Oscillation in JMI Vox AC30 Top-Boost Brilliant channel

[Rhodesplyr]

I’m at the noise-tracking stage of the restoration of a mid-1960s JMI Vox AC30, and I’ve got one of those “Is there something wrong, or is it just the way it works?” questions, having never worked on one of these before.

The Brilliant channel is noticeably more hissy than the Normal and Tremolo channels with a fizzy quality that suggests high frequency oscillation. Hooking up the scope, I did, indeed, find a small sine wave oscillation at around 50kHz (if I did the math correctly) at pin 1 of the Top Boost 12AX7. The oscillation tracks with the volume control of the Brilliant Channel and disappears when you turn it all the way down. It does not, however, appear to originate at the input gain stage.

The hiss also becomes much more noticeable when you turn the Treble control so that the wiper is towards the 50pF capacitor end. (As an aside, the Treble control appears to work backwards in terms of what one would usually expect in terms of direction of rotation.) I tried tube substitutions to see if the oscillation was more likely with one brand or type of 12AX7, but that made no difference. Moving the leads slightly had no effect, either.

What I’m wondering is if a certain level of background hiss is normal for the Brilliant channel of a Top-Boost AC30 due to the additional high-gain stage. Even if it is typical, would this be a good place to add either a grid-stop resistor right at the control grid tube socket terminal of the high gain Top-Boost stage--or a Miller capacitor of a few picofarads--to tame the tendency to oscillate at inaudible frequencies?

I’ve replaced all the electrolytics and a couple of other problematic components, including all the black, tubular Hunts caps, and a chopstick helped me track a few problematic original solder joints. I used a polystyrene cap for the 500pF coupling cap to the Top Boost tube. (The original was leaking, as everyone predicted.) I checked to see that the shell of the Brilliant Channel volume control has a good connection to ground. The hiss in the Brilliant channel was there before this work was done, but, unlike many of the other odd noises the amp was making, it did not go away when I was done. I’ve also checked the integrity of all ground paths with my Blue ESR meter.

Thanks for any helpful suggestions!

[Rhodesplyr]

Two things to add:

More investigation showed that the hiss does originate in the input stage, V1. The only thing that completely stops it is hooking an alligator-clip lead between ground and the junction of the two 68k resistors at the input jack. I'm not sure what that reveals about the origin of the noise, other than it directly grounds the input tube's control grid, rather than having it ground-referenced via two 68k resistors in parallel. The input signal runs via a shielded cable to V1. Is noise getting on the shield via the grounding system? Hard to say.

However, I experimented with a 39pF capacitor between the plate and control grid of the top-boost 12AX7. That dropped the noise quite a bit, crossing the line between annoying and acceptable. It doesn't seem to be enough additional Miller capacitance to impinge on guitar frequencies--to my ear at least.

That said, is there any reason to prefer a grid-stopper resistor or a combination of a grid-stopper and a Miller capacitor to just a capacitor, other than the potential for a plate-grid short in the event the cap failed? The capacitor is somewhat easier to add. And how high can one go in terms of a Miller capacitor before you start to cut noticeably into frequencies that a guitar speaker can reproduce?

[Chuck H]

And how high can one go in terms of a Miller capacitor before you start to cut noticeably into frequencies that a guitar speaker can reproduce?

Sorry I can't say about the rest... It's really a bench issue that's hard to troubleshoot remotely. But on this matter I can say that VERY small capacitors can cut well into the audible frequencies. With the more typical 100k plate load and about 200V on the plate I can easily hear the effect with a 4.7pf. Which I do use on one of my designs. I can easily hear the difference between 4.7pf, 6pf and 10pf. I tried them all, moving down in value as I tested all the way down to no cap at all. I've read that the actual capacitance of this circuit is affected by plate voltage. And of course the knee is affected by plate impedance. Since the Vox TB has a higher plate impedance and a lower voltage your results could be different. I wouldn't think it would be different by much though.