What next?
I’ve been listening to the Baby Huey in the last few weeks. All in all, I am extremely pleased with how it sounds. This has been a great first DIY experience for me. I am also beginning to develop a feel for how to adjust the amp to my taste. There are a few minor things I’d like to try – like an auto bias circuit that maintains all the bias currents (and allows me to switch between bias points depending on the music and on my mood), a PSU that can deliver enough power, maybe even a chassis, etc.
While I do that, I am ready to think about my next amp. One thing I feel the need for is more power. The Baby Huey is loud enough as is. But I think it sounds a bit strained at times driving my B&W 804 in the living room (15’x30’ including an adjacent open kitchen).
I have been looking around in the various forums. It seems that a pair of KT88 may be a logical next step? A pair of EL34 may be good too. I just thought KT88 may give me a bit more head room to experiment with. I don’t know how close parallel pairs need to match to work well. I have a feeling that if I don’t have to, I may be better off sticking to a single pair. Otherwise, I’m not afraid of complexity. I’m not biased against solid states outside of the signal path. I’m not even biased against global feedback. I just want the best circuit.
I’d like to ask for recommendations for a higher power push-pull design worthy as a follow on project to the Baby Huey (at least 20W; I don’t think I need more than 50W).
I’ve been listening to the Baby Huey in the last few weeks. All in all, I am extremely pleased with how it sounds. This has been a great first DIY experience for me. I am also beginning to develop a feel for how to adjust the amp to my taste. There are a few minor things I’d like to try – like an auto bias circuit that maintains all the bias currents (and allows me to switch between bias points depending on the music and on my mood), a PSU that can deliver enough power, maybe even a chassis, etc.
While I do that, I am ready to think about my next amp. One thing I feel the need for is more power. The Baby Huey is loud enough as is. But I think it sounds a bit strained at times driving my B&W 804 in the living room (15’x30’ including an adjacent open kitchen).
I have been looking around in the various forums. It seems that a pair of KT88 may be a logical next step? A pair of EL34 may be good too. I just thought KT88 may give me a bit more head room to experiment with. I don’t know how close parallel pairs need to match to work well. I have a feeling that if I don’t have to, I may be better off sticking to a single pair. Otherwise, I’m not afraid of complexity. I’m not biased against solid states outside of the signal path. I’m not even biased against global feedback. I just want the best circuit.
I’d like to ask for recommendations for a higher power push-pull design worthy as a follow on project to the Baby Huey (at least 20W; I don’t think I need more than 50W).
Consider extrapolating "Baby Huey". The drive requirements of the 7591 are only a tad greater than that of the EL84. Avery Fisher used a common cathode 'X7 section feeding a 2nd 'X7 section "concertina" phase splitter to drive PP 7591s. You'll need a B+ rail upwards of 400 V., but Gingertube's small signal circuitry rates to be just fine. Edcor's 6.6 KOhm/CT O/P trafos should get the job done. Look here.
I have a question... When PCB's are ready, I will build this DAC It is a balanced design, and I would like to keep the chain balanced up to the power amp. For the power amp I am building the Baby Huey (I already have one channel prototyped). I know this circuit (diff pair) allows the use of a balanced input (as Allen Wright has done on the PP-1C http://www.vacuumstate.com/schematics/pp-1c_s.gif , but now there is no place (grid of second triode) to apply the feedback from the OPT secondary. Allen Wright solved the problem just skipping feedback...can this also be done for the baby huey? (gain will be very large using the ECC83...maybe I should use a lower gain valve, as a 6SN7...)
ErikdeBest,
You can certainly run the "Baby Huey" with zero global feedback and hence free up that -ve input for differential input. After much listening and adjusting I'm running the original prototype with zero global feedback and 33K for the cross coupled shunt feedback set resistor.
Cheers,
Ian
You can certainly run the "Baby Huey" with zero global feedback and hence free up that -ve input for differential input. After much listening and adjusting I'm running the original prototype with zero global feedback and 33K for the cross coupled shunt feedback set resistor.
Cheers,
Ian
Since I'm going to build this design for a guy in my street...I was wondering about the bias for the output tubes...how are these different from say ONE ccs?...(since there are two bias circuits and no potentiometer in them how do they balance the two output tubes?)
Can I just substitute them with one DN2540 mosfet as a CCS? Or will I end up with a very different sounding amp?
Can I just substitute them with one DN2540 mosfet as a CCS? Or will I end up with a very different sounding amp?
Bas Horneman said:
Hi Bas,
Dunno, but the original design used fixed bias
http://www.dissident-audio.com/PP_ECL86/Page.html
I love having the cathodes of power tubes firmly grounded
Yves.
Output Tube CCS's
Appologies for getting to this late. I've spent all day running network cables and have only just got internet access back up.
Bas,
A single CCS connected to commoned cathodes is no good - that would enforce Class A only operation.
Separate CCS as per posted schematic with electrolytic bypass allows Class AB. Currents will be balanced to within about 1mA providing you use 1% metal films for those 16R current sense resistors in the "Ring of Two" CCSs on 6the EL84 cathodes. My prototype is currently running 2 x 470uF/50V Blackgate + 1uF PPS as the bypass on each CCS.
You can run fixed bias instead of using the current sources in each EL84 cathode. The 470K G1 to AC ground might have to be reduced (check EL84 data sheeet for max G1 resisitance in fixed bias). If this looks like loading the front end too much then you can add MOSFET Source Followers as I did in the 6V6 version.
An alternative if you have suitable power supply rail(s) is to run OpAmp bias servo which will allow using the cathode bias Rg1 value.
Cheers,
Ian
Appologies for getting to this late. I've spent all day running network cables and have only just got internet access back up.
Bas,
A single CCS connected to commoned cathodes is no good - that would enforce Class A only operation.
Separate CCS as per posted schematic with electrolytic bypass allows Class AB. Currents will be balanced to within about 1mA providing you use 1% metal films for those 16R current sense resistors in the "Ring of Two" CCSs on 6the EL84 cathodes. My prototype is currently running 2 x 470uF/50V Blackgate + 1uF PPS as the bypass on each CCS.
You can run fixed bias instead of using the current sources in each EL84 cathode. The 470K G1 to AC ground might have to be reduced (check EL84 data sheeet for max G1 resisitance in fixed bias). If this looks like loading the front end too much then you can add MOSFET Source Followers as I did in the 6V6 version.
An alternative if you have suitable power supply rail(s) is to run OpAmp bias servo which will allow using the cathode bias Rg1 value.
Cheers,
Ian
The guy I was going to build the amplifier has changed his mind and now wants an EL34 amplifier. At first my mind started wondering about what design to use. But after a while I thought to myself...why not take the Baby Huey design or rather the driver stage.
Would the essentially ecc83 tube be able to drive the EL34's well enough to still sound great and not need an additional gain stage
anywhere in the chain. (Source = CD player)
Would the essentially ecc83 tube be able to drive the EL34's well enough to still sound great and not need an additional gain stage
anywhere in the chain. (Source = CD player)
Bas Horneman said:
I'd say that the ECC83 (quite similar to a 12AX7A) will probably have enough gain margin. The drive capability, however, looks marginal. The current sourcing is just barely adequate in meeting the "Rule of Five" at a frequency of 30KHz. You just might be running into slew rate limiting at the high end.
It will never do for pushing the EL34s into Class AB2, so you will probably be seeing problems related to transient distortion. I would suggest looking into gingertube's 6V6 modification that includes MOSFET drivers.
Bas Horneman said:
Comes from solid state design practice. Once you've determined the base current of the driven stage, you make the collector current of the driver at least five times greater. With the afrementioned EL34:
Cgp= 0.5pF
Ci= 10.8pF
Figure a voltage gain of 19, and you have a combined Ci + Cmiller + Cstray of:
10.8 + (20)0.5 + 30= 50.8pF. At 30KHz that gives Xc= 104.4K. Figure about 15Vp of grid drive, and you have: I= 0.14mA. So the driver plate current should be at least: (5)0.14= 0.72mA. 0.72mA of plate current is, more or less, typical of a 12AX7A. So you're right on the edge. Maybe the slew rate will be adequate, and maybe it won't. Depends on how your stray capacitance goes. You can get that down to about 30pF, but not much lower. It could be higher depending on how bad your layout and construction skills are, or what sorts of hardware you use.
Of course, this presumes that a transient that forces a positive grid voltage never occurs. Now that's a big assume. With a 12AX7A, it will roll over and die with just about any grid current on the final(s). If the NFB collaspes, that, in turn, will give some really nasty clipping behaviour.
Bas,
I have a Chinese "Music Angel" PP EL34 which I have rebuilt to the "Baby Huey" circuit. Appologies for the pdf but I had trouble getting file size of a jpg down.
With EL34s in Pentode Mode I needed 5dB of global feedback.
With EL34s in Triode Mode I ditched the global feedback and dropped the 39K local feedback set resistor to 22K.
Note that this amp has an additional gain stage (the 6N1P) on the front. That was already on the Music Angel so I just kept it BUT I think you could get away without it.
Cheers,
Ian
I have a Chinese "Music Angel" PP EL34 which I have rebuilt to the "Baby Huey" circuit. Appologies for the pdf but I had trouble getting file size of a jpg down.
With EL34s in Pentode Mode I needed 5dB of global feedback.
With EL34s in Triode Mode I ditched the global feedback and dropped the 39K local feedback set resistor to 22K.
Note that this amp has an additional gain stage (the 6N1P) on the front. That was already on the Music Angel so I just kept it BUT I think you could get away without it.
Cheers,
Ian
Hi Ian and all.
Very interesting subject. I’ll try to build the Baby Huey. And, if I’ll like it, even the el34 version.
Now, I’d like to share some thoughts and doubts regarding the front end and the mosfet followers+fixed bias option.
As you said, speaking about the modified “music angel”, the amount of shunt feedback you can use with the 6sl7 front end is limited due to the increasing signal swing needed to drive the el34, even in pentode mode.
Question: if you’ll rebuilt it from scratch and not modifying an existing one, would you take into consideration a cascode differential front end? Higher Rp, higher gain, better linearity even swinging large signal. With a B+ of more than 400V and a cascade of 6922 this should be possible.
Another option, as you said in the 3d, should be to use a fet and an hybrid cascode…
A crazy idea for the Baby huey: an ecc86 cascade in the front end direct coupled to the power stage… lifting the 84s cathode potential of 24/30V this probably could works. As I said…only a crazy idea ;-)
However, which are the problems the front end has to face in this kind of circuit (shunt feedback)?
Now the mosfet follower. There is a modify in you approach that I find logical, indeed so logical that I’ve found it suspected…maybe I can’t figured out exactly what happen in this topology!
However, this is the issue: If we use the follower in conjunction with a negative supply for the bias, why don’t we use a differential supply to feed the mos and direct joined the follower to the power stage? We can leave the RC coupling between the voltage amp and the follower … its high input impedance and relatively low input capacitance shouldn’t be a problem for the I/V stage.
Driving an output stage that works in class A/B, we should achieve an improvement in terms of blocking distortion and recovery time. Another “extra” is the option for a lower voltage in the power supply (using a low voltage mos, with, probably, lower input capacitance) and higher bias current with the same or even lower heat dissipation.
I’m missing something? What do you thing?
Thanks
mark
Very interesting subject. I’ll try to build the Baby Huey. And, if I’ll like it, even the el34 version.
Now, I’d like to share some thoughts and doubts regarding the front end and the mosfet followers+fixed bias option.
As you said, speaking about the modified “music angel”, the amount of shunt feedback you can use with the 6sl7 front end is limited due to the increasing signal swing needed to drive the el34, even in pentode mode.
Question: if you’ll rebuilt it from scratch and not modifying an existing one, would you take into consideration a cascode differential front end? Higher Rp, higher gain, better linearity even swinging large signal. With a B+ of more than 400V and a cascade of 6922 this should be possible.
Another option, as you said in the 3d, should be to use a fet and an hybrid cascode…
A crazy idea for the Baby huey: an ecc86 cascade in the front end direct coupled to the power stage… lifting the 84s cathode potential of 24/30V this probably could works. As I said…only a crazy idea ;-)
However, which are the problems the front end has to face in this kind of circuit (shunt feedback)?
Now the mosfet follower. There is a modify in you approach that I find logical, indeed so logical that I’ve found it suspected…maybe I can’t figured out exactly what happen in this topology!
However, this is the issue: If we use the follower in conjunction with a negative supply for the bias, why don’t we use a differential supply to feed the mos and direct joined the follower to the power stage? We can leave the RC coupling between the voltage amp and the follower … its high input impedance and relatively low input capacitance shouldn’t be a problem for the I/V stage.
Driving an output stage that works in class A/B, we should achieve an improvement in terms of blocking distortion and recovery time. Another “extra” is the option for a lower voltage in the power supply (using a low voltage mos, with, probably, lower input capacitance) and higher bias current with the same or even lower heat dissipation.
I’m missing something? What do you thing?
Thanks
mark
Mark,
Both the cascode diff amp front end and the direct coupling of the source follower output to the output tube grids with bias applied at the source follower grid are good ideas. In fact, if we were to get really fancy, then that bias would be applied using a bias servo. Witha "clean sheet" design all things are possible and its a "golden ruleZ" of design engineering that MKII is always better than MKI (COZ hindsight is better than foresight).
For bias servos I haven't relly seen anything simpler/better than the Norman Koren "TENA" scheme.
Cheers,
Ian
Both the cascode diff amp front end and the direct coupling of the source follower output to the output tube grids with bias applied at the source follower grid are good ideas. In fact, if we were to get really fancy, then that bias would be applied using a bias servo. Witha "clean sheet" design all things are possible and its a "golden ruleZ" of design engineering that MKII is always better than MKI (COZ hindsight is better than foresight).
For bias servos I haven't relly seen anything simpler/better than the Norman Koren "TENA" scheme.
Cheers,
Ian