Lowering noise floor on fender and some Marshall style amps

[mikepukmel]

Hi All,
I had found a post by an amp guy on a Facebook page, and can't find it any longer (as Facebook goes). Sure you guys have run into this. In Merlin's book and other articles, they mention that the 68k grid stops on the input jacks are responsible for most (can't remember the percentage) of the johnson noise (hiss?).
Anyway, the guy, instead of two 68k resistors in the standard configuration, put I think 1 x 10k grid stop right on the tube socket, then two more resistors in the standard configuration (the 'low' jack hits both in parallel, the 'high' jack hits one resistor). I can't remember the size of the two on the jacks. But he mentioned that it lowered the noise floor, and as far as the grid stop, performed as well.
I sorta, kinda, sorta get the idea that smaller resistance values or physically larger resistors will have lower johnson noise. But I didn't quite understand why the configuration would work as well, for the grid stop/oscillation and lower the noise as well. If the resistors total have lower resistance, won't they be less effective at stopping rf and oscillation?

Thanks,
Mike

[Leo_Gnardo]

If you can stand having only one input jack per channel, ditch the #2 inputs, run your input wire straight almost to the grid pin, then connect to as small a resistor as you can get away with as a stopper right on the grid pin. Yes those pesky 68K's are noise generators. I've used 33K and 22K in this way, less noise but otherwise no problem. You might be able to reduce to a yet lower value. There may come a point where you start to hear scratch noises as you turn the volume pot on the guitar, then you know to back up one step and that should do it.

[eschertron]

There may come a point where you start to hear scratch noises as you turn the volume pot on the guitar, then you know to back up one step and that should do it.

Oh! I've encountered that in my builds with a 10k at the input. I've some experimenting to do...

[Leo_Gnardo]

Oh! I've encountered that in my builds with a 10k at the input. I've some experimenting to do...

Sometimes you get that because of preamp tubes that are leaking grid current. Old Fender amps with conductive circuit board material can also be the culprit but obviously not the case on a fresh build I'm sure. You can also beat that problem by borrowing a trick from grounded-cathode amps: pass the signal thru a cap say 0.01 uF, make sure you have a grid leak resistor on the tube side of that cap, the usual 1 meg is fine but you can pick whatever suits you. The larger the value the brighter. We had a discussion on this here a couple years ago, and IIRC the value 6.8 megohm was mentioned as a maximum. In any case do have a grid leak because your pre tube will go bazerko without one.

[Chuck H]

Grid leak bias with a grounded cathode should be great for a lot of things. sensitivity is maxed and noise is minimum. Maybe good for an effects loop or reverb recovery? But I'm not sure the input stage of a modern guitar amp is a good place to use it. They tend to get wonky with higher input drives, such as boost pedals or other dirt boxes adjusted to give some boost. Translate = does not accept pedals well.?. I think this is why most amp builders that were using grid leak bias first gain stages in the early years changed to the more typical cathode biased stage. I don't have personal experience with it, but I've read that even hot humbuckers can cause strange behaviors in grid leak bias first gain stages. In any case, what I've read has put me off of trying them in that role.

I think Merlin suggests a 10k grid stop with a capacitor to ground. The capacitor is valued to reject RF without notable detriment to the guitar signal. I'm sure it's here (somewhere) and on his site.

Like Leo I've used 22k and enjoyed reduced noise without any bleeding RF. But if 10k and a cap is better.?. I haven't tried it yet. I probably will sooner or later.

[the fatch]

Hi All,
I had found a post by an amp guy on a Facebook page, and can't find it any longer (as Facebook goes). Sure you guys have run into this. In Merlin's book and other articles, they mention that the 68k grid stops on the input jacks are responsible for most (can't remember the percentage) of the johnson noise (hiss?).
Anyway, the guy, instead of two 68k resistors in the standard configuration, put I think 1 x 10k grid stop right on the tube socket, then two more resistors in the standard configuration (the 'low' jack hits both in parallel, the 'high' jack hits one resistor). I can't remember the size of the two on the jacks. But he mentioned that it lowered the noise floor, and as far as the grid stop, performed as well.
I sorta, kinda, sorta get the idea that smaller resistance values or physically larger resistors will have lower johnson noise. But I didn't quite understand why the configuration would work as well, for the grid stop/oscillation and lower the noise as well. If the resistors total have lower resistance, won't they be less effective at stopping rf and oscillation?

Thanks,
Mike

In the Fender input circuit, input 1 consists of a 1M grid leak followed by a 34k grid stopper (two 68k in parallel). Input 2 consists of a 68k grid stopper followed by a 68k grid leak. As Merlin points out, the grid stopper forms a low pass RC filter with the input capacitance of the triode to ‘stop’ radio frequencies appearing at the grid. Reducing the value of the grid stopper increases the cut off frequency of the filter. If you take Merlin's value of 200pF as the input capacitance, then a 34k grid stopper cuts frequencies below 23kHz. A 10k grid stopper begins to cut at 80kHz. As Merlin says, radio frequency interference below this is rare, but adding extra capacitance in the form of a 470pF capacitor from grid pin to cathode pin brings the cut off frequency back down to around 28kHz. It boils down to whether you need input 2 to be usable. If not, you could simply replace the 1M with a 1M metal film, the two 68k with two 22k metal film and add the 470pF cap. You could also replace the plate resistor with a metal film. Ideally, the input should go straight to the 1M grid leak and from there by the shortest route via a shielded cable (with its shield grounded at the input jack) to a 10k grid stopper connected as close as possible to the grid pin. Merlin also points out that reducing the grid stopper below 10k has little effect on noise, and that many stomp boxes are unhappy driving less than 10k.

[Enzo]

Don't forget the real purpose of the two 68k resistors was to form a 2/1 voltage divider on the #2 jack. A 6db pad at the input.

[the fatch]

Don't forget the real purpose of the two 68k resistors was to form a 2/1 voltage divider on the #2 jack. A 6db pad at the input.

The 68k/68k pad is only active when using input 2. I was assuming that input 2 was not going to be used. With my simpler mod, input 1 has a 1M grid leak and 1.1k grid stopper to work with the 470pF cap. Input 2 now has a 2.2k grid leak and 2.2k grid stopper which will pass a signal to the grid, but not exactly a useful one with the resonant peak of the pickup completely damped by a 4.4k input impedance. The idea was to keep it simple by substituting three resistors (or four including the plate resistor) in their current physical location in the amp.

[mikepukmel]

Thanks everyone, yeah, will never use two jacks on one channel. Probably wil never have two instruments plugged into this amp. Can live with just the 'low' input for certain. So, going to, say two 47's in parallel (at the jacks) = 23.5 + 10k at the tube = 33.5k total won't change the noise floor over stock two 68k in parallel = 34k. its the total resistance from the jack tip to the grid on the tube that determines the noise?

[eschertron]

So, going to, say two 47's in parallel (at the jacks) = 23.5 + 10k at the tube = 33.5k total won't change the noise floor over stock two 68k in parallel = 34k.

See what Enzo mentioned, there is no reason to use any other resistors at all with a single jack - besides the 1M grid leak and the (optional) grid stopper. The two 68k resistors formed a voltage divider when the low input was used. If you want a 'low' input only, you'll want to not go any lower than the 68ks for dividers, to avoid loading the input down too much. Note the topology difference: in a modern 'high' input, the leak is before the stopper, in the classic Fender 'low' input, the leak is after the stopper, to form the divider.

[Leo_Gnardo]

Thanks everyone, yeah, will never use two jacks on one channel. Probably wil never have two instruments plugged into this amp. Can live with just the 'low' input for certain. So, going to, say two 47's in parallel (at the jacks) = 23.5 + 10k at the tube = 33.5k total won't change the noise floor over stock two 68k in parallel = 34k. its the total resistance from the jack tip to the grid on the tube that determines the noise?

Not much point in adding an extra 10K at the tube. Remember, the stopper is best located as close to the tube as possible, right on the socket. Fender (and everyone else) has been putting them at the wrong end of the wire since forever. Just because it's "traditional" doesn't make it right. In mods I've done, and occasional scratch builds, stoppers AT the socket, even if it is the traditional pair of 68K's.

[Enzo]

fatch, I don't disagree, but I felt it was left out of the description of the basic Fender circuit.


mike - it isn't about two instruments in one channel, the two jacks are so you can have one with a pad, the lower gain #2 jack. You may never use that either, but good to know why it was there. The low input is the one that makes the voltage divider.

[Mike Sulzer]

fatch, I don't disagree, but I felt it was left out of the description of the basic Fender circuit.


mike - it isn't about two instruments in one channel, the two jacks are so you can have one with a pad, the lower gain #2 jack. You may never use that either, but good to know why it was there. The low input is the one that makes the voltage divider.

I thought it was intended to have the Hi or Lo function when used with one instrument, or add the two inputs through 68K resistors when two instruments were used.

[Leo_Gnardo]

I thought it was intended to have the Hi or Lo function when used with one instrument, or add the two inputs through 68K resistors when two instruments were used.

My explanation, others may support or disagree:

I can't remember having two instruments plugged in that way since I took part in low-tech jam sessions in college. I s'pose, if you're in a pinch... And I do have one customer who plugs 3 keyboards into his Bassman head, two into one channel that way. But it is rare that our muso friends do that.

The padded input comes in handy if you have a guitar or bass with a built in preamp, or pass signals thru effects that boost volume to the point the input pre is "crushed", IOW overloaded to the point of distorting. 6 dB isn't a lot, but at least it's something. Back in the 50's when this type of input arrangement became popular, nobody was thinking about guitars with preamps nor effects pedals.

Guitar straight to amp? No preamps/pedals? The second input not only pads 6 dB, it presents the guitar with a load of 136 Kohms, obviously loading down the pickups, robbing high end clarity and probably pinching off some lows too. I never much liked it until I started working for a band that specialized in Philly soul & Motown style music. That second input comes in real handy for rhythm guitar tracks, automatically giving them a period-authentic low-fi mid rangey "boxed in" tone. So... it's good for something.

Hope this helps!

[the fatch]

OK, Enzo, point taken. When input 2 is in use, the two 68k resistors form the divider you mentioned, but when input 1 is in use they are connected in parallel and become a 34k grid stopper. If they are replaced with 2.2k it forms a 1.1k grid stopper. Ok, they’re not in the ideal position the at the grid pin, but along with the 470pF cap they will reject RF interference just the same as the original 68k while generating much less noise. (Of course, you can't do this mod if you need input 2 to be usable.)