Is it worth biasing with a distortion analyzer ???

[bembel]

Hi all,

I just finished fixing my new (acquired) good ol' Marconi distortion factor analyzer.
(some Ge Xtor deads, a few electrolytics and some solder joints... Then it was very amazing to see that the AC voltmeter was still calibrated for a 40 years old device).
First constatation, my heathkit AF generator is not better 2.5% THD even after alignement !
Then I decided to make a Test CD, I downloaded a shareware app. called "Test Tone Generator" changing it's parameters for CD quality, done some different personal test tones & added the ones of the very regreted Fred Nachbaur (http://www.dogstar.dantimax.dk/tubestuf/index.htm). then I put a old sony CDP-70 on the bench ---> Now better THD=0.02% mostly 2nd H (I was starting to get afraid that the distortion meter wasn't totally fixed yet!), tried a chinese "hi-end" cheap discount DVD player by curiosity THD=0.08% mostly noise

This thing is much than I thought an unvaluable tool !!! see the look of distortion...

Now the real question ? is it worth biasing whith a distortion analyser ? Of course it's measurement is dependent of the "set ref level function" and biasing changes gain. But it's still usefull to balance push-pull pairs is one shot ? isn't it ?

Any experiences ?

[cerrem]

Biasing has got to be the most "subjective" topic with setting up amplifiers....
First thing is to define terminology...
If were refering to a Class AB1 audio amp ,then to be accurate there is no "cross-over" distortion in Class AB output stage....
The term "cross-over" officially belongs to Class B circuits where you have a small angle of no conduction from either side of push-pull.....But that will start a big argument....since a Class AB amp when biased on the colder side will exhibit a "notch" at the point of inflection in the sine wave cycle..... This "notch" is a result of the extremely non-linear region of the dynamic transfer curve right "before" cut-off..... But over the years the term "notch" and "cross-over" became overlapped....
The standard our forefathers set for audio amps was to bias into a resistive dummy load and to bias the amp at the point when you achived maximum clean sine-wave output...right on the hairy edge before you notice any flattening of the sine wave on the scope..... Then I use the scope to eliminate the notch in the sine wave by starting cold and slowly adjusting hotter.... but that is subjective by the quality of the waveform on the scope screen and the quality of my eye-sight.....
Now if you bias any further, it is considered lowering the efficiency of the amp and running the output tubes hotter than need be....or not??? This is when I use the spectrum analyzer to see whats happening.... You are free to bias from this point where the notch fades out up till the maximum plate dissipation for that given valve......At this point you have a trade-off, of running the tubes hotter and shortening thier life vs. distortion products that you measure vs. if you can hear the difference......
Some folks like to use the 2-tone Inter-Mod test using a 4:1 amplitude over a decade.... Then you specifically look at the intermod freq and adjust to a spec or to what the amp does best... If the amp does not meet spec, than the tubes are tossed till you get a really good matched set of tubes...The matching is half the battle...since you may have perfectly matched brand new valves, but the transconductance may be low...You need strong transconductance matched valves...but those days are long gone when you had tons of Sylvanias and RCA,GE tubes to match up....good luck with todays power valves...
In theory, running the tubes as hot as you can "usually" is the most linear operation... But in practice the power supply may take a dip and may be lower in B+ than expected..... with increased loading comes increased ripple on the B+ rail...ect..ect..
What I can recomend....find the % distortion you can accept listening to and then adjust the amp to that spot...
The only rule is you bias above where the notch fades and not go past the MAX plate dissipation....you will find there is a small spot
that meets these requirements....and will have to live with what the tubes are capable of....

Chris

[rdf]

Originally posted by cerrem
In theory, running the tubes as hot as you can "usually" is the most linear operation...

Chris

Hi Chris. In my experience it depends highly on the application and the definiton of linear. Typically I find running tubes hotter lowers the second harmonic at the cost of increasing higher order components. In SE the tradeoff depends on subjective balance of the different components. In the case of 'perfect' push-pull however, if you're stuck with distortion the ideal scenario is second harmonic only. On the bench right now I have a very lightly loaded single-ended 6C45 into a cathode follower swinging about 160 volts p-p with second harmonic ~60 dB down, 3rd ~90 db down and no other harmonic visible above -125 dB. If each half of a 6C45 p-p circuit maintained the same performance under the same operating parameters in each leg, the maximum performance assuming perfect balance is completely cancelled 2nd and 3rd harmonic about 85 dB down, or ~0.006% THD. In this case raising the bias would degrade the THD figure by raising the third, the reduction in 2nd in each leg would have no impact because it's being cancelled.

As you can tell, yes indeed I do indeed lean heavily on a spectrum analyser when tuning circuits.

[HighVividity]

In my opinion, THD analyzer is not very useful for the power audio amplifier. THD analyzer analyzes the THD with a pure sin wave input. This does not tell you too much, as music is highly dynamic. As pointed out by Norman Koren, "In a well-designed high fidelity power amplifier, the lower the total harmonic distortion at rated power, the harder the clipping; hence the worse the sound." ( http://www.normankoren.com/Audio/FeedbackFidelity2.html ) I believe he is right. Please note the condition, "at rated power". Otherwise his statement will surely be misunderstood.

So let's talk about High Vividity, not High Fidelity (to run after low THD at rated power).

[consort_ee_um]

Well done fixing the analyser. I used one of those many years ago.
Nice summary from Cerrem with a slight improvement I can suggest for Class B amps
There is an easier way to adjust the bias on solid state amps. With the amp under load, if you monitor the VAS stage with an oscilloscope and put a sine wave in you can see the pertubation where the upper and lower stages switch. Because you are monitoring inside the feedback loop the amp is compensating for the non-linearity which is visible. You adjust for the best approach to a straight line. Presumably the same trick could be used on a valve amp, although you have to be careful with the voltages!
With other bias schemes the distortion analyser would be really useful

[EC8010]

Glad to hear you got the Marconi going. I think you'll find it more useful on stages other than the output stage to help you determine the optimum bias point or configuration - at least, that's my experience.

[bembel]

Thanks all, for your enlightenment.

Having fixed that stuff, I try to figure out some other uses (than the obvious one) to the THD analyzer.
It seems that his tool is more useful to some than others ...
Personally, it remembers me my first oscilloscope, and ... the end of blindness that came with it.
It is so instructive to me to see the look of distortion (THD analyzer scope output) , that I'm thinking for "dynamic procedures". just to learn what's happen.

Maybe a drawback I didn't mention, is the somewhat low input impedance of the Marconi (10k-100k) depending on the scale.
It should be a problem when monitoring interstages with InputZ in // with big grid resistors. Maybe >1Mohm voltage divider can help here ?

Thanks again.


PS: I've just found someone that made me a copy of the original service manual, expecting it for the end of next week, then a small alignement won't be bad. Maybe it include some specials measurement techniques.

[richwalters]

Despite the emphasis on power stages, I can thoroughly recommend a THD analyser for setting up earlier stages, viewing vast differences between triode and pent designs, especially driver stages. In my 40 yrs experience it's often here that a poor sounding amp starts and not as so often always blaming the output stage.

Intuitively it's a dammned good way to grasp what's actually going on, and with it sticks in the mind. The thing to watchout that the loading Z of any probe and screened lead can be capacitively signifigant on tube circuits when it comes to measuring harmonics. The pitfall is that high impedance often misleads.

richj

[rdf]

Originally posted by richwalters
Intuitively it's a dammned good way to grasp what's actually going on, and with it sticks in the mind.

Couldn't agree more.

[bembel]

thx Rich,
I have to add, that the Marconi input scale is limited to 30V, a bit problematic for high voltage outputs probing.

The thing to watchout that the loading Z of any probe and screened lead can be capacitively signifigant on tube circuits when it comes to measuring harmonics.

I didn't thought about that. A good thing to keep in mind! thanks.