• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

EL84 Amp - Baby Huey

Rhone,
The 100R pot (diff amp balance) can be a single turn, choose something mechanically strong.
The 50K set the bias for the output tubes and so you want to make sure that this is "ultra" reliable, For that reason, and for the additional "fine" adjustment they provide, I use Bourns 10 turn for these.
The 2u2 across the 2 100K bias pots is there to make absolutely sure that no noise (no AC signal) is coupled back into the audio path via the bias pots. A polyester would probably be OK but since I had some polypropylene in my parts bins I used them. Actually you could go to a 3u3 of even 4u7 for this capacitor. It doesn't need to be a 630V rated part, it only has to withstand the bias supply voltage, a 100V capacitor would be fine, even a 63V would do.

Cheers,
Ian
 
Rhone,

There is ONLY a "guideline". When using a cathode follower (MOSFET Source follower in this case to direct couple to output tube grids then it is recommended that the negative rail is at least 3 times the bias voltage. Bias voltage for EL84 is about -12V so you want a rail of at least -36V.

All,
I had a long play with the original (CCS biased) Baby Huey on the weekend.
The change to run the feedback from the Ultralinear taps rather than the EL84 anodes was something more significant than at first thought. With feedback from the anodes I could ONLY push the feedback level so far before starting to loose stereo imaging and the sound starting to be a bit messy. So we ended up with a fairly "loose" setting.
There has been an "accepted" view that the diffamp was just not coping with the demand for more current swing with higher levels of feedback and that this was the limiting factor. That has turned out to not be the case. With feedback from the Ultralinear Taps it appears that the feedback circuit is now much faster and I was able to push up the feedback levels quite a bit without any loss of imaging or introduction of too much diffamp distortion. Instead of the 10K + 9K1 + 10K, I was messing about with values of 22K, 27K and 33K in place of that 9K1, That is about 3 to 4 times as much feedback. When I got to be using 10K + 33K + 10K I decided to try 6K8 + 22K + 6K8 (identical feedback setting) instead to see if it might be a drive impedance issue. I thought that that did make a very small improvement. At the end of the weekend I ended up running 6K8 + 20K + 6K8 and not really able to make up my mind as to go one stage tighter (22K), one stage looser (19K1) or leave it alone. 15K was too loose and after some extended listening I decided 18K was not quite enough either. This has taken the "Baby" and given it better manners, BH is "growing up".
I haven't tried to apply this to the fixed biased version with the source followers yet - it may not be as applicable to that version.

Cheers,
Ian
 
And another hint from last nights experiments.

At higher feedback levels it was still sounding a little aggressive in its top end frequencies.

It suddenly clicked - the reference (AC 0V) for the shunt feedback (taken from the Ultralinear Taps) is the output transformer centre tap. I checked what I was running as an AC bypass on the B+ supply to this node and found that in this particular amp I had just a 100uF/400V electrolytic. Simply soldering in a 470nF/630V polypropylene cap across that 100uF cleaned up the feedback and got rid of 90% of that aggressive edge. I would guess that the self resonant frequency of that 100uF was well down into the audio band. Improving the AC ground at the Output Tranny centre tap had the secondary effect of cleaning up the feedback since its AC 0V reference was better.

This was a big step forward.

Cheers,
Ian
 
frugal-phile™
Joined 2001
Paid Member
gingertube said:
I checked what I was running as an AC bypass on the B+ supply to this node and found that in this particular amp I had just a 100uF/400V electrolytic. Simply soldering in a 470nF/630V polypropylene cap across that 100uF cleaned up the feedback and got rid of 90% of that aggressive edge. I would guess that the self resonant frequency of that 100uF was well down into the audio band. Improving the AC ground at the Output Tranny centre tap had the secondary effect of cleaning up the feedback since its AC 0V reference was better.

Have you tried an amp with all poly power supply caps?

dave
 
Dave,
A couple of the latter versions of the BH had significant (5 and 10uF) poly bypasses but I haven't built one with ALL poly cap supply. I will be doing just that on a PPP 300B (a quad of 300B per channel) Amp which is in the design phase.

The aim right now is to take the original BH (CCS biased circuit) and make it stunning without recourse to expensive bits. To that end I have fitted "cheap" Hammond 1608 Output Trannies back in this amp. With the last weeks or so of "fiddling" it is one of the best amps (mine or anyone elses, DIY or Commercial) I have ever heard. There is just 1 or 2% left (which I'm trying to bridge) between damn good and absolutely stunning. I'm exploring its limits if you like, and finding that the limits are often not where I originally thought, certainly the last week or 2 of mods/experiments has brought the performance up significantly with just the change to feedback from the UL taps, 3 resistor changes to increase the shunt feedback level and 2 added bypass caps - about a 100% performance boost for $5 worth of components.

Cheers,
Ian
 
gingertube, i have not done the mods to the feedback resisors yet, but i would just like to say, this amp has impressed so many that have listened. There is something about it that sets it apart. Its not the most neutral or detailed, but so many of my hypercritical mates have gone away very impressed.
 
One page back autoteck1 suggested trying 6973 for the outputs.

I bought a quad of Electroharmonix 6973 to try. On the weekend I sat down to look at putting these in the BH. 1st problem they need a complete rewire to the tube socket. 2nd problem, their gm is half that of EL84 so I don't think they will suit the BH that well.

Accordingly, the EH6973 have been put on my shelf, they will suit a little guitar amp.

So not recommended for a Baby Huey.

I had the feedback settings up and down all over the place. What suited some music did'nt cut it for others. I'm running 6K8 + 17K8 + 6K8 now. (might as well be 18K for that feedback set resistor).

Cheers,
Ian
 
Thank you very much Gingertube for this schematic.. it really is a wonderful sounding amp... now on to the cry for help

I finished my baby huey build and have run into an issue....

My B+ is 340. I read that Gingertube has said that 340 is the max, but what is the down side? It really sounds wonderful so I don't think I'm having any issues. Then again I'm a worrier.

am I going to kill tubes in a month?

Basically is it worth adding a power resister to get down to 330, or should I just live with it?
 
Just live with it.

You are still within the tube ratings.

Did you build the original version with the CCSs in the EL84 Cathodes?
If so you can change the current sense resistors (4 off) in the CCS'es from 16 Ohms up to 18 Ohms. That will reduce the output tube idle current and hence its power dissipation slightly and give you longer tube life.
The 18 Ohms is a bit better value when you have 340V B+.

If a fixed bias version then just set bias to get 35mA idle per tube.

Cheers,
Ian
 
Member
Joined 2007
Paid Member
UL-feedback instability

Gingertube,

I'm building my 2nd BabyHuey using James JS-6228H transformers.

First, I tested the fixed bias variant with feedback from the EL84-anodes.
The scope-trace on a square-wave showed no instability and a quick listening test confirmed that everything sounded ok.

Then, I converted to feedback from the UL-tap and modified the resistors to the 10k-9k1-10k scheme.
Here I'm having excessive ringing on square-waves I can't get rid of.
The shown scope trace is at medium power level, turning up the wick worsens the oscillation. The probe is attached to an 8R resistor
connected to the OPT's 8 ohm tap. The oscillation is independant
of the square-wave frequency.

Things I tried and checked:
- Grid-stoppers are right at the EL84 socket, a variation in value up to 1k had no effect on the oscillation.
- 100uF+220nF are soldered right at the OPT B+ tap, this addition had no effect.
- Swapped the EL84 tubes, this has no effect
- Increased/Decreased the value of the 9k1 feedback resistor, this had no effect.
I also tried your later 6k8-22k-6k8 settings to no avail.
- Suppression of the 10R common cathode resistor of the EL84. This has had the most notable effect so far:
- The suppression of the 10R provoked an oscillation of the upper *and* lower part of the square-wave
- The addition of the 10R made the oscillation on the *upper* part disappear
- An increase to 22R had no effect.


Gingertube, did you encounter similar issues during testing you UL-feedback variant ? Any more checks/tweaks come to mind?


I remember having had a similar issue a while back when testing the cheap Hammond 125E OPT in my 1st BabyHuey.
When the 125E was replaced by a better Hammond 1650E everything was ok then.

A test I've still on my list is to swap out the James for a known good Reinhöfer-OPT of german manufacture.
If the problem still persists, my wiring is perhaps at fault. If not, the James JS-6228 though having a good reputation, doesn't like
operating in this topology with UL-feedback.

Kind regards,

Yves
 

Attachments

  • bh-fixed-ul-ringgg.jpg
    bh-fixed-ul-ringgg.jpg
    19 KB · Views: 1,075
Yves,

Traditionally zobel networks were used with Ultralinear, wired on each side between anode and screen tap. Additionally g2 stoppers in the screen grid connection were often required and counter intuitively it seemed that the better the output transformer the more the g2 stoppers (higher value resistor) were required. Insufficient g2 stoppers usually shows up as "squegging", burst of RF just after the turn overs on +ve and/or -ve peaks, so that does not sound like the problem you have observed.

In the unit I am running at the moment I found that taking the feedback from the UL taps meant that the speaker impedance curve was more critical than when feedback was taken from the anodes. I fitted 10R + 1uF zobels across the 4 Ohm speaker taps. This helps linearise the speaker impedance with frequency BUT also its reflected impedance did at least part of the job the zobels from anode to screen would normally do. You could try this. You can also try increasing the g2 resistors.
You could also try zobels from anode to screen tap, I'd try a resistor value of 1K and capacitor of 1nF as a start point and then just change the cap value trying 470pF or 2n2 as required.

The other thing I've done in the unit I'm running right now is an interim thing. I added a Maida reg from B+ and then added 390K from the Maida Reg Output to each anode of the diff amp and doubled the diff amp current. That was just to get the diff amp triodes running in a more linear part of their Ia curve so that it did'nt "grunge up" as much when pushed hard. Eventually those 390K will be replaced with current sources or I may try to see if a can implement current mirror loads on the diff amp.

I have been a bit quiet while I built a new electronics workshop. That is now complete and I should get back to more work shortly.

Cheers,
Ian