Biasing Unmatched tubes - some questions about

[kagliostro]

If I'm not wrong years ago there was not the consuetude to match tubes (may be I'm wrong or may be they don't do because they can't do)

However now everybody is looking for well matched tubes

I've some question about biasing unmatched tubes

Assuming I must do that because I use some tubes that I find here and there that I've in few quantity and I can't try to match

First question:

Which circuit is more affected using unmatched tubes ?

the cathode bias circuit or the fixed bias circuit ?

Second question:

often I see fixed bias circuit with doubled bias pot to be able to regulate the bias voltage for each tube, question: which is the best regulation procedure to obtain best results ?

Third question:

I drew a solution for the fixed cathode bias using a wire pot (enough big & with apropiated resistance) can this be a solution as to use unmatched tubes in cathode bias amps ?

Thanks

Kagliostro

[kagliostro]

For fixed bias amps

My idea is to adjust the two valves tubes (unpaired) with the same negative voltage bias (the reference of the original scheme), then measure the consumption in mA and adjust the valve that is consuming more in order to consume as the valve which has the lower consumption

No one has experience in this or can say if this can be done with good results ?

Thanks in advance

Kagliostro

[LooseChange]

Lots of experience with Fixed adjustable bias for individual tubes.  Talked about in TUT.
Attached is a drawing.. Not sure about the -41v, but you'll get the idea.

As far as cathode bias and mismatched tubes... As long as they don't hum, the are a go.

[xm52]

There is another topology to adjust a fixed bias amp. Take the pot as you have it in your cathode bias diagram and insert it between the cathodes on the fixed bias diagram. The center tap goes to ground. You can remove the two other bias pots. This allows you to use a single pot. You sometimes see this called a balance pot. Similar to the cathode balance scheme.

In general when matching tubes, they match transconductance (gm) and plate current at a fixed B+. Ideally, you want the B+ to be the same as that of the amp.
Tubes should be matched at the same bias. The other approach is to match the tubes for maximum gm and plate current but this requires a different bias for each tube.

The idea of setting the bias for each tube is good. It cost more to implement and maintain so manufacturers avoid it.

There are different ways to wire a bias pot into a circuit. With any bias scheme that uses a wiper, you should ask yourself, what will happen to the bias of the tube if the wiper fails. See this schematic https://taweber.powweb.com/store/biasboard_schem.jpg for an example.

[HotBluePlates]

First question:

Which circuit is more affected using unmatched tubes ?

the cathode bias circuit or the fixed bias circuit ?

This is the wrong question. The Library of Information has RDH3 and 4 available for download. RDH4, chapter 13, section 5(v) on pages 580-581 discusses the effects of mismatching.

What you really need to ask is, "what class of operation and tube type is most affected by mismatching?"

The effects are most serious when you're using pentodes/beam power tubes in class AB, but the result will be some higher amount of distortion (most likely increased even harmonic distortion) and less power output. The section of RDH4 cited discusses class A triodes. In it, they use an example of using two entirely different types of tubes on either end of a push-pull output stage; one of the tubes has twice the transconductance and half the plate resistance of the other tube. This would be equivalent to having one new tube, and one nearly-dead tube in an amp.

The result? 2nd harmonic distortion, which would be zero in a perfectly-matched and balanced output stage, was only 5% which is a common design target for "low distortion". So there is no serious effect, with class A triodes.

We're not given info in RDH4 about how much power output was reduced compared to matched triodes.

So now on to your question: I'd suggest mismatching is more of a "problem" in a fixed bias circuit.

You can't assume a class of operation by the bias method, but it is more likely to see cathode bias used on class A or hot class AB amps, and fixed bias on cooler class AB amps. That's largely because cathode bias is inherently safer, cheaper to incorporate, and does work well deep into class AB. For deep class AB and higher power outputs, you generally need to use fixed bias to prevent the bias from changing materially with large plate current swings.

Further, if you use a pair of tubes with a single cathode bias resistor, they tend to find an average point where both tubes are happy. If the cathode resistor is adequately bypassed, then the differences of plate current swing won't upset the bias. Again, you might get a little more distortion, and a little less output power. It might not be enough of either to hear a difference (or only a very slight difference).

Very severe mismatching (one tube in the socket, the other tube not in the socket) could lead to power supply hum and output transformer saturation, because the hum cancelling effect of push-pull is based on equal hum currents flowing in both halves of the transformer primary. The possible increased saturation would be because push-pull OTs are not made for unbalanced d.c., and the balanced d.c. implied by push-pull allows for a smaller core for the same output power. Most push-pull OTs allow for some amount of unbalance, however.

Second question:

often I see fixed bias circuit with doubled bias pot to be able to regulate the bias voltage for each tube, question: which is the best regulation procedure to obtain best results ?

What LooseChange posted. If you are concerned, you can have a bias adjust pot for each tube.

It occurs to me that even with the idle currents made equal by the individual bias pots, if the tubes are very mismatched, you will still have the same issues I mentioned. "Very mismatched" implies very different bias voltages required for equal idle current, which means each tube can handle a different amount of input signal before serious distortion. So you might not get every last fraction of a watt. It might, however, sound good to have each tube beginning to distort at a different time, and by a different amount.

[jjasilli]

As far as cathode bias and mismatched tubes... As long as they don't hum, the are a go. Ditto to LS.

The ultimate purpose is equal and opposite (out of phase) signals pushed by the power tubes through the OT primary.  This kills 1/2 the hum & noise.  50% noise reduction is HUGE.  If the amplitudes of the 2 out-of-phase waveforms get out of whack, then hum & noise cancellation decreases, to the point where it becomes objectionable. So, the PP power tubes should put out equally strong waveforms.  Ea tube is unique, so they can be balanced in their output by tweaking their individual bias points.

How far to go in this quest?  IMHO Loosechange nails it for guitar amps.  Some slop probably contributes to mojo. Hi-fi will probably want more exacting standards.

But, no matter how perfectly you design & balance (blueprint) your bias circuitry the 2 OT primary windings will never be equal.  Hence, perfectly matched tubes, or bias balanced mismatched tubes, ultimately will not do the trick.  The bias needs to be mismatched to compensate for the mismatched primary windings in the OT.  I think a scope is the only way to measure this. 

[kagliostro]

Many Many thanks friends

I really need your councils and explanations

I must read with attention your explanation

also I'll read the RDH4 chapt 13

thanks

Kagliostro

[jjasilli]

The bias needs to be mismatched to compensate for the mismatched primary windings in the OT.  I think a scope is the only way to measure this. 

Please note that I am not advocating that anyone actually do this, just pointing it out.  I should also have mentioned that the two signal outputs from the PI also need to be perfectly balanced.  Simply equalizing the the power tubes, by itself, does not mean that the waveforms are balanced.