Tube Amp connections : 0V vs EARTH vs HEAT vs CURRENT

[systemtruck]

Tomorrow I will try a test from the opposite perspective, which is to just test the isolated performed of the output stage…. I’ll connect a balanced audio output from an interface to some caps and then to the grid inputs of the EL84’s. I’ll leave the grid stoppers in place too, which are at 82k right now.
Will this work ok with impedance relationships? The audio source would probably only have like 50ohms, but those grid stoppers add 82k. Is there some impedance interaction with previous stage that the EL84’s depend on to function?
Maybe I can successfully get the output to perform as it should, and if it does and the audio sounds as one would expect, then I think further reinforces the idea that the preamp and PI sections are being distorted simply because they’re losing their rails quite dramatically.

 

[Cqwet Dbdfte]

82k "grid stoppers"?
That is not a grid stopper. Try a typical value.

 

[systemtruck]

82k "grid stoppers"?
That is not a grid stopper. Try a typical value.

I’ll bump them down.

 

[systemtruck]

I ran a test of just the output EL84/XFO stage. I was able to provide an additional 42dB of gain after the +4dBu audio interface output, using two audio transformers in a somewhat weird fashion. The audio passed clean through them so i know it functioned enough for this basic test.

I think it was just enough volume to get the EL84/XFO to clip in the ugly fashion. It was actually close, i think my levels were just on the verge of being as loud as the PI output level required to clip the output stage. So, comparing the output of interface, which I patched back in to check, the EL84/XFO stage was clipping when the interface output was not. There was only like 5dB of room to play though, before the interface itself clipped at its output.

But, i was able to get the EL84’s/XFO to clip. I switched up the EL84’s since I have 4 of them, and it clips just the same.

Anyways, something is up. The output stage clips, and it draws too much current I believe. So i disconnected the EL84’s again, and measured impedance from the top and bottom legs of XFO to the center tap. Each side is almost exactly the same, about 164 ohms.

Also, in the recent past i already confirmed that the HV coming out of each leg is identical, both with the EL84 connected and without it connected. Without the EL84’s, it’s much higher of course, up in the 400’s instead of mid 300’s.

I could try buying yet another set of EL84’s. Or i could try hooking up a different transformer. But if i need a different transformer it’s getting to the point where I’ll have built an entirely custom amp with no original iron haha.

Oh well, sometimes you just don’t win! Maybe it’s time to give up and continue working on builds that have gone perfectly recently..

 

[CJ]

Are you converting this thing into a guitar amp?

If so why not build a VOX AC 15?

That is a wonderful transformer for guitar, it is what I use in a Princeton Reverb.

I think,

https://groupdiy.com/threads/fisher-x-100-3-output-transformer.73750/#post-1069456

 

[ccaudle]

The output stage clips, and it draws too much current I believe.

This is where an oscope would tell you right away what is going on.
Presumably you have changed the gain structure of the amplifier compared to the original hi-fi configuration. Have you calculated or measured what the amplitude of the PI stage is, and checked what that level will do going in to the output stage? Is it just pushing the grids positive and getting into grid distortion?

 

[systemtruck]

Are you converting this thing into a guitar amp?

If so why not build a VOX AC 15?

That is a wonderful transformer for guitar, it is what I use in a Princeton Reverb.

I think,

https://groupdiy.com/threads/fisher-x-100-3-output-transformer.73750/#post-1069456

This is where an oscope would tell you right away what is going on.
Presumably you have changed the gain structure of the amplifier compared to the original hi-fi configuration. Have you calculated or measured what the amplitude of the PI stage is, and checked what that level will do going in to the output stage? Is it just pushing the grids positive and getting into grid distortion?

All right folks, I have one piece of hard evidence that i didn’t have before, so i could use some expert detective work! …

I separated the cathode biases of the EL84’s. I used a fudged set of 340R high watts, and 50uF….

One of the tubes, an older one that did NOT red plate, reads about 0.04V DC on the cathode no matter what. When I swap the same two tubes from one side to the other, that same tube still reads about 0.04V DC at the other cathode now and the opposite is now healthy. It follows this tube. Again, this was a NON red plating old original Mullard.

All 3 of the other tubes read a healthy 9.5V-10V at its cathode no matter which way they’re installed.

Interestingly, when the cathode biases were unified as one segment before… i was getting heathy DC readings in the ballpark of 9V-12V. So, does this mean that one tube was constantly pulling voltage upwards or something like that? To compensate for the 0V tube constantly pulling down?

So i was able to then pick two tubes that provided a good cathode reading of 9.5V ish. But still, the distortion occurs. I had high hopes that when i got two tubes holding good cathode bias that it would solve the distortion.

Maybe this solves red plating, haven’t tested that yet.

And is it possible that maybe this situation has already damaged the other EL84’s so i just need a fresh set again? A new pair of JJ’s are 30 bucks right now, so maybe I’ll just order those to have a clean slate. I do worry about max voltage though, in the new tubes vs NOS.

Yes, I’m building a guitar amp. It would be done weeks ago if not for this. I’m starting to think that maybe the output stage just needs to be adjusted for this context instead of Hi-Fi. It already sounds fantastic in lower volumes, just a warm round rich clean tone. I don’t quite understand how i could be causing grid distortion if I’m just sending sound into the EL84’s at low to moderate volumes. It’s really not very loud in the room when it clips.

And yeah I would be super bummed to find out that the output transformer is fried, because i knew well that this generation of Fisher output transformers was going to be terrific which is why i started this project. But my rudimentary transformer DMM tests keep coming back fine. I tested the secondary too, which reads around 0.7 ohms DC and as far as i can tell that is pretty normal. And considering the fact that audio passes beautifully at lower volumes, it doesn’t seem that the OT is fried. Fingers crossed.

I wish i hadn’t just thrown out my old analog scope. It was just too faulty to deal with and i’ve been shopping around for a digital one. But even just listening to the audio in the same exact point in circuit, one time with the whole circuit connection in tact and another time with the PI outputs snipped and recorded from there, the difference is alarmingly clear. This is the audio test just posted yesterday.

Pardon my newbie level of tube stuff but here’s a stupid question… but what would you try changing about the EL84 design to see if gain/design of this stage is the issue? I’ve tried running the varic lower so that the B+ is only 300 not 340, but that doesn’t solve it either.

 

[ruffrecords]

The fact it reads so low indicates it is not passing any current which explains why it did not red plate. Sounds dead to me.

Cheers

Ian

 

[ccaudle]

i was getting heathy DC readings in the ballpark of 9V-12V. So, does this mean that one tube was constantly pulling voltage upwards or something like that? To compensate for the 0V tube constantly pulling down?

That is not how it works, one tube isn't "pulling down" anything. The two tubes are sourcing current, and voltage across a resistor is proportional to current through the resistor (Ohm's Law, V=current x resistance). If the resistor was sized so that current from both matched tubes added together and created a 10V drop, if one tube is not sourcing any current and the voltage across the resistor is still 10V, that means the one operating tube was supplying double the designed level of current.
Having the separate cathode resistors makes that immediately apparent, which is why it just seems like a better idea to me. With the resistors bypassed by capacitors it does not a change AC operation, but makes it possible to verify biasing (and will be obvious right away how closely the tubes are matched by how closely the cathode voltages match, assuming you use precision or matched resistors).
I'm not sure where the current was coming from that was being sourced into the grid which kept the grid voltage rising when you had the cathode voltage which started to continually increase. Maybe that is a separate issue, or maybe the grid leak current is proportional to plate current, so when the plate current was double the grid leakage current was higher as well.

 

[systemtruck]

That is not how it works, one tube isn't "pulling down" anything. The two tubes are sourcing current, and voltage across a resistor is proportional to current through the resistor (Ohm's Law, V=current x resistance). If the resistor was sized so that current from both matched tubes added together and created a 10V drop, if one tube is not sourcing any current and the voltage across the resistor is still 10V, that means the one operating tube was supplying double the designed level of current.
Having the separate cathode resistors makes that immediately apparent, which is why it just seems like a better idea to me. With the resistors bypassed by capacitors it does not a change AC operation, but makes it possible to verify biasing (and will be obvious right away how closely the tubes are matched by how closely the cathode voltages match, assuming you use precision or matched resistors).
I'm not sure where the current was coming from that was being sourced into the grid which kept the grid voltage rising when you had the cathode voltage which started to continually increase. Maybe that is a separate issue, or maybe the grid leak current is proportional to plate current, so when the plate current was double the grid leakage current was higher as well.

That makes sense, thank you for that.
I got lucky with power resistor values I had on hand, and was able to make one be about 342 and the other be about 340. Or something like that. I know it’s less than 1% or pretty close, just by luck.
When I use my NOS “matched pair”, I get my most even cathode reading. But the cathode DC runaway train upwards still occurs, after quite a long time, and I’m able to measure them each and they have the same behavior…. After power up phase they both slow/settle down in the mid 9V range, but don’t truly stop but do come close to stopping . They keep crawling and pick up speed more and more as they approach 13V and at that point it’s going at about .5V per second so I know it’ll keep going up. It takes minutes for this to occur after the 9V range was settled into.

Interestingly, neither one really got to red plate status. I think I saw some pink start. Im guessing that they both would have soon after I shut it down. I’m guessing the fact that they are pretty balanced means they’re sharing the work so red plating will take much much longer.

And just like in all my other tests (except for the pair of cold tubes, one of which has now been deemed dead), the B+ collapses at the same rate as above.

It’s funny to think that I should just build an AC15 because I took a look at the schematics and it’s almost identical to mine, ignoring the preamp. The long tailed pair is almost identical to my new PI I just converted to. It has an extra tremolo circuit on the other unused grid, where as mine is the standard cap to ground there. But other than that it’s pretty much the same with minor value differences. The 470k’s in mine are 1M in the AC15, which I guess is notable.
Then the output stage is nearly identical too, except that grid leak resistors are mid 200k instead of the 470K ones in mine. I think that’s an important change I should make, especially since I believe the datasheets suggest a max of mid 200Ks. The B+ of AC15 is 310 and mine is 340. So I’ve been brining mine up to only 310 today just to see if that ever helped and it doesn’t make any difference.
So I kind of already have an AC15, using Fisher transformers and Mullard tubes.
The very very first AC15 design had 100r resistors on the plates and no resistors on the grid2’s. They swapped it on all following designs. I wonder if the first one was a drawing mistake. There is a mistake in the long tailed pair, so maybe so.

I ordered a pair of JJ EL84’s. Very affordable. I’m thinking that if my issues continue with those, I just have to give up on the output transformer and try another one that I pull from an amp I don’t use much. I just can’t see any other cause at this point, for all the distortion and the voltage collapse. It’s really weird that the one pair of tubes that didn’t collapse were the two cold ones, one of which is dead. Could that be a clue?

 

[ccaudle]

give up on the output transformer

At DC the transformer is just providing power to the plate resistor, there isn't much that can go wrong if the primary winding isn't open or shorted.
The unstable bias seems like the source of the problem, but I don't see how that is related to the transformer.

 

[CJ]

If your pulling your power supply down and getting distortion then that points to ultrasonic oscillation.

You need a scope, or you can try moving wires around with a chopstick and see if it goes away. Is there negative feedback coming off the OPT secondary? Try disconnecting it and see if there is any change.
.