66 Vibrolux Bias

[TerryD]

When I do the caps, I would also like to deal with this problem.  My present bias is crazy (to me).   Let me just give you the idea as I don't remember the exact figures.  At so many volts my bias should be say 33 ma per tube.  I'm getting like 47 on one tube and 19 on the other...may not be exact, but that's the idea=far apart.  Amp sounds ok but my "balance' is probably running one a little cool and the other a little more than hot.
So.....I am going to:
Switch tubes to see if the tubes are the problem
Then (my idea) check the phase inverter tube to see if that's ok
Then check the two resisters that feed the two 6l6 plates.
 
Anything else I can do to get a better balance?  Hope my language isn't too confusing.
 
Thanks,
Terry

[alerich]

Yes, swap the output tubes first. That will confirm if it is the tubes themselves simply drawing different amounts of current (most likely) or some issue with the sockets and/or the supporting circuitry (less likely but still possible).

I ordinarily wouldn't be too suspicious of the PI since the coupling caps between the PI and the power tube grids keep DC from the PI from getting to the grids of the power tubes but you could always have a leaky coupling cap. Easy way to check that is to simply disconnect one side of the caps to isolate the PI from the power amp. Whenever I build an amp I always build the power amp and bias section first and then power up and let it cook at idle so that I know those two areas work as intended.

I also wouldn't be too suspicious of the resistors you mentioned (I assume you mean the resistors feeding bias to the power tube grids - you mentioned something about plates). The bias circuit doesn't draw much current. It is largely a static control voltage. Do make sure that you have the same negative voltage measurement at both grids. You can even do this with no power tubes installed to eliminate them from the equation.

As always, inspect all of the solder joints for integrity. My money is on the tubes themselves. Happens all the time - even with brand new "matched" tubes.

[HotBluePlates]

Anything else I can do to get a better balance?  Hope my language isn't too confusing.

Add a bias balance control, as found in Silverface Fenders. You can have a balance control in addition to an existing bias adjust control.

I agree with Alerich that there's maybe 95% chance the difference in tube current is simply a difference in the transconductance of the individual tubes.

[eleventeen]

You have exactly the right algorithm for troubleshooting your bias issue. Yes, it's probably the tubes. I would be surprised to find badly mismatched resistors from the bias source to the pin 5s (grids) I'd also be surprised to find leaky coupling caps or anything with the PI tube, just as Alerich said.


I once acquired a batch of about 30 NOS 6L6GA tubes, WW2 vintage. Nice small bowling-pin shaped tubes, JAN Sylvania, not really advisable to place them in a Fender, they are not rated > 350 volts IIRC. Nevertheless being 4 lifetimes' supply, I thought I could sell them for more money if they were grouped into matched pairs. So I built a tube matcher.






I was astounded how far apart some were, although NONE were as far apart as you are reporting. Yeah, I found plenty of pairs 10-15 mils apart but not ~~25 mils like you are saying. I could get 25 mils difference if I took the single "worst" and the single "best". You could have a baddish tube.

[TerryD]

Please bear with me as I wont dig into this until I do my cap job. 
So my beautiful "old" set of Sylvania tubes that came with the amp can be desperately mismatched, which will be a cinch to tell.  Could I run them, if they are as bad as I stated or is there a "farthest apart" scenario of say 15ma which would be safe to run?
Any slick place to find that bias balance control mod/addition?

[HotBluePlates]

... So.....I am going to:
...
Then check the two resisters that feed the two 6l6 plates. ...

I would be surprised to find badly mismatched resistors from the bias source to the pin 5s (grids) ...

It should be remembered that unless an output tube is gassy or a coupling cap is leaking d.c., that no current is drawn through any of the resistors connected to an output tube's grid at idle.

Therefore, one tube could have a 220kΩ bias feed and a 1.5kΩ grid stopper, while the other side could have a 10kΩ bias feed and no grid stopper. If the tubes are matched for idle current, are not gassy and a single bias voltage is applied to the junction of the 220kΩ and 10kΩ bias feed resistors, then both output tubes will pass the same idle current.

An unlikely scenario that would cause matched tube to pass different idle currents would be a severly drifted screen resistor. If the screen voltages of the tubes are radically different (420vdc on one, 100vdc on the other), the the plate currents of the tubes will be very different even though the same bias voltage is applied to the grids. Since the screen draws current through the screen resistor, there will be a voltage dropped across that resistor. But idle screen current is typically small (a few mA's), so it would take a major fault in the screen resistor (470kΩ instead of 470Ω) to create a major difference in screen voltage and thus tube idle current.

Could I run them, if they are as bad as I stated or is there a "farthest apart" scenario of say 15ma which would be safe to run?

The most significant impact of mismatched output tubes is you will have less clean output power. Which also means the output section will distort faster. For the most part, people worry about tube matching more than it's worth.

"Matching" of output tubes is somewhat related to the tube's transconductance (Gm or milliamps-per-volt, mA/V). It is really about making sure the tubes pass the same current at all conditions of grid voltage, which then matches the power output into a load. I cited RDH4 in the thread linked below for an authoritative, historical reference on the effect of tube matching, as quoted below.

An interesting section in RDH4 is Chapter 13.5 v, Matching and the effects of mismatching, pp 580-582.

My Hickok tube tester has a Good-Bad "English" scale, where the bottom edge of the Good range is equal to 60% of the Gm for a bogey tube of the type being tested.

Anyway, RDH4 has an example of the effects of mismatching wherein they pair a 2 different output tubes (a 2A3 and a type 45) in a push-pull stage to see what will happen. Amplification factor for these output triodes is 12% apart, but there is a more than 50% difference in Gm and plate resistance. By the standard of the Hickok tester, 50% low Gm would be the same as a "Bad tube".

The end result was 5% 2nd harmonic distortion (2nd normally cancels in matched push-pull stages), and a shift in tube bias due to rectification. All-in-all, you call it "less clean output power".

What other effects might happen in radically mismatched tubes?
...
So if you need maximum clean power & maximum bass authority, then well-matched output tubes is a good thing. But for most practical purposes, in a guitar amp (where we intentionally have ENORMOUS distortion by hi-fi standards) it just doesn't seem to matter much.

[HotBluePlates]

Any slick place to find that bias balance control mod/addition?

On this forum. See this post on Dual Bias, and look at the schematic fragment on the right labeled "Bias Balance."

The "Range" control is your raw bias adjust, and the "Balance" control allows different voltage to each output tube for idle balancing. Read the rest of the thread for discussion on whether this matters for any situation other than idle. The 22kΩ pots shown would both be linear pots, and 25kΩ would be suitable for both.

The 22kΩ resistor between the pots might need to be adjusted in value to obtain the needed range of bias voltage adjustment, or any series resistance in your raw bias supply might need adjustment to obtain the needed range (which would be in effect the same as the information presented in the Hoffman Library of Information on altering the range of bias in a bias adjust circuit).

The point that says "-bias v" is the rest of your stock bias circuit (including diode, caps, etc) prior to the original bias adjust pot & resistor to ground.Sluckey also shows a link to a schematic with 2 independent bias adjust controls, one for each tube, as an alternative to an adjust/balance approach.

[PRR]

Bring bias hotter until one tube is at 110% rated dissipation. Perhaps 80mA.

If the other one is within 20% (70mA), bring it back down to 70% (more or less) and rock on. The tubes are close-enough at BIG outputs.

If they don't come within 20% at high current, the lower one is probably sick and not pulling its share of the load. Get new tubes. There's a reason we don't weld them into the circuit.