input jack wiring to kill hum

[uneumann]

OK - so I'm probably hitting a topic that's been covered extensively before, but I have a particular angle on this question and would like to hear opinions and experiences. I don't think that there is a single right answer.

The issue is wiring between the input jacks and the first tube. There are a few common resistor networks that run between the jacks and the tube - and for the sake of this question, let's use the common Fender network. Let's also assume that the input tube has a bypassed cathode bias. That probably covers about 80% of the amps out there.

My question is this... how best to arrange and wire the resistors and RC bias network on a chassis to minimize hum?

I've drawn three possibilities below. Each circuit is exactly the same, other than how the parts are positioned and wired in the chassis.



Option (a) shows the jacks to the left and three wires to the RCs in the center and two wires to the tube. This is meant to convey that the RC components are on a circuit board somewhere with wires to the jacks and wires to the tube. This is pretty much the layout for hand-wired eyelet boards. Take the Bassman 5F6a with an eyelet board as a specific example. There is no shielded cable and all wires are reasonably short and direct. This is how Fender did it. In my experience, they all hum when you max the volume with no input connected.

Option (b) is what many have described as an improvement over (a). In this case resistors are mounted on the input jacks and shielded wire is used to make the run over to the tube grid. The dotted line denotes a shielded line with the shield grounded only at the jack end. The cathode bias RC, however is still mounted on the board. I've done this and it seems an improvement over (a), but it still hums at max volume.

Option (c) is what I recently discovered to work even better than (b). In this case all the RC components are mounted on the jacks and a two-conductor shielded cable makes the two connections to the tube - so nothing is mounted on the eyelet board. In discovering this, I found it eliminated all but the faintest traces of hum to the point where a low level of hiss is now the dominant remaining noise.

I know there was a recent discussion about hiss reduction, but in most amps I find hum outweighs hiss as a problem. I could go into my rational for why I think option (c) works best, but I'd rather get some reactions to this -- Is this commonly known? Did I just not get the memo? I'd particularly appreciate comments from anyone using other options that they find work well.

[Gingertube]

You use an isolated from chassis input jack.
From that you use shielded cable back to the V1 gridstop resistor (hot) and the 0V end of the cathode bias resistor (shield). This (tube end) will also have the Rg1 (1 M) to 0V connections.
Given decent earthing etc then the 68K grid stop is the major source of noise in a guitar amp. Reduce the 68K to 10K and add extra capacitance from grid 1 to 0V (minimum of 100pF max of 680 pF, 470pF is good))
to keep high frequency roll off much the same as before. Use a 10nF Ceramic cap to tie the "cold" side of the isolated input jack to chassis at RF right at the jack - this has the effect of extending the RF shileding effect of the chassis out along the length of the guitar input cable.
To keep hum low, use a centre tapped or Psuedo-centre tap heater wiring feed OR DC power to input tube heaters. Use a BIG bypass cap on the cathode bias resistor (low immpedance at 50/60 Hz) or if wanting to use that capacitor (low value) for tone shaping then use a DC heater. Use DC elevation of the heater this needs to be at least 20V but up to 75V can be used safely.
Cheers,
Ian

[Alan0354]

Learn what BF and SF Fender did, do not isolate ground. Have good ground connection to the chassis. If you have the tube towards the back of the amp like Fender and with the turret board between the input jack and the tube, have the cathode bypass cap grounded on the chassis close to the input jack.

If you have the input and first stage at one end of the chassis and the power amp on the other end of the chassis like Fender, as long as you propagate the signal step by step from input to the power amp, Just ground onto the chassis along the way. DO NOT DO STAR GROUND along the way. Only the ground that carry high current need to be careful. The ground of the reservoir cap, the screen grid cap have high current flowing, they need to be close to the cathode grounding of the power tube. I would tie all the cathode grounding into one star to the chassis in the middle where the power tubes are and put the ground of both the Reservoir and screen cap onto that star. With that, the high current never get onto the chassis. Your chassis essentially become a quiet signal ground. The second less important is the ground of the PI filter should go close to the cathode grounding of the PI. But this is much less important, just at the vicinity is good already.

You start cutting up ground, staring out ground at different points, you'll be playing with fire. I just did a high gain amp, I verified the only hum is from the filament of the first preamp tube and I got rid of most of it just by a rectified and filtered DC on the filament, not even regulated DC. I have absolutely ZERO problem with the hum......except from the guitar. I remove the grid stop resistor and the amp is very quiet when I remove the jack and input grounded.

[uneumann]

Thanks - all good points, Ian, Alan. This wiring issue was the last in a series of hum reduction measures I pursued. I never considered the cathode circuit a hum risk so I did everything else first... I put in a DC heater (regulated). I elevated all the heaters using +60v using a diode/cap from the bias tap. The rest of the amp from volume control to speaker was and still is dead quiet - you literally can not tell it is on. Yet the hum from V1 (pre-volume stage) was still there - I could see about 3-4 mvpp on the plate of V1 with the input shorted.

I think the lesson learned is that the cathode connection to the bias RC is also important to either keep short or shield. It's common to shield the grid line, but it seems that the same care is needed on the cathode line as well. I had about 4 inches of wire from the tube socket (pin3) to the bias RC on the eyelet board. That's all it takes to pickup that hum. By moving the bias RC to the jack (short wires to ground) and shielding the wire (about 5-6 inches) to the cathode - the hum is gone. No component changes - no cost - just layout changes.

I suspect moving the bias RC to the tube socket and running shield to the jack, as you suggest Ian, is also a good method (maybe better if there are some spare pins around the tube socket). In either case, the connection from cathode to bias RC has to be short or shielded. I not going to undo the DC heater or elevated supply, but I suspect they were never the problem. As I made those changes there was no noticeable change in the level of hum. Anyway, I think this is something worth remembering and sharing since it's simple, costs nothing, and makes a big difference.

If you're interested my now "quiet" bassman schematic - it's below...

[Alan0354]

Running unshielded wire 4" or more on the cathode or grid is inviting problem. If you have the tube towards the back of the amp like Fender where the turret board is between the input jack and the first tube, you should be able to have the cathode line very short onto the turret board, then the other end of the bypass cap grounded right on the input jack side.

In you design your own amp, you should put the first tube close to the input jack, that's what I did on my high gain amp. All the ground are at the same vicinity and leads are almost point to point. that's what's so important to do the layout before building.

FYI, from the way you described, you can move the cathode wire and change the hum. The wire of the input to the grid and the cathode wire form a loop antenna. Any EM wave passing through the loop will cause a current flow in the loop.....which creates a voltage. This is simple transformer. The voltage (EMF) is proportion to the TOTAL flux going through the loop, which is equal the flux density B X Area of the loop. The longer the wire, the farther apart the grid wire and the cathode wire, the larger the area enclosed inside the loop, the more induced voltage in pick up. Usually if you tie the grid and cathode wire close together, you minimize the loop area and reduce the noise. But now you might cause other problem like increasing capacitance from grid to cathode etc.

Having one of the two wire shielded does not help as the shield form it's own loop also.

Best is to have wires short. I design my amp into a fender bassman 100 chassis. I actually drilled holes for preamp tube butt up to the input jack. As I said, I have absolutely zero hum problem.

[Chuck H]

If you want to kill hum at the input my experience is that a standard shielded lead for the grid only is best. Other things that may be done would include using good quality, low impedance decoupling capacitors, fully bypassing the first gain stage cathode (again with a good quality, low impedance capacitor) and run grounds on their own leads without daisy chaining. ie: The cathode ground for the first gain stage is a single lead to the ground point, the input jack - lug is a single lead going to the ground point, The shield ground is a single lead going to the ground point, etc. Works for me every time.

Regarding fully bypassing the first gain stage cathode... There are other ways to get +5dB (or whatever depending on your resistor value) shelving for a given frequency that don't sacrifice any more gain than partial bypassing. Fully bypassing reduces hum from the heater/cathode circuit. This as it applies to AC filaments.

If you have DC filaments I'm not sure what sort of hum you're trying to eliminate. If it's random EMF then the absolute only answer is shielding the grid lead by itself. Combining other leads within the shield should do nothing to reduce EMF noise, UNLESS!!! There is a possibility that two out of phase leads carrying the same noise and held in close proximity MAY reduce noise through phase cancellation. Considering the circuit operation any noise cancellation phenomenon would have an equal affect causing signal cancellation. So while hum may indeed be audibly reduced, not noticed is that the signal is too. So the signal to noise ratio remains the same. There is therefor no relevant reduction in hum that couldn't be achieved by simply reducing gain.