50 watt EL34 build questions

[WRC34]

Hey you guys!

I am a beginner builder who has just "almost" finished a 50 watt amp that uses an EF86 tube in V1 and a 12AX7 phase inverter matched with 18 watt Lite iib style tone & volume controls (with pot values changed to 1M) into a JTM50 based EL34 power section.

The power transformer is an old Stancor PC8411. It's HV secondary is rated for 375-0-375 but that's with 117 VAC coming into the primary. I expected it to be higher with todays' US wall voltages and I was right, but I have been quite surprised at how much the wall voltage fluctuates throughout the day. Maybe it has something to do with my downstairs neighbor ripping his power tools in the garage.

The 5V winding is 3A so 5U4GB is safe to run as a way to drop some voltage but I have been using a 5R4GY instead to hopefully drop even a couple more. First filter cap is 30uf and I haven't had any problems yet (fingers crossed).

Initially I used 10K 5W resistors between B+ stages but that didn't get it down enough. I put in some 8.2K 3W resistors in series with the 10Ks and that helped. The voltages to the PI and preamp are still high, but not anything that will start a fire. With the 5R4 I'm getting anywhere between 460-490 VDC to the EL34 plates depending on the time of day, around 340 to the PI and 160ish to the plate & screen of the EF86. I have heard people say an EF86 should be much lower but I'm within spec of what the data sheets say for the tube so I'm not too concerned. Should I be?

I like to keep things as simple as possible (hence the simple volume & tone controls in lieu of a full tone stack), so my initial design omitted a presence pot and had a 'fixed' 33K negative feedback resistor coming off the tail resistor of the PI then attached directly to the 8ohm tap of the OT*. Once I got the amp powered up it was absolutely dead quiet, but very clean and sterile sounding. I replaced the 33K with a 120K in hopes of loosening the chains a bit and the result was ok, but not as dramatic as I hoped for. It sounded good and very punchy, but still VERY clean. To further the experiment, I said to hell with it and removed the NFB loop altogether by grounding the tail resistor (and .1uf cap from pin 7) of the PI and...WHOAH! Holy Schneebs, the thing breathed fire! Too much even.

So, I suppose I'm wondering if anyone has messed around with this type of NFB set up before and if so has any sagely advice. Like say, if I reinstated the NFB loop with a 240K resistor would I be closer to my desired 'halfway between heaven and hell'? Because I was in the goddam Catholic boy's choir with the NFB and without it I'm raping and pillaging at a pace that the most seasoned heathen would balk at

Also, another strange thing has been in biasing the amp. Out of curiosity I set my bias to 70% dissipation of the EL34s based on my plate voltage, left the DMM on and watched the reading as I hit a chord, it sank. The meter would say 37mA for example and upon hitting a note or chord hard would drop to 30mA or even lower and then stay there even after the chord had run out. Playing lightly did not have this affect, only when 'digging in'.

*OT is an even older Stancor of unknown origin and specification. The model no. is 79B80. Good luck trying to find any info on that!

Any thoughts or advice would be great. I will post photos tomorrow!

[tele_player]

I'd install a 1M linear pot in the feedback loop, and look for the sweet spot. Then maybe replace it with an equivalent resistor.

Robert

[pdf64]

What method are you using to assess the bias suitability, ie how are you measuring 30mA?

[WRC34]

I'd install a 1M linear pot in the feedback loop, and look for the sweet spot. Then maybe replace it with an equivalent resistor.

Great idea, thanks!

pdf64, I have the bias part of the circuit set up like a JTM50 with a 25K Linear bias pot. I'm using a bias tool with my DMM set to mA. After taking a plate voltage I divide my aimed-for-dissipation of 17 watts per EL34 by the plate voltage.

[pdf64]

It seems unusual for the cathode current to drop to a lower level after the amp has had a 'kerrang'.
More usually the cathode current rises a little over the 1st 20 mins as everything warms up and reaches an equilibrium.
Is your bias tool rated to carry full power cathode current, maybe ~250mA?
Perhaps its sensing resistor is overheating significantly under signal conditions, resulting in the value shifting, and consequent measurement error?

Depending on the degree of negative feedback, operating your amp with the loop closed then open can result in a big change in gain, as you noted.
But it also changes the character of the amp, from being a low output impedance voltage amp to be more akin to a high output impedance transconductance amp (voltage to current).
This tends to give a boost to bass and treble, tracking the speaker's impedance / frequency curve.
Open loop is great, it's my preference. If the gain increase is too much, then consider reducing the pre-amp gain somehow, maybe use a 12AY7 in V1, remove cathode bypass caps, use a split load plate resistor etc.
Rather than introducing negative feedback on the power amp, and potentially moving it towards the sterile quality it had before.

[jazbo8]

+1 I agree with pdf64, the cathode current should go up not down when you hit it hard, something is not right with the reading. Also, did you figure out the winding ratio of the OPT? What's the Ra-a for the EL34s?

[WRC34]

It seems unusual for the cathode current to drop to a lower level after the amp has had a 'kerrang'.
More usually the cathode current rises a little over the 1st 20 mins as everything warms up and reaches an equilibrium.
Is your bias tool rated to carry full power cathode current, maybe ~250mA?
Perhaps its sensing resistor is overheating significantly under signal conditions, resulting in the value shifting, and consequent measurement error?

I didn't think of this, good point. Most likely I will end up installing the 1 ohm resistors from pins 1 & 8 to ground to measure bias without the tool.

Open loop is great, it's my preference. If the gain increase is too much, then consider reducing the pre-amp gain somehow, maybe use a 12AY7 in V1, remove cathode bypass caps, use a split load plate resistor etc.

Yes, I want the amp to be pretty aggressive, and I do like the possibility of keeping the NFB out of the picture. Pardon my ignorance, but how would I implement a "split load plate resistor"? I'm guessing you mean in the preamp/PI?

Also, did you figure out the winding ratio of the OPT? What's the Ra-a for the EL34s?

Good question! I have not figured this out myself. The guy I bought it from said it would be perfect for a "pair of 6L6s, EL34s or a quad of 6V6s" which leads me to believe it's primary is somewhere in the 3.4K to 8K range. BUT of course it could me way different.

[martin manning]

The guy I bought it from said it would be perfect for a "pair of 6L6s, EL34s or a quad of 6V6s" which leads me to believe it's primary is somewhere in the 3.4K to 8K range.

That would say 3.4-4k to me. It's easy enough to measure something to get you in the ballpark.

[WRC34]

Here are some photos:

http://thatistheman.tumblr.com/post/105698212038

Something else that has occurred to me - there is a similarity in this amp to the Dr. Z Route 66/KT45 amps with the EF86 and lack of NFB. It is said that they use a higher tail resistor (like a 47K that you'd see in a Lite iib w/o NFB). Currently I have a 15K tail resistor in place, would raising it to a higher value help tame the insanity while allowing me to keep the NFB loop removed?

[pdf64]

That will reduce both the gain and the available voltage swing, so as you increase its value you may want to make sure that the EL34 can still be fully driven, ie to the point of grid clipping, even when the power amp is being heavily overdriven and the screen grids are sucking lots of current (thereby reducing the VB+ downstream).

[WRC34]

Ok great, I guess the best way to find out is to try a few different values. I'll post results soon.

Thanks everyone, I appreciate the replies!

[pdf64]

For an example of a split plate load, see the 6G3 Deluxe 2nd stage http://bmamps.com/Schematics/fender/del ... _schem.pdf

[roberto]

I see different options:

If you go to NFB with fixed presence, just take a look at Mesa configuration as a starting point (don't look at the depth things).

If you want a no-NFB amp, you can decrease the gain of the PI, or decrease the gain before (voltage dividers, tube configuration, tonestack).

[WRC34]

I have ordered a 1M Linear pot as I do not have one on hand. However, in the meantime I dropped in a few different value of resistors and upon closer observation it seems that any resistor of ANY value as NFB resistor has the same affect on the amp. I tried everything from 20K all the way up to 370K - same sound, same behavior, same affect - super clean and reduced volume.

Now, I am a novice so I fully admit I could have wanked it up somehow. You can see in the gut shots posted that the tail resistor is grounded along with the .1uf cap coming off of pin 7 of the PI - this is the NON NEGATIVE FEEDBACK configuration. In my experimenting with the NFB resistor I lifted the tail resistor and cap from ground and connected them directly to whichever value resistor I was testing, and from there the 'supposed' NFB resistor was connected directly to the speaker jack terminal which is tied to the 8 ohm tap on the OT secondary. What the hell am I doing wrong?

I do very much like the aggression of the amp without NFB, but it is just too crazy. I once owned a Dr. Z Mazerati GT and anything past 4 or so on the volume knob was almost unuseable (too hairy, blasted out beyond any reasonable limit, amp taking off into feedback the instant I took my hand off the strings, etc) and this reminds me of that. It's promising but frustrating.

If you want a no-NFB amp, you can decrease the gain of the PI

roberto, I like this idea best. Any advice on how to achieve this? Reduce value of PI plate resistors? Increase value of tail resistor? Increase value of PI cathode resistor? (it is currently approx 780R, an old carbon comp 680 that has drifted - I thought favorably but maybe not!)

[roberto]

Have you done something like this?
http://www.tubefreak.com/50501.gif

Can you draw a schematic?