Bias question

[jeff]

 Trying to understand bias a little more. Another thread got me thinking about this but I didn't want to get to far off topic on that thread so I started a new one.

 Say you were given an amp to bias.

 -The power tubes were unmarked and there was no way to identify them,
 -You don't have any spec sheet and the Max Dissapation is unknown.
 -The only thing you know for a fact is that the amp was designed to use these tubes.
(EDITED to adress questions brought up in first responce)
- it's a push pull fixed bias amp
-The tubes look like nothing you ever seen before, size, shape, etc.
-everything in the amp is working correctly, nothing blown or damaged, nothing has been rewired
-the amp was just running and biased correctly, some one just shut it off and turned the bias pot to maximum neg voltage. and said "bias this amp".

How would you go about biasing it?
Would it be possible to bias it correctly just by running tests and observing how the amp responds?
What would those tests be?

[eleventeen]

It's kind of odd that you have no output tube specs in your "thought experiment" but I will play along.

I guess we assume that the amp is wired correctly and that tubes are in serviceable condition. (Don't laugh; I once worked in an electronics factory fixing stuff right off the ass'y line after it failed the first "power on" test, and under those conditions, you have no right to assume that the device *ever* worked properly)
 
You would probably start by looking at the tubes and comparing them to known types. If they appeared to be beam power tubes or tetrodes, I would look to see the voltage ratings of the p/s caps. I would examine the circuitry and see if it's something resembling a push-pull output (you did say tubeS plural)

I'd insert 1 ohm resistors in the cathode circuits of the output tubes.

Assuming all that achieves a picture of a generally typical push-pull output section, and again, we are assuming the amp once worked and has not been randomly rewired, thus there is some sort of bias circuit there) Maybe it is fixed, maybe is variable. At some point, I guess I would have the confidence to power up the amp, hopefully there is a standby, and I'd set the bias to a very high (negative, of course) voltage to where I'd be pretty sure the output tubes would be shut off. Turn on B+. Await smoke. Or, if the bias was fixed but I had no confidence that the output tubes wouldn't overcurrent, I would have to build a simple bias supply and hotwire it into the circuit such that I could turn on B+ and be quite sure that the output tubes wouldn't fry. Give 'em -100 volts or so negative bias.

And basically, feed the amp a sine wave of an amplitude typical of how the amp would be used, and crank down the negative bias (eg; make it more positive, closer to zero, output section more and more turned on) while looking at the output on a scope and monitoring cathode current. At some point, the thing starts to amplify...or not...but assuming that I eventually get some kind of signal on my output, I would turn on the output section more and more until the cathode current (if the tube was the size of a 6V6) landed into 6V6 zone, or if larger, then larger. Run it for a while into a dummy load, check in to see if anything was burning or smelling funny.

If I was looking for tweed-style breakup, I might be looking for one thing. If I wanted "high-fidelity" I'd be sensitive to when distortion started showing up (which you can barely see on a scope)

As my confidence in the rig grew, I might bias the amp hotter and hotter until distortion showed up and then back off. I myself bias my amps a little on the cold side, but I am a jazz player and use a clean tone. I'd rather my tubes lasted longer, but that's just me. Tweed amps are sort of not useful to me. YMMV.

[PRR]

> bias it correctly

What is "correct"?

[jeff]

I don't know. I guess that's what I'm trying to figure out.

[frus]

You increase the bias until the tubes start redplating, then measure plate voltage and current and thus you get max dissipation?

[stratele52]

First thing , you have to identify tube. If not you work like a blind.  Looking for red plating ; good look . It could red plating after  1 to 30 minutes. And this is not the good way to do IMO unless you are a pro in fixing amp.

You can find enough power tubes's photo on the web to see what you have. And there are not many power tube used.

What the year of the amp , country of origin ? It is acheap or good quality amp ? This help to find wich tube.

[labb]

Watch part two of this video.

Marshall DSL100 repair at Earhart Electronics, part 2

[PRR]

It is an interesting thought-problem, although artificial and unrealistic.

There is a range of "correct" biases. If battery-power speech/music, you want to bias very cold for maximum battery life, even so that soft sounds are rough. For sweetest low-level sound you want a hot bias. For most uses, anywhere in the middle is close enough.

> until the tubes start redplating

Not always a good guide. It is possible to make a tube with materials which would not survive red-heat. And there are tubes designed to run red-hot which go bad if run too cool.

It helps if you have assumed the amp was known to work fine before the bias upset, and you know the correct load and connections and useful frequency response.

Bring the amp up to clipping. With the bias mis-set, this could be clipping in the driver, so gently turn the bias and see if output clipping changes. You don't have to be exact; at some point the output levels off.

Now take it to half that power level. Measure input DC power (DC current and voltage) and audio output power. The difference is plate dissipation. If the tubes are appropriate to the job, this dissipation is very near the rated dissipation, or at least perfectly safe.

If electric cost and heat are not issues, set idle bias between 50% and 90% of that dissipation. Compare the two extremes, for roughness or blistering Tolex.

[jeff]

"It is an interesting thought-problem, although artificial and unrealistic.

There is a range of "correct" biases."

My inital reason for the thought-problem was trying to understand bias better. I was figuring there was a "correct" bias, (I gotta realize that there isn't). 
In other words IF there was one correct bias setting and IF the bias pot was turned up, I was thinking the only way to rebias it "correctly" was to restore it to that orignal bias setting. My question was an attempt to find how one would go about that if 70% of the max diss was unknown.

Thanks for the replies, I need to start running some experiments.

Jeff