Thanks Ian for your response.
I'm beginning to understand why the zobels were fitted in the classic UL-tube amp designs. So far I never needed them in the amps i built.
Here in the Baby Huey UL-feedback case they are likely to be required. Ultralinear in general is a bit more critical than I always thought.
I already increased the screen grid-stoppers to 1k with no noticeable effect. I'll try the zobels on the anode-screen to see if it cleans up the response.
About your Maida-reg experiments: Sounds interesting. Recently I use the classic TL783 HV regulators, they withstand a delta of 125V over the chip. Easy to put into operation for quick checks, reliable and stable.
The more you add, the more you loose of the simplicity of the original BabyHuey (a bit my concern when tweaking it).
YvesM: I know you're negatively biased
concerning UL. 
I'll take your point and keep it in the back of my head.
Kind regards,
Yves
I'm beginning to understand why the zobels were fitted in the classic UL-tube amp designs. So far I never needed them in the amps i built.
Here in the Baby Huey UL-feedback case they are likely to be required. Ultralinear in general is a bit more critical than I always thought.
I already increased the screen grid-stoppers to 1k with no noticeable effect. I'll try the zobels on the anode-screen to see if it cleans up the response.
About your Maida-reg experiments: Sounds interesting. Recently I use the classic TL783 HV regulators, they withstand a delta of 125V over the chip. Easy to put into operation for quick checks, reliable and stable.
The more you add, the more you loose of the simplicity of the original BabyHuey (a bit my concern when tweaking it).
YvesM: I know you're negatively biased
concerning UL. I'll take your point and keep it in the back of my head.Kind regards,
Yves
Yves,
"No worries mate" as we say in Oz.
For teh anode screen tap zobels - these should go direct across the transformer wires.
anode to screen is 0.3 of total pimary or 0.09 of impedance - so with 8K total primary that gives 720 Ohms impedance of one anode to screen winding.
Zobel traditionally use a resistor value of 1.4 to 2 times the impedance they are in parallel with so 1K to 1K5 for the resistor. 2W Metal Film resistors. They can dump a lot of HF power.
With 1K then 1nF will have roll off above 159kHz, probably too high.
That suggests to me that 2n2 would be better, then roll off is above 72kHz.
I know that the little Hammond 1608 had resonances at about 68kHz
Cheers,
Ian
"No worries mate" as we say in Oz.
For teh anode screen tap zobels - these should go direct across the transformer wires.
anode to screen is 0.3 of total pimary or 0.09 of impedance - so with 8K total primary that gives 720 Ohms impedance of one anode to screen winding.
Zobel traditionally use a resistor value of 1.4 to 2 times the impedance they are in parallel with so 1K to 1K5 for the resistor. 2W Metal Film resistors. They can dump a lot of HF power.
With 1K then 1nF will have roll off above 159kHz, probably too high.
That suggests to me that 2n2 would be better, then roll off is above 72kHz.
I know that the little Hammond 1608 had resonances at about 68kHz
Cheers,
Ian
Success
Ian : Feedback about zobels
The 2n2-1k zobels across anode and screen tap took care
of the oscillations completely. See attached trace measured under
the same circumstances.
Even the one channel which was only marginally stable is cured completely.
The signal now drops in a linear way at 20kHz way down to
60kHz. Without the zobels there were several rises/drops in the 30kHz..60kHz band, although according to the spec. sheet the resonance of the James 6228 is at around 100kHz.
Time for a listening session...
PS: I've found one nice substitute for the EL84: The E80L (with lower power 9W@250V). A bit less gm, but quite comparable. You've to rewire the g2 and connect the g3, run it at 250V-24mA. I didn't do a full listening test but noticed the balanced sound of this tube. The build quality is exceptional as it is an industrial variant with gold pins.
Another tube that could do is the EL821, a video pentode with a higher gm and about 9W. The STC (Australia!) spec sheet says
that higher distortion is to be expected in audio service. I'll check
this one too if time permits.
Bye
Yves
Ian : Feedback about zobels
The 2n2-1k zobels across anode and screen tap took care
of the oscillations completely. See attached trace measured under
the same circumstances.
Even the one channel which was only marginally stable is cured completely.
The signal now drops in a linear way at 20kHz way down to
60kHz. Without the zobels there were several rises/drops in the 30kHz..60kHz band, although according to the spec. sheet the resonance of the James 6228 is at around 100kHz.
Time for a listening session...
PS: I've found one nice substitute for the EL84: The E80L (with lower power 9W@250V). A bit less gm, but quite comparable. You've to rewire the g2 and connect the g3, run it at 250V-24mA. I didn't do a full listening test but noticed the balanced sound of this tube. The build quality is exceptional as it is an industrial variant with gold pins.
Another tube that could do is the EL821, a video pentode with a higher gm and about 9W. The STC (Australia!) spec sheet says
that higher distortion is to be expected in audio service. I'll check
this one too if time permits.
Bye
Yves
Attachments
UL Zobels
Yves (dupyv),
I implemented the anode to screen zobels on my prototype BH and thought they were very worthwhile. I was a bit concerned about getting these zobels "right" as a too agressive zobel will kill the top end.
I think I may have made a "blue" (OZ vernacular for "screw up") BUT I did it this way.
I set feedback to zero by shorting across the shunt feedback set resistor. I then adjusted the Cap value (R value left at 1k) until there was just the slightest overshoot but no more than about a half cycle of damped ring with a 10kHz square wave into a pure resistive dummy load.
I came up with a value of 820pF for the series cap. I'm wondering if the 2n2 I originally suggested and you implemented might be too much. I'm also wondering if I should have set the zobel with the normal level of feedback in place.
I have tried to do this (setting of anode to screen zobels) "scientifically" in the past using this article from VoltSecond as guidance.
http://www.siteswithstyle.com/VoltSecond/Damping_ringing_XFMRS/Damping_ringing_in_xfmrs.html
I had very limited success. At that time I was applying this to a Hammond 1650T Output Tranny driven by a quad (PPP) of KT88 in Ultralinear Mode. The Push Pull sides seem to want different zobels
BUT this was probably just an artifact of their interaction. This is such a fundamental thing that I may do some more work on it. Getting the series R in the screen and the R and C values between anode and screen taps should be amenable to proper test/analysis (if ONLY I had a better understanding of what I'm doing).
Netbug,
Yes change the series screen resistors to a higher (than 33 Ohms) value. 150 Ohms would be minimum I would use with B+ of 340V. I'd be inclined to go to 270 Ohms (use 2W Metal Films).
Cheers,
Ian
Yves (dupyv),
I implemented the anode to screen zobels on my prototype BH and thought they were very worthwhile. I was a bit concerned about getting these zobels "right" as a too agressive zobel will kill the top end.
I think I may have made a "blue" (OZ vernacular for "screw up") BUT I did it this way.
I set feedback to zero by shorting across the shunt feedback set resistor. I then adjusted the Cap value (R value left at 1k) until there was just the slightest overshoot but no more than about a half cycle of damped ring with a 10kHz square wave into a pure resistive dummy load.
I came up with a value of 820pF for the series cap. I'm wondering if the 2n2 I originally suggested and you implemented might be too much. I'm also wondering if I should have set the zobel with the normal level of feedback in place.
I have tried to do this (setting of anode to screen zobels) "scientifically" in the past using this article from VoltSecond as guidance.
http://www.siteswithstyle.com/VoltSecond/Damping_ringing_XFMRS/Damping_ringing_in_xfmrs.html
I had very limited success. At that time I was applying this to a Hammond 1650T Output Tranny driven by a quad (PPP) of KT88 in Ultralinear Mode. The Push Pull sides seem to want different zobels
BUT this was probably just an artifact of their interaction. This is such a fundamental thing that I may do some more work on it. Getting the series R in the screen and the R and C values between anode and screen taps should be amenable to proper test/analysis (if ONLY I had a better understanding of what I'm doing).
Netbug,
Yes change the series screen resistors to a higher (than 33 Ohms) value. 150 Ohms would be minimum I would use with B+ of 340V. I'd be inclined to go to 270 Ohms (use 2W Metal Films).
Cheers,
Ian
Last edited:
Ian (Gingertube),
On the UL-feedback variant, I lowered the C in the zobel to 1n5 to get back a minimal overshoot on the square-wave. Right now it's not over- or underdamped, it's on spot.
I went back to anode-feedback on my current build of the baby huey.
With UL-feedback, I was unable to get the same focus and soundstage.
I went though several values of the feedback-resistor but each time voices
remained dull and flat. I tested this on two Fostex fullrange speakers/horns, with identical results. I suspect that 2way/3way speakers might respond
in a different way to UL-feedback, hence the better results you obtained
with UL-feedback.
A a test-record i used the album "Cewbeagappic" from Beady Belle, excellent jazz with lots of spatial information and tight focus. Play track 6-Shadow and in 5 seconds you know if the amp sounds right or not.
I did some tube rolling in the baby huey (anode-feedback version ):
- EL821/6CH6 (12watts): nothing exceptional, better than some russian EL84's, quite laid back.
- E80L /6227 (9watts): 250V - 24ma idle, feedback resistor set to 14.1k, nice voices & bass, better than most standard EL84's. Has a lower power
rating than the EL84. I wouldn't push it to 300V, you'll have to live with around 10W power output
Comes on place 2 of my favorite tubes for the baby huey, the 1st being a Tungsram EL84.
Cheers,
Yves
On the UL-feedback variant, I lowered the C in the zobel to 1n5 to get back a minimal overshoot on the square-wave. Right now it's not over- or underdamped, it's on spot.
I went back to anode-feedback on my current build of the baby huey.
With UL-feedback, I was unable to get the same focus and soundstage.
I went though several values of the feedback-resistor but each time voices
remained dull and flat. I tested this on two Fostex fullrange speakers/horns, with identical results. I suspect that 2way/3way speakers might respond
in a different way to UL-feedback, hence the better results you obtained
with UL-feedback.
A a test-record i used the album "Cewbeagappic" from Beady Belle, excellent jazz with lots of spatial information and tight focus. Play track 6-Shadow and in 5 seconds you know if the amp sounds right or not.
I did some tube rolling in the baby huey (anode-feedback version ):
- EL821/6CH6 (12watts): nothing exceptional, better than some russian EL84's, quite laid back.
- E80L /6227 (9watts): 250V - 24ma idle, feedback resistor set to 14.1k, nice voices & bass, better than most standard EL84's. Has a lower power
rating than the EL84. I wouldn't push it to 300V, you'll have to live with around 10W power output
Comes on place 2 of my favorite tubes for the baby huey, the 1st being a Tungsram EL84.
Cheers,
Yves
Hi
I would like to try making the baby huey as in post
http://www.diyaudio.com/forums/showpost.php?p=1669805&postcount=604
Does anyone have a list of recommended parts. I dont want to spend millions but dont want to skimp either.
Thanks
Brad
I would like to try making the baby huey as in post
http://www.diyaudio.com/forums/showpost.php?p=1669805&postcount=604
Does anyone have a list of recommended parts. I dont want to spend millions but dont want to skimp either.
Thanks
Brad
Did moving from UL-Feedback to Anode-Feedback require new feedback resistor values? I'd like to compare the two modes, so I was thinking about a wiring up a switch to quickly change between UL and Anode.
Netbug,
Going from UL-feedback to Anode-feedback will require a change of 3 resistors per channel and a different hookup of the ECC83 anode supply.
The anode-feedback variant is on post 602: http://www.diyaudio.com/forums/showpost.php?p=1669029&postcount=602
The UL-feedback is on post 604.
Compare and you see that instead of 10k-18k-10k you'll have to use 47k-16k-47k where 16k is the feedback resistor.
Kind regards,
Yves
Going from UL-feedback to Anode-feedback will require a change of 3 resistors per channel and a different hookup of the ECC83 anode supply.
The anode-feedback variant is on post 602: http://www.diyaudio.com/forums/showpost.php?p=1669029&postcount=602
The UL-feedback is on post 604.
Compare and you see that instead of 10k-18k-10k you'll have to use 47k-16k-47k where 16k is the feedback resistor.
Kind regards,
Yves
bummer.
I really need to build a second baby huey in a much larger chassis. Lay everything out so that its easy to get to and change around. My current build is a real pain to make changes to.
oh well..
I have just finished building (one channel of) the original EL84 baby huey, and i must say, i'm impressed
Instead of the ecc803s, I used the russian 6n2p, wich seems to work very well.
Another addition I made was a simple stabilised power supply with one mosfet and one high voltage transistor, if anyone is interessed, I can upload the schematic.
Also, the gNFB is not used.
Pictures and measurements will have to wait untill I finish the second channel.
Instead of the ecc803s, I used the russian 6n2p, wich seems to work very well.
Another addition I made was a simple stabilised power supply with one mosfet and one high voltage transistor, if anyone is interessed, I can upload the schematic.
Also, the gNFB is not used.
Pictures and measurements will have to wait untill I finish the second channel.
Random Ravings
Yves (dupyv),
I messed about some more with the original CCS biased BH on the weekend. I changed back to sourcing the shunt feedback from the EL84 anodes as you have done. I decided I like the shunt feedback from the UL taps better BUT ONLY when I have linearised the speaker impedance by addition of a 10 Ohm + 2u2 zobel across the 4 Ohm secondary tap. (The speakers are VAF Research DC-X, 2 x 10" bass/mid with tweeter mounted between them "time aligned", and a simple series cap for the tweeter feed). That is I have optimised the amp / speaker interface for just one set of speakers. The amp seems to be much more susceptible to speaker impedance changes with frequency when using the UL taps so in general I think sticking with the anode connections makes sense, especially with speakers with complex impedance profiles or where you want to use the amp with different sets of speakers. Also keep in mind I deliberately went back to Hammond 1608 Output Transformers on this unit, trying to keep things simple and cheap.
P.S. I greatly appreciate your testing, suggestions and insights into the Baby Huey. Thank You. I hope you are enjoying your BH Amps.
hidnplayr,
Tube amps (in particular) suffer Intermodulation distortion. Using a simple voltage reg is something yves recommended (and posted his circuit) and it makes sense since you remove the ripple from the power supply and there is then one less "signal" for the sound signal to intermodulate with. I have'nt tried this yet but it makes good sense.
Cheers,
Ian
Yves (dupyv),
I messed about some more with the original CCS biased BH on the weekend. I changed back to sourcing the shunt feedback from the EL84 anodes as you have done. I decided I like the shunt feedback from the UL taps better BUT ONLY when I have linearised the speaker impedance by addition of a 10 Ohm + 2u2 zobel across the 4 Ohm secondary tap. (The speakers are VAF Research DC-X, 2 x 10" bass/mid with tweeter mounted between them "time aligned", and a simple series cap for the tweeter feed). That is I have optimised the amp / speaker interface for just one set of speakers. The amp seems to be much more susceptible to speaker impedance changes with frequency when using the UL taps so in general I think sticking with the anode connections makes sense, especially with speakers with complex impedance profiles or where you want to use the amp with different sets of speakers. Also keep in mind I deliberately went back to Hammond 1608 Output Transformers on this unit, trying to keep things simple and cheap.
P.S. I greatly appreciate your testing, suggestions and insights into the Baby Huey. Thank You. I hope you are enjoying your BH Amps.
hidnplayr,
Tube amps (in particular) suffer Intermodulation distortion. Using a simple voltage reg is something yves recommended (and posted his circuit) and it makes sense since you remove the ripple from the power supply and there is then one less "signal" for the sound signal to intermodulate with. I have'nt tried this yet but it makes good sense.
Cheers,
Ian
Ian (Gingertube),
>The amp seems to be much more susceptible to speaker impedance changes with frequency when using the UL taps so in general
>I think sticking with the anode connections makes sense, especially with speakers with complex impedance profiles
This seems to confirm the experiences I made with my Fullrange speakers, they don't have a linear impedance over 40Hz..20kHz.
The anode connection seemsto be the more universal & easy setup. To setup the UL-connection correctly, you need
an oscilloscope to check the signal, moreover it responds more critically to the type of OPT and speaker at the end.
>P.S. I greatly appreciate your testing, suggestions and insights into the Baby Huey. Thank You. I hope you are enjoying your BH Amps.
Thanks for your design Ian (and YvesM who I hope to see at ETF2009). Since I have the BabyHueys, I didn't touch my 2A3SE and 2A3PP much... (does that say enough?).
Just for the kick, I'm tempted to try that 6AV5-pentode screen-driven design yielding 80W that tubelab.com presented on this board.
It's a class B design, I doubt it can beat the BH sonically. But it has power (call it a muscle amp, with all the drawbacks).
Kind regards,
Yves
>The amp seems to be much more susceptible to speaker impedance changes with frequency when using the UL taps so in general
>I think sticking with the anode connections makes sense, especially with speakers with complex impedance profiles
This seems to confirm the experiences I made with my Fullrange speakers, they don't have a linear impedance over 40Hz..20kHz.
The anode connection seemsto be the more universal & easy setup. To setup the UL-connection correctly, you need
an oscilloscope to check the signal, moreover it responds more critically to the type of OPT and speaker at the end.
>P.S. I greatly appreciate your testing, suggestions and insights into the Baby Huey. Thank You. I hope you are enjoying your BH Amps.
Thanks for your design Ian (and YvesM who I hope to see at ETF2009). Since I have the BabyHueys, I didn't touch my 2A3SE and 2A3PP much... (does that say enough?).
Just for the kick, I'm tempted to try that 6AV5-pentode screen-driven design yielding 80W that tubelab.com presented on this board.
It's a class B design, I doubt it can beat the BH sonically. But it has power (call it a muscle amp, with all the drawbacks).
Kind regards,
Yves
I understand, "the simplest way to go".
The original design goal was to do not use UL tapped OPT (wich are not always"ideal") while having the power of pentodes and the damping of triodes.
Bas, do you plan to move at ETF09 ?
http://www.triodefestival.net/index.php?page=etf-2009
Yves, sure, I will be too !
Yves.
The original design goal was to do not use UL tapped OPT (wich are not always"ideal") while having the power of pentodes and the damping of triodes.
Bas, do you plan to move at ETF09 ?
http://www.triodefestival.net/index.php?page=etf-2009
Yves, sure, I will be too !
Yves.
Yes that is exactly the reason why I decided to rebuild this old el84PP amp I bought.
The electrolyts where dried out, I replaced the rectifiers and put in some newer, bigger caps, but this raised the B+ to 390V and sound still wasnt as good as it should.
I didnt have any room to fit a regulated power supply in the original case.
Then I saw your circuit on here and decided to do a complete rebuild using your circuit,
wich also fixed the sound quality problems
Is the amount of feedback the same with both, anode and UL connections?
Could it be that the UL FB connection is giving you less feedback and lower
damping factor? The Zout must be higher for you to observer the greater
dependence on load impedance.
Are the g2 connected the same way in both cases?
Michael