In my amp the little itty bit resistor is replaced with a wopping 1meg resistor. The idea is that current can only travel differentially between the tube cathodes. I wounder how this relates to the situation described by Allen Wright ?
When I implemented it I definately noticed a tightening up of the overall sound over each cap going straight to earth. Does anyone think the smaller resistor would produce better results ?
Shoog
Guys,
The reduction in 3rd harmonic distortion and intermodulation products is consistent with claims made about this sort of circuit.
The thinking behind this idea grew out of some serious tube gurus looking at circuits such as the Western Electric WE86 Amplifier and what they called the "Harmonic Equaliser".
My very limited understanding suggests that it is a mixing effect which is also producing a common mode feedback voltage via the common resistor in the cathode circuit.
When you mix 2 signals in a square law device (the output tubes + output tranny) you get:
(sin(x) + siny
)^2
Doing the algebra this reduces to:
sin(x+y) + sin(x-y)
That is then being applied as a feedback signal which is why (my guess) we see a reducion in intermodulation products. Also the mixing function cancels 2nd harmonics and leaves us with 3rd harmonics so thats why you also see a reduction in 3H distortion.
A gross simplification which points to the fact that I don't understand it either.
Also if you do searches for original papers and patents you will eventually find a 1936 paper by Nadell the designer (or one of them at least) of the WE86 Amplifier and trying to wade through that paper suggests to me at least that Nadell did'nt understand it either , or perhaps I do him a dis-service, he might have understood it but just could'nt explain it to others in simple english.
Smarter guys than me have trouble understanding whats going on BUT haveing observed that it works are just happy to use it.
Cheers,
Ian
The reduction in 3rd harmonic distortion and intermodulation products is consistent with claims made about this sort of circuit.
The thinking behind this idea grew out of some serious tube gurus looking at circuits such as the Western Electric WE86 Amplifier and what they called the "Harmonic Equaliser".
My very limited understanding suggests that it is a mixing effect which is also producing a common mode feedback voltage via the common resistor in the cathode circuit.
When you mix 2 signals in a square law device (the output tubes + output tranny) you get:
(sin(x) + siny
)^2Doing the algebra this reduces to:
sin(x+y) + sin(x-y)
That is then being applied as a feedback signal which is why (my guess) we see a reducion in intermodulation products. Also the mixing function cancels 2nd harmonics and leaves us with 3rd harmonics so thats why you also see a reduction in 3H distortion.
A gross simplification which points to the fact that I don't understand it either.
Also if you do searches for original papers and patents you will eventually find a 1936 paper by Nadell the designer (or one of them at least) of the WE86 Amplifier and trying to wade through that paper suggests to me at least that Nadell did'nt understand it either , or perhaps I do him a dis-service, he might have understood it but just could'nt explain it to others in simple english.
Smarter guys than me have trouble understanding whats going on BUT haveing observed that it works are just happy to use it.
Cheers,
Ian
Thanks Ian.
I have had my head in the books for the last few weeks with school, and now I have a licence renewal on Sunday for work, so quite a few more hours of study ahead. Has left me with a head full of porridge!
Just to make sure I have things straight...
Lift the zero volt leg of the bypass caps in each bias block. Tie these zero volt cap legs of each push pull pair together. Join these tied leads back to old zero volt through 33R resistor. In other words, one 33R resistor then required per channel.
Sorry for the dumb questions, but with a head full of putty at the moment, I am bound to screw it up if given half a chance
Cheers,
Chris
I have had my head in the books for the last few weeks with school, and now I have a licence renewal on Sunday for work, so quite a few more hours of study ahead. Has left me with a head full of porridge!
Just to make sure I have things straight...
Lift the zero volt leg of the bypass caps in each bias block. Tie these zero volt cap legs of each push pull pair together. Join these tied leads back to old zero volt through 33R resistor. In other words, one 33R resistor then required per channel.
Sorry for the dumb questions, but with a head full of putty at the moment, I am bound to screw it up if given half a chance
Cheers,
Chris
Chris,
You have it right.
The idea is to get the AC push and pull currents which flow through the CCS bypass caps to pass through a small common resistance (the added resistor). That generates a small voltage which effectively becomes a feedback voltage to both the push and pull sides.
Remember the following:
DIY Moto #1: There are no "dumb questions" just "dumb answers".
DIY Moto #2: There are no "screw ups" just "learning experiences"
Professional Engineers responses to the Technical Manager:
"We have advanced to new and surprising levels of bafflement"
"I now know six ways how not to do it"
Cheers,
Ian
You have it right.
The idea is to get the AC push and pull currents which flow through the CCS bypass caps to pass through a small common resistance (the added resistor). That generates a small voltage which effectively becomes a feedback voltage to both the push and pull sides.
Remember the following:
DIY Moto #1: There are no "dumb questions" just "dumb answers".
DIY Moto #2: There are no "screw ups" just "learning experiences"
Professional Engineers responses to the Technical Manager:
"We have advanced to new and surprising levels of bafflement"
"I now know six ways how not to do it"
Cheers,
Ian
More from AW
Guys,
as AW has been mentioned I thought I would add the following though I suspect everyone already knows this.
AW went from the PP1 circuit as posted with the harmonic reduction resister to a full differential output stage with 300B amp.
He also advised, at one stage, that he was using a combination of cathode and fixed bias. The cathode bias part was the voltage dropped across the CCS, with fixed bias for the rest.
Maybe something to try?
tim
Guys,
as AW has been mentioned I thought I would add the following though I suspect everyone already knows this.
AW went from the PP1 circuit as posted with the harmonic reduction resister to a full differential output stage with 300B amp.
He also advised, at one stage, that he was using a combination of cathode and fixed bias. The cathode bias part was the voltage dropped across the CCS, with fixed bias for the rest.
Maybe something to try?
tim
Thanks Ian,
In my case, it is: no such thing as a stupid question, just stupid people!
So, in that spirit, another question:
No 33Rs in the parts bin. Can parallel two 2 watt 68R carbon film, or series two 1 watt metal film 18R.
Any choice superior to the other? Will take a break shortly and pop them in....
Thanks again,
Chris.
In my case, it is: no such thing as a stupid question, just stupid people!
So, in that spirit, another question:
No 33Rs in the parts bin. Can parallel two 2 watt 68R carbon film, or series two 1 watt metal film 18R.
Any choice superior to the other? Will take a break shortly and pop them in....
Thanks again,
Chris.
Tim,
Quite right - I've been avoiding going to full differential mode (cathodes tied together with just one CCS to 0V) because it forces strict Class A operation and output power is significantly reduced. It can certainly be done but it would'nt be a Baby Huey anymore. I am however considering it for my latest design, Parallel Push Pull 300B (A 300B quad) monoblocks.
Cheers,
Ian
Chris,
Parallel 68 Ohms will be fine.
Quite right - I've been avoiding going to full differential mode (cathodes tied together with just one CCS to 0V) because it forces strict Class A operation and output power is significantly reduced. It can certainly be done but it would'nt be a Baby Huey anymore. I am however considering it for my latest design, Parallel Push Pull 300B (A 300B quad) monoblocks.
Cheers,
Ian
Chris,
Parallel 68 Ohms will be fine.
Shoog and Sheldon.
In Shoog's circuit (as I understand it) he effectively ties the 2 cathodes together with back to back caps and then just adds a 1M "DC Bleed" to 0V.
The whole idea of separate CCSs and separate bypass caps is that the bypass caps allow the transition to Class B.
With Shoog's arrangement then the maximum peak AC current will only be 2 x idle current as set by the CCS (one tube off and its CCS conducting extra AC signal current equal to its normal idle current from the other side) - That is, its Class A ONLY. I don't see any real difference between that and tying the cathodes together directly and using a single CCS set for 2 x the idle current of each tube - except that DC idle current balance will be ensured using Shoog's scheme.
I don't think you can compare the 2 schemes - operation is quite different. Certainly there is no convenient point (except maybe the ground side of the CCSs) where you could add the common AC path resistor.
Cheers,
Ian
In Shoog's circuit (as I understand it) he effectively ties the 2 cathodes together with back to back caps and then just adds a 1M "DC Bleed" to 0V.
The whole idea of separate CCSs and separate bypass caps is that the bypass caps allow the transition to Class B.
With Shoog's arrangement then the maximum peak AC current will only be 2 x idle current as set by the CCS (one tube off and its CCS conducting extra AC signal current equal to its normal idle current from the other side) - That is, its Class A ONLY. I don't see any real difference between that and tying the cathodes together directly and using a single CCS set for 2 x the idle current of each tube - except that DC idle current balance will be ensured using Shoog's scheme.
I don't think you can compare the 2 schemes - operation is quite different. Certainly there is no convenient point (except maybe the ground side of the CCSs) where you could add the common AC path resistor.
Cheers,
Ian
Gary Pimm
Hi Ian,
I presume you are familiar with Gary Pimm's PP amps with CCS and anode to anode feedback (but without the Baby Huey cross coupled feedback) and his experiments from cathode bias to ccs bias.
Just incase a couple of links.
http://db.audioasylum.com/cgi/m.mpl?forum=tubediy&n=78781&highlight=current+Gary+P&r=&session=
http://home.pacifier.com/~gpimm/47.htm
best wishes
tim
Hi Ian,
I presume you are familiar with Gary Pimm's PP amps with CCS and anode to anode feedback (but without the Baby Huey cross coupled feedback) and his experiments from cathode bias to ccs bias.
Just incase a couple of links.
http://db.audioasylum.com/cgi/m.mpl?forum=tubediy&n=78781&highlight=current+Gary+P&r=&session=
http://home.pacifier.com/~gpimm/47.htm
best wishes
tim
You have it exactly right. I am happy with class A because I don't need the watts and I think it sounds better than having each ties to earth. Current balance is the driving force because I use toroidals and any imbalance is unacceptable, so one CCS wouldn't do it for me.
I always thought having CCS in the cathode always forced class A operation.
Thats very interesting and suggests a little experiment may be in order on one of my amps. The question would be, would the move from differential be a backwards step. I suppose that even the use of a 40R resistor would preserve most of the differential behaviour because the cap to cap current flow would be much easier than the cathode node to earth. Unfortunately I have no harmonic test equipment so it would just have to be a by ear comparison.
Shoog
Opened up Huey to try the suggested mod, but thought before doing this I might have a look at the square wave response and see if there was a difference before/after. When I built the amp, I only did a sine wave test. It looked clean, so I hooked it up to my speakers and enjoyed!
Well, imagine my surprise when I see that this is the square wave output at around 1kHz:
Does not look that great, and much worse than my EL34 SE amp.
Any hints for trying to work out the problem? grid of the top input triode is clean, as you would expect, but every where else I probe I get the same picture. Something that may be of some help, when I had the scope hooked up to the grid of the bottom input triode, I get the same picture, but when I turn the power off on the amp, after a couple of seconds, the trace comes back to a clean square wave, then dies out as the amp completely loses power.
Tried bypassing the constant current source with a resistor, but get pretty much the same picture still.
Any hints on how to attack this would be great.
Thanks,
Chris
Well, imagine my surprise when I see that this is the square wave output at around 1kHz:
An externally hosted image should be here but it was not working when we last tested it.
Does not look that great, and much worse than my EL34 SE amp.
Any hints for trying to work out the problem? grid of the top input triode is clean, as you would expect, but every where else I probe I get the same picture. Something that may be of some help, when I had the scope hooked up to the grid of the bottom input triode, I get the same picture, but when I turn the power off on the amp, after a couple of seconds, the trace comes back to a clean square wave, then dies out as the amp completely loses power.
Tried bypassing the constant current source with a resistor, but get pretty much the same picture still.
Any hints on how to attack this would be great.
Thanks,
Chris
gingertube said:
Yes, this is my understanding too. At my operating point, my amp is class A until clipping.
gingertube said:
At one extreme (no common resistor), it's two separate CCS's and two separate bypass caps. At the other extreme (a high value common resistor), you are correct, it's the same as having a single unbypassed CCS - except for the balance issue and maybe the added sonic signature of the caps. But, I don't see it as a binary function. You can operate anywhere between those extremes.
Shoog said:
All you need is a computer with a duplex sound card, an 8R (or whatever) power resistor, some cables, plugs, resistors, a pot, and jacks and you have harmonic test equipment. I'd also recommend, for safety, some antiparallel diodes the across the input to the soundcard to limit the voltage to about 4V. Download the free version of RMAA (Rightmark Audio Analyser). You can watch a rough picture of the harmonic content while you adjust. For the actual output numbers, the card averages a number of data points. Signal to noise will be a function of the card quality, but any card will give you useful info.
Sheldon
BTW, RMAA will tell you if your card will work. Almost all recent cards are full duplex.
Shoog,
Did you try that experiment with various resistor value to find a minimum?
I remember reading about splitting the cathode resistor into two - non bypassed R & bypassed one in series. There was an explanation of how to split the R. Has anybody tried this? Does anybody know the write-up I'm talking about? Tried to search Tubecad but found nada
Did you try that experiment with various resistor value to find a minimum?
I remember reading about splitting the cathode resistor into two - non bypassed R & bypassed one in series. There was an explanation of how to split the R. Has anybody tried this? Does anybody know the write-up I'm talking about? Tried to search Tubecad but found nada
jkeny said:
For cathode degeneration to work, the unbypassed section of the total cathode resistance needs to be at least equal to 1/Gm, otherwise, why bother?
There is also an experimental method. Connect a pot to the cathode with a large DC isolating capacitor. Then adjust the pot until you get the measurements and/or sound you're looking for. Remove pot from circuit, and measure the resistance. Calculate the parallel resistance, and that value becomes the unbypassed section with the rest bypassed.
Gents,
Happy and surprised to see others working with my cathode arrangement as used in the PP-1C.
Th eoriginal idea was to tie the cathodes together for AC with big back to back caps, forcing a quasi differential operation, butwith each tube having it's own cathode R to (sort of) enforce some sort of DC balance. Never tried CCs's but they could properly enforce DC current balance, which I consider far more vital than do many other designers.
Worked OK but by adding the extra R to ground from the caps center point gave a major power increase because it could move out of pure ClassA1 on peaks, and better measurements.
I went to pure differential operation on our production dpa300B amp, because I find it sounds stunningly better that even this trick - although it takes quite some tricks to keep the DC balance optimised. Sort of like long term balancing a knife on it's point...
Someone asked my on AA DIY tubes, why the huge cap size - the reason being that it sounds better - far better bass quality, not bass quantity but definition/tonality.
Regards, Allen
Happy and surprised to see others working with my cathode arrangement as used in the PP-1C.
Th eoriginal idea was to tie the cathodes together for AC with big back to back caps, forcing a quasi differential operation, butwith each tube having it's own cathode R to (sort of) enforce some sort of DC balance. Never tried CCs's but they could properly enforce DC current balance, which I consider far more vital than do many other designers.
Worked OK but by adding the extra R to ground from the caps center point gave a major power increase because it could move out of pure ClassA1 on peaks, and better measurements.
I went to pure differential operation on our production dpa300B amp, because I find it sounds stunningly better that even this trick - although it takes quite some tricks to keep the DC balance optimised. Sort of like long term balancing a knife on it's point...
Someone asked my on AA DIY tubes, why the huge cap size - the reason being that it sounds better - far better bass quality, not bass quantity but definition/tonality.
Regards, Allen
Thanks Miles,
So is it called partial cathode degeneration? Have you tried it? Any findings?
Shoog,
Hopefully, the solar panels will repay your time many fold - although this country is pretty unreliable for sun! STill looking forward to hearing your Tabor clone when you get the time! I'm building A Beam Deflection Tube preamp integrated into the Baby Huey with preamp out options to drive a Zeus which I'm also building. So next meet should be interesting
Allen,
So full differential with CCS seems the way to go? I just happen to be putting together some Gary Pimms CCS's on veroboard - I think I'll give this a try then on the Baby Huey outputs!
So is it called partial cathode degeneration? Have you tried it? Any findings?
Shoog,
Hopefully, the solar panels will repay your time many fold - although this country is pretty unreliable for sun! STill looking forward to hearing your Tabor clone when you get the time! I'm building A Beam Deflection Tube preamp integrated into the Baby Huey with preamp out options to drive a Zeus which I'm also building. So next meet should be interesting
Allen,
So full differential with CCS seems the way to go? I just happen to be putting together some Gary Pimms CCS's on veroboard - I think I'll give this a try then on the Baby Huey outputs!
Allen Wright said:
This issue has had me intrigued, since Gingertube pointed it out and I played with it a little. I have some bits and pieces left over for making a little PP amp, and I was planning to try fully differential. I noted the reference on your site to the fully differential output stage, noting also the lack of detail. Obviously you have some tricks up your sleeve, that you may want to keep there. With my limited experience, I've been trying to see what I can come up with to control DC in each output tube, but allow AC current to swing from tube. Big caps between the cathodes would do it, but trying to come up with something a bit more elegant. Nothing so far.
Sheldon