Announcement

Collapse
No announcement yet.

Why Does CMOS Get Such a Bad Rap ?

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

  • Why Does CMOS Get Such a Bad Rap ?

    My first experiments with CMOS started around 1980. I got a copy of the National Semiconductor CMOS data book which had a copy of AP Note AN-88 in the back. I posted a copy in post #4 of this thread:
    https://music-electronics-forum.com/...in-the-circuit

    I call this the Reagan Fuzz because he was elected President around that time. My first version used an opamp as a power supply regulator and a CD4007UB as the active element. The hard part was keeping the current draw as low as possible (250uA) so I used a programmable opamp, (LM4250 I think) and a reverse biased PNP junction as a low power zener diode. I still have that thing somewhere, got it back from the steel player I sold it to, but I'd have to trace out the schematic.

    The next version came about after I moved to Texas in 83. It dispensed with the opamp and used a JFET current source and a big cap as a power supply for the CMOS. When you turn it on, it takes several seconds for the cap to charge up before it will pass a signal, kinda like a tube amp. For this version I used a CD4069UB and connected all the gates in parallel. Since I was using the CMOS at a very low current, I used a JFET Source Follower on the output to drive the real world since any load would reduce the gain of the CMOS. To protect the CMOS, I also used a JFET Source Follower on the input. The schematic is posted below.

    The next version (also attached below) came somewhat later. I used the fact that all the MOSFETs in a CMOS chip match each-other pretty close so you can take one or two gates, tie input to output and that Voltage can be used to bias the other gates and you can run them without feedback. You can even stack the chips on top of each-other to up the gate count. More gates in parallel should lower the noise.

    Some of the non-ideal characteristics of CMOS that you may have to deal with: The gain actually goes up as the power supply goes down. At 3V, gain for a single stage is around 300. At 5V it's around 100 and in the 10V to 15V range gain is about 30. The bandwidth also goes down with lower power supplies. It can go below 10KHz near 3V and up to around 5MHz at 15V. Noise is generally low, but can go up with big resistors in the inverting feedback configuration. Another thing is that when CMOS is banging rail to rail, the current it draws goes down. You can put a resistive load on it, but that will reduce the gain somewhat.

    One problem I didn't solve is fizz. Under some circumstances, the CMOS fuzz can sound fizzy. I don't know if it was always there and I just didn't hear it, or if certain guitars that I don't own cause it. I have yet to hear it when I'm playing. Maybe if I recorded my playing, I would hear it in the playback.

    My next experiments came around 2010. I wanted to use CMOS to more closely emulate a tube preamp. This was around the time I started using a scope in the X-Y mode to look at how a tube bends in guitar signal. I started looking at all the different ways a CD4007 could be used. I ended up using the 3 input NOR gate. With feedback it comes real close to a 12AX7. The 2 input NOR is pretty close, but I liked the 3 input better. A preamp was built, the 3 input NOR is the first stage followed by an opamp stage to boost the output of the CMOS. Next is a tweed style Volume and Tone control, then a three stage CMOS amp that I don't remember what I was thinking. Last is a bass cut control and an opamp buffer for the output. I don't think I ever sat down and tweeked that last part with a power amp connected.

    Another thing that came out of all these experiments is what I call the "P Series N Series Inverter". It's a configuration of the CD4007 that isn't a normal configuration. Sort of like a NOR and NAND gate melted together. The advantage is it doesn't draw as much current as the normal inverter. At 15V, a normal inverter draws around 11mA. The PSNS inverter draws about 4.5mA, less than half. Clipping is a little softer too.

    Attached Files
    WARNING! Musical Instrument amplifiers contain lethal voltages and can retain them even when unplugged. Refer service to qualified personnel.
    REMEMBER: Everybody knows that smokin' ain't allowed in school !

  • #2
    CMOS distortion is something I want to get back to experimenting with. I breadboarded some ideas a while back and hit on an outstanding range of sounds from a relatively simple circuit. Instead of leaving it at that and documenting it, the next day I started to think that if it's this good it can only get better - what if it can do this, what if it can do that. But in changing too many parts I lost my original build and never found my way back. Fizz was a characteristic (for me) of trying to get a simple control layout that went from almost clean to full-on extreme distortion. That was overcome by switching several component values to give gain ranges and tonal variations. I planned to use quad analog switches to do this, driven of simple CMOS logic. I already did this with another preamp build along with an LED single digit display and two foot switches - one for up and one for down selections. I'd though for a long time to build a (large-ish) stompbox using a CMOS-based preamp running into a tiny 100W class D amp module that I have that runs off 24v. This thing is insanely loud, cost less than $5 and doesn't get even remotely hot at full output.
    Last edited by Mick Bailey; 01-28-2021, 10:46 AM.

    Comment


    • #3
      Nobody gives them a bad rap
      Unless somebody literally believes the "tube distortion" myth and gets dissappointed.

      I briefly experimented with them long ago, not bad sounding at all, not tubey either, very SS with a twist.
      Not a mystery, SUNN and Laney amps have been reliably delivering that sound for decades and some players (Doom, Stoner styles) swear by them.
      One "problem" is that, as is, they clip way too symmetrically which is "woody/boring" , so some have added "bias" to shift them fom halfway between supply rails to provide a non symmetrical waveform, the idea has its merit.

      I donīt get the parallelling many gates thing, unless you want to drive a reverb tank or something (headphones?) , way back then I *cascaded* a few to get high gain, seem to remember than above 3 or 4 stages it became hard to control, never ever could use all 6 in series, not even 5. Oh well.

      I think they deserve more experimentation; one problem is that nobody (I know) went real deep; most circuits were absolute raw, no EQ, specially after the clipping (some Laney added mid/treble boost before full gain) .
      Fuzzy memories, seem to remember Gallien Krueger also did something similar, a little pre EQ.

      They definitely deserve further development.
      Juan Manuel Fahey

      Comment


      • #4
        To add to the collection I found this in my archives.

        Click image for larger version

Name:	Craig Anderton Tube Sound Fuzz.gif
Views:	311
Size:	5.1 KB
ID:	923731

        Experience is something you get, just after you really needed it.

        Comment


        • #5
          Laney and Darkglass have exploited them quite a lot, Laney probably most extensively. Laney circuits typically cascade several as "opamp-ish" gain stages while introducing a lot of interstage frequency response mangling. All in all the CMOS linear amp is just a very soft and rather symmetric peak clipping stage, no frills about it. If a circuit sounds boring don't blame the CMOS, (it does its job alright), blame other details.

          These things sometimes are critisized for clipping TOO soft. One solution is to cascade a few stages under global loop: open loop gain goes up, clipping turns harder. Gain also follows "opamp-ish" rules more properly this way.

          Couple of interesting designs to check out are Lab Series L3 (not rest of the series) and Sunn SL260 / Alpha series. L3 for genius exploitation of discrete MOSFETs within as generic gain stages, all eventually driving a single linear amp. The circuit thus mimicks typical tube amp architecture with SE preamp stages and (overdriven) push-pull power amp as final stage. Sunn for genious implementation of tracking HT rails so that power supply sag correspondingly decreases clipping threshold of the final CMOS peak limiter stage. (No. This scheme is not featured in the most popular example of Sunn amps with CMOS linear amps, the Beta series).

          Comment


          • #6
            Yes.
            I think the (unjustified) bad rep comes precisely from the Craig Anderton design, not that itīs "bad", as Teemu says they do their job, but the "Tube" word in the name scratches some people the bad way.
            They definitely deserve development.
            So, Loudthud, now itīs up to you
            Juan Manuel Fahey

            Comment


            • #7
              The Sunn Beta Bass Amp is one that has the CMOS stages. I had recently built a foot pedal for it, and when it was ready for pickup, the owner (who had been the Guitar Dept mgr at CenterStaging where my shop is, dialed up the settings so being able to add just a enough of the overdrive it achieves for a nice taste of growl and grit. Now that I'm using the Ampeg SVT4-Pro with my Symetrix 501 Compressor/Limiter and SE-400 Parametric in the Effects Loop, I might see about building a CMOS device to add to it, maybe in the sidechain of the 501. Interesting thread!
              Logic is an organized way of going wrong with confidence

              Comment


              • #8
                Yes, saying that CMOS has "tube tone" is just based on myth that tube amp tone would be result of soft clipping, and nothing but soft clipping. Which of course isn't true.
                Slapping a CMOS linear amp into circuit with broad bandwidth and no pre or post emphasis for distortion (e.g. "tube sound fuzz") doesn't magically convert everything to sound like 2203, 5150 or SLO100, tube amps particularly NOT identified as "fuzzy". It may sound slightly like a Laney Klipp amp, which has plain symmetrically and softly clipping tube distortion effect with no particular emphasis and therefore sounds more like an old fuzz unit than a modern tube amp designed from entirely different basis.

                Of course "fuzz sound fuzz" sounds strikingly less interesting than "tube sound fuzz".

                Comment


                • #9
                  I'm not very familiar with Laney amps, don't see them down in Texas much. If anyone can site specific models that used CMOS and has links to schematics, that would be cool. I'm much more familiar with Sunn. The SL160/260, the Coliseum 300 and Alpha series amps had the soft clipping feature where the signal was clipped by CMOS just before the power amp that had just enough gain to show the soft clipping on the power amp output. The power supply for the CMOS is tied to the rails of the power amp so as the supply sags, the CMOS rails also sag. As Teemuk says, I don't think this soft clipping feature makes an amp sound like a tube amp. A low damping factor is much more effective. The Sunn Beta amps use CMOS as variable gain stages in the preamp as well as channel switching logic.

                  Link to SL160-260 thread: https://music-electronics-forum.com/...-supply-issues
                  Sunn Alpha and Beta schematics attached below.

                  That Anderton foot pedal probably consumed close to 10mA which would drain a 9V battery pretty fast. But it probably sounds better when the battery is close to dead. My pedals are all around 250uA.

                  I dug out the prototype I made of the preamp I posted in the first post above. Took a little work to attach it to a power amp I'm working on. It didn't sound very good, not near as good as the JFET preamps I've made recently. Kind of noisy and an unpleasant tone. The first stage with the NOR gate should be OK, I'll see if I can come up with something better for the 2nd stage.

                  Attached Files
                  WARNING! Musical Instrument amplifiers contain lethal voltages and can retain them even when unplugged. Refer service to qualified personnel.
                  REMEMBER: Everybody knows that smokin' ain't allowed in school !

                  Comment


                  • #10
                    At least the Laney HC and HCM series feature CMOS linear amps. Google search finds several schematics in an eyeblink.

                    Comment


                    • #11
                      The old Linebacker does, schematic calls it a proprietary chip.

                      Laney_Linebacker_50R,_65R,_100R.pdf
                      Education is what you're left with after you have forgotten what you have learned.

                      Comment


                      • #12
                        A totally novice question... Why do some circuits use external half supply (?) DC reference (i.e. old Linebacker & early 80's Fender amps) while many other schemes are simply biased via feedback?

                        Comment


                        • #13
                          Originally posted by teemuk View Post
                          A totally novice question... Why do some circuits use external half supply (?) DC reference (i.e. old Linebacker & early 80's Fender amps) while many other schemes are simply biased via feedback?
                          I'm not 100% sure I know what you are talking about, please post examples or provide links to schematics. Circuits without feedback have more gain and more of the non-linear distortion and softer clipping. An example is in the two "Reagan Fuzz" schematics in post 1 above. Also, the high value resistors required tend to add noise.
                          WARNING! Musical Instrument amplifiers contain lethal voltages and can retain them even when unplugged. Refer service to qualified personnel.
                          REMEMBER: Everybody knows that smokin' ain't allowed in school !

                          Comment


                          • #14
                            I mean they all use feedback but some circuits use an external bias voltage reference and some simply have the CMOS bias itself via feedback.

                            You can compare e.g. Laney Linebacker vs. HC65R circuits (links to at least LB posted earlier, google finds the second easily): Both have unipolar CMOS supply so it can't be that. Linebacker uses additional DC reference (with trimmer adjustment even) while HC65R doesn't. Is it simply for asymmetry to output, a chip architecture thing, or something else?

                            Comment


                            • #15
                              Originally posted by nickb View Post
                              To add to the collection I found this in my archives.

                              Click image for larger version

Name:	Craig Anderton Tube Sound Fuzz.gif
Views:	311
Size:	5.1 KB
ID:	923731
                              I built this many years ago and liked the sound I got but it was a battery killer.
                              --Jim


                              He's like a new set of strings... he just needs to be stretched a bit.

                              Comment

                              Working...
                              X