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

[systemtruck]

Do you read 0V on both cathodes, when measured right at the tube socket? The 125 ohm cathode resistor I assume measures fine by itself?

Yes and yes. The cathodes are connected so it’s just a wire between the pair of socket pin 3’s.
Also, somewhat surprisingly, the 125R to ground measures fine WITH the 50uF cap in place too. I measured that before I snipped the cap. I thought for sure it would read as a short circuit since I figured the cap was dead short, but it measured 125R before and after snipping the cap. So I guess the cap is fine.
So the very notable thing here is that both tubes have 0V cathode bias somehow, but only one tube is red plating. I have basically zero knowledge about what could go bad inside a tube, let alone to cause this, so I’m basically in the dark. If I had a spare pair of EL84’s I would just pop them in and take a chance to see if that solved it.

 

[Matador]

What do you read on the other nodes (plate, screen, and grid)? If you remove the tube, do you read 125R between the actual tube socket cathode connection and ground?

Seems highly suspicious that you read exactly 0V at the cathode, as there's aren't any other internal tube elements which would be grounded directly in a typical beam pentode, unless you happen to be running DC heaters.

You said 'pin 3' so this is an EL84 amp?

 

[systemtruck]

What do you read on the other nodes (plate, screen, and grid)? If you remove the tube, do you read 125R between the actual tube socket cathode connection and ground?

Seems highly suspicious that you read exactly 0V at the cathode, as there's aren't any other internal tube elements which would be grounded directly in a typical beam pentode, unless you happen to be running DC heaters.

You said 'pin 3' so this is an EL84 amp?

Yeah the fact that I’m getting a precise 0V makes me think something is shorted mechanically, like a wiring mistake somewhere or something is rubbing up against another joint. But i’ve been through a visual inspection a number of times now. I’ll do it again though.
Yes EL84 amp.
It’s AC heaters.
And yes, i successfully read 125R between the pin3 situation and ground. That ALSO is confounding, because I’m sitting here thinking pin3 is accidentally shorted to 0V but I’m reading a correct 125R between pin3 and 0V, so clearly it’s not shorted.
So it’s like it’s own little floating 0V, not the main 0V. Super weird.

 

[ccaudle]

And would this one glow red where as the other EL84 would not?

The circuit configuration you describe sounds like what is known generically as a long tail pair, or differential pair, which is very sensitive to device balance unless there is DC feedback to force the devices to balance. Maybe pentodes are different, but a transistor long tail pair will very easily switch between all current through one device or all current through the other device without DC feedback, or degeneration resistors in the emitter/cathode to reduce gain and make the devices easier to balance.

By way, I find this section of the schematic a little confusing:


Typically a schematic will use a dot to make crossing nets unambiguous regarding whether they actually connect or not, like this:


Also, EL84 is a pentode, but I only see 4 terminals on your power tube symbol. Shouldn't there be another grid?

 

[systemtruck]

The circuit configuration you describe sounds like what is known generically as a long tail pair, or differential pair, which is very sensitive to device balance unless there is DC feedback to force the devices to balance. Maybe pentodes are different, but a transistor long tail pair will very easily switch between all current through one device or all current through the other device without DC feedback, or degeneration resistors in the emitter/cathode to reduce gain and make the devices easier to balance.

By way, I find this section of the schematic a little confusing:
View attachment 145604

Typically a schematic will use a dot to make crossing nets unambiguous regarding whether they actually connect or not, like this:
View attachment 145605

Also, EL84 is a pentode, but I only see 4 terminals on your power tube symbol. Shouldn't there be another grid?

Ah, i meant to add NODES! Yes it is terrible to read without nodes, my apologies. Will add in next version.

Hmm, that’s really REALLY interesting about imbalancing and it sounds like that’s what is happening…
I deleted the audio NFB in this circuit because the sound/drive is a huge improvement without it. So perhaps i need to implement one of the balancing solutions you mention. But…

I’ve attached the original schematic. That modification is not mine. It must be someone’s old modification that got uploaded. My 125R 7W / 50uF 25V cathode setup is what was installed in the amp when i took it apart, with pin 3s combined. I got new replacement parts but kept those values. But should i try doing what they did and have independent cathode setups?

And as you can see, Fisher drew only 2 grids for the EL84, so that’s why i only drew 2 grids.

One other difference in the hardware VS the schematic is that they used 68 R resistors to feed the rail to pin 9 of each EL84. The Fisher had them installed and they looked factory, so i kept them in mine and that’s a difference you’ll also see in my schematic vs this one.

How would DC feedback or degeneration resistors be implemented into this amp? And what are the benefits or either vs the other?

 

[Matador]

Also, EL84 is a pentode, but I only see 4 terminals on your power tube symbol. Shouldn't there be another grid?

In an EL84 the suppressor grid is internally tied to cathode.

 

[ccaudle]

In an EL84 the suppressor grid is internally tied to cathode.

Thanks, my pentode inexperience shining through.

 

[Matador]

As a test, you can replace your tube with ~6.8K, 10W resistor, placed directly in each tube socket between pins 3 (cathode) and pin 7 (anode). If your anode is at 300V, and you're expecting ~10V at the cathode, then you want to draw 40mA through your 'tube', so (300V - 10V) / 40mA = 7.2K. 6.8K is close, meaning you'll draw a few more mA than 40mA, but you should see something close the typical bias voltage at the cathode if everything is working correctly (except the tube).

Dissipation is ~12W, so you'll need to make your measurements fairly quickly and not leave it running forever.

 

[merlin]

Considering the OP started the thread with confused ideas about grounding, I wouldn't be surprised if he has mistakenly grounded the EL84 cathodes (or wired the EL84s up plain wrong), or has created a grounding error that makes it *appear* there is 0V on the cathodes because (e.g.) he is measuring between unconnected circuits...

 

[systemtruck]

Considering the OP started the thread with confused ideas about grounding, I wouldn't be surprised if he has mistakenly grounded the EL84 cathodes (or wired the EL84s up plain wrong), or has created a grounding error that makes it *appear* there is 0V on the cathodes because (e.g.) he is measuring between unconnected circuits...

Humbly speaking, as I am writing this directly to the author of a book I am waiting for the new release of, there is a very large difference between someone admitting they need help understanding the noise differences of 0V and Earth in a tube circuit (as is the purpose of this thread), and someone just plain connecting an unrelated signal to ground out of willful ignorance. Those are pretty different things.
A third possibility would be accidentally shorting this pin to ground by physical mistake, such as one joint slightly bending over and touching something nearby, something we all do at all levels of expertise and have to correct once in a while. I wish that were the case here and will happily admit that if it happens, and breath a sigh of relief that it’s not something more difficult to repair.
But this mystery is beyond that, as I measure 125R between ground and cathode, which is to be expected, yet the cathode carries 0V.
I don’t have time to get back in the workbench today but will update later in the week when I can do more prodding. I’m also going to hook up my variac from here on so I can stop pummeling my circuit with so much voltage during startup, unless someone here wants to suggest that it’s ok for these 450V capacitors to see 480V volts on a regular basis but not sustained for long.
But other than all this, the amp sounds incredibly quiet already. The bus bar method, using local reservoir capacitors, is very nice.

 

[merlin]

I certainly look forward to finding out what the issue is, because it's a head scratcher right now!
BTW you can use two caps in series, if you can't find 500V caps.

 

[systemtruck]

I managed to squeeze in some time this morning before work, and here’s an updated report…
After checking all the tube wiring and once again finding no issues, I hooked up my variac and brought it all up to voltage slowly, with the EL84’s swapped this time, and after some time, the cathode voltage rose to about 9V DC. It took like 20 seconds to even budge from a perfect 0V DC and a quite a while after that to slowly come up to 9V. I’m wondering if my 0V reading was from a time based issue… maybe i wasn’t letting the circuit fully come up before taking those readings. I certainly was operating under fear of damaging things and shutting it down quickly. That’s the most logical explanation here. Now that I’m bringing it up with control, it all had enough time. It’s pretty interesting to me though, that it could remain at a rock solid 0.000V for a while after power up, and not show a flicker of voltage right at power up.

The reason I swapped the EL84’s was to see if the hotter tube follows the circuit or if it follows itself. It does follow itself, so maybe it’s the tube. However i will say that the red plating appears less severe. But, after a few moments after power off.. the problem tube still is much much hotter than the other tube. So even if it wasn’t visibly red plating as much this time, it sure is running a lot hotter. I think it’s unlikely the red plating was less this time, i think maybe the lighting/angle in the room was just different today and i wasn’t seeing it as clearly. The screen printing and grey glassy stuff on the tubes hide a lot of the insides from one angle and i happen to have my sockets in opposite rotation, so rotating the tubes makes it much more difficult to inspect.

I then powered down and also reconnected the NFB path as original. The amp is less awesome sounding haha, but still healthy per se…. but the ugly high pitch farty clipping still occurs (and to further describe, it stops very suddenly once the guitar note decays to a certain amplitude), and the tube still runs hot. So I don’t think the NFB is causing some instability. Just like yesterday, a couple minutes after power down i could in theory remove the colder tube by hand but the hot tube I can’t even touch for more than a tap.

I do think maybe one tube is damaged. Could it have been over voltage from power up a bunch of times now? It could have just always been bad. I do vaguely recall some ugly distortion in this thing when i used it as original radio before starting this project. But i am not certain about that recollection. So i guess it’s worth buying a decent tested pair of EL84’s and just tossing them in to see how it goes. Or maybe just one? Anyways my common sense is leaving me at that conclusion. The fact that I’m getting a decent cathode voltage now points me to that too. Right?

 

[merlin]

Mystery solved! It's probably just mismatched tubes. This is exacerbated by using only a single, shared bias resistor. If both tubes had their own bias resistors (twice the original value) there would be some measure of self-balancing in the circuit. Moreover, 125R sounds like not enough for EL84s running at 360V, no wonder one of them is suffering. I think you need closer to 180 - 200 ohms (or twice that if you go for for individual bias resistors).

 

[Matador]

It’s pretty interesting to me though, that it could remain at a rock solid 0.000V for a while after power up, and not show a flicker of voltage right at power up.

You will measure 0V until the tube can establish it's operating current, which can take anywhere from 10-20 seconds from a cold start. There can't be any current through the tube until the cathode has reached a sufficient temperature that emission can take place (and the high positive anode voltage can start pulling electrons through the control grid).

Could it have been over voltage from power up a bunch of times now? It could have just always been bad.

Keep in mind that the EL84 is specified at maximum 300V anode to cathode voltage. A sudden high voltage on start up is far less damaging that operating the tube sustained at over 300V (and at the limit of dissipation).

Many guitar amp makers abuse this specification in order to squeeze more wattage from this tube, but it comes at the cost of tube longevity, and the need to pre-screen tubes at this higher voltage. Buying EL84's meant for use in Mesa Boogie's (probably the #1 abuser of EL84's) is probably your best bet.

Even assuming you are at 300V exactly at the anode to cathode limit, 100% max dissipation is at 40mA, which is exactly where your 125R cathode resistor is going to bias the tubes. Given you are likely operating closer to 350V this bias is over the max dissipation which is pretty abusive of this particular tube.

 

[Matador]

Dammit Merlin keeps beating me to the post.

 

[systemtruck]

Alright! At least I’m getting somewhere. Thank you fellas for the help.
Would it make sense that the ugly clipping I’m hearing has to do with this?
Maybe I’ll try beefing up that cathode resistor to 225R and even 325R. I have some high watt 100ohm’s on hand that I could add to the 125R. In theory i could also try 175 total by adding a 50ohm parallel pair of those.
Would increasing this value possibly let me test to see if i can eliminate the clipping, if it’s the imbalance-ness that’s causing the clipping? I’m guessing this re-biasing will just downshift the juice and will allow the hotter tube to be brought down into the non-clipping zone?
Side question, will the 9V go up or down when increasing the 125R cathode resistor?

It’s not the nice tube drive sound I’m referring to, it’s pretty similar to the sound of clipping the inputs of an audio interface.

I don’t quite have enough parts to create separate bias resistor/cap circuits, so it would be nice to be able to adjust the single setup to at least confirm this is the cause of clipping.

 

[merlin]

'Ugly clipping' is a bit vague. But since you have a cathodyne PI, check out the bottom of this page for one possible cause:
https://valvewizard.co.uk/cathodyne.html

Side question, will the 9V go up or down when increasing the 125R cathode resistor?

It should go up, to around 15 to 17V

 

[systemtruck]

'Ugly clipping' is a bit vague. But since you have a cathodyne PI, check out the bottom of this page for one possible cause:
https://valvewizard.co.uk/cathodyne.html


It should go up, to around 15 to 17V

I’ve taken DC measurements of everything in the PI and EL84 area, and have also made some sound recordings of the clipping. Uploading audio somewhere is a pain, but basically the clipping definitely occurs at the 12AX7 PI pin 2. (Side note, it also occurs when swapping in a known good 12AX7). It does seems to also slightly occur just prior at the PI pin 7 also, but to a much much lesser degree. And the further back you go, like prior to EQ and within the Preamp, it is not really clipping in this manner at all.
I also made a recording of the speaker output using a Weber speaker attenuator, and the clipping is basically the same as the PI pin 2 but is a lot brighter/clearer. The reason for this is my equipment though, it’s not the circuit. For tapping into audio within the circuit i am using an old tube signal tracer that i restored some years ago. It’s got poor high frequency response, but is very handy, and it’s the only thing i have on hand right now that has a high enough impedance to not load down the circuit too much. I recently threw away my analog scope and am shopping for a digital one haha. Point is, although my audio recording at PI pin 2 is much darker than speaker sound, the clipping occurs in both these points in the same exact severity in relation to amplitude.
All of this clipping occurs at low guitar volumes. It is not in breakup territory of the output stage at all. Once i get there, the clipping is sort of masked by all of that, for guitar cab purposes anyways. But these low volumes are maddening to hear.
Perhaps I’ll upload audio soon, but for now i have some readings to present. I noted the measured DC after the circuit fully settled down, and i also noted the Spec’d voltage that is listed in the Fisher service manual for each point.
There are some pretty notable differences between Spec’d and measured voltages. I wonder if any of this has anything to do with clipping.
On the “hotter” EL84, there is a 30mV reading of DC on the pin 2 input. On the other EL84, it is a perfect 0.000 V on pin 2 input.
On the 12AX7 PI, the rails are quite a bit higher than Spec’d. the cathode voltage on PI stage is also a good bit high.

I read that whole article you linked, and I’m ready to start trying to throw parts at this. Maybe just a 1M grid stopper resistor that follows C49 and hits pin 2 of PI?

 

[systemtruck]

Here’s the “ugly” clipping for some point of reference.
Pin 2 of the PI for 3 notes, then the speaker output for another 3 notes. Essentially the same clipping, but as mentioned in the prior post, the monitoring equipment imposed a dark loss of high frequency material so the speaker output will sound much brighter.
https://www.dropbox.com/scl/fi/1ozx...ey=trv3c95s3epx9xkvyhxzv978q&st=ggpkaw5c&dl=0