Hi, after the request of @OldHector , I post here my own version of the Baby Huey with EL34 (that fits well with KT77 and other tubes).
I will also post some guidelines for other tubes like 6550 (I would avoid KT88 because the original pcbs have the sockets too close for KT88).
Let's start with a bit of hystory: this kind of amps have ultralinear and Schade feedback combined.
Schade original feedback (1938) can be found here:
https://www.dos4ever.com/uTracer3/Schade.pdf
Crowhurst reposted that kind of anote-anode feedback in 1952 here:
http://www.tubebooks.org/Books/Crowhurst - Audio Handbook No2 - Feedback 1952.pdf
Yves Monmagnon posted the PP Schade feedback (and phase inverter supply) in pentode configuration plus conventional gnfb here:
http://www.dissident-audio.com/PP_ECL86/Page.html
The EL84 development has been done in 2006 by @gingertube here using 40%UL and Schade feedback:
The EL34 version has been started by bandol (RIP) in 2018 here with 40% UL and Schade feedback:
Here I posted a similar thread with noted on the EL84 version:
Here I will report my takes on the EL34 version, but before a small introduction on Schade feedback:
Numerous tests have been done during the last 90 years, but just to report real measurements, I link some sources:
One is Broskie in 2001:
https://www.tubecad.com/march2001/
One is Bertola with his plts on Schade feedback only.
in 2013 he has tested on the curve tracer the 807a ( https://frank.pocnet.net/sheets/136/3/307A.pdf ) showing how Schade alone can triodify the curves:
source: https://www.bartola.co.uk/valves/2013/03/16/307a-dht-in-triode-and-schade-feedback/#more-1331
Some italians have done tests on the curve tracers too in 2021:
source: https://www.sb-lab.eu/schadeode-alternativa-ultralineare-triodi-virtuali/
There are many others curve tracer plots, but I will only report the ones above.
I will also post some guidelines for other tubes like 6550 (I would avoid KT88 because the original pcbs have the sockets too close for KT88).
Let's start with a bit of hystory: this kind of amps have ultralinear and Schade feedback combined.
Schade original feedback (1938) can be found here:
https://www.dos4ever.com/uTracer3/Schade.pdf
Crowhurst reposted that kind of anote-anode feedback in 1952 here:
http://www.tubebooks.org/Books/Crowhurst - Audio Handbook No2 - Feedback 1952.pdf
Yves Monmagnon posted the PP Schade feedback (and phase inverter supply) in pentode configuration plus conventional gnfb here:
http://www.dissident-audio.com/PP_ECL86/Page.html
The EL84 development has been done in 2006 by @gingertube here using 40%UL and Schade feedback:
The EL34 version has been started by bandol (RIP) in 2018 here with 40% UL and Schade feedback:
Here I posted a similar thread with noted on the EL84 version:
Here I will report my takes on the EL34 version, but before a small introduction on Schade feedback:
Numerous tests have been done during the last 90 years, but just to report real measurements, I link some sources:
One is Broskie in 2001:
https://www.tubecad.com/march2001/
One is Bertola with his plts on Schade feedback only.
in 2013 he has tested on the curve tracer the 807a ( https://frank.pocnet.net/sheets/136/3/307A.pdf ) showing how Schade alone can triodify the curves:
source: https://www.bartola.co.uk/valves/2013/03/16/307a-dht-in-triode-and-schade-feedback/#more-1331
Some italians have done tests on the curve tracers too in 2021:
source: https://www.sb-lab.eu/schadeode-alternativa-ultralineare-triodi-virtuali/
There are many others curve tracer plots, but I will only report the ones above.
I will use, as a reference the schematic share by Bandol on the first post of the original EL34 thread:
https://www.diyaudio.com/community/attachments/schema-el34-rev-3-12-july-2018-pdf.701184/
First of all, I don't use the version with the dedicated pcb for PSU, because I personally don't find it very useful based on my experience:
the negative rail voltage regulation is not strictly needed because all negative supplies pass through cascoded CCS, so highest possible noise rejection.
Low voltage positive supply could benefit from a 7824, but even without it I've found it not noisy at all.
Looking at the Mullard datasheet of EL34 ( https://frank.pocnet.net/sheets/129/e/EL34.pdf ), I've chosen to work at 23%UL for different reasons.
I will then talk later on about a second feedback I applied, but it would be off-topic here.
About R33 and R38:
to fully drive the EL34 I need around 67 Vpp up to ten times (to be conservative) the bandwidth, so 200 kHz, so slewrate will be:
2 x pi x f x V = 2 x 3,14 x 200 kHz x 67 = 84 V/us
One EL34 has an input capacitance of 15 pF.
C = i x dt / V
dV/dt = 84 V/us
then
15 pF = i / slewrate
so
i = 15 pF x 84 V/us = 1.26 mA
The current of the CCS is 0,7 / R33 (or R38), so we'll need that resistor to be less than 1.5 kOhm. I will use 330 Ohm.
Powerdrive AC voltage supply
Being the negative side of the PI's CCS connected with the negative rail of the Powerdrive circuit, the voltage is set by this one at 3 times the bias voltage.
So for a EL34 amp with approximately -33 Vdc it will be around -100 so, dividing by squared root of 2, a 80 Vac winding is needed to supply it (it's called BIAS on the pcb).
About R16 and R17:
On the BOM and the PCB they are shown as 15 kOhm and 10 kOhm respectively.
The guideline is to keep the current of the led of the PI's CCS around 2 mA.
To improve the CCS, R17 can be sostitued with a second red led. R16 will then be 47 kOhm.
The table here below can be followed for different negative supplies:
About R3 and R4:
On the BOM and the PCB they are shown as 1 kOhm, but this is quite low as a value.
The old guidelines for triodes were to use 8/gm to set the grid stopper value, so with a 12AX7 which gm is 1.6 mA/V the value would be around 5 kOhm. In my guitar amps I always use at least 10 kOhm because gm of the tubes goes down with age, and I want to be sure there are no issues even when tubes are almost gone.
With an avarage value of 12AX7's input capacitance (Miller included) in the order of 150 pF, 10 kOhm gives a low pass filter with -3db point at 106.1 kHz. I would say it's plenty acceptable.
IMPORTANT NOTE: carbon comp here is the best choice, because they are the resistors with lowest capacitance and inductance, and that is what is needed in this position. I used carbon film, the second place in the choice.
About R7 and R8:
In the BOM they are 600 mW resistors, but they need to be 1 or 2 Watts, because of the voltage they see across them (remember that they are connected to the shunt feedback system that is directly connected to the primary of the output transformer.
Having higher B+, I can use higher values here without going into grid current area. I indeed use 270 kOhm to further improve linearity and gain of the phase inverter.
About R14 and R15:
On the pcb they are shown as 270 Ohm, but I will follow datasheet safer value of 1kOhm.
https://www.diyaudio.com/community/attachments/schema-el34-rev-3-12-july-2018-pdf.701184/
First of all, I don't use the version with the dedicated pcb for PSU, because I personally don't find it very useful based on my experience:
the negative rail voltage regulation is not strictly needed because all negative supplies pass through cascoded CCS, so highest possible noise rejection.
Low voltage positive supply could benefit from a 7824, but even without it I've found it not noisy at all.
Looking at the Mullard datasheet of EL34 ( https://frank.pocnet.net/sheets/129/e/EL34.pdf ), I've chosen to work at 23%UL for different reasons.
- Allows to get some more power compared to the 40% UL;
- Requires less swing to drive it;
- Works better at higher voltages;
- It ensures a good reduction of internal resistance of the tube;
- I've been very happy with 23% for EL84, and I wanted to use it for EL34 too;
- It has an input capacitance similar to the pentode, so easier to drive;
- There's more voltage on the phase inverter, that helps with linearity.
I will then talk later on about a second feedback I applied, but it would be off-topic here.
About R33 and R38:
to fully drive the EL34 I need around 67 Vpp up to ten times (to be conservative) the bandwidth, so 200 kHz, so slewrate will be:
2 x pi x f x V = 2 x 3,14 x 200 kHz x 67 = 84 V/us
One EL34 has an input capacitance of 15 pF.
C = i x dt / V
dV/dt = 84 V/us
then
15 pF = i / slewrate
so
i = 15 pF x 84 V/us = 1.26 mA
The current of the CCS is 0,7 / R33 (or R38), so we'll need that resistor to be less than 1.5 kOhm. I will use 330 Ohm.
Powerdrive AC voltage supply
Being the negative side of the PI's CCS connected with the negative rail of the Powerdrive circuit, the voltage is set by this one at 3 times the bias voltage.
So for a EL34 amp with approximately -33 Vdc it will be around -100 so, dividing by squared root of 2, a 80 Vac winding is needed to supply it (it's called BIAS on the pcb).
About R16 and R17:
On the BOM and the PCB they are shown as 15 kOhm and 10 kOhm respectively.
The guideline is to keep the current of the led of the PI's CCS around 2 mA.
To improve the CCS, R17 can be sostitued with a second red led. R16 will then be 47 kOhm.
The table here below can be followed for different negative supplies:
Code:
B- [V] R17 R16 [kOhm]
15 red led 5,8
20 red led 8,3
25 red led 10,8
30 red led 13,3
35 red led 15,8
40 red led 18,3
45 red led 20,8
50 red led 23,3
55 red led 25,8
60 red led 28,3
65 red led 30,8
70 red led 33,3
75 red led 35,8
80 red led 38,3
85 red led 40,8
90 red led 43,3
95 red led 45,8
100 red led 48,3
105 red led 50,8
110 red led 53,3
115 red led 55,8
120 red led 58,3
125 red led 60,8
130 red led 63,3
135 red led 65,8
140 red led 68,3
145 red led 70,8
150 red led 73,3About R3 and R4:
On the BOM and the PCB they are shown as 1 kOhm, but this is quite low as a value.
The old guidelines for triodes were to use 8/gm to set the grid stopper value, so with a 12AX7 which gm is 1.6 mA/V the value would be around 5 kOhm. In my guitar amps I always use at least 10 kOhm because gm of the tubes goes down with age, and I want to be sure there are no issues even when tubes are almost gone.
With an avarage value of 12AX7's input capacitance (Miller included) in the order of 150 pF, 10 kOhm gives a low pass filter with -3db point at 106.1 kHz. I would say it's plenty acceptable.
IMPORTANT NOTE: carbon comp here is the best choice, because they are the resistors with lowest capacitance and inductance, and that is what is needed in this position. I used carbon film, the second place in the choice.
About R7 and R8:
In the BOM they are 600 mW resistors, but they need to be 1 or 2 Watts, because of the voltage they see across them (remember that they are connected to the shunt feedback system that is directly connected to the primary of the output transformer.
Having higher B+, I can use higher values here without going into grid current area. I indeed use 270 kOhm to further improve linearity and gain of the phase inverter.
About R14 and R15:
On the pcb they are shown as 270 Ohm, but I will follow datasheet safer value of 1kOhm.
Last edited:
Output Transformer:
Looking at this document: http://www.oestex.com/tubes/Rudolf Moers - Appendices.pdf
And this one: http://primarywindings.com/wp-content/uploads/2017/04/Mullard-Circuits-for-Audio-Amplifiers.pdf (page 22-23)
About R13
I tried different values here, and at the end I used 39k that worked best for all listenings.
About R18
I use 620 Ohm because that way current through the two triodes can be higher but still the 12AX7 will be far from grid current zone (with that tube it starts at around Vgk=-0,9V).
Looking at this document: http://www.oestex.com/tubes/Rudolf Moers - Appendices.pdf
And this one: http://primarywindings.com/wp-content/uploads/2017/04/Mullard-Circuits-for-Audio-Amplifiers.pdf (page 22-23)
About R13
I tried different values here, and at the end I used 39k that worked best for all listenings.
About R18
I use 620 Ohm because that way current through the two triodes can be higher but still the 12AX7 will be far from grid current zone (with that tube it starts at around Vgk=-0,9V).
Last edited:
Dear @zintolo,
Thank you for taking the time to do this nice summery of your “personal interpretation” of BH EL34. I have not read it all. but wil no doubt enjoy it when I do and learn some new things.
Thanks very much.
Thank you for taking the time to do this nice summery of your “personal interpretation” of BH EL34. I have not read it all. but wil no doubt enjoy it when I do and learn some new things.
Thanks very much.
Thank you @Francois G and @neodymium ,
I've built several Baby Hueys, and I have to say I like how it can be simplified further more.
I've even used it for instrument amplification (reversed) with very good results.
One note I want to add is a guideline for the impedence of the CCS that @Mark Johnson gave me time ago:
(CCS impedance) x (DC constant current) > 40 MOhm*mA
I've built several Baby Hueys, and I have to say I like how it can be simplified further more.
I've even used it for instrument amplification (reversed) with very good results.
One note I want to add is a guideline for the impedence of the CCS that @Mark Johnson gave me time ago:
(CCS impedance) x (DC constant current) > 40 MOhm*mA
I have Baby Huey octal and EL84 boards and I'm hoping to get them built sometime this year. Looking at the schematic I find that the mosfets in the power drive circuit are NLA, at least at Mouser, but listed as no longer manufactured on DigiKey. I can order them from DigiKey but would rather use something that is more likely to be available for a little while. Is there a good substitute that anyone has used?
Thanks
Thanks
It would have been something I would have talked about later on, but you can use the STF3LN80K5.
It has an input capacitance around 1 pF, similar to the tube socket alone, and can get full voltages without issues. Being TO220 (with voltage insulation), you can raise the current without issues.
I have some STF3LN80K5 in use too (and some in stock as well). They work very well indeed with low CRSS.
In threads involving "Schade feedback" almost always missing is any thought about the effect this type of feedback has on the input impedance of the power tube involved.
See for an explanation of this effect pages 396 to 399 of this book:
Fundamentals of Radio Valve-Technique, Deketh, Philips, 1949
The conclusion of this explanation is that applying this type of feedback to a triode in the stage preceding the power tube could very well be a bad choice (since the load for AC this triode sees is its plate resistor paralleled to the input impedance of the power tube).
Note that in the reference linked to in post #1 in this thread, the "Schade feedback" is applied to the secondary of an interstage transformer. So, by choosing an appropriate transformer ratio, the loading of the driver can be made right.
But the majority of examples I see of "Schade feedback" on this, and many other forums, has a triode driving a power tube without an interstage transformer with a correct ratio to deal with the above.
Like explained in the passage in the book I linked to: Use a pentode as a driver when not using an interstage transformer.
See for an explanation of this effect pages 396 to 399 of this book:
Fundamentals of Radio Valve-Technique, Deketh, Philips, 1949
The conclusion of this explanation is that applying this type of feedback to a triode in the stage preceding the power tube could very well be a bad choice (since the load for AC this triode sees is its plate resistor paralleled to the input impedance of the power tube).
Note that in the reference linked to in post #1 in this thread, the "Schade feedback" is applied to the secondary of an interstage transformer. So, by choosing an appropriate transformer ratio, the loading of the driver can be made right.
But the majority of examples I see of "Schade feedback" on this, and many other forums, has a triode driving a power tube without an interstage transformer with a correct ratio to deal with the above.
Like explained in the passage in the book I linked to: Use a pentode as a driver when not using an interstage transformer.
@Robert H. Gribnau
thanks for the link.
The load of the previous stage is very clear to everyone who has ever used Schade feedback.
Every feedback has his own drawbacks, and the limit in this kind of feedback is the capability of the previous stage to drive a low impedance.
That's why it is usually used with 12AT7 or similar tubes.
In this specific case the EL34 anode swings approximately from 390 to 510 V with g1 swinging from -43 to -32 V. That's a gain of -10,9x.
The 12AX7 works on 270k + 47k with around 750 uA per side, so its anode is at around 210V.
The center of the 39k has no signal, because on both sides we have a signal of same amplitude and opposite phase (hyp: no distortion).
This makes a voltage divider from the EL34 anode that is 47k to 19,5k, so around 1/2,41x.
So the anode of the 12AX7 sees {(270k + 47k||19,5k)/[(10,9/2,41)+1]} || 1M= [(270k + 13,8k)/5,52] || 1M = 51,4 k || 1 M = 48,9 kOhm
Still acceptable for a 12AX7 and very good sounding with the right considerations.
thanks for the link.
The load of the previous stage is very clear to everyone who has ever used Schade feedback.
Every feedback has his own drawbacks, and the limit in this kind of feedback is the capability of the previous stage to drive a low impedance.
That's why it is usually used with 12AT7 or similar tubes.
In this specific case the EL34 anode swings approximately from 390 to 510 V with g1 swinging from -43 to -32 V. That's a gain of -10,9x.
The 12AX7 works on 270k + 47k with around 750 uA per side, so its anode is at around 210V.
The center of the 39k has no signal, because on both sides we have a signal of same amplitude and opposite phase (hyp: no distortion).
This makes a voltage divider from the EL34 anode that is 47k to 19,5k, so around 1/2,41x.
So the anode of the 12AX7 sees {(270k + 47k||19,5k)/[(10,9/2,41)+1]} || 1M= [(270k + 13,8k)/5,52] || 1M = 51,4 k || 1 M = 48,9 kOhm
Still acceptable for a 12AX7 and very good sounding with the right considerations.