• 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.

Troubleshooting a restored Edison 12 valve amplifier

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

I just completed a restoration/repair of an old Edison 12 valve amplifier, designed by Erik Andersson (www.erikasson.se).

It was not working at all when I got it.

I have replaced a few components, including every electrolytic capacitor that was in there. All resistors have been checked and those that were out of spec or looked charred got replaced, the output tubes (4x EL84) got replaced with brand new Sovtek tubes.

Now, the amp works. It plays.

But there are some issues...

The same sound comes out on both channels. I confirmed with a left/right audio test. If I disconnect one of the RCA plugs, I only get sound for the remaining channel - but it plays in both speakers.

Another thing. The sound is very, very weak. The bass is almost non-existent, and the treble is very strong. If I connect a pre-amp and use "loudness", there's some bass, but the sound clips strongly. It never plays loud, even with maximum volume.

Another little thing is that it outputs some 50Hz hum into the speakers, not loud but hearable. It gets considerably worse when there's no source connected.

Comparing the amp to the schematic, it's not fully original. It has been modified in a few places. I've not yet detected all the differences yet, but here's something:

The 100R resistors of the output tube pairs are not connected to "black" on the output transformer, but rather - they're connected to yellow and orange, respectively. Triode connection, I think (this is the first tube amp I've worked on, certainly not an expert in the field).

Does anybody have any idea of what might be causing this? My initial suspicion was the ECC88 tubes (they are of unknown age/wear) and I tried switching them around, but that did not change anything. I also clipped off the "feedback" wires (at least I think they are), because they were connected to the opposite channel. No difference after cutting them off, though.

I think I will try replacing the ECC88s, but I would have to order new ones from abroad, which will take some time.

I've also read about a modification on this model, which involves disabling the first ECC88 altogether, by moving the input to the second. I'll report on what happens after I've tried that.

Any other ideas would be greatly appreciated. :)

Here's a schematic. It might not be 100% accurate, I had to touch up the original quite a bit because the text was barely legible. Click to zoom.

An externally hosted image should be here but it was not working when we last tested it.
One possibility is that both OPTs have been damaged by being used without a load. This might explain the weak and distorted sound.

It won't explain the poor channel isolation. That may be a wiring error - perhaps a missing ground connection. Double-check what you did. Triple-check whatever mods have been done by someone else! Try to draw a circuit diagram of what is actually there. Check DC voltages - post them here for comment. We can at least tell you what is plausible and what is suspicious.
Thank you for your reply.

That might be the cause, but I feel it might be unlikely. I do know the previous (and only) owner as well as the cause (sort of) of the failure. I don't think the amp's been used without a load.

What happened was that one of the 150R resistors started smoking, as well as the 100µF capacitor across it blew up (these components are between the two EL84 valves on the schematic). This was caused by a faulty valve. Replacing the valve solved that issue. The amp still worked with the faulty valve (with new resistor/capacitor), but only for a short while, before stuff would blow up again. That's how I discovered what was wrong.

The "mono sound" problem might have been caused by me, during the restoration, or it has been there all along, without anyone noticing it (it took a while for me to find it too, and that was purely by accident). I will try to re-wire the rca connections as well as moving the input signal. I will also try to document all the differences from the schematic that I can find.

One last thought: could it be possible that the two signals cancel each other out, making the sound weak?
Welcome said:
One last thought: could it be possible that the two signals cancel each other out, making the sound weak?
That would give the symptoms, but difficult to see how you could have achieved this accidentally unless the phase splitters have got entangled between channels. How are the ECC88 triodes assigned from envelopes to circuit positions?

There is a difference between most Western EL84 and the Russian equivalent. One pin (2?) is unused and not connected internally, versus being tied to another electrode. All OK provided that tag on the valveholder has not been used as a convenient place to mount a few components. Check this - I can't remember the details.
Okay, this was simple. Well, kinda.

Turns out, the thing really, really wasn't running right before. Apparently, (looking at the schematic) the point between "2" on the top (from the psu) and the 22K resistor was cut off, it wasn't easy to detect that because the board is in really bad shape. I soldered a wire across the trace and now, everything is different. (PS: the 15K resistor on the "1" point from the psu is actually connected to the "2" point (only). Is this bad?)

It now plays loud, really loud, even on minimum volume on the computer I'm using as a source. Bass is there too, and quite strong (too strong?).

So that is sorted. But now I have another problem. Before there was just a slight hum in the background, now that hum is stronger, and accompanied by a rather nasty hiss as well!

I might be able to sort that out, though. I am in the process of "reverse engineering" this heavily modded circuit, and will try to restore everything according to the schematic. Maybe that'll fix something. It seems really weird in íts current state, some things don't really make sense to me. I'll post my edited schematic when I'm finished.

I didn't have time to test and see if the left/right issue is still there, because I chickened out of the whole thing. I thought it was running hot before, but now is totally different. I have no clue about this sort of stuff, is it supposed to smell like it'll could burst into flame at any minute (note: there's no dust or dirt anywhere, I have cleaned it ultra-thoroughly)?

Would really like some input on that last one. :eek: As for the rest, I'll keep investigating!
Last edited:
It's funny that you report that the connection between "2" and the plate load to the middle triode was broken. First thing that made it to my mind was that you were missing somehow one of the PS filter caps which would allow coupling between the channels, but really did not explain the "weak" sound. Your finding does, I assume that the two 22K resistors (their "top") were connected together but not to "2". If that is the case, your lack of separation is likely fixed also.
The heat issue does need to be resolved quickly, as DF96 kindly posted. If a part is underrated or something is not connected as expected or if you have a failing component then extended operation must not be allowed.
Populating your schematic with voltages with no input but with a load would be helpful, along with the power ratings of all the resistors used. Do you have a scope? If so, please connect it to the output and make sure your amp is not oscillating at some ultrasonic frequency, that can be a quick killer also.
Best luck!
Hi guys, thanks a lot for your thoughts and suggestions.

I've not tested further since my last post, but here's some additional info regarding what you've said:

All the remaining original parts have been working up until now, so their voltage and temperature ratings should be okay. Of course, something may not be working right due to age and wear. I did replace many seemingly "critical" components, such as all the bigger resistors. I used appropriately rated components (metal oxide resistors, 700 VDC, 3 W - originals were 2 W). I also replaced the wiring with 700VDC / 125ºC, etc.

I tested the amp with a scope before I fixed the broken trace, and then there was no oscillation going on. It simply detected a very weak 50Hz hum (only a few mVs p-p). I'll have to try it again, I would like to have a look at the new post-trace fix issue, that hiss. Could that be caused by bad triodes?

I'll try to probe around for voltages, although the thought does make me a bit nervous. This is not a point to point wired amp, it has a regular PCB. I'll have to use my mini grabbers.

There are no bleeder caps across the main filter caps, but they still somehow drain out completely when you turn the amp off. All the other bigger (100µF) caps in the circuit are rated at 25 and 35 V only.
The first two stages, taken together, are running at very high gains, I'm estimating (just that, no calcs) the cascaded gain at some 500+. The feedback does, of course, reduce the overall gain of the amplifier but the gain within those 2 stages remains high. Perhaps you don't need that much gain, you've kind of alluded to that already.

You could place first a dead short between the grid of the first (input) tube and ground and note any effects; then do the same with the second triode, dead short to ground, and note the results. If the hiss is reduced substantially with the second shorted grid then the first stage is responsible, though that does not mean that it is defective necessarily, just could be excessive gain.
That being the case, it may be enough to remove the first cathode bypass cap to reduce that stage's gain, though you'd have to make sure that the rise in output Z is not going to be a problem.

First of all, left and right is left and right again, so that's sorted. :)

I have not yet done what you suggested, centraltexastubeguy. I will, though. But right now I have a much bigger issue. I finished comparing the amp to the schematic, and decided to try something, namely removing the new (as in: not on the schematic) capacitor that's sitting between pin 6 and 7 on the middle ECC88 triode. I also disconnected pin 9 on all triodes and let it float in the air.

Testing it resulted in the amp having the same hum and hiss problems, with a new addition - random honks. I turned the thing off and reversed my changes. Now it constantly sounds like a lawnmower or something (with the old hiss and hum still present in the background). Did I break it? How come reversing my changes did not restore the previous functionality?

About that schematic...

I've made a picture comparing everything that's different, it's highlighted in blue. Some components are missing, and some are added. Some have different values and some wiring's been changed. Some things aren't on the schematic at all, so I have no idea what it was originally, such as pin 9 on the middle triode being connected to the 270R resistor underneath it. Also note the two added capacitors with the bizarre codes W107 and 312 on them. These are small film caps but I have no idea what these values are supposed to be.

Diyaudio scales it down. Right click on it and select "view image" or "open link in new tab" for full size:

An externally hosted image should be here but it was not working when we last tested it.
Last edited:
ECC88 pin 9 should always be grounded (or connected to the most negative cathode). If not the internal screen can accumulate an electric charge which cause strange behaviour.

The changes you highlight are:
output in triode mode instead of pentode
lead-lag HF network removed from first anode and replaced by Miller cap on second stage - could cause (or cure) HF loop instability
lead-lag LF network removed from first-second coupling - could cause (or cure) LF loop instability
feedback network changes - could cause (or cure) loop instability and change gain

It seems to me you have two options:
- try to understand it and fix it as it is - but we don't know how experienced the modder was
- revert back to the original circuit before the mods were done - we can probably assume that it originally worked OK
I'd go with DF96's second option, and revert to original circuit - but including grounding that internal screen on pin 9 (this was probably done originally, but being a "routine" connection, just not shown on a basic circuit diagram - it may have just been connected to a central spigot on the valveholder if there is one and it's grounded). You may have got hold of an amp which someone decided to play with and tried mods that weren't successful, but then just got bored with it, and given up without bothering to undo his mods.
Last edited:
Actually, the amp worked well for many years, until it decided to blow up (presumably due to a bad valve). It was then left unused (and untouched, internally) in a basement for many, many years until I took it in for restoration.

DF96 mentioned western vs Russian valves. I actually replaced the original Philips EL84 valves with brand new (manufactured in 2011) EL84 / 6BQ5 Sovtek valves. I'll check if there's any difference here.

Oh, and another thing. The various stability mods that have been done might have been to suit the triodes that aren't original. These, too, are Sovteks - 6922 to be specific - and not ECC88 as was mentioned in the schematic. I have not changed these, so they are the same ones that were in when the amp was working, so many years ago.
Sorry for the delay in getting back to you, just got home from a day trip to Houston.
Your description sounds a lot like "motorboating". Most commonly, this malady is caused by interstage coupling thru the power supply itself. You reported earlier that you found a broken connection relating to power to the top of one of the 22K resistors. My suggestion is that you go back and check, minutely, every connection to any and all of the power supply caps, plus to make sure those caps are indeed good. The second comment about whether the caps are good is hard to reconcile with your having put it all back the way it was but the fact that you did rearrange wiring could have resulted in an inadvertent component disconnection.
The blue lines show the output tubes triode strapped while the non highlighted version shows conventional pentode connection. I can't quite make out the filter cap values and don't know why they are highlighted.
The twin triodes you are using were originally designed for VHF RF, i believe pin 9 is an internal shield to help isolate one section from the other. I would not think that shield terribly important for audio work but, can't assure it's not either. Can't see where it hurts to ground the shield.
Motorboating due to lost power supply decoupling is my vote.
Good luck!
Not sure, but look at the schematic of the mods attached to post #10 above. Isn't that mod to the anode HT supplies for 3rd triode a bit strange? I can't see what it could achieve, but it does seem to remove effective anode decoupling between triodes 2 and 3. I think that's the first thing I would restore. But, then, apparently this arrangement was working before the catastrophic incident that caused all these problems. I assume identical mods were made to both channels, but we don't know that - thus my question. Surely the same catastrophe didn't occur simultaneously on both channels - unless it was common power supply. I feel we don't have enough info...
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.