Massive crosstalk in stereo LA-2A

[mjrippe]

I've got a stereo LA-2A clone here made by Tubetronix (they don't exist anymore).  It was supposedly working fine before shipping but arrived with bent rack ears, a busted VU, and crazy crosstalk between channels.  When either channel is at +4dBm output, the other one has the same signal at -8dBm!

Construction is point to point on tagboard, except for the solid state sidechain & metering section on PCBs.  So each channel audio path is essentially transformer, opto, 12AX7, 12BH7, transformer.  They followed the Urei schematic fairly closely and all components I have checked seem ok (tubes, caps, resistors).  Voltages are a bit higher but not abnormal.  Wiring is fairly organized and moving things around causes no change.

One odd thing, if I pull the 12BH7 from the channel with an input signal, the adjacent channel signal rises to -1dBm!  Pulling just the 12AX7 from the channel with input kills the output of both channels.  So of course I checked everything around the 12AX7 but no clues.

Any ideas or general things to check when there is crosstalk in a stereo tube unit?

 

[radardoug]

Do you have a proper ground on it? Do you know that your input and output wiring is done right? Transformer balanced in and out device, you need to wire it right.

 

[mjrippe]

Hi Doug,

Checked grounds and all seem good.  This is on my test bench attached to my Amber 5500 analyzer, so wiring is correct.  Only odd thing is pin one on the input XLRs was not connected to anything - not even the chassis!  I tried attaching to chassis with no change, but might leave it as a matter of good practice.

Mike

 

[sircletus]

I can see the output level changing when you pull a tube out of the other channel.  If B+ is unregulated (which it is on the originals, and probably is here) pulling a tube can raise the plate voltage on other tubes and potentially change signal level.

The only way I can think of that pulling one 12AX7 would kill output on both channels is if the heaters of the 12AX7s on the two channels were wired in series.  Pull one, the other one goes cold and turns off.

Any pics of the guts?

 

[mjrippe]

Hi sircletus, I am actually thinking that the crosstalk is occurring at the 12AX7 stage, hence yanking the one with signal applied silences the one without.  Filaments are all parallel.

Might get some photos next time I have it on the bench, but much is obscured under the tagboards so it is hard to follow visually. 

I guess my question is, what causes audio coupling through a shared power supply and how do we  avoid/fix it?  I know that the coupling caps between stages block DC, but what keeps the AC signal out of the PSU?

 

[PRR]

> what keeps the AC signal out of the PSU?

B+ filter caps.

 

[mjrippe]

> what keeps the AC signal out of the PSU?

B+ filter caps.

Hi PRR,

Pardon my ignorance, but we are talking about the AC audio signal not 60Hz power, correct?  How do the filter caps block that?  My thought was that the 220k plate resistors presented a higher impedance path than the coupling caps/grids so the audio would take the "easy route".  However the plate resistors are fine but audio is clearly coupling through the PSU.  I can scope the + side of the filter caps and see the waveform. 

Thanks,
Mike

 

[radardoug]

If you can see audio signal; on your power supply decoupling caps, then they are not doing their job. Either the cap is faulty, or its negative terminal is not connected to ground.

 

[[silent:arts]]

Don't know if this helps, but I have never heard of any crosstalk problems in D-LA2As, and they share the PSU transformer and the first filter caps. You can find the schematic here:
https://groupdiy.com/index.php?topic=32677.0

 

[PRR]

> the 220k plate resistors presented a higher impedance path than the coupling caps/grids so the audio would take the "easy route".

Huh? The plate resistor is 220K. The grid resistor is, say, 1Meg. The signa current splits in proportion, The "easy route" (like 4X easier) is through the 220K to the filter cap.

The filter caps BOTH reduce wall-power ripple from the B+ source AND reduce audio signals sneaking around through the B+ rail. See attached.

Why have you not already tacked a 40uFd across that filter cap to see if it makes a difference?