Power tube loses bias at full output.

[leadfootdriver]

If anyone would know, it would be the big brains on this board!

A fellow Trainwrecker was monitoring the bias on his Express clone and noticed that the 1st power tube swings high on the bias but the 2nd power tube loses bias. Is this normal?

Here's a vid of the situation and thanks!

[g1]

Bias is for idle current. Normally we don't look at the current under signal conditions unless there is a problem we are trying to find.
Since he's looking anyway, swap the tubes and see if the problem follows the tube or stays with the socket/probe.

[leadfootdriver]

Thanks.. He switched them and the result was the same. It must be the way that amp responds cuz it's in full power clip to get it's distortion.

[Chuck H]

Thanks.. He switched them and the result was the same. It must be the way that amp responds cuz it's in full power clip to get it's distortion.

If by "the same" you mean that the bias shift remained with the socket and didn't follow the tube I'm actually surprised. I could reconcile the tubes actual dissipation being grossly different in heavy clipping. I've seen this a lot in modern tubes. They're matched at idle and that's it. What they do when they're actually amplifying, when it matters, is all over the map even with "matched" tubes now. I don't think there's anything about the circuit that should cause that phenomenon.

[pdf64]

It would be interesting to scope the wave along the amp's signal chain.
I guess that when overdriven it gets asymmetrical, and the output mark space ratio would be rather uneven.
I think it was the Express that had its overdrive characteristics exhaustively examined and analysed in a thread over on TAG, maybe by katopan.
It turns out that it's one of the very few amps where, under sustained overdrive, the LTP clips, rather than the power tube grid, on one of the p-p 'sides' at least.

[Chuck H]

Pete has it. I woke up with the answer. Must have been mulling it over n my sleep

The offset begins with the PI. The LTP PI clips asymmetrically. There's also an impedance difference between the inverted and non inverted PI outputs under heavy clipping that makes the time constant unequal for coupling cap discharge so one power tube grid DOES end up biased colder. This is probably exacerbated by the third stage cold clipper which only clips on the same cold swing of the PI and the colder side power tube grid while the other swing is free to keep driving harder. I suppose I'd have to call this behavior "normal" then. But I'd look on a scope just to see if the duty cycle looks the "normal" amount of uneven. Indeed, different PI tubes and different power tubes could balance things a little, but never completely because the circuit IS imbalanced WRT it's intended operation. That is, clipping hard most of the time.

In this light it might be a good idea with those amps to swap the power tubes occasionally for more even wear. I'd certainly qualify the clip as sounding right for the design

I linked a site that has the scope shots:

Trainwreck Express Clone Oscilloscope Photos

[J M Fahey]

What kind of PI does it have?
LTP tends tobe quite symmetrical; cathodine is horribly unsymmetrical as far as current driving capability, since one output comes from a plate (high impedance), the other from a cathode (very low impedance).

Performance as seen is horrible, but since a tube guitar amp is not an amplifier but an effects box, anything is possible.

Sound is good, who cares the amp emits cold neutrons and sterilizes anybody within 20 yards?

FWIW: current rising a lot is normal in an AB class amplifier; staying the same is not and indicates tons of grid rectification.
Signal out must be quite assymmetrical, we *deserve* a scope capture, showing just a couple needles is teasing followed by leg crossing.

[Chuck H]

IMHE the LTP PI can become asymmetric under heavy clipping. I can't site the mechanisms on a technical level, but I expect it's for the same reasons you usually see different plate loads on the circuit as an attempt to pad output from the inverted side. Local NFB in the circuit is always trying to rebalance things, but maybe since the inverting side already has an advantage the circuit can become more asymmetric when the clipped signal no longer matches the input signal. But I'm just guessing on that.

I DO know that I've done experiments to balance LTP circuits, and therefor the drive to the power tube grids. With the circuit I kludged up I was able to adjust symmetry on the fly. I wouldn't say balancing did the actual tone of the amp any justice. Different maybe. Not better or worse. Since the asymmetry is part of the TW sound any such attempts at balancing the currents would likely ruin the tone.

[nickb]

It would be interesting to scope the wave along the amp's signal chain.
I guess that when overdriven it gets asymmetrical, and the output mark space ratio would be rather uneven.
I think it was the Express that had its overdrive characteristics exhaustively examined and analysed in a thread over on TAG, maybe by katopan.
It turns out that it's one of the very few amps where, under sustained overdrive, the LTP clips, rather than the power tube grid, on one of the p-p 'sides' at least.

All true but I find it hard to imagine that a tube producing useful output power has a lower average current than when idle. That is what the meter in the OP is saying. Given the peak current is likely a few hundred mA and the waveform looks like a pulse, the mark-space ratio would have to be <1:10 to make this happen. I'm not buying that is normal.

Do what Pete says a put a scope on both grids on the power tubes and let's see what's really going on. Anything else is just speculation.

[g1]

Also flip the bias probes around to make sure there is not some issue with the probes themselves.

[leadfootdriver]

Thanks for all the replies! I had my guy switch the bias probes, and the outcome was the same. So it's unique to this amp which is a Ceriatone build.

[nickb]

Here is a set of (somewhat idealized) waveforms I'd expect you see as the input is raised from just clipping to full blown hellfire level drive. The top trace shows an approx 2:1 ratio in average currents in the two EL34s. The asymmetry happens due to grid current flowing in the LTPP input grid. You'll notice that both of them go UP, not one up and the other down as the drive is increased.



PS (edit)

.and here is what happens if you change the PI grid resistor from 1meg to 180K. You completely change the character of the amp to be much more symmetrical.

[g1]

In the absence of a scope, something to try.
Use a sinewave at the input. Set the amp up for a basic clean sound.
Slowly bring up the level of the sinewave, the current shown on the meters should increase on both sides.

Have you checked the screen resistors?

[J M Fahey]

Loudthud was kind enough to link a Trainwreck schematic for me, and no, it´s not a cathodyne but a very conventional LTP inverter followed by conventionally coupled power tubes, so whatever it is, it´s not coming from there.
There *is* a cold biased gain stage before it which does clip assymmetrically, but it´s way too early in the chain to have such a big influence on the power tubes
Have we checked that the difference stays in the socket or follows the tubes?

[Pedro Vecino]

That behavior in vumeters seems completely abnormal. The impact of standby on one of them, too. On the other hand, I am the only one who hears the sound of ripped paper? With long, sustained notes/chords would be better appreciated.
I d´ont know that model, but I would try the following:
- Make sure the preamp output is connected to the appropriate input in the PI.
- Check their PI resistors values again (with multimeter).
- Disconnect the feedback.
- Retest inverting output transformer primary to identify abnormal operation in the transformer (if so)

[nickb]

The impact of standby on one of them, too.

+1
That's a great observation. That has to be a big clue.

[loudthud]

TrainWrecks are prone to oscillation. Even with the "approved" layout, lead dress can be a problem. Without a scope, you might never find it. Any odd behavior makes me suspect an ultrasonic oscillation. It's possible it only oscillates when a signal is present.

[Mike Sulzer]

What kind of PI does it have?
LTP tends tobe quite symmetrical; cathodine is horribly unsymmetrical as far as current driving capability, since one output comes from a plate (high impedance), the other from a cathode (very low impedance).

Output impedance and current driving capability are not the same thing. In general, you cannot get more AC current out of a cathode follower than a common cathode circuit. The significance of the low impedance is that the voltage follows accurately and then clips suddenly when the limit is reached. This is why I do not like cathode followers in a guitar amp except in situations where they cannot overload: the clipping onset is too rapid.

The situation with a cathodyne is even more complicated since the two outputs have what some people call "identical differential output impedance" (if I have the terminology right).

[Merlinb]

Using this layout: http://www.ceriatone.com/ceriatone/w...nCeriatone.jpg
Looking from the back, the left tube would be the one driven by the LTP's inverting output. With a warm-biased LTP the duty cycle from the inverting output will increase under overdrive, i.e. per cycle it spends more time positive than negative. This will drive the corresponding power tube harder, and vice versa for the other tube. This is indeed the trend we're seeing in the video. However, the effect appears to be quite extreme in this case, perhaps compounded by the GNFB loop whose influence I can't figure out in my head just now!

[DRH1958]

On the other hand, I am the only one who hears the sound of ripped paper? With long, sustained notes/chords would be better appreciated.

I heard this as well. Listen at 1:19. I wanted to post this yesterday but my computer crashed.

[Chuck H]

... The impact of standby on one of them, too...

I think we need to revisit this. So far the explanations have been excellent. But they continue to focus on the signal conducting behavior of the tubes. Here there is evidence of a problem that isn't signal dependent.

[Malcolm Irving]

Have you checked the screen resistors?

I like this suggestion. Was it followed up? If the screen is 'floating' on one of the tubes all bets are off.

[nickb]

I like this suggestion. Was it followed up? If the screen is 'floating' on one of the tubes all bets are off.

It's a great idea but got lost in the noise. I think we need some measurements from the OP now...

[Merlinb]

I think we need to revisit this. So far the explanations have been excellent. But they continue to focus on the signal conducting behavior of the tubes. Here there is evidence of a problem that isn't signal dependent.

It doesn't strike me as unusual. When you power on, the inverting triode will be turned on hard as the input coupling cap charges up, but the non-inverting triode will be held cold until the cap on the 'unused' grid charges up. This in turn means the output coupling cap from the non-inverting side will charge up faster, turning the corresponding power tube on harder and quicker, while the other side takes a little longer, as we see in the video.

[Chuck H]

So it's like I said in post #6 ?

There's also an impedance difference between the inverted and non inverted PI outputs under heavy clipping that makes the time constant unequal for coupling cap discharge so one power tube grid DOES end up biased colder. This is probably exacerbated by the third stage cold clipper which only clips on the same cold swing of the PI and the colder side power tube grid while the other swing is free to keep driving harder.

How about this from post #8 ?

Local NFB in the circuit is always trying to rebalance things, but maybe since the inverting side already has an advantage the circuit can become more asymmetric when the clipped signal no longer matches the input signal. But I'm just guessing on that.

I'm not being snarky. I actually just want to see if I got this right.

In any case... It's been reasoned so far that the behavior is probably normal? If that's the case I'll reiterate what I said before. It's probably a good idea to swap the power tubes every now and then in these amps since they pretty much live at a full grind.

My logical brain is siding with Nick in post #12. That is, I wouldn't expect the current to actually decrease in the weaker tube. But unless the amp is bench tested there isn't likely to be any definitive answers.

[Mike Sulzer]

It's a great idea but got lost in the noise. I think we need some measurements from the OP now...

"I like this suggestion. Was it followed up? If the screen is 'floating' on one of the tubes all bets are off."

If one screen was not supplied with the correct voltage, I think the idle current would not be right.

My interpretation of the video is that as soon as the output is large enough to move the operating point out of class A into B the right hand tube goes bananas. As suggested, oscillation is a real possibility. But there are some simple things to try . New tubes? Swap the output tubes?

Then lots of measurements.

[g1]

If one screen was not supplied with the correct voltage, I think the idle current would not be right.

Agree but I was thinking of a screen resistor (or connection) that was opening up beyond idle, not completely open. Also like the oscillation idea, but not sure how it could cause the current in one tube to decrease.
In any case, the test with a sine wave gradually increasing in volume would be instructive.

[HilltopExplosion2]

Hi guys! Leadfootdriver has pointed me this direction for help, and instead of aggravating him, I joined. I do not have a vast knowledge of electronics, just the basics, so please bear with me.

My friend brought this amp over for me to try out, and I liked it, so a deal was struck. Shortly after, I lost a power tube. The one that is going negative on the bias meter. I asked if there had been any problems before and he said no, but the tubes had been in it for a while. I replaced the set with a new set and all was ok, no hum, sounds fine. When I measured across pins 8 and 3 of the power tubes, it was 400 VDC on both. When I checked the right one ( the tube that goes negative) my voltmeter started clicking. I then figured 40 ma would be a safe place to start and see how the amp sounds. That is when I noticed the right tube go in the negative on the bias. I'm new to this amp, and this occurrence. The same tubes in a 100 watt amp, along with the same PI, did not have the same effect. I pulled 2 tubes and turned the amp to 8 ohms to rule out the tubes. They were very close all the way up to 80ma.

To answer a few questions, I checked the resistors, and the showed 989, and 996 ohms.

The reason for the bad recording is I just wanted to show what the amp was doing, so I just used my phone..

I do not have an oscilloscope, but if I can't figure this out, I think will take this to a repair shop to get it checked.

When the amp's volume is at 9 o'clock, and the guitar volume rolled back, it is very clean. I can then play a chord lightly with not much movement of the bias needle. When I hit the string hard, or chord hard, the needles will move in opposite directions. As of today, the amp has developed a hum, and sounds funny. With the amps volume at 9 o'clock (which is where it was in my vid) and the guitar volume maxed, there is a PEEEOWW to the attack of the string, or chord. It almost sounds like you hit a tight cable with a wrench.. also my voltmeter still clicks when I check pins 8 and 3 on the right power tube..

Any and all help are appreciated! With my limited tools, any suggestions of where and what to check would be appreciated. Thank you!

[Chuck H]

Well I think this confirms that something is amiss. Even if someone wanted to find flaw in the design, there are too many out there working fine to be that glib. So, yeah, take it to someone with bench gear to test and check things at a professional level. I won't be surprised if there's a blown screen resistor (as has been surmised), probably taken out with the tube failure. I hope the new symptom are nothing serious.

[nickb]

It really sounds like instability might be the issue. Tough to be sure about it and fix without a scope.


Just sort of throwing ideas out here:

If is has always done it then it could be a layout or component value issue. You could try upping the EL34's grid stoppers, no more than 47K. It's quite common to see a small 630V ceramic capacitor between the plates of the phase inverter - try 100pF.

If it just started then something has changed, wires got moved, solder joint, bad grounds esp around the filter caps - you did say the hum had increased recently too.

[Gingertube]

What nickb said.
Open the feedback loop first and see if the problem goes away, If so feedback phase is wrong.
If NOT:
EL34 are high transconductance (gm) so grid stops need to be 10K minimum. 1K5 are for lower gm tubes like 6L6 but even for them this "traditional" value is a bit low.
Make sure the grid stop resistor body is hard up against the tube socket pin.

Ditto for the screen grid resistors, they have a grid stop function as well as a screen protection function.

When you suspect bias shifts always check the grid leak resistors. For cathode biased outputs it is sufficient to check that the resistance to 0V is correct.
In fixed biased amps it is necessary to check the entire chain back to the -ve bias voltage and while you are at it check that the "bypass" capacitor which establishes the signal ground for that bias supply is wired the correct polarity (-ve to the bias supply and +ve to ground).

If the impedance to ground is too high then grid current in the output tube will cause a bias shift (decreased bias) in the increased tube current direction, which increases grid current, round and round .poo. melted output tube.

If all that is OK. Then the PI might be oscillating.
His suggestion of the 100pF plate to plate is worth trying
BUT POSSIBLE LIGHT GLOBE MOMENT
Since it was mentioned above, I built a TrainWreck Clone with pretty much the circuit for the power amp which we are discussing (I used 6CA7 in lieu of EL34) and I DID have PI stability problems. I tried a heap of things (not the 100pF between anodes) but what worked in the end was fitting 10K grid stops to the phase inverter triode grids (both).

One more basic:
Have you checked for a leaky coupling cap from the Phase Inverter to the Output tube grid circuit?

The fact that the problem stays with the same output tube socket suggests a basic build problem, or component failure problem, but unfortunately does not eliminate the other problems outlined above.
So replacing the coupling caps might be worth a try too.

Cheers,
Ian

[JoeK]

Had a similar problem at idle with EL34 power amp and it ended up being the plate load resistor drifting on one side. Bias drifted and current went way up.

[loudthud]

leadfootdriver (the OP) has not been back here for over a week.

[nickb]

I think his buddy HillTopExplosion2 took up the slack. Still, it has been a while, even then.

[HilltopExplosion2]

Thanks for all the advice, and troubleshooting help! I went through a whole box of preamp tubes to replace v1 with the quietest tube I could find.. next I replaced the JJ EL34's with some EH 6CA7's that were in my other amp.. I KNOW they are good. I also replaced the rest of the pre tubes with tubes that I know are good. The hum is still there, but not as bad.

Without a guitar plugged in, I can turn everything up to 10 (Bass, treb, etc). As I increase the volume the amp starts to oscillate. Everything is fine until the volume approaches noon, then here comes the noise.

Prior to replacing the tubes, I called a friend and he helped with some troubleshooting over the phone. Poor guy was on there for almost an hour. We started with the power tubes, pulled them, then turned the amp on.. no hum. Put 'em back in, next was PI, no hum. Then V2, no hum. V1, and there it is. He had me check a few resistors, and everything was o.k. When I had the amp on, I tapped the blue 2200 pF ceramic capacitor, (on the board between V1 and V2) and it was very loud. I could also tap anything around it and it has a very glassy sound to it.. like a microphonic tube, even though that preamp tube (V1) is not microphonic in any of my other amps. I had the same result no matter what tube I put in there. He had me cut 1 leg of the cap and turn the amp back on. THAT took care of a lot of the noise. That said, my next plan of action is to replace the capacitor, and see if that helps. If not, it goes to his shop on his bench. He said it could also be a ground loop. I don't know, I'm ignorant when it comes to working on amps.

Here's a gut shot.