4007 Chopper-Trem (who needs another trem design?)

[anotherjim]

I decided my old trem pedal needed a refurb. A vactrol based job, no matter what I tried, there was some LFO tick and the deep chop wasn't definite enough. So I needed a new circuit to suit the old case with just 2 knobs and a switch.
On top of that, I've been messing with CMOS CD4007 chips a lot and thought I could get it all in the one chip.
This is the scheme I built...

So that it isn't just wires and a blank DIL chip, I've added discrete MOSFET equivalents to show real workings.

The hardest part was getting the modulator just right. The mosfet needs to be only just "off" with the depth pot minimum, and a trimmer adjusts for that. The DC volts around the mosfet are critical. It's switch threshold varies from chip to chip (it's about 2volt) and the source voltage needs to be set something stable in relation to the threshold. 2 diode drops do that  - enough to keep the signal from hitting the MOSFET body diode (which connects to pin7, not the source as it does in discrete MOSFETs). A late addition was the Zener providing the gate bias as I found the 9volt supply volts only had to be a bit different
to change the trimmer setting and the trem was either too deep at minimum or didn't come in soon enough with increasing depth.

The LFO is a twin-T oscillator, the only type I could get working on a single inverter. There isn't enough gain for a phase shift type. Speed range is poor, but I only like Trem about 6 or 7 Hz anyway. The output is a Sine wave. Chopping is achieved simply be overdriving the modulator gate. The slope of the sine appears to be slow enough to stop it clicking the edges.
The oscillator is easily loaded down. It wouldn't drive a standard LED without a buffering from a BJT. High efficiency LEDs could be directly fed, so long as you have at least 22k series resistance. My old case didn't have a rate indicator, and I didn't fit one on my build but I did try it out on the breadboard.

The input preamp has a little treble boost over 3k - I find it gives the chopper effect an edge.

The effect is inverting (not phase accurate) - Oh dear...

Soundcloud clips here...
https://soundcloud.com/ashdalestudio/sets/4007-chopper-trem

[maartendh]

Nice sounds in your clips, Jim. Now I have to find my 4007's again...

[PRR]

> LFO is a twin-T oscillator, the only type I could get working on a single inverter.

Reduce supply voltage to that stage? CMOS gain goes up as voltage drops, to ~~300 at ~~3V. You only need Gv=27 for a phase shift.

The 10K resistances for depth seem lower than needed, since you are only pushing a gate-oxide. 1Meg ballpark here is some less loading on the LFO gain.

[anotherjim]

I considered trying a PSO at 5v, but the connections choices of the 4007 would mean a reshuffle to run an inverter off a lower voltage than the rest of the chip and could mean losing the source resistor in the pre-amp although I don't think that would be fatal. I also considered running the whole thing off a lower regulated voltage which I might try if I revisit something similar again. A regulator on board would mean all the reference voltages could be resistor dividers.

The low resistance about the Depth control was a development trap. I thought I needed a low impedance reference at the Depth trimmer and did not want to fit a large cap. At 10k, the oscillator doesn't wobble it noticeably. The oscillator is then padded down 100k:10k (Depth pot) at tremolo range and it probably could have been 1M:100k instead - but I got it working and built as is. Besides, I didn't have a 100k panel pot to hand!
The 10k switched in for Chopping is also lower than necessary in retrospect. Switched out the Tremolo can get fairly deep anyway at maximum depth, the idea was to switch to maximum cut. Switched in, the depth control has only subtle effect.

By messing about with the details, you might be able to shave 2mA off the supply current. As it is, I think 5mA is just about ok for battery. For myself, I'm starting to delete battery provision, although I could fit one retrospectively to this one.

[duck_arse]

very interesting. does the 4007 run warm?

[anotherjim]

No me-ducks. Only the oscillator inverter is biased linear like in all the 4049/4069 distortions etc. When oscillating, it only lives in the "Both mosfets on" hot condition while it swings thru the mid supply point.

[duck_arse]

jim, this twin tee business. I have one connection error on the BB, but it's leds instead of 914's, so I don't think it's the problem, but - my osc was dying at the fast end, and the "range" is only 6.7~5 Hz. a bit stingy, but only the first try.

about this lower voltaged invertor - it has to be moved from the pin7/14 inverter if we want to lower the voltage to increase the gain, yes? and how much current does the osc draw? cause I was thinking about shoving a rate led into the Vdd to drop some V for a pso, might that work?

[anotherjim]

I can't see how to shuffle the inverters around easily. Input amp wants a source resistor to keep it under some control. Modulator wants it's source other than ground. The one with the sources tied to Vdd & Vss in the chip is the only one free.

But, the whole thing might be fine at a lower voltage. I did a delay using 4007 with the lot off 5V.

The osc can die at the fast end if the load current is too high. That I think is because the rate pot is minimum and there's a comparatively low resistance path to ground via it. That's why I had to fit as high as 33k in series and why any extra load on the output can't be much.
Maybe it's the other T that would be better to adjust, but that calls for a dual gang pot (or does it?).

I did try different 4007's & never failed to osc, so I hope it's not a select-a-part job.

Rules for oscillation...
C6 = C10
C8 = 2*C6
R15 = R19
R17+18 ~ R15/10
Which I stole from a web page I found.

[duck_arse]

orrlrite, here's some observations. the first was also the last - C11 is 22nF, not 22uF. a slight difference .... also, hmmm.

I couldn't phase shift it to save my life, right down to 3V. so I went back to 5V reg and the twin T. had to up the depth pot to 100k to get anything percolating, but then managed 4~11.6Hz at reasonably even output level, using a 250k pot and 6k8 stop. replaced the bias diodes with 1k5 gave (a soft) 620mV bias, removed the 'zener', increased R21 to 4k7 (see first obvs above).

distortion - clipping on the pick attack. and, it sounds rather bright without the 22uF bypass. I'll have to have at that again tomorrow. also, the depth pot seems no, no, some, all, all, all, all - either S1 setting.

[anotherjim]

For sure C11 bigger will boost the input. Leave it off if you don't like the treble from 22n.

Is pin10 anything like 2.5v on 5v power now?

Wot, you got a PSO working? If so, help yourself to virtual Bara Brith & butter all the way from old Wales.

So, the modulator. Taking this as 5v supply now?
That N should be normally off. The actual threshold gate-source may now be only 1v-ish, but to start to turn it on, the gate voltage must be higher than the source and so with depth 0, the gate will be about 1v-ish more than Vref. 
She calls!

[duck_arse]

thanks for the cake^*: what was it for again?

I couldn't phase shift it to save my life ....

well, I take that back now, and more cake please. all 5V supply now - phase has been shifted, but I added a fourth cap, as is my want. with 4x220nf and 12k/250k pot (and 2x180k pulldowns, 3M9 feedback), we get 2.8~7.1Hz. that's an extra 0.1 of your favourite speed! and it's about 4V5 pp, steady across the range. it only has the led drive connected at the moment, so I'm not sure how it will go with bias dangling.

and the input section, pin 10 sits at 3V [when the pin 12 connection is correctly made], and w/ 300mV input, no bypass cap, out is about 420mV. with the 22nF cap, I see no boost w/ sig gen till over 2kHz. lots of cmos hiss up there? I'll get at the modulator soonest.



* pardon any cultural insensitivities.

[anotherjim]

Who said you can't have your virtual cake & eat it?

R21 & R22 may need changing for 5v. Pin 12 should be good anywhere 2 - 3v without clipping.  I think I'd want at least x3 gain without C11. Boost from 22n is high up, it gave a slightly synthy edge to the sound on chop mode, especially with clean chords. If R22 is changed, it changes C11 anyway.

BTW - Sorry for any confusion on my scheme -  the detail drawings have different part numbers from the main because I let Eagle renumber. I refer to the main diagram numbers.

Bias will probably want to be 1 - 1.5v above whatever Vref ends up being. The modulation only works downwards, so at low depth, it's only a half wave dip from the positive peak of the LFO.

The LFO output to depth path can probably be scaled up from the 100k//10k pot. As mentioned before, maybe even x10 to 1M//100k pot, giving negligible load on the osc.

[duck_arse]

todays findings - the 10k depth pot was too loading the osc, so it's now B100k - full osc range. UB is a 10k//1k5. with R22 now 15k and R21 1k, no bypass, I get x3 gain, no load.

the sound - I thought it was still a bit distorted/dirty on the low chop range, but on the full chop, it was not so noticable. and so I gave the bias pot a tweek, and with R12 it seems the depth pot goes from none to full chop quite nicely. perhaps the chop switch could go, or be moved to a bypass cap.

and an odd thing, after a few minutes of twang, I went for a fuzz, as you do. first to hand was a fuzzrite copy. turned up good, with low trem depth, it seems to overload the tremming action, which then comes up on the decay, such as it is with the fuzz. perhaps an interesting feature/point of difference.

[anotherjim]

Does Ub need a bypass cap? I had one originally, but it didn't seem to make much difference. I thought maybe it's because the 2 diodes I had were worth only about 60ohm, but an R only network is worth more?


So, is it time for soundclips?

[duck_arse]

So, is it time for soundclips?

oh, please, don't make me laugh. I can hear a few other members wincing at the thought, too. you can have a soundclip if you send someone compentent to do the playing.

as for Ub, I thought you said you didn't want a cap - somewhere. I looked on the cro at an earlier bb stage, at it was a little - willing? soft might be the word, as opposed to stiff. and if it ain't stiff ......

how does yours handle big complex input/low depth?

[anotherjim]

On the bias trim I didn't want a big cap - it would have to be low-Z at the LFO freq, maybe 100uF?

I think I had a Keeley DS-1 in front of it at one point. I'll recheck.

[duck_arse]

I hung the cro on Ub tonight, and it was very soft, modulated by tremmed audio. I tried a 10uF viagra cap, could hear no difference (but was then picking much harder for a few hours....). and then I whacked a different fuzz in front, and I think it is a real thing. possibly a normal trem thing, maybe psychoacoustics, maybe arse talk, but there is a difference. with fuzz level at 23:30, everything gets tremmed, with level at 00:30, it seems to me to hold-off tremm till the decay.

and I couldn't see so much bias level shifting to say it was bias level switching, so, level dependant tremolo? I didn't bother messing w/ the depth bias.

[edit :] oh, a 100R series resistor upstream the 5V reg showed the total current draw was 1.7mA.

[anotherjim]

"[edit :] oh, a 100R series resistor upstream the 5V reg showed the total current draw was 1.7mA"
Excellent! I would have been pleased with anything under 3mA. Definitely battery friendly.

Publish and be damned!

And I think trem coming up during decay out to be cool. Probably is a psycho effect - the brain is not really listening if you hit it with a lot of sound. That's what mp3 does isn't it? Chops out data for a short time after every strong transient?

[duck_arse]

here is my bucket, with a 5V revision of anotherjim's circuit, a PSO osc, and a bonus perf layout. [black on white versions on request]
http://imgur.com/a/BqbWO

I have this on the BB, it has an odd asymetric distortion which I don't know is useful, so we need another someone to have a swing at it and give it a listen.

anyone?

[anotherjim]

Old dog learned new trick. I open the "Goth" scheme in MS Pa(i)nt(s) and found an invert colour command. So now I have the pale rider version!