Odd biasing problem with SE EL84 amp

[David McGown]

I am trying to debug an amp that I helped Jason Garel build last year. The circuit is based on the DECWARE SE Stereo ZKit1, but point-to-point wired instead of using (buying) the circuit board and kit components. It is cathode biased with a shared cathode resistor for both channels (150 ohm) for the two EL84s. The driver (6DJ8 plate resistor loaded common cathode) is capacitor coupled to the grid of the EL84, with a 330K grid leak and 1K grid stopper resistor.

Problem Description:
With the amp started, both output tubes bias up with 12.5V across the shared cathode resistor, i.e., 83mA total current. After a few minutes, I am seeing the voltage at the grid of one tube (at only one of the two sockets) rise from nominally 0V with respect to ground until it eventually after several minutes rise to the cathode voltage. Naturally, at this point the tube is conducting heavily with a bit of an orange glow on the plate. Overall voltage across the shared cathode resistor rises to about 20V, with a toasty 133 mA total current.

Swapped tube, and the problem followed the socket, not the tube, I checked all connections, everything reads A-OK in terms of value and layout. Tested the coupling cap, with the output tubes removed, and the capacitor passes no DC from the driver tube plate. Resoldered all the connections just to make sure there was not a bad solder joint.

Still, the problem is there. The tube sockets appear to be fine, new bright condition. The amp was setup in UL mode AND triode mode exhibiting the same problem. One thing of note, the voltage is a bit high for an EL84, i.e. 325V on the plate, but using Russian EL84s (New Sensor "Mullard"s). Under normal operating conditions, the plate dissipation would be 13W, which is a bit high. Oh, there is a 1K screen stopper installed, and pin 6 is not connected to anything (should be internally connected to cathode in the tube)

The puzzling thing is both channels are absolutely identical in layout, but only one socket has the problem, and the problem follows the socket, not the tube. Therefore, it is not a case of a mismatched tube.

Any ideas?

David

[Roscoe Primrose]

Double-check the value of the grid leak resistor on that channel. If it's higher than it should be, a small amount of grid current could be causing the problem...

330k is fairly high, you might try a lower value unless the coupling cap is quite small...

Roscoe

[David McGown]

Double-check the value of the grid leak resistor on that channel. If it's higher than it should be, a small amount of grid current could be causing the problem...

330k is fairly high, you might try a lower value unless the coupling cap is quite small...

Roscoe

Roscoe,

I guess I should "triple check" it but that is easy enough to do. The coupling cap is a 0.1uF Multicap. One thing, the leads are a bit taut, it is at least possible as the amp warms up, there "could" be thermal stress on the leads, causing a connection problem at the resistor end cap. But this is just speculation, and will have to check when I get home.

David

[J-ROB]

Max grid resistor for 6BQ5 self bias is one meg. Of course, this presumes good NOS-grade tubes. New Sensy Mullards aren't actually Mullards in any way.

In other words, try a different pair of tubes that speak an Indo-European language.

Try pulling the output tube when it goes into overconduction and test for DC leakage in the coupling cap at that point, rather than when cold. A Multicap shouldn't leak, but who knows?

And if the grid rises to exactly the cathode value, mistrust your tube socket....sure there's not a micro hair of solder bridging the terminals? Check with loupe, scrape and clean. If it is a junky Chinese socket, remove it, scrub it carefully with concentrated muratic acid, and if it doesn't dissolve into sludge, throw it away anyway.

[dberning]

Could be contamination of some sort. I think I would remove the tubes and apply a high voltage on the grid pin of the socket to see if you can clear any contamination. While the most likely source of such leakage would be from the plate pin or the screen pin to the grid pin, these are not next to one another in the EL84. Leakage to the cathode is less likely because of the low voltage on the cathode, but the pins are next to one another.

David

[Stuart Polansky]

A Multicap shouldn't leak, but who knows?[/quote]

I had a 4.0uF Multicap fail without provocation. Put ~200 volts to the DH500 Black Gate bipolar cap, which in turn conducted, causing me to spend days repairing the delaminated voice coil wires in one speaker. Multicaps do fail.

Stuart

[tomp]

Stuart:

Here is a speaker protection circuit that may help prevent voice coil burnouts.

Tom

http://www.ebay.com/itm/Assembled-board ... 3413wt_960

[David McGown]

Thanks to all for your good suggestions of things to check. I'll have you know, I failed Troubleshooting 101 by assuming the output tubes were actually good, since the problem seemed to be attached to one socket. Swapped in a matched pair of Sovtek EL84Ms, and the amp biased properly and stayed stable. I did check the grid leak and grid stopper resistor prior to new tubes, and resoldered the connection just to be doublely sure to no avail, only cleared up with the new tubes.

I guess I have been away from my soldering iron too long, or the effects of the lead is finally catching up with me.

David

[Guy]

But David McG, the problem is gone - that's the good news! And it's good for me to know that I'm not the only one here to make an occasional slip.
A good thread - I liked how it progressed. My thought was to replace the socket. It's like reading a mystery novel!