"Sander deWaal" wrote in message

"Arny Krueger" said:

OTOH, tubed amps are more prone to EMI problems due to unique
tube-only problems like grid rectification, Miller capacitance and
their high input impedances. I hard a lot more RFI problems during
the days of tubes than I do now.

Grid rectification: the voltage span of an average power amp input
tube is about 2 volts.

A serious problem because of the high impedances involved and the
fact that tubes are frequently run without local feedback.

A cathode resistor isn't local feedback?

Not if its bypasssed.

You're not thinking of auto-grid bias, are you?

No.

A method that isn't used anymore in "modern" tube amps, at least not
that I know of.

Agreed.

This might also explain why you think tube inputs have such extremely
high input resistors, 10 megohm wasn't unusual is such an arrangement.

470K still is common. that's about 50 times the last number you mentioned,
Sander. Hunt around, sooner or later you'll find a believable number.

The voltage span of an average base-emitter junction is what...20
mV?

Irrelevant because transistor inputs stages are always far lower
than with tubes and almost always run with emitter degeneration.

Partly agreed, see above for the same rule that applies to cathode
resistors.

Look up "Ebers-Moll" for a better understanding about driving bipolar
transistors.

Here we go, the big tube eggshpert is now talking down his nose at me.

Miller capacitance: dependent on circuitry.

Also dependent on the tubes themselves.

Also agreed, but certainly not just "due to unique tube problems like
Miller capacitance [.....] ".

Fess up Sander, Miller capacitance is a far more significant issue with
tubes.

*Every* amplifying device with inverting output suffers from mr.
Miller.

It's all in the quantification, and the quantities are vastly
differnt for tubes and SS.

Also agreed, but certainly not just "due to unique tube problems like
Miller capacitance [.....] ".

Yawn.

Examples of said circuits would be CCC, CEC, CSC.

High input impedance: dependent on circuitry.

Of course, but as a matter of course tubed circuts have far higher
input impedances.

Which is compensated for by the much karger grid voltage span.

Nope.

A tube amp with a 10 k resistor at the grid to ground won't be more
susceptible (sp?) to RFI than an average bipolar input.

Perhaps, but irrelevant due to the common practice of using far
higher grid resistors than 10K.

They're 25 kohms in my tube amps.

That's a choice you get to make. But, it says nothing about tubed amps in
general.

Not that I'm alone in this, you will note.

Not that you set the rule or the average, either.

Let alone FETs or MOSFETs.

Oh my goodness, it's a red herring. Actually two of them.

Actually, no.

Yawn.

One of the main *advantages* of (MOS)FETs is their voltage-driven
nature.

What was the first transistor again in your FM-frontend, Arny? ;-)

Yet another red herring. BTW, the cross-modulation characteristics
of SS FM front ends are generally better than what we had when tubes
were king.

Only due to clamping diodes.

Nope. Cross modulation is about nonlinear distortion, and clamping diodes
increase that, not decrease it.

If the same technique is used with a triode- or even penthode input
circuits (NOT hexodes or pentagrid tubes!), cross-modulation is almost
not an item to worry about.

See previous comment.

Noise may be, though.

Granted that most tube amps have higher value input resistors.

Thanks for agreeing that your argument based on unrealistically low
grid resistors was irrelevant, Sander.

Not unrealistic at all considering my design views, but then again,
*I* am able to put off my blinders ;-)

What blinders? The vacuum tube horse has been dead in the mainstream of hi
fi for about 30 years. Putting blinders on a dead horse makes very little
sense.

BTW, SS amplifiers are also inherently low pass filters.

In general, SS circuitry is capable of handling higher frequencies
than tube equipment, consider an average output transformer with a
corner freq. at 50 kHz or so.

Actually 50 KHz or so is not an uncommon corner frequency for SS
amps.

Due to what property, exactly?

Desire to have an amp that is nice to work with.


Of course, a good practice is to add an anti-slewing filter at the
input of the amp (even with tubes this might be beneficial!), but
some designers select a corner frequency of 150 or 200 kHz.

Slewing was a problem in someone's mind by the time it was
publicized.

Since Otala, Garde et al, we know better ;-)

Yes, Otala is known to have been an alarmist, to say the least.

Besides, that is mostly a simple RC 6 dB.oct filter, not steep
enough to supress real strong RF energy.

A 6 dB/oct filter cornering at 50 KHz is down over 20 dB at 500 KHz,
40 dB at 5 MHz and 60 dB down at the 50 MHz that Mike origionally
mentioned.

One wrong grounding point may wreak havoc with that 60 dB number.

How many red herrings does this make?

hence even if Shakti stones work
as described on their website, they won't do very much for tubed
systems IMO (from a technical POV that is!).

From a Sander technical view, that is. ;-)

Gratuitous personal insult ignored.

Inability to gracefully perceive a gentle stab due to a
technological disagreement noted.

Please stop being so hostile! ;-)

Yawn.

"Arny Krueger" said:

A serious problem because of the high impedances involved and the
fact that tubes are frequently run without local feedback.

A cathode resistor isn't local feedback?

Not if its bypasssed.


Ah. Another design tirck from me.
I rarely bypass cathode resistors.
My, I must be an exceptional tube amp designer ;-)


470K still is common. that's about 50 times the last number you mentioned,
Sander. Hunt around, sooner or later you'll find a believable number.


In older tube amps, such high input resistors were indeed common.
Modern tube amps, as far as i've seen them, use lower values from
10k....100 k.


Look up "Ebers-Moll" for a better understanding about driving bipolar
transistors.

Here we go, the big tube eggshpert is now talking down his nose at me.


Please find a quote by me where I pretend to be an expert (or
eggsphert) in *anything* , except tinkering.

Furthermore, the Ebers-Moll approach starts with the premise that
bipolars are voltage-driven.

But you knew this, right? ;-)


Also agreed, but certainly not just "due to unique tube problems like
Miller capacitance [.....] ".

Fess up Sander, Miller capacitance is a far more significant issue with
tubes.


I replied to your statement that those problems are "unique with
tubes, like Miller capacity etc".

Fess up Arnold, Miller capacitance is apparent in *every*
signal-inverting amplifying device.


High input impedance: dependent on circuitry.

Of course, but as a matter of course tubed circuts have far higher
input impedances.

Which is compensated for by the much karger grid voltage span.

Nope.


Yep.


Yet another red herring. BTW, the cross-modulation characteristics
of SS FM front ends are generally better than what we had when tubes
were king.

Only due to clamping diodes.

Nope. Cross modulation is about nonlinear distortion, and clamping diodes
increase that, not decrease it.


Remind me again when MOSFETs without internal clamping diodes are
available to designers.

Non-linear distortion, again due to the smaller BE voltage span as
compared to the larger GC span of a triode or penthode.


Granted that most tube amps have higher value input resistors.

Thanks for agreeing that your argument based on unrealistically low
grid resistors was irrelevant, Sander.

Not unrealistic at all considering my design views, but then again,
*I* am able to put off my blinders ;-)

What blinders? The vacuum tube horse has been dead in the mainstream of hi
fi for about 30 years. Putting blinders on a dead horse makes very little
sense.


Thanks anyway for discussing obsolete technology with me Arnold. ;-)


Actually 50 KHz or so is not an uncommon corner frequency for SS
amps.

Due to what property, exactly?

Desire to have an amp that is nice to work with.


I meant what causes the LPF character of an average SS amp?
In what stages is that most apparent?


Slewing was a problem in someone's mind by the time it was
publicized.

Since Otala, Garde et al, we know better ;-)

Yes, Otala is known to have been an alarmist, to say the least.


Otala was an optimist ;-)


Besides, that is mostly a simple RC 6 dB.oct filter, not steep
enough to supress real strong RF energy.

A 6 dB/oct filter cornering at 50 KHz is down over 20 dB at 500 KHz,
40 dB at 5 MHz and 60 dB down at the 50 MHz that Mike origionally
mentioned.

One wrong grounding point may wreak havoc with that 60 dB number.

How many red herrings does this make?


What was the minimum requirement for measuring S/N again?
When I once mentioned I couldn't measure better than -80 dB at one
time, you had a good laugh.

Think about a bipolar differential input, CMRR, and the actual signal
voltage needed to excurse it.

You may use Google, if you like ;-)

--
Sander de Waal
" SOA of a KT88? Sufficient. "

"Sander deWaal" wrote in message

"Arny Krueger" said:

A serious problem because of the high impedances involved and the
fact that tubes are frequently run without local feedback.

A cathode resistor isn't local feedback?

Not if its bypasssed.

Ah. Another design tirck from me.
I rarely bypass cathode resistors.

Good.

My, I must be an exceptional tube amp designer ;-)

Yawn.

470K still is common. that's about 50 times the last number you
mentioned, Sander. Hunt around, sooner or later you'll find a
believable number.

In older tube amps, such high input resistors were indeed common.
Modern tube amps, as far as i've seen them, use lower values from
10k....100 k.

100K being only 10 times higher than the number you previously mentioned...


First site I found with the search string tube preamp schematics:

http://www3.telus.net/public/vintage1/mu.jpg

1 meg grid resistor, bypassed cathode resistor

http://www3.telus.net/public/vintage1/phones.jpg

1 meg grid resistor, bypassed cathode resistor

http://www.triodeel.com/a340a.gif

220k grid resistor, bypassed cathode resistor

http://www.drtube.com/schematics/an/ankoru.gif

bypassed cathode resistor

http://www.drtube.com/schematics/ar/d40.gif

475 k grid resistor

http://www.drtube.com/schematics/ar/vt100.gif

100 K grid resistor

etc., etc.

Look up "Ebers-Moll" for a better understanding about driving
bipolar transistors.

Here we go, the big tube eggshpert is now talking down his nose at
me.

Please find a quote by me where I pretend to be an expert (or
eggsphert) in *anything* , except tinkering.

Sander wrote: " My, I must be an exceptional tube amp designer"

Furthermore, the Ebers-Moll approach starts with the premise that
bipolars are voltage-driven.

But you knew this, right? ;-)

Yawn.

Also agreed, but certainly not just "due to unique tube problems
like Miller capacitance [.....] ".

Fess up Sander, Miller capacitance is a far more significant issue
with tubes.

I replied to your statement that those problems are "unique with
tubes, like Miller capacity etc".

Fess up Arnold, Miller capacitance is apparent in *every*
signal-inverting amplifying device.

Can apparent be a synonum for significant? If its not significant, why
worry about it?

High input impedance: dependent on circuitry.

Of course, but as a matter of course tubed circuts have far higher
input impedances.

Which is compensated for by the much karger grid voltage span.

Nope.

Yep.

Claim of superior expertise noted. Again.


Yet another red herring. BTW, the cross-modulation characteristics
of SS FM front ends are generally better than what we had when
tubes were king.

Only due to clamping diodes.

Nope. Cross modulation is about nonlinear distortion, and clamping
diodes increase that, not decrease it.

Remind me again when MOSFETs without internal clamping diodes are
available to designers.

Remind me when all SS FM front ends used MOSFETs,

Non-linear distortion, again due to the smaller BE voltage span as
compared to the larger GC span of a triode or penthode.

Apparant self-contradiction noted. Sander claims he never bypasses cathode
resistors but apparently demands that all emitter resistors be bypassed.

Granted that most tube amps have higher value input resistors.

Thanks for agreeing that your argument based on unrealistically low
grid resistors was irrelevant, Sander.

Not unrealistic at all considering my design views, but then again,
*I* am able to put off my blinders ;-)

What blinders? The vacuum tube horse has been dead in the mainstream
of hi fi for about 30 years. Putting blinders on a dead horse makes
very little sense.

Thanks anyway for discussing obsolete technology with me Arnold. ;-)

It's a waste of time, but whatever.

Actually 50 KHz or so is not an uncommon corner frequency for SS
amps.

Due to what property, exactly?

Desire to have an amp that is nice to work with.


I meant what causes the LPF character of an average SS amp?

Pole in negative feedback loop.

In what stages is that most apparent?

The whole amp. There's often a dominant pole at some lower frequency but its
LF gain is so high that its the pole in the negative feedback loop that sets
the performance of the amp.

Slewing was a problem in someone's mind by the time it was
publicized.

Since Otala, Garde et al, we know better ;-)

Yes, Otala is known to have been an alarmist, to say the least.

Otala was an optimist ;-)

OK, joke.

Back in the real world...

Besides, that is mostly a simple RC 6 dB.oct filter, not steep
enough to supress real strong RF energy.

A 6 dB/oct filter cornering at 50 KHz is down over 20 dB at 500
KHz, 40 dB at 5 MHz and 60 dB down at the 50 MHz that Mike
origionally mentioned.

One wrong grounding point may wreak havoc with that 60 dB number.

How many red herrings does this make?

What was the minimum requirement for measuring S/N again?

Signal and noise to measure....

When I once mentioned I couldn't measure better than -80 dB at one
time, you had a good laugh.

Something about being able to routinely measure signals 20-40 dB below that.

Think about a bipolar differential input, CMRR, and the actual signal
voltage needed to excurse it.

Depends on the circuit design.

You may use Google, if you like ;-)

Yawn.

"Arny Krueger" said:

My, I must be an exceptional tube amp designer ;-)

Yawn.


Do I bore you?
Just let me know, and I'll declare that I've "won" the debate and all
is good again. ;-)


First site I found with the search string tube preamp schematics:

http://www3.telus.net/public/vintage1/mu.jpg
1 meg grid resistor, bypassed cathode resistor


Keyword: vintage.


http://www3.telus.net/public/vintage1/phones.jpg
1 meg grid resistor, bypassed cathode resistor


Keyword: vintage. With a transformer!


http://www.triodeel.com/a340a.gif
220k grid resistor, bypassed cathode resistor


Keyword: vintage.


http://www.drtube.com/schematics/an/ankoru.gif
bypassed cathode resistor


Keyword: Audio Note. 'nuff said.


http://www.drtube.com/schematics/ar/d40.gif
475 k grid resistor


Mine reads 100 k in parallel with 475 k for AC.
Still high IMO.


http://www.drtube.com/schematics/ar/vt100.gif
100 K grid resistor


And a cap of 27 pF.


etc., etc.


I already granted you that most tube designs use too high values for
the first grid resistor.

There. Satisfied? ;-)


Please find a quote by me where I pretend to be an expert (or
eggsphert) in *anything* , except tinkering.

Sander wrote: " My, I must be an exceptional tube amp designer"


And that equals a claim of expertise how, exactly?

It could as well mean I'm the worst tube amp designer. ;-)


Fess up Arnold, Miller capacitance is apparent in *every*
signal-inverting amplifying device.

Can apparent be a synonum for significant? If its not significant, why
worry about it?


When you look at the definition of "Miller effect" or "Miller
capacitance", it appears clear to me that in certain cases, this
effect can NOT be neglected. Not with tubes, not with bipolars, not
with (MOS)FETs and not with IGBTs.

Tinker a lot? ;-)


Of course, but as a matter of course tubed circuts have far higher
input impedances.

Which is compensated for by the much larger grid voltage span.

in the context of RF-induced distortions

Nope.

Yep.

Claim of superior expertise noted. Again.


Nope, just experience in tinkering and repairing.


Non-linear distortion, again due to the smaller BE voltage span as
compared to the larger GC span of a triode or penthode.

Apparant self-contradiction noted. Sander claims he never bypasses cathode
resistors but apparently demands that all emitter resistors be bypassed.


Of course, the comparison ought to be unfair .

BTW you're right Arnold. ;-)


Thanks anyway for discussing obsolete technology with me Arnold. ;-)

It's a waste of time, but whatever.


Nope, it's just fun!


Actually 50 KHz or so is not an uncommon corner frequency for SS
amps.

I meant what causes the LPF character of an average SS amp?

Pole in negative feedback loop.


Only that?


In what stages is that most apparent?

The whole amp. There's often a dominant pole at some lower frequency but its
LF gain is so high that its the pole in the negative feedback loop that sets
the performance of the amp.


For the loop feedback to be effective, enough amplification at the
power bandwidth has to be there for it to be effective.

Now, if the circuit, as you seem to imply here, has very high
bandwidth, and the only reason for a SSamp to roll off at high
frequencies is the dominant pole of the feedback loop, what will
happen if some RFI signal appears at the input of the differential,
gets processed by the amp, only to find the feedback effective at the
very frequency of said signal?
Wouldn't that hang up the entire amplifier?
If yes, why isn't this effect widely known and dealt with?
If no, why do we use anti-slewing filters at inputs, and lead- and lag
networks throughout the amp?
And why are there LPFs at the outputs even?

Hint: see Otala for some explanation. ;-)


Otala was an optimist ;-)

OK, joke.

Back in the real world...


Yup. Keep a working cell phone or portophone near the back side of one
of your amplifiers.
Feed an audio signal through it and report the difference in YHD and
IM.
Now do the same with the feedback loop removed (neglecting the
absolute value of THD and IM, just the deviation).

I've done this test with some amps, tube and SS, and am willing to
discuss the results after you've seen what happens for yourself.


When I once mentioned I couldn't measure better than -80 dB at one
time, you had a good laugh.

Something about being able to routinely measure signals 20-40 dB below that.


OK, so the average 6 dB/oct filter at the input giving some 60 dB at
50 MHz attenuation, would that suffice if the RFI signal is strong?

Are signals below say 60 dB important for our perception of reproduced
music?


Think about a bipolar differential input, CMRR, and the actual signal
voltage needed to excurse it.

Depends on the circuit design.


Yes. At -60 dB, it becomes even more important. IMHO of course,
because I'm not an expert ;-)

--
Sander de Waal
" SOA of a KT88? Sufficient. "