Interesting design.
I assume that the LM317’s ADJ pins are connected to ground; they’re missing the “dot” to show a connection. The interstage transformer ought to give good push-pull drive symmetry and little common-mode, but I do wonder how much the loading on one side by the 100K and 100pF feedback parts might unbalance things at very high frequencies. Have you tried loading the other secondary with a dummy 100K and 100pF to ground for balance? The upper secondary also sees a tiny bit of DC current (less than 5uA) from the voltage across the 470 ohm resistor in the first stage. These things might not matter much, but I don’t know.
Since you’ve presumably got good push-pull drive with no common-mode power supply noise on the input, option “E” should have less of an advantage over “H”. If you were using a concertina phase splitter, then “E” might be more helpful. If your output stage is truly Class A then it will have good inherent rejection of B+ noise, but the slight edge could go to “E” for that. When you try “E”, make sure that all four bypass caps (the two 100uF and the two 10uF caps) are all tied together at the bottom ends, and then that one node is connected via a large resistor to ground. Whether it will sound better than “H” is only knowable from listening. The shift in tone you mentioned with "E" could be due to any one of a lot of things. I don’t think you’re hearing the LM317s directly either way, since the differential mode is all that really matters here, and in that mode the caps are shorting together the two sides’ signals. The input provides virtually no common-mode to put the CCSs in play. B+ noise would play the CCSs in the “E” case, but the CCSs should only improve rejection in that case.
Can you also try connecting a smaller film cap from 6AS7 directly from cathode to cathode? Maybe something in the range of 0.1 to 1.0uF?
I assume that the LM317’s ADJ pins are connected to ground; they’re missing the “dot” to show a connection. The interstage transformer ought to give good push-pull drive symmetry and little common-mode, but I do wonder how much the loading on one side by the 100K and 100pF feedback parts might unbalance things at very high frequencies. Have you tried loading the other secondary with a dummy 100K and 100pF to ground for balance? The upper secondary also sees a tiny bit of DC current (less than 5uA) from the voltage across the 470 ohm resistor in the first stage. These things might not matter much, but I don’t know.
Since you’ve presumably got good push-pull drive with no common-mode power supply noise on the input, option “E” should have less of an advantage over “H”. If you were using a concertina phase splitter, then “E” might be more helpful. If your output stage is truly Class A then it will have good inherent rejection of B+ noise, but the slight edge could go to “E” for that. When you try “E”, make sure that all four bypass caps (the two 100uF and the two 10uF caps) are all tied together at the bottom ends, and then that one node is connected via a large resistor to ground. Whether it will sound better than “H” is only knowable from listening. The shift in tone you mentioned with "E" could be due to any one of a lot of things. I don’t think you’re hearing the LM317s directly either way, since the differential mode is all that really matters here, and in that mode the caps are shorting together the two sides’ signals. The input provides virtually no common-mode to put the CCSs in play. B+ noise would play the CCSs in the “E” case, but the CCSs should only improve rejection in that case.
Can you also try connecting a smaller film cap from 6AS7 directly from cathode to cathode? Maybe something in the range of 0.1 to 1.0uF?
Indeed they are.
I have modified the amp since that schematic and have dropped that loop altogether. This was made possible by driving the pentode harder.
Overall the amp is wisper quiet with sensitive speakers. PS noise rejection is not really an issue.
These caps are all hanging down from the valve base and tied together. The big cap is 1000uf not 100uf.
I didn't yet try the output stage with a film cap between cathodes, I wanted to hear the result quickly and adding the film cap is a little more tricky..
Its entirely possible that the ECL82 is producing the harshness, its my small experience that this valve tends to harshness. It could be that the extra clarity that pushing the output stage into true differential mode is simply showing up the shortcomings of the driver.
planet10 said:
While most of the recent posts in this thread are over my head technically, and seem to be related to DIY / reverse engineering the design, I can attest to what Dave mentioned above.
It's been almost 4 years since I heard the original pair of these amps ( hard to believe), driving a pair of Terry Cain's Abby's, and nobody's acoustic memory is that good, but I did spend a lot of time the immediate weeks following the show at a local dealer with more than a few higher priced tubed systems on demo.
I'm sure there are several hundred 300B designs of varying vintage, complexity and pricing that I'll never get a chance to audition, but Allan's is the only one I've heard to date that I think I could live with.
Hi Brian,
Following on from my thought about the driver been shown up as been inadequate, I decided to try removing the plate to plate feedback. With such low current going through the triode it was always going to be marginal. I took this out and the whole sound cleaned up considerably. So I though I would try plan "E" again, with the additional film cap between cathodes.
The Result is amazing. I though this amp sounded good originally, but now it is both smooth and detailed. Seems that a differential output pair is the way to go - along with an overall Zero feedback amp.
Marvellous !!
Thanks Brian for your invaluable input and thanks to both Allen and Lynn for the original inspiration.
Shoog
Following on from my thought about the driver been shown up as been inadequate, I decided to try removing the plate to plate feedback. With such low current going through the triode it was always going to be marginal. I took this out and the whole sound cleaned up considerably. So I though I would try plan "E" again, with the additional film cap between cathodes.
The Result is amazing. I though this amp sounded good originally, but now it is both smooth and detailed. Seems that a differential output pair is the way to go - along with an overall Zero feedback amp.
Marvellous !!
Thanks Brian for your invaluable input and thanks to both Allen and Lynn for the original inspiration.
Shoog
Hi Shoog,
Looks like a deceptively simple schematic to me although it contains lots of innovative tricks & tweaks (CCS on input & output tubes, LED biasing, toroid trafos) It also appeals to my frugophile nature so I'm hooked. From what you say it sounds like an excellent sonic result - I'm keen to try it out & about ready in my understanding of tube circuits to have a go.
I have built the Baby Huey circuit, find it much better than the Rogers Cadet circuit it replaced (using the same ECL86 tubes). Whether this is as a result of the diffferential pair phase splitter or the shunt feedback from the output tubes, I don't know. Anyway I am using it to try out various CCS circuits on input & output. I wondered if the differential pair drivers (ECL86 triode sections) would have enough drive for the interstage trafo? Or could one of these ECL86 be used as replacement for ECL82(with circuit value mods)?
I also have two toroid transformers which may work as interstage trafos but they're 160VA 25+25V:230 - any good? As I will have to buy output tubes & trafos & will be driving Jordan JX92S which I think is 6 ohm impedance (not given on the Jordan site) would the 6V dual secondary trafos suffice - what VA rating do you suggest?
You seem to have gone zero global feedback in your latest mods - is this possible as a result of the 6AS7 tubes and are these the tubes you would now recommend?
Sorry for all the questions but that's what us noobs are meant to do. Thanks for any help you provide
John
Looks like a deceptively simple schematic to me although it contains lots of innovative tricks & tweaks (CCS on input & output tubes, LED biasing, toroid trafos) It also appeals to my frugophile nature so I'm hooked. From what you say it sounds like an excellent sonic result - I'm keen to try it out & about ready in my understanding of tube circuits to have a go.
I have built the Baby Huey circuit, find it much better than the Rogers Cadet circuit it replaced (using the same ECL86 tubes). Whether this is as a result of the diffferential pair phase splitter or the shunt feedback from the output tubes, I don't know. Anyway I am using it to try out various CCS circuits on input & output. I wondered if the differential pair drivers (ECL86 triode sections) would have enough drive for the interstage trafo? Or could one of these ECL86 be used as replacement for ECL82(with circuit value mods)?
I also have two toroid transformers which may work as interstage trafos but they're 160VA 25+25V:230 - any good? As I will have to buy output tubes & trafos & will be driving Jordan JX92S which I think is 6 ohm impedance (not given on the Jordan site) would the 6V dual secondary trafos suffice - what VA rating do you suggest?
You seem to have gone zero global feedback in your latest mods - is this possible as a result of the 6AS7 tubes and are these the tubes you would now recommend?
Sorry for all the questions but that's what us noobs are meant to do. Thanks for any help you provide
John
Hi again Shoog,
Some more questions - am I right that the input stage a standard concertina phase splitter? (with the pentode triode strapped and CCS on the plate & LED biasing). If so I could probably use the ECL86 concertina circuit from the Rogers attached.
What are you using for this CCS circuit?
Any special considerations for PS? You have 170V on drivers & 125V on output tubes - are these from the same PS trafo? Do you use back to back toroids in the PS?
Thanks
John
Edit: Just re-read thread & noticed that you got rid of pentode CCS. Would it be possible to post the latest schematic?
Some more questions - am I right that the input stage a standard concertina phase splitter? (with the pentode triode strapped and CCS on the plate & LED biasing). If so I could probably use the ECL86 concertina circuit from the Rogers attached.
What are you using for this CCS circuit?
Any special considerations for PS? You have 170V on drivers & 125V on output tubes - are these from the same PS trafo? Do you use back to back toroids in the PS?
Thanks
John
Edit: Just re-read thread & noticed that you got rid of pentode CCS. Would it be possible to post the latest schematic?
Attachments
Hi Jkeny,
You could easily use the ECL86 in substitution for the ECL82.
The phase splitting is achieved with the interstage toroid. The toroid you have would work but only as a step down. This has advantages as its twice as easy to drive. However the down side is that you will need twice the +B for the drivers. This is not such a bad thing as a seperate driver power supply is a good idea in this setup. The triode section would be totally inadequate for driving the interstage transformers, I am pushing 40mA through the triode strapped pentode section. With a step down situation you could probably get away with 20-25mA, but this is still a job for the pentode.
Having removed all the feedback my amp has masses to much gain, so if I was doing it again I would consider a single driver tube, capable of delivering 40mA (or 20mA in your case), and with a mu of at least 20. possibly a ECC99 or 12B4.
I have, and would, drop the LED bias as at the currents I am using its risky. I also wanted a bit of cathode feeback and a slight softening of the sound.
I haven't dropped the pentode CCS, but have increased the current. The pentode stage CCS I use is the diyaudio standard cascaded one which a simple search of the site will pull up.
I use a single 60V+60V transformer and use silicone rectification. I take the driver stage power off after the first 8uf cap. I use a light ballast for driver smoothing and a MOV for output stage smoothing. I had the transformer left over from a PA amp, and the nearest commercial equivalent would be 55V
I recone that you could buy all of the iron for this for the price of a regular PP transformer.
We aught to try to get together some time, there aren't many of us audio nuts in Ireland - to busy looking at cows in my neck of the woods.
Shoog
You could easily use the ECL86 in substitution for the ECL82.
The phase splitting is achieved with the interstage toroid. The toroid you have would work but only as a step down. This has advantages as its twice as easy to drive. However the down side is that you will need twice the +B for the drivers. This is not such a bad thing as a seperate driver power supply is a good idea in this setup. The triode section would be totally inadequate for driving the interstage transformers, I am pushing 40mA through the triode strapped pentode section. With a step down situation you could probably get away with 20-25mA, but this is still a job for the pentode.
Having removed all the feedback my amp has masses to much gain, so if I was doing it again I would consider a single driver tube, capable of delivering 40mA (or 20mA in your case), and with a mu of at least 20. possibly a ECC99 or 12B4.
I have, and would, drop the LED bias as at the currents I am using its risky. I also wanted a bit of cathode feeback and a slight softening of the sound.
I haven't dropped the pentode CCS, but have increased the current. The pentode stage CCS I use is the diyaudio standard cascaded one which a simple search of the site will pull up.
I use a single 60V+60V transformer and use silicone rectification. I take the driver stage power off after the first 8uf cap. I use a light ballast for driver smoothing and a MOV for output stage smoothing. I had the transformer left over from a PA amp, and the nearest commercial equivalent would be 55V
I recone that you could buy all of the iron for this for the price of a regular PP transformer.
We aught to try to get together some time, there aren't many of us audio nuts in Ireland - to busy looking at cows in my neck of the woods.
Shoog
Hi Shoog,
I think I got you - let me repeat & see if I got it right:
Yes I could use ECL86 tube as driver but using triode strapped pentode section to drive the interstage toroid with about 20-25mA. I would also need about 300-350V to drive this tube. Cascoded CCS on the pentode plate with bypassed R for cathode bias (or experiment at will here).
I have a couple of salvaged PS trafos which might do the job - a 70+70V from multichannel amp, the Rogers PS trafo which is delivering 250+V at 200mA I believe.
Things I will need to source:
Toroid OPTs - 110+110V:5+5V would probably do for my Jordans but at what VA rating?
Four output tubes - 6AS7 are still the best you reckon? No matching required?
Good idea about getting together - I only see WoodturnerFran, Yourself and myself on this site although I do seem to remember some other posters from Ireland from time to time - just can't recall their names.
I was thinking, if anybody was interested, in building the LDR based lightspeed attenuator mentioned in the solid state area http://www.diyaudio.com/forums/showthread.php?s=&threadid=80194&highlight=lightspeed
and as it requires two matched pairs of Silonex LDRs NSL32SR3S, I was wondering if anybody was interested in joining in to a local group buy to purchase a volume of these to maximise the chances of matched pairs? The unsorted LDRs are 2.69 each from Farnell - I believe. The sorted ones NSL-32SR2S are £1.60 from RS UK.
From my reading I believe, even with the sorted ones, you need to buy about 10 to guarantee 2 matched pairs. I reckon the scale of numbers would work for us if three or more of us entered into a group buy. Anyway, let me know what you think or PM me if you like.
John
I think I got you - let me repeat & see if I got it right:
Yes I could use ECL86 tube as driver but using triode strapped pentode section to drive the interstage toroid with about 20-25mA. I would also need about 300-350V to drive this tube. Cascoded CCS on the pentode plate with bypassed R for cathode bias (or experiment at will here).
I have a couple of salvaged PS trafos which might do the job - a 70+70V from multichannel amp, the Rogers PS trafo which is delivering 250+V at 200mA I believe.
Things I will need to source:
Toroid OPTs - 110+110V:5+5V would probably do for my Jordans but at what VA rating?
Four output tubes - 6AS7 are still the best you reckon? No matching required?
Good idea about getting together - I only see WoodturnerFran, Yourself and myself on this site although I do seem to remember some other posters from Ireland from time to time - just can't recall their names.
I was thinking, if anybody was interested, in building the LDR based lightspeed attenuator mentioned in the solid state area http://www.diyaudio.com/forums/showthread.php?s=&threadid=80194&highlight=lightspeed
and as it requires two matched pairs of Silonex LDRs NSL32SR3S, I was wondering if anybody was interested in joining in to a local group buy to purchase a volume of these to maximise the chances of matched pairs? The unsorted LDRs are 2.69 each from Farnell - I believe. The sorted ones NSL-32SR2S are £1.60 from RS UK.
From my reading I believe, even with the sorted ones, you need to buy about 10 to guarantee 2 matched pairs. I reckon the scale of numbers would work for us if three or more of us entered into a group buy. Anyway, let me know what you think or PM me if you like.
John
John,
You weren’t asking for comments on LDRs, but since I’m already involved in this thread, I wanted to bring up something about the LDRs. I didn’t have the patience to read all 60+ pages of the thread on LDRs, but I hope someone asked about the voltage coefficient of resistance (VCR) of the photo detectors. I’m guessing that the parts you cite use CdS photo resistors. I can’t find the data right now, but I seem to recall that the VCR for CdS was somewhere around 0.2%/v or 2000ppm/v. This means that even with a steady light source, the resistance of CdS is somewhat non-linear. Compare this to a metal film resistor where VCR is only measurable at very high resistance values and even then it’s less than 0.1ppm/v. This means, if I have my CdS VCR number correct, that the resistance will change by 20,000 times more with a CdS resistor element than with a metal film resistor in response to a signal voltage. In absolute terms, if you put a 2 volt (5.66 peak-to-peak) sine wave across your CdS volume control, then the resistance will change by about 1.1% from one peak to the opposite peak. This will generate a little bit of distortion, but perhaps not enough to get fired up about. I just think that a good set of contacts switching metal film resistors in a stepped attenuator will make less distortion.
My recollection of the VCR for CdS may be wrong, and I hope someone will correct me if it is.
You weren’t asking for comments on LDRs, but since I’m already involved in this thread, I wanted to bring up something about the LDRs. I didn’t have the patience to read all 60+ pages of the thread on LDRs, but I hope someone asked about the voltage coefficient of resistance (VCR) of the photo detectors. I’m guessing that the parts you cite use CdS photo resistors. I can’t find the data right now, but I seem to recall that the VCR for CdS was somewhere around 0.2%/v or 2000ppm/v. This means that even with a steady light source, the resistance of CdS is somewhat non-linear. Compare this to a metal film resistor where VCR is only measurable at very high resistance values and even then it’s less than 0.1ppm/v. This means, if I have my CdS VCR number correct, that the resistance will change by 20,000 times more with a CdS resistor element than with a metal film resistor in response to a signal voltage. In absolute terms, if you put a 2 volt (5.66 peak-to-peak) sine wave across your CdS volume control, then the resistance will change by about 1.1% from one peak to the opposite peak. This will generate a little bit of distortion, but perhaps not enough to get fired up about. I just think that a good set of contacts switching metal film resistors in a stepped attenuator will make less distortion.
My recollection of the VCR for CdS may be wrong, and I hope someone will correct me if it is.
Hi Brian,
Maybe so - I don't want to hijack this thread as it is my main interest - but if you read the last page of the lightspeed thread you will find glowing reviews of it (by one who replaced a stepped attenuator) & Silonex have graphs showing distrortion on their site.
Bottom line - distortion seems to be a non issue & it's a high end (sonically) vol control & very much cheaper than stepped attenuator.
John
Maybe so - I don't want to hijack this thread as it is my main interest - but if you read the last page of the lightspeed thread you will find glowing reviews of it (by one who replaced a stepped attenuator) & Silonex have graphs showing distrortion on their site.
Bottom line - distortion seems to be a non issue & it's a high end (sonically) vol control & very much cheaper than stepped attenuator.
John
My glowing praise may have been a bit premature. Its now distorting at loud levels. I believe its the ECL82 but don't know which element is overloading. I would have to say that there are definately better driver elements availble and I am seriously considering pulling them and substituting some PL84's with 5687's in front. The same can be said of the ECL86 - the low current in the triode section places sever limitations on how it can be used.
Jkeny - you only need two 6AS7's as they are dual power triodes. Also your toroids which you were thinking of using as interstagers will produce a 4x step down which would probably make them unsuitable for this application. At first glance I thought would produce a 2x, but then I remembered that I am using mine with the secondaries in series and as the primary.
Drop us a mail and I will send you my phone number.
Shoog
Jkeny - you only need two 6AS7's as they are dual power triodes. Also your toroids which you were thinking of using as interstagers will produce a 4x step down which would probably make them unsuitable for this application. At first glance I thought would produce a 2x, but then I remembered that I am using mine with the secondaries in series and as the primary.
Drop us a mail and I will send you my phone number.
Shoog
jkeny said:
John,
I hadn't seen that link to the Silonex distortion curves, and I have to say that distortion performance is worse than I imagined. I’ve attached the Silonex distortion curve (in green) for the series/shunt LDR configuration mentioned in the thread. It can be as high as 0.025% at low attenuation settings, and it can exceed several percent THD at high attenuation settings. A shunt-only LDR has better distortion performance, but it’s still worse than a simple potentiometer. Even Silonex admits that distortion performance is “medium-to-good” compared to “very good” for potentiometers. They seem to be marketing other features besides sound quality, such as control signal isolation. I know that this simple distortion test alone doesn't guarantee a bad sound since the ear is pretty forgiving of simple harmonic distortion mechanisms. But with good switch contacts (that’s really important) and steel-free metal-film resistors, the distortion of a stepped attenuator would be all but unmeasurable. Granted, a stepped attenuator can be expensive, have coarse steps and it’s hard to motor-drive for remote control. But why introduce this level of distortion (probably exceeding the amp itself), when even an Alps Blue or Black is so much better? I didn’t set out to “rain on your parade”, but I wouldn’t be too swayed by a few favorable listening reports (several of which are from a fellow who makes LDRs attenuators “as a source of income”). There are so many other variables.
Sorry for the momentary hijack.
Attachments
Turns out the distorting element is the triode section. I took it out of the circuit. Not quite enough gain without it (but almost), much cleaner sound.
Turns out the triode section is quite blousey and the pentode section is quite lean - I suppose they balance each other out somewhat. I will look around for a driver pentode with a gain of about 15 so that I can have my two tube amp. Any suggestions of a good tube capable of 40mA and 15 gain.
Shoog
Turns out the triode section is quite blousey and the pentode section is quite lean - I suppose they balance each other out somewhat. I will look around for a driver pentode with a gain of about 15 so that I can have my two tube amp. Any suggestions of a good tube capable of 40mA and 15 gain.
Shoog
Sorry Brian,
Hope my posts don't seem rude but my internal dialogue is very like your posts and I struggle against this internal voice in order to actually move ahead. This voice tells me that there are a lot of valid reasons for not using LDRs, like distortion, the fact that it is never fully off and maybe more but it seems to hold promise.
Thanks Brian - I appreciate your valuable contribution to this thread.
John
Hope my posts don't seem rude but my internal dialogue is very like your posts and I struggle against this internal voice in order to actually move ahead. This voice tells me that there are a lot of valid reasons for not using LDRs, like distortion, the fact that it is never fully off and maybe more but it seems to hold promise.
Thanks Brian - I appreciate your valuable contribution to this thread.
John
Hi Brian Beck,
Something came up on another thread concerning your idea for a dual CCS differential output stage.
Maybe you can have a look at this schematic
http://www.triodeguy.com/6as7_pp.htm
He is only using a 10uf coupling cap between the cathodes. Seems woefully inadequate to me - but maybe i'am missing something.
In my amp I have effectively gone from 1000uf bypassing to 500uf coupling - and I think I may have lost some of my bass response.
Care to express an opinion on selecting this caps value.
Shoog
Something came up on another thread concerning your idea for a dual CCS differential output stage.
Maybe you can have a look at this schematic
http://www.triodeguy.com/6as7_pp.htm
He is only using a 10uf coupling cap between the cathodes. Seems woefully inadequate to me - but maybe i'am missing something.
In my amp I have effectively gone from 1000uf bypassing to 500uf coupling - and I think I may have lost some of my bass response.
Care to express an opinion on selecting this caps value.
Shoog
Shoog,
Thanks for the link. I now recall seeing Sear's article in AudioXpress, but didn't realize that he'd used the same dual CCS approach that we've been discussing. If you open the link on his website to his AudioXpress article, you'll see that he discussed the smallish cathode-coupling caps. Apparently, he is creating a resonance between these caps and the OPT's primary inductance as transferred to the cathodes. There is enough resistance, he claims, to adequately damp the resonance. He used a toroidal power transformer as an OPT, and perhaps the primary inductance was not very high. This situation makes me vaguely squeamish. There is, technically, always a resonant interaction between the primary inductance and the cathode bypass caps in every tube amp using cathode bypass. But, I would prefer to see the primary inductance large enough to create a lower-frequency roll-off without depending upon resonant peaking to counteract a too-high roll-off. This requires more primary inductance, and bigger caps.
Thanks for the link. I now recall seeing Sear's article in AudioXpress, but didn't realize that he'd used the same dual CCS approach that we've been discussing. If you open the link on his website to his AudioXpress article, you'll see that he discussed the smallish cathode-coupling caps. Apparently, he is creating a resonance between these caps and the OPT's primary inductance as transferred to the cathodes. There is enough resistance, he claims, to adequately damp the resonance. He used a toroidal power transformer as an OPT, and perhaps the primary inductance was not very high. This situation makes me vaguely squeamish. There is, technically, always a resonant interaction between the primary inductance and the cathode bypass caps in every tube amp using cathode bypass. But, I would prefer to see the primary inductance large enough to create a lower-frequency roll-off without depending upon resonant peaking to counteract a too-high roll-off. This requires more primary inductance, and bigger caps.
My experience with using power toroids as output is that they can sustain good response down to 10hz which suggests that the inductance is adequate for this purpose.
I can't imagine that the bypassing requirements aren't similar to a normal cathode bypass.
Shoog
I can't imagine that the bypassing requirements aren't similar to a normal cathode bypass.
Shoog
Yes, I don't what he's got going on for sure. I agree that the bypass requirements should be the same as for "normal" bypassing.