OK. Got parts, looking for a plan. 
Here's what I have:
Chassis with power transformer and power supply (about 300V B+ or so), and four 6V6 output tubes (already installed). Was a PPP mono amp... I'm wanting to convert over to PP stereo.
Output transformers- from Hammond organ amps- part # A024158. As used in the AO-44 PP 6GW8 and AO-35 PP 6BQ5 power amps. About 8K ohms CT primary: 8 ohms secondary. Probably capable of 12-15 watts.
I've built several Magnavox clone amps... looking for something different. Especially as these output transformers were originally intended for 8 ohm speaker loads, while the Magnavox outputs were made for 4 ohm loads.
I would like, if possible, to make this amp where it could TOLERATE 4 ohm loads on the outputs... which implies, to me, that it needs to, somehow, manage to decrease the effective plate impedance of the 6V6 output tubes- but without ultralinear operation (no taps on these transformers).
To that end, I was looking, briefly, at the feedback mechanism of Baby Huey... from the output plates, to the plates of the phase splitter. However, I know that it's been stated that Baby Huey probably wouldn't have enough gain to drive 6V6s in that application... my proposed solution would be to add one additional 12AT7/6201 tube as an input amp, for both channels of the amp (one side of the 12AT7 as a grounded cathode input stage, feeding into the Baby Huey-style phase splitter with feedback).
Now, I know these output transformers are NOT ultralinear... so that's another difference I've gotta deal with. But, with the amount of gain I can get from TWO gain stages (the grounded cathode input AND the modified-LTP phase splitter), I think I should be able to get back the low output impedance, fairly well...
In essence, what I'd really like to do, is try a non-ultralinear Mullard-style circuit, with triode front-end gain stage, but with Baby Huey-style feedback from the outputs to the phase splitter... anyone got any WAGs on whether this sounds like a feasible idea? Or, given the parts I listed above, is there an easier plan that would work well (to get low output impedance, with decent overall gain)?
Regards,
Gordon.
Here's what I have:
Chassis with power transformer and power supply (about 300V B+ or so), and four 6V6 output tubes (already installed). Was a PPP mono amp... I'm wanting to convert over to PP stereo.
Output transformers- from Hammond organ amps- part # A024158. As used in the AO-44 PP 6GW8 and AO-35 PP 6BQ5 power amps. About 8K ohms CT primary: 8 ohms secondary. Probably capable of 12-15 watts.
I've built several Magnavox clone amps... looking for something different. Especially as these output transformers were originally intended for 8 ohm speaker loads, while the Magnavox outputs were made for 4 ohm loads.
I would like, if possible, to make this amp where it could TOLERATE 4 ohm loads on the outputs... which implies, to me, that it needs to, somehow, manage to decrease the effective plate impedance of the 6V6 output tubes- but without ultralinear operation (no taps on these transformers).
To that end, I was looking, briefly, at the feedback mechanism of Baby Huey... from the output plates, to the plates of the phase splitter. However, I know that it's been stated that Baby Huey probably wouldn't have enough gain to drive 6V6s in that application... my proposed solution would be to add one additional 12AT7/6201 tube as an input amp, for both channels of the amp (one side of the 12AT7 as a grounded cathode input stage, feeding into the Baby Huey-style phase splitter with feedback).
Now, I know these output transformers are NOT ultralinear... so that's another difference I've gotta deal with. But, with the amount of gain I can get from TWO gain stages (the grounded cathode input AND the modified-LTP phase splitter), I think I should be able to get back the low output impedance, fairly well...
In essence, what I'd really like to do, is try a non-ultralinear Mullard-style circuit, with triode front-end gain stage, but with Baby Huey-style feedback from the outputs to the phase splitter... anyone got any WAGs on whether this sounds like a feasible idea? Or, given the parts I listed above, is there an easier plan that would work well (to get low output impedance, with decent overall gain)?
Regards,
Gordon.
hey-Hey!!!,
I've come to question the 'lots of gain needed' philosophy.
There are a few means to deliver local FB. The Gary Pimm 47 amp is one, the RCA SP20( or single pair version http://mysite.du.edu/~etuttle/electron/elect35.htm ) work really well. With U-L taps there is E-Linear, though that does best with pentodes up front, or better still Cascode.
Short story IMO is that if you build with a low gain single stage up front, it is quite easy to modify with a higher gain element later if you've guessed badly wrong( not likely ).
cheers,
Douglas
I've come to question the 'lots of gain needed' philosophy.
There are a few means to deliver local FB. The Gary Pimm 47 amp is one, the RCA SP20( or single pair version http://mysite.du.edu/~etuttle/electron/elect35.htm ) work really well. With U-L taps there is E-Linear, though that does best with pentodes up front, or better still Cascode.
Short story IMO is that if you build with a low gain single stage up front, it is quite easy to modify with a higher gain element later if you've guessed badly wrong( not likely ).
cheers,
Douglas
Andrewbee said:
Had thought about that. MIght be a way to go.
I would wonder which one would wind up with the lowest output impedance, though... the HK design or the Mullard with output anode/phase splitter anode feedback...
Also, Doug... as I said, I don't have ultralinear capability,with the parts available. Hence, the search for alternative "impedance reducing" strategies...
Regards,
Gordon.
Bandersnatch said:
Looked at that RCA design... only problem, with those pentodes, it would require more "bottles" than I have room for.
Hence, the idea of modifying a triode-front-end Mullard... that could be done in stereo, with just three twin-triodes...
Though, the basic idea of the feedback scheme on the RCA might be a good option. I'd think going from plate-to-plate might be easier to implement, however... Bricktop over at AK certainly got that to work easily enough... and it seems to have the added benefit of helping stability and bandwidth on the top end, too...
Regards,
Gordon.
hey-Hey!!!,
You can also cross couple final's plates to opposite driver grids, similar to what was used in the Citation II. I've done this once, and I am going to rig my Mk.III's this way. Principle difference being that I have room for two tubes per channel in the monoblock amps. They'll use 6NS7 LTP running open loop, feeding the LTP with cross-coupled FB from the final's plates.
cheers,
Douglas
You can also cross couple final's plates to opposite driver grids, similar to what was used in the Citation II. I've done this once, and I am going to rig my Mk.III's this way. Principle difference being that I have room for two tubes per channel in the monoblock amps. They'll use 6NS7 LTP running open loop, feeding the LTP with cross-coupled FB from the final's plates.
cheers,
Douglas
Gordon,
With a "twist", "Baby Huey" may be just fine. You fear that the 12AX7 will not provide sufficient gain. Buffer the 'X7 sections with DC coupled ZVN0545A source followers. Now, the 'X7 is working into just its anode load resistor. An added benefit of buffering is the option to use "fixed" bias. The 6V6 data sheet gives 100 KOhms as the max. allowable grid leak resistance, when "fixed" bias is employed. You could go down to 22 KOhms and the FET will not "complain".
With a "twist", "Baby Huey" may be just fine. You fear that the 12AX7 will not provide sufficient gain. Buffer the 'X7 sections with DC coupled ZVN0545A source followers. Now, the 'X7 is working into just its anode load resistor. An added benefit of buffering is the option to use "fixed" bias. The 6V6 data sheet gives 100 KOhms as the max. allowable grid leak resistance, when "fixed" bias is employed. You could go down to 22 KOhms and the FET will not "complain".
The 6AU6 is almost the ideal driver for a Partial feedback amplifier. It can handle respectable current of at least 5mA with a gain of about 90x and a suitable high output impedance. Also its cheap and sounds good.
Any triode front end in a partial feedback amp imposes great limitations on the design as current is always the loser to impedance and gain.
Another option would be to use something like the ECL82 which would give you huge amounts of gain.
Shoog
Any triode front end in a partial feedback amp imposes great limitations on the design as current is always the loser to impedance and gain.
Another option would be to use something like the ECL82 which would give you huge amounts of gain.
Shoog
Well, I did some calculations...
Looks like with my original idea (12AT7 front end/12AU7 LTP), I can get, before feedback, a gain of about 32 from the first tube (no cathode bypass cap) and about 7 from the LTP.
Comparing that to the Magnavox AMP175 clone I built before (also with 6V6 output)... it had a SINGLE gain stage (12AX7) of about 50 or so (the phase splitter is a paraphase- NO gain).
So, that gives a total preamp gain (first stage times LTP) of about 224. If you divide 50 into that, I've got about 4.5 times the gain of the Magnavox design... which had PLENTY of gain to run from any preamp I hooked up to it!
That means, I have a gain factor of about 4.5, to throw at the internal feedback loop.. about 13.5 dB. Which leaves a gain there, after internal feedback, of about 1.5... not a bad place to be, since the first stage can swing plenty of voltage (probably 40-50 volts, where I only need 25 or so, MAX). Or, I can go down to 10dB feedback to the LTP and up the feedback on the global loop 3dB over what Magnavox used... whichever sounds better...
And, the nice thing, is that with the lower gain on the first stage (12AT7 without cathode cap opposed to 12AX7 with cathode cap), I've got better bandwidth than the Maggie designs too... might not need the "crowbar" compensation cap to ground that Magnavox used after the first stage...
Regards,
Gordon.
Looks like with my original idea (12AT7 front end/12AU7 LTP), I can get, before feedback, a gain of about 32 from the first tube (no cathode bypass cap) and about 7 from the LTP.
Comparing that to the Magnavox AMP175 clone I built before (also with 6V6 output)... it had a SINGLE gain stage (12AX7) of about 50 or so (the phase splitter is a paraphase- NO gain).
So, that gives a total preamp gain (first stage times LTP) of about 224. If you divide 50 into that, I've got about 4.5 times the gain of the Magnavox design... which had PLENTY of gain to run from any preamp I hooked up to it!
That means, I have a gain factor of about 4.5, to throw at the internal feedback loop.. about 13.5 dB. Which leaves a gain there, after internal feedback, of about 1.5... not a bad place to be, since the first stage can swing plenty of voltage (probably 40-50 volts, where I only need 25 or so, MAX). Or, I can go down to 10dB feedback to the LTP and up the feedback on the global loop 3dB over what Magnavox used... whichever sounds better...
And, the nice thing, is that with the lower gain on the first stage (12AT7 without cathode cap opposed to 12AX7 with cathode cap), I've got better bandwidth than the Maggie designs too... might not need the "crowbar" compensation cap to ground that Magnavox used after the first stage...
Regards,
Gordon.
With regards to current in the input stages- another limitation I'm working with, is the original power transformer.
I am sure I can get enough power to run the 12AT7/12AU7 stages (3ma on each section of the 12AT7, 3-4ma on each section of the 12AU7)... but I would hesitate putting any more load on the power transformer.
So, that given... I've seen plenty of designs that worked well with these tubes at these current levels... some at MUCH lower, in fact...
Regards,
Gordon.
I am sure I can get enough power to run the 12AT7/12AU7 stages (3ma on each section of the 12AT7, 3-4ma on each section of the 12AU7)... but I would hesitate putting any more load on the power transformer.
So, that given... I've seen plenty of designs that worked well with these tubes at these current levels... some at MUCH lower, in fact...
Regards,
Gordon.
Gordon,
One of the Baby Huey variants I did was a modification of a Chinese Music Angel PP EL34. It was very close to what you have described. Details are here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=78712&perpage=10&pagenumber=3
You can use that for a few "ideas" to be applied to a 6V6 variation.
I have been "itching" to try the Gary Pimm PP47 arrangement with 6AU6. It fits the "Theoretical" requirements.
The clue to the Baby Huey arrangement is that the diff amp needs to be low current (say less than 1mA per side and preferrably about 0.6mA per side) so as to not load the output stage excessively.
That diff amp also needs to be high rp. The shunt feedback voltage is being divided and the "lower arm" of the voltage divider is the load resistance presented by the next stage in parallel with the impedance looking down into the anode of the diffamp sides. That means the tube stages of the diff amp need to be high rp. That means a 12AX7 or a penetode (6AU6) or a cascode. It also means that you cannot simply connect fixed biased 6V6 with Rg1 value of 100K max. directly to the diff amp anodes but must impedance buffer them with mosfet source followers. The ZVN0545A that was recommended above is a great device for this (VERY low input capacitance and Crss).
The 6V6 Baby Huey variant I did is at post #54 here
http://www.diyaudio.com/forums/showthread.php?s=&threadid=72536&perpage=25&pagenumber=3
Good luck - keep us "posted".
Cheers,
Ian
Pretty Picture:
One of the Baby Huey variants I did was a modification of a Chinese Music Angel PP EL34. It was very close to what you have described. Details are here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=78712&perpage=10&pagenumber=3
You can use that for a few "ideas" to be applied to a 6V6 variation.
I have been "itching" to try the Gary Pimm PP47 arrangement with 6AU6. It fits the "Theoretical" requirements.
The clue to the Baby Huey arrangement is that the diff amp needs to be low current (say less than 1mA per side and preferrably about 0.6mA per side) so as to not load the output stage excessively.
That diff amp also needs to be high rp. The shunt feedback voltage is being divided and the "lower arm" of the voltage divider is the load resistance presented by the next stage in parallel with the impedance looking down into the anode of the diffamp sides. That means the tube stages of the diff amp need to be high rp. That means a 12AX7 or a penetode (6AU6) or a cascode. It also means that you cannot simply connect fixed biased 6V6 with Rg1 value of 100K max. directly to the diff amp anodes but must impedance buffer them with mosfet source followers. The ZVN0545A that was recommended above is a great device for this (VERY low input capacitance and Crss).
The 6V6 Baby Huey variant I did is at post #54 here
http://www.diyaudio.com/forums/showthread.php?s=&threadid=72536&perpage=25&pagenumber=3
Good luck - keep us "posted".
Cheers,
Ian
Pretty Picture:
Attachments
The 6AU6 does everything you need with only a single gain stage. Gary Pimms implementation has excellent bandwidth and sounds great.
I built a little headphone amp using the ECL82 and coupling caps on the same principles as the Tabor. The front end passes around 1mA and is equivalent to an ECC81 (I think), it works very well with good bandwidth.
In both cases I had to use step down transformers to limit the gain to reasonable levels. Also with the range of LTP valves you are thinking of, the bias point is usually less than -1V which in my experience invites overloading and distortion (remember in partial feedback the feedback is reducing the output voltage gain and not the grid drive to the front end. If you supply 2V to a -1volt biased front end then the valve is in overload). The 6AU6 is capable of sensible bias points of a few volts. Really its the only sensible choice of partial feedback driver.
I would be inclined to try all I could to build it as a two stage design and only add a further front end if absolutely forced.
Shoog
I built a little headphone amp using the ECL82 and coupling caps on the same principles as the Tabor. The front end passes around 1mA and is equivalent to an ECC81 (I think), it works very well with good bandwidth.
In both cases I had to use step down transformers to limit the gain to reasonable levels. Also with the range of LTP valves you are thinking of, the bias point is usually less than -1V which in my experience invites overloading and distortion (remember in partial feedback the feedback is reducing the output voltage gain and not the grid drive to the front end. If you supply 2V to a -1volt biased front end then the valve is in overload). The 6AU6 is capable of sensible bias points of a few volts. Really its the only sensible choice of partial feedback driver.
I would be inclined to try all I could to build it as a two stage design and only add a further front end if absolutely forced.
Shoog
As Shoog says, the partial-feedback topology a'la Baby Huey present a difficult load for the driver stage.
This is discussed in some detail in this thread, supported by some measurements, and loadline illustrations:
Partial Feedback EL84PP - Not quite as expected (yet)
SveinB
This is discussed in some detail in this thread, supported by some measurements, and loadline illustrations:
Partial Feedback EL84PP - Not quite as expected (yet)
SveinB
How about 6V6 output with self or CCS bias?
Hi Ian (or Yves),
I understand the limitation of driving fixed biased 6V6 output stage with the low current biased 12AX7. However, will the original Baby Huey circuit work for 6V6 if we keep the same CCS bias for the output stage and Rg1 at 470K?
I did some LTSpice simulation. I can see a bit drop in open loop (and closed loop) gain with 6V6 with either CCS bias or self bias. However, the close loop THD is similar to the EL84 output stage. I did not see significant distortion at the output of 12AX7. Did the simulation miss something?
Alternatively, is there any sonic difference between fixed bias and CCS biased output for 6V6 tubes? As Yves original design and your later version of Baby Huey all used fixed bias, I am just curious about the decision to use CCS for the original Baby Huey design.
Hi Ian (or Yves),
I understand the limitation of driving fixed biased 6V6 output stage with the low current biased 12AX7. However, will the original Baby Huey circuit work for 6V6 if we keep the same CCS bias for the output stage and Rg1 at 470K?
I did some LTSpice simulation. I can see a bit drop in open loop (and closed loop) gain with 6V6 with either CCS bias or self bias. However, the close loop THD is similar to the EL84 output stage. I did not see significant distortion at the output of 12AX7. Did the simulation miss something?
Alternatively, is there any sonic difference between fixed bias and CCS biased output for 6V6 tubes? As Yves original design and your later version of Baby Huey all used fixed bias, I am just curious about the decision to use CCS for the original Baby Huey design.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.