nfb causing buzz on new 18 watt plexi

[hasserl]

Just got this thing fired up the other day and one issue is really screwing with me, and that is I'm getting a serious buzz from the nfb.

This amp is basically a plexi style preamp/pi into a pair of cathode biased EL84's. The PI and Presence control are straight JTM45 (or 5F6A) , the nfb is taken off the 16 ohm wire at the impedance switch, thru a 100k resistor to the junction of the .1uf cap (to the PI grid opposite of the input grid), the ground side of the PI tail resistor and the input of the Presence pot (5K). Nothing unusual really.

If I ground the PI tail resistor the buzz goes away. But of course, I also loose the Presence if I do that. Lifting the nfb cures the problem , and the amp sounds gorgous like that, I'm tempted to just leave the nfb off. But I know that there's got to be a reason it's causing the buzz,

I tried reversing the OT leads at the tube plates, thinking I might have positive feedback, but that just made things much worse, the added a loud squeal to the problem, so I re-reversed the plate leads, putting them back to where they were originally. I've replaced the .1uf cap, thinking perhaps it was causing the problem, that did not help. I replaced the Presence pot, that didn't help. I even took the Presence pot out of the circuit and just used a 6k8 resistor (closest I had to 5k on hand) that didn't help. It's definitely a buzz caused by the nfb into the PI, but why?

Note, in my grounding scheme I grounded the Presence pot to the smae point as the filter cap that supplies the PI, and I also grounded the isolated speaker jacks there. In troubleshotting I've moved both of those grounds around to different locations, none of which helped the buzz at all.

Any thoughts?

[Dagan8]

I don't remember if the symptoms are exactly the same but I remember reading about noise issues with the original presence control circuit on the JTM45/5F6A.

Apparently the redesigned presence circuit of the 59 Bassman LTD reissue addresses this.

Might be worth checking out a schematic.

[dai h.]

could be the grounding. Neg. fdbk. (as I understand) should be grounded on the side it's returned to. So if returned to the PI, then the PI and not with the spk. jack ground. And on Marshalls and Fender Bassmans that is how the network (27k/5k, etc.) is grounded.

the "noise issue" (I'm thinking) is probably in reference to the "scratchy pot syndrome" from the DC on the presence pot and not to do with buzz.

[hasserl]

could be the grounding. Neg. fdbk. (as I understand) should be grounded on the side it's returned to. So if returned to the PI, then the PI and not with the spk. jack ground. And on Marshalls and Fender Bassmans that is how the network (27k/5k, etc.) is grounded.

the "noise issue" (I'm thinking) is probably in reference to the "scratchy pot syndrome" from the DC on the presence pot and not to do with buzz.

Could be ground, but I did try moving the grounds to different locations which didn't change the noise problem at all. Grounding schemes is a topic that I still struggle to grasp, but in a few recent threads on the topic I got this from Randall Aiken:

if you are using global negative feedback and isolated output jacks you should also run a wire from the output jack to the ground side of the PI tail. This is because the PI is amplifying the difference between it's grid and local ground at the bottom of the cathode, as Mark is also referring to.

Since you are returning the speaker output (divided by the feedback ratio) to the grid of the PI, you need to return the ground reference to the bottom of the tail, where the PI is grounded, so it amplifies the difference between the two points. The PI tail should also be grounded to it's regular place on the star or buss or whatever.

If I understand correctly this means running the speaker jack ground to the same ground point as the PI, which technically is the ground point of the Presence pot, which serves as the RS of the voltage divider in the nfb network, in this case it's 100k/5k. The tail resistor isn't actually grounded directly, it reaches ground thru the 5k Presence pot (or in some cases, such as with a 1959 Super Lead it reaches ground thru a 4k7 resistor, the Presence pot in that case is a 25k pot fed by a .1uf cap that bypasses the 4k7 resistor). In this case I've returned that ground to the same point as the filter cap that feeds the PI.

But then while troubleshooting I moved the speaker ground directly to the chassis, that did not help. Note the Presence pot is "scratchy", but the loud hum is in addition to that. It is loud, this is not a faint noise.

[JoeM]

Have you tried a non-isolated speaker jack to see how it affects the noise?

[Dagan8]

the "noise issue" (I'm thinking) is probably in reference to the "scratchy pot syndrome" from the DC on the presence pot

Yes. You're right. The presence circuit on the 59 Bassman RI and probably the 1959 Super Lead will fix the "scratchy pot syndrome".

[dai h.]

if you are using global negative feedback and isolated output jacks you should also run a wire from the output jack to the ground side of the PI tail. This is because the PI is amplifying the difference between it's grid and local ground at the bottom of the cathode, as Mark is also referring to.

Since you are returning the speaker output (divided by the feedback ratio) to the grid of the PI, you need to return the ground reference to the bottom of the tail, where the PI is grounded, so it amplifies the difference between the two points. The PI tail should also be grounded to it's regular place on the star or buss or whatever.

this is a bit confusing since the second part seems to agree with what I'm saying (what I read) but the first part is a head scratcher. From what I gathered (reading, looking at chassis, trying things), the spk. jack ground should be grouped with the power tube cathodes (basically how it's wired on the Soldano SLO100 and what appears to be the intention on a typical Marshall with the difference that Marshall uses the chassis to connect spk. jack ground to pwr. tube cathode to next pwr. tube cathode, etc.). In the Aiken star grounding article, also, (IIRC) the spk. jack ground is grouped with the power tube cathodes for one star point.

grounding is basically that a circuit is made of a bunch of DC and AC loops (sometimes both on the same loop), and when ground is used for return parts the loops (which have various magnitudes of AC and DC) you need to connect them together (since they all have to connect) in a way which is not deleterious to circuit operation (hums, buzzes, instability). So for instance the return path or portion of the ground line which sees the high current charging pulses (rectifier to filter cap) makes a poor place to connect the input 1Meg ground even though on the surface it might look like "zero ohms" (nothing to develop a voltage drop over), there is a small finite non-zero impedance which the high current pulses (line freq. or multiples of) can develop a voltage drop ( = signal) over (which you don't want getting in to your signal).

[hasserl]

Well I agree that this may be a ground issue, so I will try again moving the grounds. However, I fly out early tomorrow AM, so this will have to wait until later in the week when I get back.

Thanks for the suggestions Dai.

[JoeM]

hasserl,

You've probably already seen this article on grounding, but just in case not, there's some more info on grounding:

http://www.freewebs.com/valvewizard2/Grounding.pdf