• 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.

Advice on refining SE EL84 pentode Amp

As I mentioned earlier I don't have any 2W resistors of the correct value for the anode to anode negative feeedback, which is the mod I'm intending to try first.

As you know the 100k feedback resistor dissipates some 0.4 W and 150k about 0.27 W, so you manage without 2 W resistors, at least when tweaking briefly to see if the result is what you want.

Any suggestions for potential improvements?

Cathode negative feedback is one possibility.

Putkivahvistinrakennusohje SE/84 Brachyera - Osa 2 | AudioVideo.fi
 
Last edited:
Thanks for the advice. I'll try some values this afternoon. The postman has been and gone but brought me no resistors, so I'll use the 1W parts until they do.

Regarding the cathode negative feedback - if I'm reading the schematic right, I disconnect the EL84 cathode resistor/bypass cap from the 0V and reconnect at the top of the OPT secondary?

I'm using the 8ohm tap, but I notice in the schem it's connected to a seperate tap. Is this essential?

Matt.
 
Here you can see the marked colors of the taps.
putkivahvistin-rakennusohje-brachyera-el84-se-vahvistinasteen-kytkentakaavio.jpg




Yes it is essential. It is maybe 16 ohms tap. (I assume you have Hammond OPT datas)

Hi,

Thanks.

Yes, I have the datasheet. The circuit in the schematic must be running a 5k primary, for 8 and 4 ohms on the yellow and green wires respectively.
I'm using 10k, so the white wire is actually 32ohm and the yellow wire is the 16ohm tap.
Can I use either?

What effect is the negative cathode feedback actually having on the circuit. I'm aware that negative feedback reduces gain and affects impedences, but I'm a little fuzzy as to exactly how. More reading on the subject is evidently in order.
It's good to know that something will work, but it's always better to know a little of the why :).

Matt.
 
You should also check if the 5k load impedance is better. 10k is at high side taking into account the operating point of the output tube.

You can try 16 and 32 ohms taps.

The negative feedback methods suggested to you will reduce distortion, extend frequency range, improve signal to noise-ratio and increase damping factor.

To accomplish all the above some amount of gain is traded off.
 
Hi,

I'll look into switching to the 5k tap, although I did design the circuit (such as it is) specifically for a 10k load. Measured voltages are somewhat higher than anticipated, so I'll take a look at the loadlines and see how a 5k load might affect things.

Single-Ended Pentode Power Amp.png

Above is the latest version of the schematic, you'll be pleased to see some changes have been made, (fixed the grid stopper, for one thing). The 220k resistor is doing a passable job of taming the mids. 470k and 330k were a little much.
I'll have a listen to some favourite pieces of music this evening and see how I like it.
Hopefully I'll have time next week to try out some negative cathode feedback as well.

Thanks for your help. I appreciate your taking the time to answer what are likely trivial questions, especially on a circuit I suspect may not be your cup of tea.

Matt.
 
I agree with Artosalo about trying a 5k OT tap.

If my calculations are correct, based on your most recent schematic, a 100k resistor for plate to plate feedback should provide approximately 10% feedback. This is the value that Schade says is optimum.

If you put resistors in series to get the desired value the power rating will be the sum of individual resistors power rating.
 

PRR

Member
Joined 2003
Paid Member
NFB monitors the Output of the amplifier and tries to make it "same" as the Input (only larger).

You really want this. Generally you un-bypass the driver cathode, and add a resistor from there over to the speaker output. If it howls real loud, reverse the OT primary leads.

No-NFB-at-all reeks of $9.98 table radios and a large class of snarly guitar amplifiers.

The real simple path is to wire the pentode as triode. Triodes have internal NFB. You will get a dmaping factor over 4 which is enough to tame the bass boom and rising highs.
 
If you put resistors in series to get the desired value the power rating will be the sum of individual resistors power rating.

Thanks.

So if I put, say, a trio of 33k, 1W resistors in series, I'd have the equivelent of a 99k, 3W resistor?

I'll recalculate my loadline for a 5k load when I get a minute. I'd been expecting a lower B+; 10k seemed like it would offer a more linear response with the expected voltages, but since the consensus says 10k is on the high side I'll certainly try halving it.

What difference would you expect switching to a 5k load to make?

Also, is the 100k anode to anode resistor optimised for a 5k load?

Matt.

NFB monitors the Output of the amplifier and tries to make it "same" as the Input (only larger).

You really want this. Generally you un-bypass the driver cathode, and add a resistor from there over to the speaker output. If it howls real loud, reverse the OT primary leads.

No-NFB-at-all reeks of $9.98 table radios and a large class of snarly guitar amplifiers.

The real simple path is to wire the pentode as triode. Triodes have internal NFB. You will get a dmaping factor over 4 which is enough to tame the bass boom and rising highs.

Thanks.

I omitted any kind of NFB not because I didn't want it but because I don't know enough to correctly implement it, hence this thread!

I'm quite fond of ten dollar radios and snarly guitar amps, but there's a time and a place for such novelties and I recognise this is neither :).
I'm listening to a few well played favourites as I type, it seems what I originally took for strident upper mids is actually an aggressive treble response, which is not tremendously pleasing.

I'm thinking that the general consensus is to switch to a 5k load, and apply global negative feedback? But how do I know how much NFB to apply?

Regarding triode strapping the EL84 - I'm all for trying it, but the datasheet quotes (from memory) 2W max dissipation in triode mode, whereas I'm dissipating over 3W at peak power. Is it safe to switch to triode mode without making other modifications?

Matt.
 
I notice that you have more than 300V B+. The maximum recommended quiescent voltage from the EL84 cathode to the plate is 300V, and from the cathode to the screen is 300V.
Suppose you have + 12V self bias on the cathode. Then the maximum voltage to the plate and the screen is 312V. If the self bias is 8V, the B+ is 308V, etc.

Triode wiring is with a resistor from the screen to the plate. You could use the 1.5k resistor you have just for the initial try, but I would really recommend using a 100 Ohm resistor there.

A typical implementation of "global" negative feedback would be from the output transformer secondary to the cathode circuit of the driver tube (feedback resistor from a secondary tap, to the top of a resistor that is between ground and the bottom of the parallel combination of the driver cathode bias resistor and cathode bypass cap.

Global negative feedback is much more tricky to do properly than the plate to plate feedback you have already tried successfully (and it made it sound better, right?).

Global negative feedback is also a lot more tricky to use than to do a triode wired configuration. That is easy to do properly.

Cathode feedback is fairly easy to implement. But like Global negative feedback,
it does include the output transformer in the feedback loop.

With the tube in pentode mode, I recommend the plate to plate feedback you already tried.

With the tube in triode mode, I would just do it without the plate to plate feedback.
You could also then try cathode feedback in triode mode, but the gain will be even lower.
I did this once with an EL34. You connect the bottom of the self bias resistor and bottom of the bypass cap to one of the output taps. The feedback gets larger as you move up to the taps that are 'further away' from the black ground transformer wire.
 
Last edited:
You're misreading the datasheet. 2 W. is the power you get from a Class "A" triode wired EL84. The max. anode dissipation for an EL84 is 12 W.

Thanks.

I think I get it now. The anode can (and probably does, single-ended) dissipate 12W of power irrespective of whether the valve is operating as a triode or pentode. The 1.5W and 5.7W figures are the expected output powers, triode and pentode respectively.
Connecting as a triode I'll loose some output power, but the valve will still be safely within it's limits.
That being the case, I'll give it a go.

Matt.
 
I notice that you have more than 300V B+. The maximum recommended quiescent voltage from the EL84 cathode to the plate is 300V, and from the cathode to the screen is 300V.
Suppose you have + 12V self bias on the cathode. Then the maximum voltage to the plate and the screen is 312V. If the self bias is 8V, the B+ is 308V, etc...

Thanks.

There's a lot to take in there, but I think I follow most of what you're saying. No time for tinkering today but I'm keen to try out some of your suggestions and will hopefully have a chance next week.

Matt.
 
So long, and thanks for all the fish...

Good afternoon all,

I've spent the last week or so digesting the information offered, and have tried most of the suggestions made.

It'll probably come as no surprise to most of you that the 'best' (ie. most musical) option has been to triode strap the EL84 and apply some cathode feedback.

Final schematic below, with measured voltages:

Single-Ended Triode Power Amp.png

There are some slightly hooky values in there - particularly in the mains filtering section - resulting from my using the parts I had to hand to get (close to) the voltages I was after, having taken 6A3sUMMER's advice and dropped anode voltages below 300V.

I auditioned the anode to anode negative feedback for a week or so. 120k did a reasonable job - especially after I dropped the output primary load to 5k - but the mid range was still weak.
Trying the triode mode with the 5k primary had a marked improvement, but adding some cathode feedback on top of that has balanced everything out nicely. There's still enough output power available to get it loud enough that I worry about upsetting the neighbours, without distorting.

It's interesting (although perhaps not surprising), that my final circuit has ended up so similar to other tried and tested designs. This leads me to a few conclusions:

1: I could have saved myself and everyone here a lot of time by simply following an existing design ;).
2: While there may be infinite theoretical paths from point A to point B, only a few of them make any sense to follow, and the shortest distance is likely still a straight line.
3: I'm not quite so terrible a mathematician as I thought I was. Hope springs eternal...:)

Hearty and heartfelt thanks to everyone who took the time to chip in and help out along the way. Without your input I'd likely be lumbered with a pile of smoking remains, or at best an expensive $10 radio ;).

Matt.

Ps. Plans for the future include upgrading the output transformers and perhaps switching to EL34's. Also, voltage regulation is on the list of things to learn more about.
 
The voltages and resistors around the ECC81 can not be true as you have listed them.
9.8V across a cathode resistor of 470 Ohms is 21mA.
21 mA across the plate resistor of 47k Ohms is a 987V drop.

Check the values of the ECC81 cathode and plate resistors (after you turn off the amp,
and make sure all the power supply capacitors are completely discharged).
And that brings up another safety point: did you put bleeder resistors in your B+ power supply, and did you calculate how long the resistance you used will take to discharge the capacitors?
 
Last edited:
The voltages and resistors around the ECC81 can not be true as you have listed them.
9.8V across a cathode resistor of 470 Ohms is 21mA.
21 mA across the plate resistor of 47k Ohms is a 987V drop.

Check the values of the ECC81 cathode and plate resistors (after you turn off the amp,
and make sure all the power supply capacitors are completely discharged).
And that brings up another safety point: did you put bleeder resistors in your B+ power supply, and did you calculate how long the resistance you used will take to discharge the capacitors?

My apologies, that's a typo. The voltage on the cathode of the ECC81 should read 1.8V.

1.8/470 = 0.0038mA
0.0038x47,000 = 180V

Those figures seem a little more plausible!

Single-Ended Triode Power Amp.png

Regarding bleeder resistors, I did not go to the trouble of fitting them. The thought never occurred to me. One to go to the top of the 'to do' list. That said it's probably a little late now, since the work is done. I'll leave well alone until I'm back 'under the hood', as it were.

Matt.
 
Last edited:
A possible bleeder:
A 30k Ohm, 10 Watt resistor will draw 10mA from 300V.
That is 3W in a 10W resistor, hot, but OK.

You have 487uF B+ filtering on your schematic.
470uF x 30k Ohms = 14.6 seconds time constant.
The discharge rate is 63% of the voltage per 1 time constant.
(or 63% of the remaining voltage in 1 more time constant, etc).
300V x .63 = 189V 300V - 189V = 111V after 15 seconds
111V x .63 = 70V 111V - 70V = 41V after 30 seconds

If after you remove power, and you take 45 seconds to take the screws out, and remove the bottom plate, you may be safe
(but only if the bleeder resistor and other circuitry is all intact).
The 3 B+ capacitors are separated by 2 dropping resistors, so if any one of them fails, the other cap(s) may not be discharged by the bleeder. That is one reason to use larger wattage ratings on those resistors.

If you use a stereo amp, and 2X the capacitance, it will take twice as long to be safe.
 
Last edited:
That is one reason to use larger wattage ratings on those resistors.

Thanks,

You think it's a good idea then, to up the wattage of the power supply resistors? I've noticed they're running quite hot, but unless I have my numbers wrong, at 5W, they're well within the nominal 'safe' limits. I had been thinking about replacing them with 10W wirewound, mounted on the chassis, but space is tight.
Top and bottom plates are 1.5mm perforated alloy sheet, so the air flow is reasonably unrestricted, as it stands.

With such a thin chassis, I'm concerened using it as a heatsink will cause more problems than it solves. It seems likely that rather than radiate the excess heat into the air, it might simply conduct it through the chassis itself, heating up areas which had previously been running cool...

DSCF4363.JPG DSCF4364.JPG

There's quite a lot going on in a relatively small space. Insufficient planning has already caused it to end up a bit of a rats nest, in places. Squeezing more parts in may be a challenge!

Matt.
 
Last edited:
Improving performance...

Evening all,

Having lived with the amp for a couple of months, it's become increasingly obvious that it's under achieving :).

I've ordered a pair of full range OT's (Hammond 1628SEA), which should go a long way towards fixing it's inherent defects, but I'm also hoping to refine the circuit.

The most obvious starting point would seem to be to switch back to pentode mode, and apply some negative feeback from the OT secondary to the driver cathode.

I've been reading up on the subject, and I think I've finally got my head around the basics. My first question is, how much NFB should I apply?
I'm aware that excessive feedback can cause instability, and that around 6dB of feedback will increase the bandwidth by an roughly octave either way.

With that in mind, I've redrawn the circuit to include 12dB of NFB, both with the driver cathode bypassed and unbypassed:

Single-Ended Pentode Bypassed 12dB NFB.png Single-Ended Pentode Unbypassed 12dB NFB.png

Do those values seem right?

If I chose to go unbypassed, can I get away with less feedback (say 6-10dB), since there's already feedback inherent in the design?

Also, most circuits add a low value cap in parallel with the feedback resistor. Is that essential, and how should I calculate it's value?

Matt.