Biasing 1987x - disparate plate currents

[dozturk]

Hi everyone - first time posting here.

I have a 1987x reissue and recently bought a new matched pair of Mesa EL34's (STR447, Gray).

After biasing the amp, I get significantly disparate plate currents: 31 vs. 38 mA. I assume 18% difference is not normal for a matched set.

I want to know if this pair is indeed not matched, or if there's some asymmetry in the amp that I'm not accounting for.

Here's what I did:
- measured plate voltages at pin#3 of each tube: both 467 V
- grid voltages at pin#5: both -45.5 V.
- OT winding resistances, measured between center tap and each plate (pin#3): 41.7 and 42 Ohms (<1% difference)
- measure voltage across same terminals, and divide by resistances to calculate plate currents: I get 31 mA and 38 mA, corresponding to 57% and 70% power dissipation.
- screen resistors are 977 and 987 Ohms (<1% difference)
- calculate screen currents via voltage drop: 3.9 and 4.9 mA.
- to rule out any effects of socket contact oxidation I swapped tube I and II, and the measured current values also moved over exactly.

Am I missing anything or can I blame the tubes?

Thanks

 

[Kuga]

Welcome.
If bias pot it's at min. value. Do you get disparate plate current?
I'm not an amp tech but looks like you have not a matched pair.

 

[Ronquest]

You can blame the tubes. Maybe let them run for a while and see if you get any changes, but with the current following the tube, I'd say your tubes aren't as matched as they should be.

 

[Pete Farrington]

Probing output valve anodes can induce oscillation, screwing up the readings.
I suggest to repeat the measurements but with the 12AX7 valve in the V3 LTP socket removed, as that will prevent oscillation.

 

[playloud]

Good suggestion by Pete above, but I wouldn't be surprised if the readings are correct. That doesn't too bad as far as matching goes. How does the amp sound?

 

[Ronquest]

Probing output valve anodes can induce oscillation, screwing up the readings.
I suggest to repeat the measurements but with the 12AX7 valve in the V3 LTP socket removed, as that will prevent oscillation.

Interesting. Whats causing the oscillation? Something in the meter or feedback from the leads to the phase inverter / preamps, or something else ? I'll have to keep this in mind.

 

[961le]

I have the same set, mine are at .036 and .029 With 453 pv. It sounds good. My other 2 pair bias much closer than this set.

 

[mickeydg5]

Hi everyone - first time posting here.

I have a 1987x reissue and recently bought a new matched pair of Mesa EL34's (STR447, Gray).

After biasing the amp, I get significantly disparate plate currents: 31 vs. 38 mA. I assume 18% difference is not normal for a matched set.

I want to know if this pair is indeed not matched, or if there's some asymmetry in the amp that I'm not accounting for.

Here's what I did:
- measured plate voltages at pin#3 of each tube: both 467 V
- grid voltages at pin#5: both -45.5 V.
- OT winding resistances, measured between center tap and each plate (pin#3): 41.7 and 42 Ohms (<1% difference)
- measure voltage across same terminals, and divide by resistances to calculate plate currents: I get 31 mA and 38 mA, corresponding to 57% and 70% power dissipation.
- screen resistors are 977 and 987 Ohms (<1% difference)
- calculate screen currents via voltage drop: 3.9 and 4.9 mA.
- to rule out any effects of socket contact oxidation I swapped tube I and II, and the measured current values also moved over exactly.

Am I missing anything or can I blame the tubes?

Thanks

First, did your try swapping the power tubes between sockets to see if the prior results follow the tubes?

Second, always clean the tube's pins and socket's prongs with a good electronic lubricating contact cleaner.

 

[dozturk]

Thanks for all your input.

Probing output valve anodes can induce oscillation, screwing up the readings.
I suggest to repeat the measurements but with the 12AX7 valve in the V3 LTP socket removed, as that will prevent oscillation.

Thanks for the suggestion. I just repeated the measurements and got very similar readings.

However, perhaps this explains the weird reading + noise I occasionally got from that low-current tube. This happened only 2-3 times, and would randomly go away. Contact of the DMM probe with pin#3, would cause a distinct mid-frequency noise, and the voltage would read +3 V (positive) rather than the typical -1.07 V (negative) for that tube (my COM is attached to OT-center tap). @Pete Farrington is this the kind of behavior you would expect from that oscillation? Or do you think something else is going on?

Welcome.
If bias pot it's at min. value. Do you get disparate plate current?
I'm not an amp tech but looks like you have not a matched pair.

As I turned the bias pot CCW, the currents decreased and the disparity grew (>25--30%). But I did not push it all the way down.

First, did your try swapping the power tubes between sockets to see if the prior results follow the tubes?

Second, always clean the tube's pins and socket's prongs with a good electronic lubricating contact cleaner.

Yes, I swapped the tubes and the current measurements followed the tubes.

How do you guys clean the sockets? I have 90% alcohol and DeOxit, but I can't get into those holes.

 

[Kuga]

How do you guys clean the sockets? I have 90% alcohol and DeOxit, but I can't get into those holes.

I use this. KONTAKT chemie tuner. It's a fast-drying precision cleaner for all types of contacts. Its gentle cleaning power makes it especially suitable for use on rubber, graphite conductors and high-frequency components. Dries in seconds without leaving residue.

 

[Pete Farrington]

Whats causing the oscillation? Something in the meter or feedback from the leads to the phase inverter / preamps

Yes, the probe / lead / meter become part of the anode circuit, and act as a transmitting antenna. Think of it as momentarily becoming the worst lead dress imaginable.
That facilitates coupling back to earlier parts of the circuit, hence promoting feedback paths to form, allowing oscillation to occur.

Contact of the DMM probe with pin#3, would cause a distinct mid-frequency noise, and the voltage would read +3 V (positive) rather than the typical -1.07 V (negative) for that tube (my COM is attached to OT-center tap). @Pete Farrington is this the kind of behavior you would expect from that oscillation? Or do you think something else is going on?

If not oscillation, then what?
Can you recreate it, first with the valve in V3, and then without?

 

[anitoli]

Why are you attaching COM to the OT center tap? Unless you are specifically looking for voltage drops across two points the ground lead ( COM) should be at ground reference.

 

[dozturk]

Why are you attaching COM to the OT center tap? Unless you are specifically looking for voltage drops across two points the ground lead ( COM) should be at ground reference.

Yes I only do that to measure the voltage drop (and calculate the plate current). This provides much better resolution than individually measuring and subtracting V-OT and V-plate with respect to the ground.

If not oscillation, then what?
Can you recreate it, first with the valve in V3, and then without?

Makes sense. It's very hard to recreate. Happened only occasionally with V3, and never happened without V3. I don't know if this was a coincidence or not.

I use this. KONTAKT chemie tuner. It's a fast-drying precision cleaner for all types of contacts. Its gentle cleaning power makes it especially suitable for use on rubber, graphite conductors and high-frequency components. Dries in seconds without leaving residue.

Thanks - but how do you mechanically wipe inside the pin holders? Is spraying sufficient?

 

[Blake F]

You will always get a small amount of current mismatch unless the bias feed resistors (220k x 2 for EL34 tubes) are matched well, but not that much mismatch. As most people have chimed in, you likely have a mismatched set or they need burn-in. However, for a 50W you can add a circuit to center the bias on mismatched tubes so that they run the same. It's a simple circuit- I have it in my 1987 amp, it sounds great and you can find the diagram easily on the internet. It's nice especially for vintage tubes being as pricey as they now are.

 

[mickeydg5]

How do you guys clean the sockets? I have 90% alcohol and DeOxit, but I can't get into those holes.

You spray the electronics lubricating cleaner into the socket hole on the prongs and a little on the tube pins.
Then work the tube in and out of the socket a few times.
It does the work for you.

 

[Matthews Guitars]

Swap the two tubes around and check the bias again. If the bias values stay the same for each tube, and not each socket, then assume that the tubes are not well matched.

Mesa's biasing system is usually pretty good but sometimes tubes drift out of their color range and into the next one. Blame it on the quality of tube manufacture today, which can be summed up as "If it functions as a tube, we ship it."

 

[mickeydg5]

Yes I only do that to measure the voltage drop (and calculate the plate current). This provides much better resolution than individually measuring and subtracting V-OT and V-plate with respect to the ground.

No.
Do as @anitoli mentioned.
I connect COMMON to chassis ground, preferably main chassis ground from power cord and do the math.
If you do not, then your entire meter becomes an antenna in the output circuit.

I have never had an issue using the OT bias method nor putting a probe on my plates.

 

[Pete Farrington]

then your entire meter becomes an antenna in the output circuit.

The AC scenario is the same, whether the meter is referenced to circuit common or to the OT CT.
Although those nodes are several hundred V DC apart, the reactive impedance of the reservoir etc caps (a few ohms at the most) makes the V AC between them effectively 0.

I have never had an issue using the OT bias method nor putting a probe on my plates.

It’s possible that unless you were also simultaneously checking for oscillation, you might not be aware of it.

 

[Pete Farrington]

You will always get a small amount of current mismatch unless the bias feed resistors (220k x 2 for EL34 tubes) are matched well

Why would that make any difference? The V DC across the grid leaks is negligible, unless the valve is faulty.

 

[Blake F]

Why would that make any difference? The V DC across the grid leaks is negligible, unless the valve is faulty.

I guess it could be a little crossover distortion manifesting itself when amplified by the power tubes if there's enough of a mismatch. Yes, those 220k's are grid leaks but they're also biasing those power tubes from a single point split in two. I think the quiescent grid current on the power tube grids would be a little bit more than a preamp tube but I'm just spitballing here without looking up the idle grid current and it might also depend on how hot or cold you're biasing them I guess. I read an article on the subject years ago so I tried swapping-in 2 that were well-matched for the ones in the amp that weren't, and I noticed a small but noticeable pleasant difference. I didn't put a scope on it, and my results certainly *could* be fooling me but I seem to be blessed (or cursed) with really good ears and I'm certain I can hear a difference. Seems like the article was by Paul Rivera and if you can find the article it may have some EE theory with it. Then of course you have the whole rabbit-hole of whether you really need to match tubes and whether a little crossover distortion actually sounds good.