moved...

[puretube]

info saved to HD...

[Tim Escobedo]

Good questions.

I've never seen any specs regarding noise figure for linear amplifier configured CMOS inverters. One thing I have noticed is that they generally are used without clipping resistors, allowing for substantial voltage swings at the output. This allows the noise floor to be rather low in comparison. Obviously, this isn't a very academic way of measuring noise, and a 741 could easily pass muster under the same conditions.

What I did regarding input impedance was either decouple AC feedback, allowing the input resistor to be raised significantly (to control gain of the stage), or even eliminate it (for a inverter run full gain).

Another possibility is using a resistive divider to bias the inverter into linear region, possibly using a high value resistor between the input and divider.

[gez]

anybody can tell me if cmos inverters wired up for linear amplification purposes are more noisy than "regular" opamps (say TL074) in comparable configuration (e.g. gain=50x) ?

They make 'poor' linear amplifiers and have limited bandwidth.  Bandwidth decreases as voltage decreases.  This fact is commonly quoted in text books but no explanation is given.  I should think the reason for this is the high output resistance (which increases at lower voltages) forming a RC low-pass filter with gate capacitance, so high end is sliced off somewhat.  This might explain why they sound a bit wooly at low gain (clean).

Probably for the above reason (and because they're comprised of MOSFETs), I've always found these circuits to make exceptionally quiet distortion units (with basic supply filtering)

why do I never see a circuit for distortion purposes, which does use cmos inverters, taking advantage of the high input-impedance of those devices and use it for an input stage ? (seems like they all use a discrete FET source-follower, or normal opamp input-stage... or am I wrong here?)

You do see schematics taking advantage of this, I've posted a few!  
As Tim said, you can decouple AC feedback (I tend to do this) or bias them as you would a MOSFET with a divider/trimpot and large value resistor to gate.  I've always found the latter method finicky (when I did this I used a regulator to keep bias-point stable).

Incidentally, these chips are more efficient at lower voltages.  I tend to run them from 5V regulators.  Current consumption, at this voltage, decreases to about a third of what it is at 9V, open-loop gain increases two to three fold and you get more distortion due to less headroom (don't need as many stages).

You don't mean to tell me you actually build effects...er, WITHOUT tubes?!!  

[Ge_Whiz]

I always understood that one of the advantages of using CMOS inverters as linear amplifiers is that they are self-biasing(?) Certainly, I have a nice little ETI 4007-based distortion / compressor where this is the case - and the input runs straight into the first inverter stage.

[puretube]

info saved to HD...

[puretube]

info saved to HD...

[puretube]

info saved to HD...

[Paul Perry (Frostwave)]

My first Fuzz-box (1966) was a selenium device hooked up after a tube-amp! :wink:

Selenium..that takes me back... none of the current diodes smell like that

[puretube]

remember? those guys that looked like old aircooled motorcycle cylinders
(in a way, at least...)

[Mark Hammer]

I've built a couple of inverter-based distortion/overdrive units from Anderton's designs, and one of the things I found was that I liked how it sounded when there was a non-CMOS input stage that hit the first CMOS stage with a big signal, and did not like the tone if the CMOS stages were providing all the gain.  That may well have been due to factors I was and am still unaware of, but the fact of the matter is that I did not really like the old E-H Hot Tubes very much (personal taste), did not like the EPFM Tube Sound Fuzz, but *loved* the earlier op-amp-plus-2-invertor-sections overdrive that Anderton published in 76 or 77 (and still do), especially when the brunt of the drive was shifted to the opamp input stage.

Stellan Lehrberg stumbled onto my comments about this some time back, tried it out, and came to a similar conclusion, designing and posting his "Slowfinger" unit.

One of the things that has not been discussed much here is the late Charles R. Fischer's current-restriction design that was published posthumously in Electronic Musician some 8 years back.  A scan of the schematic is posted on the generalguitargadgets site (http://www.generalguitargadgets.com/v2/index.php?option=html&file=instructions/emfuzz.htm).  The design uses a current regulator to restrict supply current to a single invertor section used as clipping element, apparently to good effect.

I have not built the unit, and no one else seems to have been in a position to provide build reports or supply sound clips, but the idea still intrigues me.  I mention the design not only to pique your interest, but also to point out that there are ways of using invertor sections which would necessitate "decoupling" the supply available to an input stage from the supply available to the invertor/s doing the distorting.  I suppose one could have two different invertor chips with different current supplied, but that starts to get big and ugly.

If you haven't seen the EM Fuzz, take a look.  I certain a clever lad like yourself would get some good ideas from it. :wink:

[R.G.]

anybody can tell me if cmos inverters wired up for linear amplification purposes are more noisy than "regular" opamps (say TL074) in comparable configuration (e.g. gain=50x) ?

They are quieter than regular opamps with clipping diodes as Tim notes because of the higher headroom. You're right though - big resistor values will lead to higher thermal noise if you go that way.

why do I never see a circuit for distortion purposes, which does use cmos inverters, taking advantage of the high input-impedance of those devices and use it for an input stage ?

I believe that may be because the inverters do not have a noninverting input. The impedance at the inverting input of a feedback amp is always low impedance by the nature of inverting feedback. That leads to the use of a follower to get high input impedance and drive a lower value input resistor, which makes for an even higher value feedback resistor.

There are techniques to get very high input impedance in inverting feedback setups, but I've never seen them applied to CMOS, possibly because there is no noninverting input to tie the output firmly to a fixed reference voltage. With all the CMOS inverters in the same package doing their own self-bias voltage, you could use one of the inverters just for a reference voltage, but you're counting on internal device matching for that to be accurate within that package.

I guess overall - it could be done better, but it takes some insight into CMOS and feedback amplifiers to do it, and you get 90% of the value cheaply without the advanced stuff, so most people don't know - or care.

[puretube]

info saved to HD...

[gez]

(I kind of like the idea of having a built in lowpass, just by lowering the supply voltage, while raising the gain, and reducing supply current...)

Another forumite I'm in contact with has done this using an adjustable regulator to supply the CMOS chip (apparently to good effect!).  I have yet to experiment with this, but know from running chips from fixed regulators that different supply voltages do make a difference with the same circuit.

[puretube]

info saved to HD...

[Nasse]

That lost paper of mine claimed that somebody had indeed used the power supply pin as a feedback path :wink: when the author warned about hissy power supplies, while opamps suppress supply voltage noise, these animals can sometimes amplify them... But thinkin that phenomenom as some kind of feedback path so that Mark Hammer link is very very interesting...

Few years ago I made a one 4049 based fuzz a´la later G.A. tube sound fuzz for a relative, young bass player. I housed the thingie in very small box, just one gain knob and no bypass switch. Instead of grounding the unused inputs I connected those parallel (just wanted to try if it works) if i remeber I put four inverters parallel in first gain stage and two for second one. I believe it sounded "beefier" and "more solid" and the overall signal was "cleaner" than my earlier stock version. The difference was very subtle, and may be just my imagination, but if you need that extra push to "11" performance like Nigel Tufnel, it is worth try. This was not same physically as piggybacking opamps but piggybacking cmos inverter stages...

And what else... 10 volts was ideal supply voltage thinkin noise and headroom, and the author suggested silicon tranny emitter follower to the output when thinkin output impedance

[gez]

Instead of grounding the unused inputs I connected those parallel (just wanted to try if it works) if i remeber I put four inverters parallel in first gain stage and two for second one. I believe it sounded "beefier" and "more solid" and the overall signal was "cleaner" than my earlier stock version. The difference was very subtle, and may be just my imagination, but if you need that extra push to "11" performance like Nigel Tufnel, it is worth try. This was not same physically as piggybacking opamps but piggybacking cmos inverter stages...

That's really interesting Nasse.  I recently saw an old (Elektor?) schematic that had parallel inverters used as one stage of a linear amp (not a distortion circuit) biased from the same feedback resistor.  This was obviously done to lower output impedance.  Perhaps this is why your distortion sounded beefier?

Will have to try this sometime!  

[mikeb]

Some interesting use of CMOS devices can be found at Rene's DIY synth page here:
http://www.uni-bonn.de/~uzs159/
From memory there's a VCF (WASP filter clone), VCO and maybe more there.
JH has more info on the WASP filter here:
http://www.oldcrows.net/~jhaible/tonline_stuff/hj_wasp.html

Osamu Hoshama has some more synth circuits using CMOS:
http://www5b.biglobe.ne.jp/~houshu/synth/

All extremely interesting reading!

Mike

[puretube]

info saved to HD...

[puretube]

info saved to HD...

[puretube]

info saved to HD...