Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum. New: view Recent Searches.
New: visit Schematic Hell!
The sunn still shines online!

Listen to great tunes streaming live right now!

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

Thoughts on input stages....

9/29/1999 3:27 PM
Ken Gilbert
Thoughts on input stages....
I'm building myself the quietest preamp I possibly can. It got me to thinking about balanced guitar outputs and differential input stages.  
It wouldn't be too hard to isolate the - from the gnd lead in the guitar. You could then use a stereo 1/4" plug, with gnd on sleeve, - on ring, and + on tip, as is the standard balanced 1/4" plug. This way, if you had a SE guitar amp, you could merely use a MONO 1/4" plug, shorting - to gnd and establishing a normal connection.  
But, if you plugged into a guitar amp that had provision to take advantage (using a 2 conductor + shield input jack, wired similarly to the jack on the guitar itself), then you could easily achieve much better noise performance. Using a differential input stage with a high CMRR, any signal common to both + and - would be attenuated, and any differential signal (actualy guitar output) would pass through unaffected.  
Again, by using the stereo 1/4" plug, the ring (-) and sleeve (gnd) would be shorted together when you plugged in a normal, mono, SE guitar cord. The diff amp would operated normally in this mode, with half the gain of course.  
The input stage could be a current sourced differential (aka long-tailed) pair. By using a current source in the tail you can achive much better CMRR. This is because a common mode input (like noise) would attemp to "move" both cathodes at the same time. The current sourced tail approximates an infinite resistor fed from an infinite voltage. In our case, it's not quite that ideal of a CCS, but you can easily achive dynamic impedances of a half a megohm or so. This will stabilise the tail current, which prevents either tube from "stealing" current from the other. That's just another way of saying it's got high CMRR.  
Of course this approach will not stop any hum that is injected into the pickup itself--it will only minimize that which is picked up AFTER the pickup. Thus it may not even be worth it... but it IS worth it to talk about!  
Other things to consider would be volume/tone controls on the guitar itself, which would have to be modified as well.  
At the moment I'm thinking of a differential JFET front end, cascoded with both halves into a high gm tube like the 6N1P. This will irradicate the Miller input capacitance of the JFET (since the upper tube is operating in "current-mode"), offer good CMRR, and also offer the very high effective input gate impedance of the JFET (maybe up to a couple of megohms or so), which minimises the cable loading effect.  
The 6N1P can be operated at a lower plate voltage, say 100V, to minimise shot noise. The load resistors can be much smaller than your typical 12ax7--something on the order of 10K or less, which will minimise thermal noise from the load resistor.  
The bandwidth of this input stage will easily reach into the RF region, so that will have to be dealt with. Gain will be pretty damned high too, due to the cascode connection. Other than that, it seems like a little re-wiring and a few extra components here and there, and you've got a MUCH better designed front-end. After all, the input stage is the stage which distorts the LEAST out of all of them in the amp.  
Lemme know what you think. Maybe I'm just insane.  
9/29/1999 4:49 PM
Gil Ayan

Very thoughtful post, Ken. However, I wonder why you would ever wnat to do something like that. I don't know, in my world I use amps to play live gigs (in which case event the annoying single coil buzz gets drowned dow), or to record, and in that case there is always a way to minimize buzz (in my modest studio work, problem has always been buzz and not hum while in the studio).  
An old trick is the orientation of the mic, of course to get a good tone, but there is also one to minimize the buzz from getting onto the tape. So, with those two applications in mind, I guess any "averagely" quiet amp has worked for me.  
The use of differential amplifiers is mandatory is you're trying to reduce noise. Remember, every sinlge thing you add to the amp has a noise figure greater than 1... unless you start pulling some tricks, you will get even more noise, and unless the amp is builts very carefully on a PCB, etc., adding more components will probably increase the chance for parasitic oscillations in there somewhere.  
"Lemme know what you think. Maybe I'm just insane."
Naahhhh... although on second though... :)  
9/30/1999 12:14 PM
Andy Fuchs
you don't need all that.....
The average guitar amp input stage offers an adequate amount of gain at a reasonable level of noise. To gain a true benefit of balanced inputs, you would really need a balanced source. Okay, so center tap your pickups.... But seriously, switch the first tube to a 2SK30 ultra hi mu fet, and ditch the tube maybe. Perhaps parallel the 12AX7 halfs, and reduce the gain back to where one tube was (and buy some noise db's). But cascodes are super sensitive to outside world noises (I've tried them), and they amplify the capacitance of the input source, so your cables 'sound' will become an issue.
9/30/1999 7:16 PM
Andy said:  
"The average guitar amp input stage offers an adequate amount of gain at a reasonable level of noise."
And this is undoubtedly true. And it is also precisely why I want to try to optimise it. I'm not really in for the gain--the only thing I need gain for is to give the signal enough gumption to be a little more immune to stray noises. The input stage is the single greatest contributor of noise in the amp... it's contribution gets amplified by all subsequent stages. Thus, attention paid here to minimise noise will be most beneficial.  
I want an UNREASONABLY low level of noise. Hell, I already have an UNREASONABLY high power output capacity--600W. Now I'm seeking the space in between the notes; I want it black as the cloak of the grim reaper. I don't want it with noise gates or downward expansion or bandwidth limiting. I want it with whacked out, well-engineered design and quality components.  
Why not use the right tool for the job--a high gm stage with all WW resistors, operated low and hot?  
"To gain a true benefit of balanced inputs, you would really need a balanced source. Okay, so center tap your pickups...."
Umm, what do you mean by balanced source? I don't need to center tap the pickup at all. The differential input has no care what the reference of the signal input is--that's the beauty of it. It doesn't matter whether the source is balanced or not. It's not about absolutes at all, but rather the simple, relative difference in voltage between the two input signal leads (+ and -). Ground reference doesn't enter the equation one bit. The only ground I need is for shielding purposes, and nothing more.  
"But seriously, switch the first tube to a 2SK30 ultra hi mu fet, and ditch the tube maybe."
Methinks you threaten to offend the tube gods... "Ditch the tube???"  
"Perhaps parallel the 12AX7 halfs, and reduce the gain back to where one tube was (and buy some noise db's)."
This is an option of course. Double the gm, double the Pa, half the rp, half the Rk and Rl, mu stays the same. Resistor thermal noise drops substantially.  
"But cascodes are super sensitive to outside world noises (I've tried them),"
Hence the balanced input with as high a CMRR as I can muster. I am not sure how a cascode can be any more succeptible to "outside world noises" than another circuit with comparable gain. Could you elaborate a bit on this?  
"...and they amplify the capacitance of the input source, so your cables 'sound' will become an issue."
I'm not quite sure I follow you here. When you say amplify the capacitance, what do you mean? The voltage output from the pickup remains unchanged. The loading of the input stage can be largely ELIMINATED, using a cascoded FET--the gate is a VERY high impedance, and when cascoded there is NO Miller capacitance multiplication going on since the voltage of the drain does not change. If anything, this extremely high input impedance will LESSEN the effects of the cable construction, not increase them.  
Ken Gilbert  
10/1/1999 12:22 AM
Andy Fuchs

Firstly, the noise generated by the constant current source you would need to make an optimal differential amplifier, will most certainly be noisier than a cnventional single triode stage. A 6EJ7 (Pentode very high mu), would work well as a resistance coupled amp, and wired as a triode. (We don't want to hear the crap generated by the screen grid network do we ?). A diff-input would only prove beneficial if driven by a balanced signal source, so lose the effects of hum and extranious noises through CMRR. Now a cascode presents a new set of problems: A cascode by nature will amplify the capcitance presented to it's input grid. Low capcitance cables are a must. The second stage (flying up in the air so to speak), is very sensitive to RF pickup and electromagnetic noises. Lastly cascodes can't drive anything insofar as a load, without some form of follwer after it. So, to feed a tone stack (or anything except another hi resistance amplifier input), you would need a cathode follower or a buffer stage. Again, a simple ,conventional multi tube section in parallel triodes, is the most viable way to do this. Heck a bunch of 6DJ8s (4 with both sections in parallel) and a 25 watt plateload makes a swell low noise input stage (like was used for moving coil phono cartridges years ago. Nobody has used cascodes anywhere in musical instrument amplifiers, and I suspect this might be why. (I triode one as a reverb return amp and liked it), but never commercially. All of these reasons are based on facts in the Radiotron book, Electron Tubes By Seely, and The Audio Cyclopedia by Tremaine. Don't take this diatribe as a put-down, or being confrontational. I'm being realistic, and maybe hoping to save you some time and parts... af
9/30/1999 4:24 PM
J Epstein
Re: Thoughts on input stages....
"Lemme know what you think. Maybe I'm just insane."
I think we've alreday been over that, Ken.  
9/30/1999 5:42 PM
Don Symes

It's an interesting approach. My thoughtless opinions and questions would be:  
1) At the guitar end, if the metal pickup cover isn't connected to the pickup(-), perhaps that should get connected to the pickup/control cavity shielding you aren't quite thinking about yet.  
2) In the preamp, somewhere, you're going back to single-ended... how? transformer? push-pull triode pair?  
3) Are you familiar at all with the LesPaul Recording model? It used low-impedance pickups (don't recall if balanced) and a transformer at the preamp input to match up to the input impedance. Maybe some study of that concept, and if possible, some time using one, could save you a a load of grief.  
4) Don't let my uninformed opinion stop you.  
5) Good luck  
   Page 1 of 3 Next> Last Page>>