[Ian Bell[_2_]]

I have done some follow up work on 6SN7 mu follower distortion by
testing over 50 6CG7/6FQ7 samples. New paper on this is he

http://www.ianbell.ukfsn.org/data/6C...distortion.pdf

Cheers

Ian

[Doug Bannard]

"Ian Bell" wrote in message
...
I have done some follow up work on 6SN7 mu follower distortion by testing
over 50 6CG7/6FQ7 samples. New paper on this is he

http://www.ianbell.ukfsn.org/data/6C...distortion.pdf

Cheers

Ian

Thanks Ian. Good work! I saved myself a copy.

Best Regards : Doug Bannard

[Ian Bell[_2_]]

Doug Bannard wrote:
"Ian Bell" wrote in message
...
I have done some follow up work on 6SN7 mu follower distortion by testing
over 50 6CG7/6FQ7 samples. New paper on this is he

http://www.ianbell.ukfsn.org/data/6C...distortion.pdf

Cheers

Ian

Thanks Ian. Good work! I saved myself a copy.

Best Regards : Doug Bannard




Thanks Doug. I had long suspected there were measurable differences
depending on the way the tubes are constructed and after these tests it
seems there really is.

Cheers

Ian

[Patrick Turner]

On Oct 7, 4:00*am, Ian Bell wrote:
I have done some follow up work on 6SN7 mu follower distortion by
testing over 50 6CG7/6FQ7 samples. New paper on this is he

http://www.ianbell.ukfsn.org/data/6C...distortion.pdf

Cheers

Ian

Many thanks for your informative and exhausting account of what goes
on in a 6CG7. I use this tube in many of my amps which I custom build
for my discerning clients, and I have always found the mu-foll stage
to sound well, and better than the SRPP.

What was the exact schematic used for the test results?

What would be the distortion result for 20Vrms output when you test
the tube with CCS load?

It is interesting that you have found the recently made Russian 6CG7
to measure better than the much favoured US made NOS versions of the
tube. I'll never forget a day some years ago when 3 friends and myself
had a tube rolling session using NOS Siemans, Mullard, AWV Australian
and one new stock EH 6CG7. This was in a preamp arranged so that the
CD signal level was attenuated by a DACT switched attenuator, then
amplified for the power amps which were SE 23W SEUL with 13E1, and
speakers which were Vienna Acoustic Motzarts.
The preamp 6CG7 gain was configured as a normal resistance loaded gain
triode with bypassed cathode and 47k dc RL. This was direct coupled to
a second triode as a CF with 47k dc RL.
Ia was about 3.5mA.
Output levels needed to give good listening were always below about
0.5Vrms, and some weeks later while doing maintenance I measured the
amp and found THD 0.01% for levels 0.5Vrms.


We agreed on the day that the Seimans sounded best, followed very
closely by the AWV, with Mullard sounding nice and polite, and good
for harsh sounding females. The EH was found to be way behind and none
of us liked its sound. One would think that if we were to record low
distortion levels then the tube should sound better. Your tests would
tend to indicate the russian made 6CG7 should sound the best, but we
didn't find this and we were working the tube in regions where the THD
was 0.01% or less, and not at 20Vrms where its likely to be 0.3%
according to your tests.

If THD varies about linearly between the 20V level and the 0.5V level,
then we could expect any sample of yours which measured 0.4% at 20Vrms
to make 0.01% at 0.5V which is about what I measured just using a
normal R loaded triode. Usually, the least THD made by 2 triodes is
where one has a CCS load and is then direct coupled to a CF with CCS
cathode load, and the output is taken to a high Z input THD meter.
Then the real potential for low THD with SE signal triodes is
realised, but anyway, 0.01% is an entirely blameless performance and
with only this much THD the resulting IMD would be entirely inaudible.
0.01% is -80dB.

So, what makes different samples of the same tube number sound
different when used in preamps where the THD/IMD is plainly and simply
jolly inaudible?

At this point, I could be forgiven if I said "I'm fuct if I know".

But methinks the answer may be simpler, and its the microphonics, and
the microscopic movements of the electrodes caused by the tiny
electrostatic forces within the tube.
The usual testing of THD with a sine wave is unlikely to show these
very subtle but probably audible effects which might measure way above
THD, but because the artifacts may not be harmonically related to the
test tone used they may well be missed by the measuring gear, or put
another way nobody is looking to measure in the right way.
Mircophony in a bad triode will often manifest itself usually as a
mechanical resonance of electrodes swaying around inside the glass at
some F which would rarely ever be conveniently harmonically related to
any usual test F used.
Now some tubes have such woefully aged electrodes that they become so
microphonic that you get acoustic feedback and the amp system then
squeals or howls. But you could and you do get some tubes which have
not reached the stage of being really bad enough to allow acoustic FB
but you will be getting some injection of resonant tones into the
music signal. And then If the resonant tone F or number of them were
related to say middle C, then music played in that key should sound
well, quite euphonic in fact, and many people will die to get such
sound from vacuum tubes, and it isn't difficult to arrange if you like
DHT. But usually you have little control over the tube resonant F, so
it rarely "fits" into the music score, so some music might sound good
while other music in another key might make you want to buy a better
system.

Using adequate amounts of GNFB around stages in amplifiers will reduce
the artifacts of resonance like it reduces everything else - within
limitations of open loop BW, amount of FB and amount of THD present
without the NFB.

Some would now say I have just put BOTH my feet in my mouth when I
have described why tubes sound different, and what to do about it with
FB.

But I have found that seldom do I ever need to put much FB around any
stage with 6CG7. They are inherently bloody good triodes, and afaiac,
the little king of signal tubes; a gold standard in fact. When I have
used some shunt FB with a mu-foll stage its a mild amount to just
bring the gain down a little, maybe offer some selectable HF shelving
to tame harsh CDs, or for plain old fashioned tone controls with FB
and Baxandal networks.
I quite like the 12AU7. Its another sin to say that because the curves
look horrible compared to 6CG7, but the AU7 is a glory to me. You
don't get too much gain with it. And generally it sounds excellent,
which is why I like tubes and can make a living using them.

Patrick Turner.

[Ian Bell[_2_]]

Patrick Turner wrote:
On Oct 7, 4:00 am, Ian Bell wrote:
I have done some follow up work on 6SN7 mu follower distortion by
testing over 50 6CG7/6FQ7 samples. New paper on this is he

http://www.ianbell.ukfsn.org/data/6C...distortion.pdf

Cheers

Ian

Many thanks for your informative and exhausting account of what goes
on in a 6CG7. I use this tube in many of my amps which I custom build
for my discerning clients, and I have always found the mu-foll stage
to sound well, and better than the SRPP.

What was the exact schematic used for the test results?


The circuit is the same as for the original paper except I have change
both cathode resistors to 1K following your suggestion to try slightly
lower current operating points. Using the original 320V PSU gives an
operating current in the region of 4.5mA.

What would be the distortion result for 20Vrms output when you test
the tube with CCS load?


I don't know, but I suspect it will be very similar to the results
Morgan Jones got when he used a pentode CCS with the 6SN7.


It is interesting that you have found the recently made Russian 6CG7
to measure better than the much favoured US made NOS versions of the
tube. I'll never forget a day some years ago when 3 friends and myself
had a tube rolling session using NOS Siemans, Mullard, AWV Australian
and one new stock EH 6CG7. This was in a preamp arranged so that the
CD signal level was attenuated by a DACT switched attenuator, then
amplified for the power amps which were SE 23W SEUL with 13E1, and
speakers which were Vienna Acoustic Motzarts.
The preamp 6CG7 gain was configured as a normal resistance loaded gain
triode with bypassed cathode and 47k dc RL. This was direct coupled to
a second triode as a CF with 47k dc RL.
Ia was about 3.5mA.
Output levels needed to give good listening were always below about
0.5Vrms, and some weeks later while doing maintenance I measured the
amp and found THD 0.01% for levels 0.5Vrms.


We agreed on the day that the Seimans sounded best, followed very
closely by the AWV, with Mullard sounding nice and polite, and good
for harsh sounding females. The EH was found to be way behind and none
of us liked its sound. One would think that if we were to record low
distortion levels then the tube should sound better. Your tests would
tend to indicate the russian made 6CG7 should sound the best, but we
didn't find this and we were working the tube in regions where the THD
was 0.01% or less, and not at 20Vrms where its likely to be 0.3%
according to your tests.

If THD varies about linearly between the 20V level and the 0.5V level,
then we could expect any sample of yours which measured 0.4% at 20Vrms
to make 0.01% at 0.5V which is about what I measured just using a
normal R loaded triode. Usually, the least THD made by 2 triodes is
where one has a CCS load and is then direct coupled to a CF with CCS
cathode load, and the output is taken to a high Z input THD meter.
Then the real potential for low THD with SE signal triodes is
realised, but anyway, 0.01% is an entirely blameless performance and
with only this much THD the resulting IMD would be entirely inaudible.
0.01% is -80dB.

So, what makes different samples of the same tube number sound
different when used in preamps where the THD/IMD is plainly and simply
jolly inaudible?

At this point, I could be forgiven if I said "I'm fuct if I know".

But methinks the answer may be simpler, and its the microphonics, and
the microscopic movements of the electrodes caused by the tiny
electrostatic forces within the tube.
The usual testing of THD with a sine wave is unlikely to show these
very subtle but probably audible effects which might measure way above
THD, but because the artifacts may not be harmonically related to the
test tone used they may well be missed by the measuring gear, or put
another way nobody is looking to measure in the right way.
Mircophony in a bad triode will often manifest itself usually as a
mechanical resonance of electrodes swaying around inside the glass at
some F which would rarely ever be conveniently harmonically related to
any usual test F used.
Now some tubes have such woefully aged electrodes that they become so
microphonic that you get acoustic feedback and the amp system then
squeals or howls. But you could and you do get some tubes which have
not reached the stage of being really bad enough to allow acoustic FB
but you will be getting some injection of resonant tones into the
music signal. And then If the resonant tone F or number of them were
related to say middle C, then music played in that key should sound
well, quite euphonic in fact, and many people will die to get such
sound from vacuum tubes, and it isn't difficult to arrange if you like
DHT. But usually you have little control over the tube resonant F, so
it rarely "fits" into the music score, so some music might sound good
while other music in another key might make you want to buy a better
system.

Using adequate amounts of GNFB around stages in amplifiers will reduce
the artifacts of resonance like it reduces everything else - within
limitations of open loop BW, amount of FB and amount of THD present
without the NFB.

Some would now say I have just put BOTH my feet in my mouth when I
have described why tubes sound different, and what to do about it with
FB.

But I have found that seldom do I ever need to put much FB around any
stage with 6CG7. They are inherently bloody good triodes, and afaiac,
the little king of signal tubes; a gold standard in fact. When I have
used some shunt FB with a mu-foll stage its a mild amount to just
bring the gain down a little, maybe offer some selectable HF shelving
to tame harsh CDs, or for plain old fashioned tone controls with FB
and Baxandal networks.
I quite like the 12AU7. Its another sin to say that because the curves
look horrible compared to 6CG7, but the AU7 is a glory to me. You
don't get too much gain with it. And generally it sounds excellent,
which is why I like tubes and can make a living using them.

Patrick Turner.


Bottom line is the specs tell only part of the story. I built the
prototype mic pre using Mitsubishi 6CG7s and did some test recordings
and emailed them to cipher. He loved them. I then shipped it to him but
DHL managed to break both tubes. Cipher replaced them with GE types
which he said sounded great but were microphonic. In my tests the GE
tubes turned out worst (but I had already determined they were noisy and
more microphonic than most others.)

At the end of the day I'll fit the tubes my customers prefer the sound of.

Cheers

Ian