Ad Widget

Collapse

Announcement

Collapse
No announcement yet.

Silly-simple JFET (guitar) buffer

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

  • Silly-simple JFET (guitar) buffer

    I have a Joyo compressor pedal that works fine with my acoustic-electric guitars, but produces audible treble loss when I plug in my electric guitars. The acoustic-electric guitars have onboard preamps and low output impedance; the electric guitars don't. Evidently, the input impedance of the Joyo isn't high enough to avoid the dreaded "tone suck".

    So I needed a buffer. I happened to have a few old MPF 102 JFETs sitting in the parts box. And I wanted to build the thing quickly, since I hoped to use it the next day, so I decided to try the simplest possible circuit (attached). Keep in mind, I was looking for a completely transparent, always-on, clean buffer - this isn't intended to produce any sort of overdrive or audible coloration to the sound.

    The price one pays for this super-simple circuit is that the bias point will vary considerably from one JFET to another, and that will affect how big a signal you can feed in before the buffer clips. (Clipping might not sound good - the circuit has 100% negative voltage feedback, and usually, lots of negative feedback equates to harsher clipping.)

    To see if this would be a problem, I bread-boarded the circuit; with a 10k source resistance and 9V DC power supply, all the MPF 102 JFETs I had settled down with the source somewhere between 1.41 and 1.72 volts. That means even the worst of these should be able to handle a guitar signal up to 2.8 volts peak-to-peak, which is comfortably big enough for most guitars and playing techniques.

    I built my final version on a little scrap of proto-board, using the JFET with the largest Vgs (1.72 volts). I added a red power indicator LED and (820 ohm) series resistor, which I forgot to draw in the LTSpice schematic. I power my pedals with a One-Spot wall-wart, so I did not include any kind of on-off switch or bypass capability in my little buffer. No controls of any sort, in fact.

    I thought about adding a footswitch and output level pot, so the pedal could double as a rhythm/lead level adjustment switch. But I'm using it in front of a compressor, which is the wrong place to try and change the signal level. So I decided not to bother with the pot or switch.

    I tested it out last night, and I can't hear any audible treble loss from any of my guitars now. I can't get it to clip with any of my guitars plugged in and strummed hard, either. Eureka, it works!

    I should stress that there is nothing magical or mojo-tistic about this circuit at all. Its designed from the perspective of a lazy audio engineer (though I am not in fact an engineer, I've played that role a few times). This circuit simply does its job, using a minimum of components, no fuss, no muss, no magic. If you need a simple, completely clean and transparent guitar buffer, this one might suit you.

    -Gnobuddy
    Attached Files

  • #2
    I build these FET buffers with DC blocking cap on the input and a second resistor from gate to V+. This resistor should be identical to R1 (in your diagram), but because they effectively appear in parallel need to be doubled in value to maintain the same input impedance as before. So where you have 1M I would use 2x2M. This improves the signal output symmetry between different FETs if they have wildly differing characteristics.

    This circuit is tolerant of increasing the source resistor considerably to reduce battery drain if it's (say) built into a guitar without adversely affecting performance.

    Comment


    • #3
      Originally posted by Mick Bailey View Post
      I build these FET buffers with DC blocking cap on the input and a second resistor from gate to V+.
      Thanks for the comment! Oh, I know the "proper" way to do it. Yes, that would bias the gate to half the supply voltage, the source a bit higher, and give you near-maximum headroom for a given supply voltage. That is a bit closer to the carefully correct engineers way to build a JFET buffer.

      I built this circuit the way I did, precisely because I did not want to bother with an input coupling cap, or a second gate bias resistor, or additional power supply filtering. When you add that second gate resistor, half of any noise or ripple on the power supply appears at the gate - and at the output of the circuit. So now you have to break that upper resistor into two, and hang a filter cap off the junction, to make sure the DC bias voltage at the gate is clean. And because the input coupling cap now has DC voltage across it, you really should add yet another resistor to the input end of that cap, to keep it from popping loudly if you plug in a guitar with the JFET powered up.

      Supply noise is much less of an issue with the stupid-simple circuit I used. The JFET has considerable power supply noise rejection in source-follower mode, because small voltage fluctuations at the drain produce almost no change in source current.

      So the stupid-simple version I built gets rid of three resistors and two capacitors, compared to the "proper" way of doing things. As mentioned, the price paid for that is reduced input headroom, but it turns out not to be an issue. I ended up with an input headroom of roughly 3.4 volts, peak to peak, and that is far more than I can get out of any of my guitars, even by aggressively strumming big fat six-string cowboy chords. I tried! (Keep in mind the typical 12AX7 triode input stage has less input headroom than that!)

      BTW, I added some supply filtering anyway, to ensure stability at the end of a long 9V cable from the One-Spot power supply. I added a 470-ohm build-out resistor for the same reason.
      Originally posted by Mick Bailey View Post
      This circuit is tolerant of increasing the source resistor considerably to reduce battery drain if it's (say) built into a guitar without adversely affecting performance.
      Sounds good! In my case, I didn't have a high-efficiency LED lying around, so the LED power-on indicator draws 10 mA, and the JFET draws 0.17 mA! So my build of this circuit would actually benefit much more from a switch to a high-efficiency LED than a larger source resistor!

      -Gnobuddy

      Comment


      • #4
        Hello,

        Do you have the schematic of the circuit? I am having problems with a circuit using bf245c. My circuit has one resistor at drain of 2k2, and the source to ground without resistor, the gate is grounded with a 1 Mohm resistor, and I have added a 6n8 input capacitor, and a 47n output capacitor, and two zener diodes of 5.1 v to protect the base of the jfet. I have tried the circuit with an lm386 amp circuit, and I can barely hear anything, so I must misunderstood something. My transistor has Vgs=3.2 V and Idss=4.6 mA. I am a bit confused now.

        Comment


        • #5
          The schematic is in the first post image. here is a larger version.

          Originally posted by Enzo
          I have a sign in my shop that says, "Never think up reasons not to check something."


          Comment


          • #6
            Originally posted by OscarAV View Post
            I am having problems with a circuit using bf245c. My circuit has one resistor at drain of 2k2, and the source to ground without resistor, the gate is grounded with a 1 Mohm resistor
            It won't work like that, You have the source and gate at the same potential which will turn it hard on. If you want to use it as a gain stage by taking the signal from the drain resistor then it needs a source resistor to bias it mid rail. If you just want a unity gain buffer use the OP's circuit or the one Mick suggested.

            Comment


            • #7
              First of all, thanks a lot for the replies, since this post is a bit old. Thanks again. I managed to get the schematic in full view, since I am with the smartphone. I uploaded my shematic as well.

              Well, I understood that Vgs has to be negative or zero to bias the jfet. I thought that to have the maximum headroom or signal swing without much distortion, maybe a little bit at the top, and with the maximum current, I just could set Vgs=0 grounding the gate, and limiting the current below Idss, with a drain resistor. Then Vin would be between -|Vp| to Vp, which in my case is 3.2 V, pretty good apparently for a guitar. However I misunderstood something clearly... I am confused with the q point of the jfet and the swing of the signal, since Vgs has to be negative always, in order to amplify, as I understood...

              As far as I know, JFETS have a wide range of values of Vp and Idss, so I wonder what values does have the real jfet of the initial gnobuddy schematic. In my case, with Vp of 3.2 V, I should have Vs= 1,6 v to have the maximum headroom without clipping too much, but a typical electric guitar signal when played a bit hard has peak values of 2 V or more. Then, I would have to compute the source resistor to have a given value of Vs. And I do not know how to compute this value neither.

              Basically, I just want high impedance and unity gain or a bit more with maybe a bit of distortion, and connect this with the characteristics of the Jfet. I would like to understand to be able to make some numbers by myself... Thanks a lot...

              And just to finish, are zener diodes useful to prevent peaks at the gate of the jfet for instance with a piezo disc?
              Attached Files
              Last edited by OscarAV; 02-28-2020, 03:41 PM.

              Comment


              • #8
                Something like this would do it if you want the gain to be greater than unity.
                Adjust R1 to set the bias point. Adjust R5 to set the gain.
                Gain is 1.5 (3.5dB) in the example below.

                Click image for larger version

Name:	2n3819 scm.png
Views:	1
Size:	12.4 KB
ID:	856858

                Click image for larger version

Name:	2n3819 plot.png
Views:	1
Size:	12.1 KB
ID:	856859

                Comment


                • #9
                  Or wire the bypass cap on the wiper of a pot to get variable gain. I have a circuit where R2 is 39K, R1 a 10K pot with 1uf from wiper to ground (no R5), r4 is 100k. Rather more gain, but gives humbucker output from single coils.

                  Comment


                  • #10
                    So, just for completeness, I have found this guide:

                    http://diy.smallbearelec.com/HowTos/...ardBareAss.htm

                    What is interesting is that they do a complete study to make the circuit. Have you done something similar in completeness, or just have taken into account simple thumb's rules? Do you take into account the values of Vp and Idss? I do not understand the function of the jfet properly yet, hopefully reading the website carefully, I will understand things better... I will have to wait to complete the circuit...

                    Comment


                    • #11
                      Only when I was studying electronics at college. I usually just put something together and it works fine, as most audio circuits are based around a certain range of component values and similar designs. It's good to go through the process of understanding how things work.

                      Comment


                      • #12
                        Originally posted by OscarAV View Post
                        First of all, thanks a lot for the replies, since this post is a bit old. Thanks again. I managed to get the schematic in full view, since I am with the smartphone. I uploaded my shematic as well.

                        Well, I understood that Vgs has to be negative or zero to bias the jfet. I thought that to have the maximum headroom or signal swing without much distortion, maybe a little bit at the top, and with the maximum current, I just could set Vgs=0 grounding the gate, and limiting the current below Idss, with a drain resistor. Then Vin would be between -|Vp| to Vp, which in my case is 3.2 V, pretty good apparently for a guitar. However I misunderstood something clearly... I am confused with the q point of the jfet and the swing of the signal, since Vgs has to be negative always, in order to amplify, as I understood...

                        As far as I know, JFETS have a wide range of values of Vp and Idss, so I wonder what values does have the real jfet of the initial gnobuddy schematic. In my case, with Vp of 3.2 V, I should have Vs= 1,6 v to have the maximum headroom without clipping too much, but a typical electric guitar signal when played a bit hard has peak values of 2 V or more. Then, I would have to compute the source resistor to have a given value of Vs. And I do not know how to compute this value neither.

                        Basically, I just want high impedance and unity gain or a bit more with maybe a bit of distortion, and connect this with the characteristics of the Jfet. I would like to understand to be able to make some numbers by myself... Thanks a lot...

                        And just to finish, are zener diodes useful to prevent peaks at the gate of the jfet for instance with a piezo disc?
                        The zeners are there to protect the JFET.

                        Since JFET's do vary a lot you have to have a means of correcting for that. That means selecting resistors on test (a pain) or using a trimmer. Here's an idea that uses a second transistor to automatically correct for the differences. You can drop a wide selection if JFETS in this circuit and it just works. The source is set to be about 5V to allow for a wider range of JFETS. There is also a high cut filter on the input to reduce interference. I didn't add zeners as the 47K resistor on the input gives some measure of protection by limiting the current. No supply decoupling capacitor is needed.

                        Click image for larger version

Name:	Auto JFET Buffer.JPG
Views:	1
Size:	360.0 KB
ID:	856885
                        Last edited by nickb; 03-02-2020, 02:56 PM.
                        Experience is something you get, just after you really needed it.

                        Comment

                        Working...
                        X