I'm restoring an HP 400C AC VTVM. I've re-capped it, checked all the resistors, replaced the selenium filament rectifier, etc. I'm most of the way there. However, I started calibrating it and ran into a problem on the 100 volt scale. It was fine up until about 60 volts and then it became progressively less less linear, so that 100 volts input only registered as about 80 volts on the meter.
I checked all the other ranges and discovered that the problem exists on the 0.1 volt, 0.3 volt, 100 volt, and presumably the 300 volt ranges. I don't have an audio frequency AC source readily at hand that will go up to 300 volts, so I couldn't actually test the 300 volt range.
However, from the schematic, you can see that the 0.1 volt range shares the same input divider tap as the 100 volt range, and the 0.3 volt range shares the same tap as the 300 volt range. (Click on the thumbnail above to get the full sized schematic.) So I think the problem is common to the 300 volt range, also, though I can't verify that assumption.
I hooked my scope up to the output terminals of the HP 400C and discovered that as I increased the input voltage on the problematic scales, the waveform would distort and break into high frequency oscillation on the positive peaks.
Now the manual cautions that if the Gm of the 6AK5's is too high, the amplifier may oscillate. But the tubes are all out of either this 400C or its mate. I may have swapped the tube sets between the two units as I forgot to label them. I also forgot to label the tubes with their corresponding socket, so I may not have the 6AK5's back in their original sockets. I did try swapping V2 and V3, but didn't see any change.
The oscillation only happens on the scales where the grid of the first 6AK5 is connected to either R8 or R9. These are the two lowest-impedance taps on the input divider. So the oscillation only happens when the grid is connected to a low impedance source to ground, through C5.
I checked the signal from the cathode follower (V1) at the top of the input divider, and the waveform there is fine.
I put a scope probe on the grid of V2 (top trace, channel 1) and on the amplifier output (bottom trace, channel 2).
At about 60 volts input, little blips of oscillation start to appear on V2's grid.
As I increse the amplitude of the input, the blips get bigger ...
... and bigger ...
... and bigger, and the oscillation becomes very evident on the output waveform (bottom trace).
I tried probing the plate of V2 and the grids and plates of the succeeding stages, V3, V4, and V5, but the capacitance of the scope probe killed the parasitic oscillation and the amplifier went back to operating correctly.
Now the HP 400C amplifier circuit is unusual in that there is feedback from the plate to the screen and also to the cathode. This puts me in over my head on vacuum tube circuit design. I don't understands pentodes well enough to grok the implications of that feedback. I suppose it was done to linearize the amplifier.
I probed the screen of V2, shown as the top trace in the above photo (the bottom trace is still the amplifier output).
Here's the screen of V3, which to my naive eye looks suspiciously like the source of the oscillation, ...
... the screen of V4, ...
... and the screen of V5.
The manual has you perform various acts of voodoo to peak the high frequency response, such as varying the inductance of the ground connection of V2, V3, and the feedback network, and connecting a small capacitor between the center ground shields(!) of the sockets for V3 and V4. But unlike its earlier-serial-number mate, this particular unit doesn't have any of those bits of black magic. So there isn't anything in the circuit that is by design goosing the high-frequency response.
However, the replacement polyester film capacitors probably have much lower inductance than the original paper capacitors, which were physically huge.
I should mention that I haven't done a leakage test on the mica capactors as I was reluctant to disturb the wiring to test them out-of-circuit.
Any ideas on the cause and possible cure for this oscillation problem?
Update:
I got some advice from the friendly folks on the Test Equipment section on the Antique Radio Forum. I started looking for feedback through the power supply as a possible cause of the oscillation, and I thought I found it in the form of a bad ground. But beefing up the ground didn't cure the oscillation.
I noticed that no matter where I put my scope probe on the power or ground rails, there was a high enough parasitic inductance for me to see the oscillation on the scope - 10 mv or so. So it wasn't a poor ground, just the normal designed-in parasitic inductance in the power supply rails. (The 400C is constructed on a phenolic terminal board. There are no ground or power planes like you get with a modern multi-layer PC board.)
So I tried reducing the inductance in what the manual calls the "ground loop" by shorting points A and C as shown in Figure 3 from the manual.
I also reduced the inductance to point B by routing both legs of the loop together.
This was enough to kill the oscillation. I thought I had probably compromised the high frequency response by reducing the ground loop inductance, so I broke out the HP 651B test oscillator and ran a sweep on the 0.1 volt scale with a scope on the 400C output.
The 400C output did show a drop in response with a corner frequency of about 600 KHz or so, but you could see some phase lead kick in from around 1 or 2 MHz up to 6 MHz, and then the response dropped fast. Now the output is taken from the cathode of V5, but the 400C meter measures the voltage at the plate of V5. The meter response didn't drop at all. It was flat up to near 2 MHz, and then it rose to a peak at about 4 MHz and then rapidly fell off over 6 MHz.
As suggested by Rapaich on the Antique Radio Forum, I tried a square wave test and saw almost 100% overshoot on the leading edges. This all adds up to an amplifier with a nasty high frequency peak in its response that I somehow introduced during the restoration.
I broke out my copy of Millman and Taub and read their chapter on vacuum tube video amplifiers. The lightbulb went on when they described cathode compensation. The small capacitances across the cathode resistors are to compensate for the plate capacitance of the tube and any parasitic capacitance in the plate circuit. Now if you compare the coupling capacitors from Figure 3 above to my photo of the restored terminal board, you can see that the smaller size of the coupling capacitors, along with the re-routed wiring, introduces much less parasitic capacitance into the plate circuit. So now the compensation capacitors on the cathode resistors are too big, leading to an overcompensated frequency response with a big overshoot on the leading edge of a pulse and a big peak in the frequency response.
That peak is also probably the source of my oscillation.
My choices now are either to reduce the size of the cathode capacitors or to add capacitance to the plate circuit. I'll try the latter by adding some gimmick capacitors.