz.....

Yes, I was using the Bode straight line approximations - in practice the
real things are curves. In practice 500Hz and 2.12KHz are the 3dB point
so if you add 2*3dB to the 5dB you see, you get 11dB. However, there is
still 12dB amplitude difference on disc between frequencies removed from
the 3dB points.

There is NO flat part on the RIAA recording or playback curves between
say 20Hz and 22kHz.
In practice no but again I was using Bode approximations - see above.

From 2KHz upwards the RIAA record curve again slopes upward at
6dB/octave which, when combined with the head response of -6dB/octave
means the recorded amplitude is constant versus frequency.
So the slight change inthe rate of RIAA attenuation/boost between about
500Hz and 2,112Hz means that there is a change in amplitude between F
ranges below 500 and above 2,112?
Yes


On playback the mag cartridge has a response which rise at 6dB/octave.

The RIAA playback curve falls at 6dB/octave from 50Hz to 500Hz. So the
constant amplitude on disc plus the rising characteristic of the mag
cart means the pre-amp input signal rises at 6dB/octave between these
frequencies and is compensated for by the falling characteristic of the
RIAA replay curve.

Between 500Hz and 2.12Khz the signal from the disc falls at 6dB/octave
which combined with the 6dB/octave rising characteristic of the mag cart
means the signal at the pre-amp input is constant between these two
frequencies and the flat part of the RIAA replay curve between these
frequencies leaves the input signal unchanged.
But preamp input isn't constant between the two F.....
No, but see above.

From 2.12KHz upwards the constant amplitude from the disc combined with
the 6dB/octave rising characteristic of the cart means the signal at the
pre-amp input rises at 6dB/octave which is compensated for by the
6dB/octave falling characteristic of the RIAA replay curve.

Hope that helps.
I hear what you are saying, and basically the faster the cutter has to
move side to side to make a groove then the less it travels.

And the faster the cartridge moves side to side to track gooves, the
higher the amount of voltage output and its all a fairly linear system.

And to avoid having high bass groove swings which would take up too much
room on the record they cut the bass signal to the cutting head and to
avoid treble grooves having too small an amplitude they boost the treble
signal to the cutter. The groove amplitude is thus more or less flat
across the whole range.
That's is basically it.

But the cart then puts out little bass and a huge amount of treble
because its response is linear with F and so the RIAA playback curve
reduces treble and boost bass in reference to 1kHz.
That's right

Disc Noise is
reduced at HF above 1kHz at least, and amplifier noise somewhat more
methinks.

Amplifier noise certainly, disc noise I am not so sure about.

Definately disc noise above 1 kHz is much reduced by the boosting of HF.
Just think if the amplitude of HF were to dwindle down with no RIAA
recording HF emphasis. The HF disc noise would still be high and signal
low and snr lousy.

Agreed, but the boosting only compensates for the decline in HF due to
the cutter head to produce a flat amplitude response. In your FM radio
example, the response starts of flat and the 75uS emphasis boosts the HF
even more above the noise so at the receiver the de-emphasis improves
the S/N ration. It is well known that in music the amplitude of the
spectrum of the higher frequencies is lower than that of the lower
frequencies so you can boost the HF without danger of causing overloads.
I see no reason in principle why this could not have been applied to
vinyl with a consequent improvement in S/N.


Cheers

Ian


Someone may say that the higher HF amplitude means
more noise is generated but afaik, most experts say the disc noise is
much less when its done the way we know.

The luck with vinyl was that it was so much quieter than shellac
records.

And many people still take vinyl so seriously.

Patrick Turner.

Cheers

Ian

Patrick Turner.




Cheers

Ian
--
Regards,

John Byrns

Surf my web pages at, http://fmamradios.com/