Vanishing voltage - why is my amp leaking?

[Lectroid]

I'm rebuilding a Bell&Howell 179 conversion, and replaced the original PT with a ClassicTone 40-18021.  I followed a fairly typical new amp startup procedure with one exception: I did not form the new filter caps.  At all. Schematic, pics below.

With no tubes in, I have 750V across the rectifier tube pins.  With no LBL and a 5Y3,
B+ at A node was ~420V with a bias voltage of -34V.
 
After I put the 6V6s in, the B+ voltage dropped to around 370V at A node, right in the ballpark, but the voltage kept bouncing around in small increments.

When I tried to measure voltages at the plates, the highest voltage I measured on a given tube immediately began to drop as I watched the meter, at a rate of about 1 V/second. And as the 6V6 plate voltages dropped, the cathode bias voltage went up.  When the bias voltage headed north of 38V without slowing down, I stopped and removed all the tubes until I can get this voltage problem sorted. 

Possible issue:
Here I should mention that while my original version of this BH179 had a "D" node filter cap, I left it off this version.  The D dropping resistor drops right off the C node to power the 6SL7.
==> Is that missing "D" filter cap the cause of this steadily dropping voltage? 

Voltage is really leaking away, so I'm sure I've got a hidden connection to ground at some point but I haven't found it yet.  Is there a way to figure out electrically if I have such a hidden ground?  All the B+ dropping resistors test nominal.  I've tested the reservoir cap; it's within tolerance.  Should I unship the others and test them?

Has anyone ever seen this, or knows how I could track it down? 

Thanks in advance.

[SEL49]

When I tried to measure voltages at the plates, the highest voltage I measured on a given tube immediately began to drop as I watched the meter, at a rate of about 1 V/second. And as the 6V6 plate voltages dropped, the cathode bias voltage went up.  When the bias voltage headed north of 38V without slowing down, I stopped and removed all the tubes until I can get this voltage problem sorted. 

This is a classic example of run-away bias. And your pics reveal there are no grid leak resistors for the 6V6s. The solution is to connect a 220K resistor between pin 1 and ground for each 6V6.

[Lectroid]

SEL49,

I can't believe I left off the grid leaks!!   Thanks for the quick diagnosis and the nice answer.  I didn't know runaway bias was a thing before this, but the idea certainly explains what I've been seeing.  No connection from cathode to grid = no bias applied. 

And if my mistake is a classic example,  I am gratified that something I've contributed here rises to the level of "classic."   

Thanks again

[pullshocks]

SEL49,



And if my mistake is a classic example,  I am gratified that something I've contributed here rises to the level of "classic."   

Thanks again

 

[wsscott]

I’m confused, as usual.  Since pin 1 in a 6V6 is unused, why connect a resistor between pin 1 and ground?  Why wouldn’t it be pin 5, the control grid pin that is used?

[Lectroid]

It's an old well-known builder's trick I learned here.  It's best to be able to solder a resistor between two stable points.  Pin 1 on a 6V6 is usually NC, no connection, unused.  We can use that pin just like a terminal strip.  The 4.7K grid stopper on each tube is soldered from pin 1, right across the socket to pin 5, the actual grid pin.  Electrically, connecting to pin 1 is the same as connecting right before the grip stopper, which is right where my missing 220K grid leaks should have been attached.  That's why SEL49 suggested I run them from pin 1 to ground. 

Which did solve the runaway bias condition (thanks again SEL49).  The amp still has no sound but at least I can test voltages and trace a signal now.

[wsscott]

So did you run a separate wire from Pin 1 to a ground point on the chassis or somewhere else?

[Lectroid]

So did you run a separate wire from Pin 1 to a ground point on the chassis or somewhere else?

Not in the sense of a separate wire.  What I added was a 270K grid leak resistor from each 6V6 grid stopper (4.7K). 

On each 6V6 socket, at pin 1, the signal from the PI comes together with the 4.7K grid stopper resistor going to the grid, and the 270K grid leak resistor going to ground. 

On the schematic, find the black connector dot on the schematic, next to each 6V6's grid pin.  It is where three connections come together at one point pointr:

• the signals coming from the PI to the 6V6s grids
• the 4.7K grid stopper leading to each tube's grid
• the 270K grid leak resistors leading off to GND

That dot on the schematic, is "Pin 1".

The picture below shows the two 6V6 sockets in the foreground, and the new grid leaks (with arrows) stretching from pin 1 of each to ground. Also from each pin 1 you can the 4.7K grid stopper resistors soldered between pin 1 and pin 5, that we were discussing; they're blue.

Hth.

[wsscott]

Gotcha-Slucky and you referred to a 220k, and the schematic shows a 270k, so I wasn't sure if you left something out during the build, or needed to add additional resistors.  I couldn't tell from the photos where the ground was located.

Your new photo clears it up for me.  Thanks for taking the time to respond.