PP amp crossover distortion

[db!]

I recently (re)built one of my amps and I've come across a small problem. When I initially listened to the amp, it sounded like crap; loads of distortion. Upon scoping it, there was a large amount of crossover distortion.

The design and schematics can be seen here: audio diy

The circuit was built exactly as seen in the schematic with the exception of the 270r cathode resistor and the inclusion of CRC filter in the power supply. The cathode resistor was replaced with a simple LM317 CCS.

The test signal is a 70hz sine wave 350mv P-P. I'm not sure of the exact quiescent current but the voltage from cathode to ground is approx 28v.

Does anyone have any ideas where/what the problem might be?

[DavesNotHere]

the cathode is not low enough impedance. divide the push pull cathode and use 2 ccs (one per cathode circuit) or use the resistor-grouped cathode method like you had originally.

I don't like using ccs in the final stage since it effects the dynamic headroom of the output stage.

[nigelwright7557]

You need a way of adjusting the cathode current to get rid of cross over distortion.
Too little current and you get cross over distortion.
Too much current and you lose headroom.

[db!]

Originally Posted by DavesNotHere

the cathode is not low enough impedance. divide the push pull cathode and use 2 ccs (one per cathode circuit) or use the resistor-grouped cathode method like you had originally.

I don't like using ccs in the final stage since it effects the dynamic headroom of the output stage.

The 832A is a dual beam tetrode with a common cathode within one envelope. There is no way to separate the cathodes.

You need a way of adjusting the cathode current to get rid of cross over distortion.

I am using a small trimpot with the LM317 CCS. Increasing and reducing current has no real affect on distortion.

Is it possible that the distortion is coming from the driver stage? I haven't had time to scope this section yet.

[artosalo]

In the picture of your first post, there is no cross over distortion, which should appear symmetrically at rising and declinig side of the half cycle.
Instead this is unsymmetrical and possibly caused by saturation of the OPT's core.

The test signal is a 70hz sine wave 350mv P-P.

At input ? What is output level ?

I haven't had time to scope this section yet.

Take that time. With a scope you will see immediately where the problem comes from.

[DavesNotHere]

the only way I see is to get the impedance down in the cathode (if it was ok, then substituting the ccs caused the xover distortion).

use a mosfet based ccs instead of a bjt based like the lm317 voltage reg.

otherwise, I would directly ground the cathodes and run grid bias, but I'm kinda of a brute force guy.

grid leak resistor method- tieing the grid resistor 390k to the top of the ccs-cathode might give you that extra bit of electrons from grid leakage, but the mosfet based would be more of an academically sound choice.

[Jebem]

What kind of scope are You using?
Also, where are You checking the output signal?
Is the problem happening on both channels?

As a general troubleshooting steps, I would:
- Replace the output tube just in case.
- Check the components for wrong values or connections, restoring the original cathode resistor and cap to start with.
- Then check all the the voltage points.
- Then disconnect the feedback loop for a while to make further tests.
- Use a sine wave and a square ware at 1Khz (one at a time), 100mV at the input, so one can compare the results.
- Then use a scope to read the signals at anode of the amplifier section 12AT7 to see how clean is the signal.
- Then look for the signals at the inverter outputs (anode and cathode) of the second section of the 12AT7and compare them; it should be of the same amplitude but in opposite phase.

Good luck!

[tricomp]

Originally Posted by artosalo

In the picture of your first post, there is no cross over distortion

Exactly; this is not cross-over distortion.
What OP-transformer are you using?

A simple way to check if the CCS is to blame is of course to just replace it with the original cathode RC-complex.

With the cathode at 28V at present it seems to me that you're running the tube with too high bias getting into the non-linear region; check curves. It won't cause the distortion you're experiencing but it also won't sound very good. Probably somewhere around 20-25V is suitable, but don't go above 25V.
Pushing this tube into AB2-operation with a fixed-bias, grounded cathode and a suitable CF/CCS-drive seems more rewarding, judging from the curves. You're getting (should be getting) around 7-8 Watts from your design, but this tube holds potential for a lot more imho.

rgds,

/tri-comp

[SemperFi]

I agree with Artsolo, that is not x-over dist. The first thing I would do is check the testsignal at input tube, then grids of the 832. If all is well there, I'd disconnect the feedback, then send in a testsignal 500-1kHz, 100mV like Jebem said, and go from there.
At 70Hz the output transformer may be either giving you a phase shift (which may explain the non-symmetric waveform) or even be saturating (not likely), so using a testsignal that rules out the transformer is a better idea. Once you've figured out the amp at midfrequency, you can check out the low- and high extremes.

[DF96]

A CCS in this circuit maintains a constant mean current. What you actually need is a constant quiescent current. Not the same thing at all. You either need to run the valves hot with no signal, or accept crossover distortion with large signal as the bias shifts due to second-order distortion. Simple cathode resistor bias is better than CCS bias, as it partly compensates for bias shift with signal. Best of all is voltage bias.