Announcement

Collapse
No announcement yet.

understanding fet noise curves

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by Gnobuddy View Post

    Interestingly, though, I have never noticed any of my guitar amps hissing more loudly when I turn down the guitar volume (though many times I've noticed the dreaded "tone suck", treble loss caused by the same increase in source resistance).

    Perhaps input noise is dominated by something else, perhaps I don't hear it because I tend to play using relatively low gain, and at modest volume.

    All the back-of-the-envelope calculations I've done invariably point to JFETs as being the quietest input devices at typical guitar source impedances. Not all JFETs, though, as we also want low flicker noise down to 80 Hz.

    -Gnobuddy
    You are right; you have to use a very quiet input stage and then you can hear it. The low pass effect of the cable capacitance removes lot of pot hiss. So that 33k grid resistor matters.

    Comment


    • #17
      Originally posted by nickb View Post
      The voltage gain Av was 24.7dB
      => Av= 10**(1.235) = 17.1 not 295
      So Ceff= (Gds*(Av+1) + Cgs) = (5 * 18.1 +20) =110pF
      and F-3db = 14468.6
      I thought db was defined in such a way that you get the same number whether you are referring to power or voltage. So I would say that your gain is 12.35 db, and then agree with your math..

      Comment


      • #18
        Originally posted by Mick Bailey View Post
        I do have one guitar that gets really noisy as the volume control approaches halfway - more hum and hiss that isn't there on full. I thought the hum was the ground-referencing, but I never could rationalize the hiss that crept in.
        Or you might have a bad pot.

        Comment


        • #19
          Originally posted by Mike Sulzer View Post
          I thought db was defined in such a way that you get the same number whether you are referring to power or voltage. So I would say that your gain is 12.35 db, and then agree with your math..
          Voltages are what we are interested in here. It comes from the definition https://en.wikipedia.org/wiki/Decibe...wer_quantities

          24.7dB is the correct figure.
          Experience is something you get, just after you really needed it.

          Comment


          • #20
            OK, you are right 24.7 is the gain, corresponding to a voltage gain of 17.

            Sorry, Nick,
            Last edited by Mike Sulzer; 08-14-2017, 05:34 PM. Reason: got it all wrong

            Comment


            • #21
              Great info here. I'm much more comfortable with all of this now. Thanks to all.
              I've built the attached test circuit. The J113 is in cascode with the LND150. I've used the LND150 by itself just to confirm how horribly noisy it is. It's horrible. In cascode however, it's fine and the overall circuit is very quiet. I've fed this into a SS PA and I can get really (uncomfortable) loud levels with barely a trace of hiss or hum.

              The guitar PU noise (even humbuckers) far exceeds the circuit noise. Hiss at zero guitar vol is almost inaudible. Hum+noise at max volume is MUCH louder.

              Click image for larger version

Name:	Capture.JPG
Views:	1
Size:	112.9 KB
ID:	846175

              The test configuration gives about the same gain as a standard 12ax7 stage (response curves below). I don't have an LSK170 yet, but the spice model shows similar gains to that of the J113. When I get a LSK170 I can compare noise levels.

              Click image for larger version

Name:	Capture2.JPG
Views:	1
Size:	163.0 KB
ID:	846176

              I've also verified some relationships experimentally and they now agree with my understanding of all of this (thanks again).

              - R11 and R8 both produce increased hiss as their values increase. This is no surprise, but now that fits with my new understanding of noise figure curves.

              - Noise in the cascode device (Q3) has minimal impact. I've read this before and the arguments make sense now. The test circuit is really quiet, so the evidence support that too.

              - I have not done a precise measure of tube vs FET noise (I'll work on that) but I've found plenty of arguments for FETs producing less noise. I found some numbers to back that up. Seems like the data is plausible. Maybe someone here can vouch for it.
              “If you have integrity, nothing else matters. If you don't have integrity, nothing else matters.”
              -Alan K. Simpson, U.S. Senator, Wyoming, 1979-97

              Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.

              https://sites.google.com/site/stringsandfrets/

              Comment


              • #22
                Originally posted by uneumann View Post
                I've found plenty of arguments for FETs producing less noise. I found some numbers to back that up. Seems like the data is plausible. Maybe someone here can vouch for it.
                Broad rule of thumb is 1uV EIN for a garden variety triode, 0.3uV EIN for a garden variety JFET, both assuming negligible source resistance.
                Click image for larger version

Name:	fox.jpg
Views:	1
Size:	124.3 KB
ID:	846180

                Comment


                • #23
                  Originally posted by Merlinb View Post
                  Broad rule of thumb is 1uV EIN for a garden variety triode, 0.3uV EIN for a garden variety JFET, both assuming negligible source resistance.
                  [ATTACH=CONFIG]44505[/ATTACH]
                  Thanks! The pieces are all falling into place and I'm left wondering if I'll ever have reason to use a triode input stage.
                  “If you have integrity, nothing else matters. If you don't have integrity, nothing else matters.”
                  -Alan K. Simpson, U.S. Senator, Wyoming, 1979-97

                  Hofstadter's Law: It always takes longer than you expect, even when you take into account Hofstadter's Law.

                  https://sites.google.com/site/stringsandfrets/

                  Comment


                  • #24
                    Originally posted by uneumann View Post
                    Great info here. I'm much more comfortable with all of this now. Thanks to all.
                    I've built the attached test circuit. The J113 is in cascode with the LND150. I've used the LND150 by itself just to confirm how horribly noisy it is. It's horrible. In cascode however, it's fine and the overall circuit is very quiet. I've fed this into a SS PA and I can get really (uncomfortable) loud levels with barely a trace of hiss or hum.

                    The guitar PU noise (even humbuckers) far exceeds the circuit noise. Hiss at zero guitar vol is almost inaudible. Hum+noise at max volume is MUCH louder.

                    This is an interesting circuit. The idea of using two FETs with different transconductances in order to suppress the noise of the second (upper) one in the cascode is described here: http://nvlpubs.nist.gov/nistpubs/jre...n6p383_a1b.pdf. I do not know if this is the first place it was described, but it is the first I have seen. The LND150 has a transconductance as much as 10 times lower than the J113, and so the suppression can be quite good. The cost is a bit more input capacitance from a small Miller effect C, but it is not a problem. And the LND150 can take the high voltage, which you must have in order to be able to utilize the same gain as a triode without overload given the high output of some guitar pickups. I think this is a very good way to replace a triode input stage.

                    Of course tube purists could use a tube-tube cascode where the first triode would have high transconductance and the second would have low. I suppose that if you look around, you would find that someone did this in 1939.

                    Edit: One more thing. I do not think that it would help to replace the J113 with a lower noise FET. The 2K source resistor would have more much more noise than the FET. It might be the limit with the J113 as well. With that 2K resistor, the noise is no lower than a 12AX7, but a good 12AX7 stage without the usual 33K resistor in series with the grid. (It is possible that you could use a small resistor and still achieve proper biasing with some FETs.)
                    Last edited by Mike Sulzer; 08-15-2017, 05:48 PM.

                    Comment


                    • #25
                      Another possibility (which I assume that someone has tried?) is to modify the FET/FET cascode back in the other direction: use a low noise, high transconductance FET in the bottom part of the circuit, and a low transconductance triode in the top, say the usual 12AX7, although you might need a bit more plate current than it can provide for some FETs. Then you keep the output characteristics (including some degree of nonlinearity), of the usual tube, but get reduced noise (although in this case probably not as good as a 170 is capable of, but still very good). Of course you also get a much reduced Miller effect, and thus a reasonably small Miller effect input C, although not as small as a cascode circuit using devices with equal transcendence.

                      (You would have to fool around with the source resistor to get the right biasing, and perhaps use two resistors in series, one bypassed wit a C in order to get the best compromise between noise and effective transconductance in order to not overload the circuit with a high output guitar pickup.)

                      Comment


                      • #26
                        Originally posted by uneumann View Post
                        Thanks! The pieces are all falling into place and I'm left wondering if I'll ever have reason to use a triode input stage.
                        I still like the sound of a triode input stage. Contrary to some published opinions, I found that, with almost all my electric guitars, there was a slight, but clearly audible, difference between a triode input stage, and a truly clean (lots of NFB) solid-state one. The triode produced a more attractive, more "valvey" or "tubey" clean, i.e., there were audible amounts of low-order harmonic distortion.

                        The exception was one Squier guitar with very low-output single coil pickups (read that as: Fender cheaped out on this model), which evidently didn't drive the triode hard enough to create audible amounts of harmonic distortion.

                        Of course your JFET also inherently produces 2nd harmonic distortion, and you might very well find it sounds just as good as a triode input stage. I don't know, I've never tried a JFET front-end, though it's been on my list of things to do for some years now. If you want, you could bypass that JFET source resistor with a capacitor to remove local negative feedback and maximise JFET nonlinearity.

                        -Gnobuddy

                        Comment


                        • #27
                          Originally posted by Mike Sulzer View Post
                          This is an interesting circuit.
                          <snip>
                          I think this is a very good way to replace a triode input stage.
                          Evidently, you're not the only one with that opinion. On a different forum, someone posted a commercial guitar amp schematic that had exactly the same input configuration (a small signal N-channel JFET cascoded with an LND150 MOSFET). I forget exactly what amp that was.

                          I'm still curious about sonic differences, if any, between a JFET input stage and a triode input stage. Aside from the audible triode nonlinearity I mentioned in my last post, a "classic" 12AX7 input stage operates at Vgk only about (-1.5 V). We know grid current starts to flow in a 12AX7 if the grid reaches maybe (-0.9 V), and we know plenty of guitar pickups can put out peak voltages of 1 V or bigger.

                          This suggests to me that some signal clipping may actually occur right there at the input grid of the first 12AX7 triode that the guitar sees, because the high source impedance of a guitar is not going to supply the grid current drawn by the triode without having the peaks flattened. I don't know if this is audible or not, but I wouldn't be surprised if it turns out to be a factor in the sound of, say, heavily strummed chords played through a triode vs a typical solid-state guitar amp. To me, almost every solid-state guitar amp sounds much harsher (than a valve amp), even when played entirely clean, i.e., not audibly overdriven.

                          It is possible that choosing a JFET with the right Vgs (and carefully chosen gate resistor) might allow one to rougly replicate the grid current effect of an input triode; the gate-source diode will also start to conduct once Vgs gets to around (+ 0.5 V) or so.

                          Originally posted by Mike Sulzer View Post
                          One more thing. I do not think that it would help to replace the J113 with a lower noise FET. The 2K source resistor would have more much more noise than the FET. It might be the limit with the J113 as well.
                          A few months ago, I decided to buy a few JFETs before they went entirely extinct. There were only a pitiful few part numbers to choose from at Mouser, and after looking at a few datasheets, the J111, J112 and J113 were ones I thought were good candidates, because all of them looked to have very good noise figures, and good flicker noise down through the entire guitar bandwidth.

                          If the input resistor is eliminated entirely, it might be necessary to take some other steps to keep RF and high-frequency hash out of the input. Maybe a ferrite bead with a few turns of wire wrapped around it.

                          -Gnobuddy

                          Comment


                          • #28
                            Originally posted by uneumann View Post
                            Thanks! The pieces are all falling into place and I'm left wondering if I'll ever have reason to use a triode input stage.
                            Because even a triode is quiet enough for guitar, and it likes high voltage, which is what you conveniently have in the amp already. Noise in guitar amps normally comes not from the valves but from the resistors, especially the input grid stopper which is commonly rather large. Reduce it to 10k or even less (with suitable extra shunt capacitance if necessary) and you can approach the irriducible noise of your pickups.

                            Mike Sulzer
                            Another possibility is to modify the FET/FET cascode back in the other direction: use a low noise, high transconductance FET in the bottom part of the circuit, and a low transconductance triode in the top
                            Yep. But you don't need to fret about the relative transconductances in a cascode (well, maybe if you're designing RF antenna amplifiers, but not guitar amps). Even if you use two identical devices, EIN will always be doninated by the lower device.
                            Click image for larger version

Name:	wes.jpg
Views:	1
Size:	97.8 KB
ID:	846188

                            Comment


                            • #29
                              Originally posted by Merlinb View Post
                              Even if you use two identical devices, EIN will always be doninated by the lower device.
                              I do not see how you arrive at that conclusion. The analysis in the paper (http://nvlpubs.nist.gov/nistpubs/jre...n6p383_a1b.pdf.) shows that the two contribute equally (with identical gms) unless I am grossly misreading something. In any case, the obvious approximate analysis, without writing anything down, is that the lower device has unity gain (gm/gm) and that therefore its noise contributes the same as the top device. If you have a better analysis, I would be happy to listen.

                              And do not be concerned with my fretting. That is just how we "internet experts" are.

                              Comment


                              • #30
                                Originally posted by Merlinb View Post
                                Because even a triode is quiet enough for guitar, and it likes high voltage, which is what you conveniently have in the amp already. Noise in guitar amps normally comes not from the valves but from the resistors, especially the input grid stopper which is commonly rather large. Reduce it to 10k or even less (with suitable extra shunt capacitance if necessary) and you can approach the irriducible noise of your pickups.
                                But it is not irreducible.

                                There are two ways that I use to reduce it:

                                1. Medium impedance pickups: Just use fewer turns of larger wire and use Cs and Rs to get a switched flexible tone. For example, you can lower the impedance a factor of ten while losing a factor of about three in signal voltage. You do not need a transformer or a preamp if you improve the noise performance of the amp (assuming that you are playing without pedals).

                                2. Pickups with a better magnetic return path. For example, you can make a single coil size pickup with the normal inductance, but a factor of three lower resistance. You can use lower value pots, or you can set up the circuit so that the tone control is normally nearer the middle like nearly all other audio devices.

                                Even with 250K pots, if you play with the volume control of one pickup very close to zero in order to get a nearly clean sound while the other pickup is at full volume (you need a volume pedal at the amp master volume to make this work right) you need lower than "normal" noise.

                                Sure, most guitarists just do the standard thing, but in this forum we discuss ways of pushing that.

                                Comment

                                Working...
                                X