• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Broskie auto bias

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am in the process of building an amplifier with toroidal output transformers and had heard that they were more sensitive to DC imbalance. I could adjust bias all the time but I thought it might be better if a circuit just did it for me.

I was researching the subject and came across this blog entry at TubeCad. Folks have mentioned that it has been built and works well.

Anyway, I laid out a PCB for this. Anyone see any problems with how I have done this?

One tube is the master and the other just tries to maintain the same current as the other. I added a relay to put the circuit in a defined state before the HT- supply comes up. I want bias voltage to start out at the negative rail.

Thanks in advance for your comments/critiques.
 

Attachments

  • Bias_servo_sch.pdf
    21.6 KB · Views: 942
  • Bias_servo_brd.pdf
    58.7 KB · Views: 477
I missed that excellent article. I always enjoy reading Broskie's work.

I like Blumlein's original circuit. It seems very clever. Broskie's premise is that the original circuit (that seems to perform very well) can be improved with solid state components, but did he really close on that point? If he did, I missed it.

Broskie made this statement: "Blumlein didn’t have access to transistors and OpAmps; we do. (I know that some are already getting nervous. But why not take advantage of a technology that’s cheap, effective, doesn’t wear out, and, most importantly, isn’t in the audio signal path? Anxiety or arrogance? Fashion or foolishness?)". I find the original circuit to be elegant in it's simplicity and with just a pair of cathode resistors rather than a single and a length of wire, it too is cheap, effective and doesn't wear out.

It would be fun to bread board some of those and compare the results.
 
Last edited:
Blumlein garter works really well and sounds excellent. i used it in a headphone amp with toroidal outputs.
The problem with Broskies bias circuits are that they slave one triode to the others bias. It maybe that the one is running very hot and this then forces the other really hot. In the end this maybe fairly unstable and hard on valves.

Broskie discusses this in a subsequent article where he goes back and looks at the garter again - and concludes that it is intrinsically superior because it avaerages the imbalance between the two halves. He then goes onto design a version where one of the resistors is replaced with a transistor which preserves all of the benefits of the circuit without the hit of doubling the bias voltage.

Shoog
 
I like Blumlein's original circuit. It seems very clever. Broskie's premise is that the original circuit (that seems to perform very well) can be improved with solid state components, but did he really close on that point? If he did, I missed it.

Blumlein's circuit is very elegant in its simplicity and offers pretty good performance, but not as good as the circuits Broskie fleshes out. In the end, gain has to be used to get the imbalance down to near zero. If you missed it, it is all of the stuff after "A Modern Approach to Bias Balance." All of those approaches will give essentially zero imbalance as opposed to the 8.5% imbalance of the Blumlein circuit in the example.

I have an aversion to cathode bias since it allows bias voltage to vary with signal, particularly in a class AB output stage(which is what this is going in). Also, it must be noted the Blumlein's circuit doubles the cathode resistor power dissipation. That's the price paid for the performance increase.
 
I think the original Blumlein Garter circuit would be an excellent choice for this application. IMNSHO no need for fancy and potentially unreliable solid state circuitry in this application.

This is going in an amp with a McIntosh-style Unity Coupled output stage with a source follower driver. I think it would be an awkward choice for this application, fixed bias seems much more natural to me, especially since the negative supply is mandatory for the source followers already.

As far as reliability goes, I'm not asking much of the semiconductors. My guess is they would outlast a hot resistor...
 
The example given is an extreme example - which illustrates how effective it really is. In normal operation, with the same triode on each half, expect the imbalance to be less than 1%. As I said I use it in exactly the scenario you intend to and it works excellently.

I would look at the second implementation of the transistor garter as a compromise which offers the best of both worlds without undue complexity.

Shoog
 
Blumlein garter works really well and sounds excellent. i used it in a headphone amp with toroidal outputs.
The problem with Broskies bias circuits are that they slave one triode to the others bias. It maybe that the one is running very hot and this then forces the other really hot. In the end this maybe fairly unstable and hard on valves.

Broskie discusses this in a subsequent article where he goes back and looks at the garter again - and concludes that it is intrinsically superior because it avaerages the imbalance between the two halves. He then goes onto design a version where one of the resistors is replaced with a transistor which preserves all of the benefits of the circuit without the hit of doubling the bias voltage.

Shoog

That's not really a problem with the design I posted. The bias will be set on the master to a particular value. Insert a KT88, 6L6, EL34, doesn't matter. There should be no 'running hot' as any correctly functioning tube will settle at the same current in the master slot. The slave will match that current.

As far as safety goes(a point Broskie brings up in the garter bias revisited article), I have a separate comparator that will shut the amp down should any output tube draw too much current.
 
If you missed it, it is all of the stuff after "A Modern Approach to Bias Balance." All of those approaches will give essentially zero imbalance as opposed to the 8.5% imbalance of the Blumlein circuit in the example.

That (two parallel tubes) was on a rather extreme example, no? With a typical imbalance I'd expect that the error would be very much smaller and so the advantage to the SS circuits reduced in that regard.

After reading the second (followup) article, I came to better appreciate the advantage that would apply when adding this circuit to an existing design. Avoiding the loss of 50 volts of plate voltage seems almost a necessity in a retrofit.
 
The example given is an extreme example - which illustrates how effective it really is. In normal operation, with the same triode on each half, expect the imbalance to be less than 1%. As I said I use it in exactly the scenario you intend to and it works excellently.

I would look at the second implementation of the transistor garter as a compromise which offers the best of both worlds without undue complexity.

Shoog

Are you saying you use it with a toroidal OPT or that you use it in an amp with cathode feedback windings and a DC coupled driver? I'd be interested to see how you did it if you have a schematic.

Sure, I agree garter bias is an excellent solution but I was just pointing out that it is unfair to say that gain doesn't improve the performance.

I am still not liking the solid-state garter bias since it is still cathode bias, and a high-powered class AB power stage seems like the worst place for cathode bias. This amp will play loud electronic music with long sustained bass notes, etc. I don't just listen to female jazz vocals, etc. I think that bias voltage will fluctuate significantly with signal and since I am planning on running output tubes as pentodes, tubes will be very sensitive to bias voltage fluctuations due to the high gain.

This is a link to the thread for the amp I am building.
 
I have used it with toroidal outputs but not cathode feedback, though I do not see why that would be an issue.
Unfortunately you have just pointed out that it is DC coupled which makes garter an impossible option. I have been building direct coupled amps for a while now and have thought about how garter might be used direct coupled - can't be done as far as I can see.

As for AB - it all depends on what size of cathode bypass you are prepared to tollerate really. I use cathode to cathode bypassing which is slightly different in terms of overload behaviour - I cannot say I have encountered problems and get good response down to 10hz.

Shoog
 
That (two parallel tubes) was on a rather extreme example, no? With a typical imbalance I'd expect that the error would be very much smaller and so the advantage to the SS circuits reduced in that regard.

You seemed to be implying that there was no advantage when there was in fact a tiny advantage. Sorry, I'm an engineer.
 
Me too, but they don't let me blow the horn anymore.

As an engineer, it is important to evaluate the cost to benefit ratio, no? At what cost to complexity and reliability comes a benefit that may (or may not) be significant. It is that aspect I was questioning.

Shoog, you are correct in that observation, but also the practical advantage may be found when adding the Garter to an existing design. For that it is significant.
 
I have used it with toroidal outputs but not cathode feedback, though I do not see why that would be an issue.
Unfortunately you have just pointed out that it is DC coupled which makes garter an impossible option. I have been building direct coupled amps for a while now and have thought about how garter might be used direct coupled - can't be done as far as I can see.

As for AB - it all depends on what size of cathode bypass you are prepared to tollerate really. I use cathode to cathode bypassing which is slightly different in terms of overload behaviour - I cannot say I have encountered problems and get good response down to 10hz.

Shoog

Actually, mine would be easy since the DC coupled driver is a follower which is AC coupled to the previous VA stage. Just send the bias to the gate of the follower instead of the power tube grid.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.