Reply
 
Thread Tools Display Modes
  #1   Report Post  
Posted to rec.audio.tubes
patrick-turner patrick-turner is offline
external usenet poster
 
Posts: 119
Default valveradio.net

On Sunday, 24 February 2013 23:19:59 UTC+11, Alex Pogossov wrote:
On the dusk of my days I decided to follow an example of some folks and launched my website: www.valveradio.net It is supposed to cover some not very common (though not new) circuit ideas related to radio and audio. The site is pretty much empty, with only one relevant article so far: http://www.valveradio.net/en/home/au...-follower.html I will be adding more. Regards, Alex


Keep up the good work Alex. Most of the world quickly tires of having anything explained to them. But I have the time for those few who look deeper, who take longer, and who think every detail matters.

May I make some constructive suggestions?

Where you have schematics, it'd be nice to know what Vac exists at points around the schematic. For example, where you have a "boosted cathode follower" with a bjt in cathode circuit of a triode, just what voltage and current flows in/out of the bjt base? most bjts have very low input Z and less than 1k which would shunt any voltage production at the 33k anode resistance. There may be good reason to use a j-fet instead of the bjt. But then one has to watch out the circuit doesn't become an oscillator.
Your boosted follower looks very much like a White Follower, but the White is using an anode resistor in the top tube follower to create a drive voltage for the bottom tube. as the load on the follower becomes lower and Iac rises, then the bottom tube does more and it works in PP with top tube. Its not a perfect kind of PP follower, and for such an animal, more tubes are needed, and it all gets so complex ya wouldn't wanna do it.

With the REAL boosted follower, you have would have 3 triodes. One might use a 6CG7, and 6BQ5 in triode.
The boosted follower is actually a cascade amp with the 2 halves of 6CG7, and V1 has a lowish value RLa, maybe 10k. Its cathode R goes to a negative rail, maybe -200V, maybe it is 5k0. Signal into V1 generates a voltage at V1a to drive V2 which is normnal common cathode gain stage with gain about 16.. The V2a drives the V3 6BQ5 grid with a signal much higher than Vin at V1. The cathode of V3 is connected to V1 cathode. So what is happening in the circuit is that the internal gain of the 6BQ5 is boosted maybe 30 times by V1 and V2, and the cathode voltage at V1 is then much more determined to follower V1 grid signal, and Rout is many times lower than the Rout of a normal 6BQ5. If the 6BQ5 has gm = 0.005A/V, Rout = RL // ( 1/ Gm ) = 5k0 // ( 1/0.005 ) = 200 ohms approx.
Where RLk is many times 1 / gM, it may be neglected.

Distortion reduction in all followers Dn' = Dn / ( 1 + [ A x ß ] ) where Dn' is the distortion measured with the load value used when tube is in follower mode, Dn is the distortion if the tube where the load used is in a common cathode gain stage, and A = Va-k / Vg-k, and ß = fraction fed back. In the case of a follower, ß = 1.0, because all the output is fed back and is in series with input grid signal. So the follower formula for Dn' becomes more simply, Dn' = Dn / 1 / ( 1 + A ).

The same principles apply to j-fets, mosfets or bjts. A common Garden Variety of signal bjt with Ic = 10mA and emitter load of 2k0 may be found to have voltage gain of its gM x RL = 0.2A/V x 2,000 = 400. ( this is approximate ).
Suppose the bjt generated 5% THD at 5Vrms when 2k0 is between collector and B+ of +40V. AND suppose the emitter was grounded. Then you'd find Vbe needed to generate 10V = 5V / 400 = 0.0125Vac. A tiny voltage. Well, with the 2k0 in emitter circuit, Vbe remains at 0.0125V, and Vb needed is 5.0125Vac, hardly different to Ve output. The THD is reduced from 5% to 5% / 401, = 0.0124 approx.I'm not sure anyone needs to know the formulas for emitter follower distortion that Alex has, but then RDH has formulas as well for tube distortions but hey, why spend all day making mistakes with formulas, or ****ing around with a simulation program when it would be more realistic to build a circuit, try it out, and MEASURE real world performance. I have found the simple formula above holds true for folowers, and I never bother to quantify follower harmonics because followers have such low THD. However, with any bjt, the input resistance to base is low, and non linear, like a diode turn on threshold voltage. The bjt operates over a small voltage range from about 0.45V to 0.75V, only 0.3V range. So when you measure performance of the bjt, one must be fair to your mind and drive the base with extremely low impedance or drive the base with a current source with high Z and because bjts have a fairly linear relationship of base current to collector current, the current drive gives a more linear performance. But any change to collector current affects the base current. From my point of view based on years of experience, I am extremely reluctant to trust any bjt with providing active gain to any circuit anywhere with audio. But my disdain for bjts hasn't stopped me using them as constant current sources where they excel as slaves to the current requirement of the tubes. Not only that, I've just spent 3 weeks of spare time re-building my bench AC volt meter. I did try some CA1430 "fast" op-amps choko-blocko with bjts plus a few input fets, but I found them too slow for over 1MHz of bandwidth so I made the wideband amps and metering amp using split rails, +/-15Vdc, discrete bjts arranged with LTP and CCS, and output bjts with CCS, and all DC coupled so that the devices make a well regulated 0V at their Vo. I MHz at 7Vrms was dead easy, with good gain reduction with NFB from about 2,500 to 10.

Anyway, to conclude, there is a lot more you could tell us about with your boosted cathode follower, but have you asked yourself if anyone knows how a bjts works? They are very tricky things to use, and very prone to oscillations. In my case the self made meter op-amp drives a bridge rectifier to produce 1Madc for the 50 ohm meter for full swing. The rectifier is in series with 90 ohms and join goes to FB connection of amp. My amp wanted to oscillate at 15MHz and it needed 200pF between Q1 collector and VAS amp base to reduce HF gain. Most people posting up circuits for bjt stuff often leave out many of the finer points about the pain and agony it takes to get something working properly.

What I made is better than what it replaced, which was what I made in 1993.
I can now measure 1mV with a nice big meter face, accurately, so I can see 0.1mV, and for all F between 2Hz and 1MHz. Defeating noise was done using an input stage with 2SK369.

It wasn't until I ws 1/2 way done with the meter re-build that I realise there are piles of available fast opamps with rise times of up to 6,000V/us.
I managed to get my op-amps to give 59V/uS, a vast improvement over using normal op-amps such as NE5534, TLO72, etc. Such "audio" op-amps have their place, but in my mind they don't belong in test gear which needs to have extended HF performance above 200kHz.

I recall my first web-page in 2001. Boy, what a dumb page that was! I was told I'd make a lot of money with such a simple page. Maybe I got 2 emails in the first year from guys wanting to pay peanuts. Then I decided that if my website was never to make me rich, I may as well make it useful for those who wish to learn about tube gear worth using with music. I've currently got 3 big new pages underway...... soon be 40MB.....All free, no copywrite. What happens is that some get ideas and build something without copying every single one of maybe 2,000 details in an an average tube power amp. They don't bother to learn basic properties of devices and other stuff. So if they get smoke it ain't my fault. So the free info has a catch.

Smoke is the reward for those who have the confidence of the un-skilled.

Patrick Turner
  #2   Report Post  
Posted to rec.audio.tubes
Alex Pogossov Alex Pogossov is offline
external usenet poster
 
Posts: 95
Default valveradio.net

Thanks Patrick,

It is a useful advice to specify AC voltages where applicable. Will try next
time.
We all are free to publish in any style we like.
My general direction was (I thought) to promote relatively simple solutions
where adding a transistor dramatically improves tubed circuit without adding
complexity..

With this boosted cathode follower, you almost always can solder the bjt
into an existing pure cathode follower. Literally, only one extra solder
lug -- to support the base of the bjt -- will have to be added. And the
output impedance and distortion reduces significantly.

And a few quick comments:

Using a bjt for boosting is better. It is crucial to use bjt as a current
amplifier, not voltage amplifier.
JFET as a boosting element will give worse results both in distortion and
output impedance. Besides it will be prone to oscillation, as you correctly
mentioned, due to a pole in the hi-Z gate node.

Interestingly that a tube with a bjt for boosting gives the best results and
the widest range of applications -- can work up to high signal voltages with
low distortion and low output impedance. Other combinations are inferior,
with only perhaps boosted emitter follower being better on distortion, but
things may go wrong if the signal source has large impedance and the "main"
following transistor has Early effect (beta modulation).

----------------------

"patrick-turner" wrote in message
...
On Sunday, 24 February 2013 23:19:59 UTC+11, Alex Pogossov wrote:
On the dusk of my days I decided to follow an example of some folks and
launched my website: www.valveradio.net It is supposed to cover some not
very common (though not new) circuit ideas related to radio and audio. The
site is pretty much empty, with only one relevant article so far:
http://www.valveradio.net/en/home/au...-follower.html I
will be adding more. Regards, Alex


Keep up the good work Alex. Most of the world quickly tires of having
anything explained to them. But I have the time for those few who look
deeper, who take longer, and who think every detail matters.

May I make some constructive suggestions?

Where you have schematics, it'd be nice to know what Vac exists at points
around the schematic. For example, where you have a "boosted cathode
follower" with a bjt in cathode circuit of a triode, just what voltage and
current flows in/out of the bjt base? most bjts have very low input Z and
less than 1k which would shunt any voltage production at the 33k anode
resistance. There may be good reason to use a j-fet instead of the bjt. But
then one has to watch out the circuit doesn't become an oscillator.
Your boosted follower looks very much like a White Follower, but the White
is using an anode resistor in the top tube follower to create a drive
voltage for the bottom tube. as the load on the follower becomes lower and
Iac rises, then the bottom tube does more and it works in PP with top tube.
Its not a perfect kind of PP follower, and for such an animal, more tubes
are needed, and it all gets so complex ya wouldn't wanna do it.

With the REAL boosted follower, you have would have 3 triodes. One might use
a 6CG7, and 6BQ5 in triode.
The boosted follower is actually a cascade amp with the 2 halves of 6CG7,
and V1 has a lowish value RLa, maybe 10k. Its cathode R goes to a negative
rail, maybe -200V, maybe it is 5k0. Signal into V1 generates a voltage at
V1a to drive V2 which is normnal common cathode gain stage with gain about
16. The V2a drives the V3 6BQ5 grid with a signal much higher than Vin at
V1. The cathode of V3 is connected to V1 cathode. So what is happening in
the circuit is that the internal gain of the 6BQ5 is boosted maybe 30 times
by V1 and V2, and the cathode voltage at V1 is then much more determined to
follower V1 grid signal, and Rout is many times lower than the Rout of a
normal 6BQ5. If the 6BQ5 has gm = 0.005A/V, Rout = RL // ( 1/ Gm ) = 5k0 //
( 1/0.005 ) = 200 ohms approx.
Where RLk is many times 1 / gM, it may be neglected.

Distortion reduction in all followers Dn' = Dn / ( 1 + [ A x ß ] ) where Dn'
is the distortion measured with the load value used when tube is in follower
mode, Dn is the distortion if the tube where the load used is in a common
cathode gain stage, and A = Va-k / Vg-k, and ß = fraction fed back. In the
case of a follower, ß = 1.0, because all the output is fed back and is in
series with input grid signal. So the follower formula for Dn' becomes more
simply, Dn' = Dn / 1 / ( 1 + A ).

The same principles apply to j-fets, mosfets or bjts. A common Garden
Variety of signal bjt with Ic = 10mA and emitter load of 2k0 may be found to
have voltage gain of its gM x RL = 0.2A/V x 2,000 = 400. ( this is
approximate ).
Suppose the bjt generated 5% THD at 5Vrms when 2k0 is between collector and
B+ of +40V. AND suppose the emitter was grounded. Then you'd find Vbe needed
to generate 10V = 5V / 400 = 0.0125Vac. A tiny voltage. Well, with the 2k0
in emitter circuit, Vbe remains at 0.0125V, and Vb needed is 5.0125Vac,
hardly different to Ve output. The THD is reduced from 5% to 5% / 401, =
0.0124 approx.I'm not sure anyone needs to know the formulas for emitter
follower distortion that Alex has, but then RDH has formulas as well for
tube distortions but hey, why spend all day making mistakes with formulas,
or ****ing around with a simulation program when it would be more realistic
to build a circuit, try it out, and MEASURE real world performance. I have
found the simple formula above holds true for folowers, and I never bother
to quantify follower harmonics because followers have such low THD. However,
with any bjt, the input resistance to base is low, and non linear, like a
diode turn on threshold voltage. The bjt operates over a small voltage range
from about 0.45V to 0.75V, only 0.3V range. So when you measure performance
of the bjt, one must be fair to your mind and drive the base with extremely
low impedance or drive the base with a current source with high Z and
because bjts have a fairly linear relationship of base current to collector
current, the current drive gives a more linear performance. But any change
to collector current affects the base current. From my point of view based
on years of experience, I am extremely reluctant to trust any bjt with
providing active gain to any circuit anywhere with audio. But my disdain for
bjts hasn't stopped me using them as constant current sources where they
excel as slaves to the current requirement of the tubes. Not only that, I've
just spent 3 weeks of spare time re-building my bench AC volt meter. I did
try some CA1430 "fast" op-amps choko-blocko with bjts plus a few input fets,
but I found them too slow for over 1MHz of bandwidth so I made the wideband
amps and metering amp using split rails, +/-15Vdc, discrete bjts arranged
with LTP and CCS, and output bjts with CCS, and all DC coupled so that the
devices make a well regulated 0V at their Vo. I MHz at 7Vrms was dead easy,
with good gain reduction with NFB from about 2,500 to 10.

Anyway, to conclude, there is a lot more you could tell us about with your
boosted cathode follower, but have you asked yourself if anyone knows how a
bjts works? They are very tricky things to use, and very prone to
oscillations. In my case the self made meter op-amp drives a bridge
rectifier to produce 1Madc for the 50 ohm meter for full swing. The
rectifier is in series with 90 ohms and join goes to FB connection of amp.
My amp wanted to oscillate at 15MHz and it needed 200pF between Q1 collector
and VAS amp base to reduce HF gain. Most people posting up circuits for bjt
stuff often leave out many of the finer points about the pain and agony it
takes to get something working properly.

What I made is better than what it replaced, which was what I made in 1993.
I can now measure 1mV with a nice big meter face, accurately, so I can see
0.1mV, and for all F between 2Hz and 1MHz. Defeating noise was done using an
input stage with 2SK369.

It wasn't until I ws 1/2 way done with the meter re-build that I realise
there are piles of available fast opamps with rise times of up to 6,000V/us.
I managed to get my op-amps to give 59V/uS, a vast improvement over using
normal op-amps such as NE5534, TLO72, etc. Such "audio" op-amps have their
place, but in my mind they don't belong in test gear which needs to have
extended HF performance above 200kHz.

I recall my first web-page in 2001. Boy, what a dumb page that was! I was
told I'd make a lot of money with such a simple page. Maybe I got 2 emails
in the first year from guys wanting to pay peanuts. Then I decided that if
my website was never to make me rich, I may as well make it useful for those
who wish to learn about tube gear worth using with music. I've currently got
3 big new pages underway...... soon be 40MB.....All free, no copywrite. What
happens is that some get ideas and build something without copying every
single one of maybe 2,000 details in an an average tube power amp. They
don't bother to learn basic properties of devices and other stuff. So if
they get smoke it ain't my fault. So the free info has a catch.

Smoke is the reward for those who have the confidence of the un-skilled.

Patrick Turner


  #3   Report Post  
Posted to rec.audio.tubes
patrick-turner patrick-turner is offline
external usenet poster
 
Posts: 119
Default valveradio.net

On Wednesday, 27 February 2013 23:11:52 UTC+11, Alex Pogossov wrote:
Thanks Patrick, It is a useful advice to specify AC voltages where applicable. Will try next time. We all are free to publish in any style we like. My general direction was (I thought) to promote relatively simple solutions where adding a transistor dramatically improves tubed circuit without adding complexity.. With this boosted cathode follower, you almost always can solder the bjt into an existing pure cathode follower. Literally, only one extra solder lug -- to support the base of the bjt -- will have to be added. And the output impedance and distortion reduces significantly. And a few quick comments: Using a bjt for boosting is better. It is crucial to use bjt as a current amplifier, not voltage amplifier.


***Don't let me spoil your posting style. And don't let me ever stop you asking questions. If I had a quid for every time someone said to me " Patrick, for heaven's sake man, please be clearer" then I'd be rich. A building site manager once said to me, 24 yo, and with a tendency to bull****, "Son, say what you mean, and mean what you say..." The words are still echoing around in my head, some 43 years later. Now, sometimes ppl might think I'm tactless, no empathy either. Nope, It just means I've left out the waffle&bull****.

From what little I could see, the bjt in your booster does act as a current amp because you have a higher source drive from 33k // 12AX7 Ra to drive the bjt base with grounded emitter, so Rb input is very low. So the 12AX7 anode voltage change is low, probably negligible, thus allowing the 12AX7 to work as a follower.

JFET as a boosting element will give worse results both in distortion and output impedance.

***Well, sure there would be a big difference between bjt and j-fet. J-fet Rin at gate is high like a triode grid. J-fet Gm is much lower than a bjt.

But for me to really assess what you are saying means I have to build your circuit and measure it carefully. I'm time limited, and have a very full list of other interests, and, I've never been keen to do other people's homework when I have such huge amounts of my own.

Besides it will be prone to oscillation, as you correctly mentioned, due to a pole in the hi-Z gate node. Interestingly that a tube with a bjt for boosting gives the best results and the widest range of applications -- can work up to high signal voltages with low distortion and low output impedance. Other combinations are inferior, with only perhaps boosted emitter follower being better on distortion, but things may go wrong if the signal source has large impedance and the "main" following transistor has Early effect (beta modulation).

**It all seems like you are using a tiny tube current change to change a base current in bjt to get a bigger collector current change so bjt acts like an active current source load for triode, and its sort of PCFB which is then regulated by the NFB of the follower connection. But with just a 12AX7 with say idle Ia = 1mA, then class A current change can only ever be +/- 1mA peak, and anything more gives high THD. What you have is a tube white follower, super-charged with a bjt which has high ac current gain and hence strong action in the 0.7mArms region. One has to wonder why more ppl don't use white followers, or your idea of boosting. More than one idea of boosting can and do exist.
Patrick Turner.

Reply
Thread Tools
Display Modes

Posting Rules

Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT +1. The time now is 10:35 PM.

Powered by: vBulletin
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
Copyright ©2004-2024 AudioBanter.com.
The comments are property of their posters.
 

About Us

"It's about Audio and hi-fi"