DOD 690 VCO Question

[CheapPedalCollector]

I'd like to know how to calculate the operating frequency of the VCO design in the DOD 690 and other DOD pedals. I can't seem to find the circuit on the web, any of my books or datasheets.

[PRR]

> how to calculate the operating frequency  ...  I can't seem to find the circuit on the web, any of my books or datasheets.

For the practical problem: I'd be lazy and just build it. It can't come out so wrong that a few cap-trials won't get it in a zone.

The new kidz @ TI have totally gerfarkled the classic datasheets. Fortunately Futurlec has the 2000 sheet archived. Page 13 has this plan with minor transposition. But no theory.

Aside from added Vc filtering and raised impedance, DOD changed the hysteresis ratio. Which changes the speed but hurts my brain. I'd assume "about one time constant" each way, so 390k * 0.1uFd makes 0.039 second each way, 0.078 both ways, 12.8Hz. Which is in the ballpark. A pocket full of caps will get you where you want to go.

The R & R/2 structure looks familiar. If it is not due to Bob Pease, he cudda tole you all about it. He might say "Voltage to frequency converter" VCF.

[CheapPedalCollector]

Oh I know how the LFO works, I mean the VCO with the CD4001 and 25pf capacitor. The diode is there to change the duty cycle. That's the circuit I mean, sorry for not being clear.

The LFO changes the Vd voltage which changes the high frequency oscillator and sweeps it from about 12ms to 25ms delay, and I guess the oscillator frequency is ms*1000 according to an article by Craig Anderton in a polyphony article. I've just not been able to find this particular oscillator design, it's pretty interesting. Some of their other pedals use an opamp hooked to a 4013 to do it. I guess you can use a 4011 in the same way, which is also intereting. CMOS is cool sometimes, and it gives me some interesting ideas. I'm just curious about the math and wondering what kind of diode network can be used to change pulse width or not, and it if it will work without the diode.

[Rob Strand]

Yes, there's two parts: the LFO and the VCO.  To work out the frequency you need to know the swing of the LFO.  I'm getting about 2.3V to 11.7V at LM324 pin 7.

The LFO output is combined with a DC offset depending on the width control setting.    The DC offset at the 750 ohm resistor is about 1V.

When the width control is full you will get a swing of about 0.81V to 1.9V at the top of the 150k.  The following divider will drop that by a factor of 16 (150k & 10k?) giving a swing at the input to the VCO (LM324 pin 12) of 51mV to 119mV.


The VCO timing has two parts:  One where the cap is reset quickly through the diode and another where the current source charges the VCO cap depending on the VCO input voltage.   Small VCO input voltage => low charge current => long charge period => low VCO freq.  Similarly High VCO input voltage => high charge current => short charge period => high VCO frequency.   The charge current is I = VCO_in / 3.3k  (the LM324 pin 14 etc + 2N4124 transistor + 3k3 resistor form voltage to current converter).

When the diode is conducting, CD4001 pin 11 is high, and the time is approximately,

    t1 = R * C * ln(2)

Roughly, R = 330 ohm and C = 25pF  so t1 = 6ns.
I've also ignored gate output resistances and stray capacitance.   
This is way faster than the CD4001 so would have to expect the actual time to be longer, maybe 300ns.

When the cap is charging through the current source, CD4001 pin 11 output is low:

   t2 = C*(Vdd/2 + Vd) / I            ; Vdd = 15V, Vd = 0.7V

where,

        I = VCO_in / 3.3k

When the width is full, VCO_in = 51mV to 119mV so I = 15.5uA to 36.1uA.
So with C=25pF, Vdd=15V, Vd =0.7V,  t2 = 13.2us to 5.7us.

Again we are ignoring other capacitances so we could be talking the actual t2 being 20% longer.

The idea is t1 is fairly small compared to t2 so the VCO input voltage determines the VCO period.
period T = t1 + t2 ~ 6us + 1us = 7us to ~13us + 1us = 14us.   About 2 to 1 span.

The SAD512 is 512 stages so the delay is Td  = (512/2)*Tc = 1.8ms to 3.6ms.
The SAD512D is 512 stages and also divides the clock by 2, so the delay is Td  = 2*(512/2)*Tc = 3.6ms to 7.2ms

Reality will be a little slower than that.

If you go here you can see Craig Anderton gets a 2:1 span.  I think he measured the clock but his delay calculations are way off, looks totally wrong.  Then he does a mod based on the wrong numbers.

https://www.muzines.co.uk/articles/dod-mod-ii-chorus/5290

It's also possible my numbers are wrong.  I have not simulated this circuit.  IMHO 20ms from a SAD512 is asking a bit much, so I'm going with mine.

If you plug in a square-wave input you can sometimes see the audio signal delay time on the oscilloscope.   You could also measure the clock and calculate the delay from that.

I'll let do you the case where the Width pot is on minimum.  Same idea with different VCO_in voltage.

Since that unit is using an SAD512D, not SAD512; forgot that subtlety.  IIRC that doubles the delay as it has a divide by 2 on the clock.
Can someone check?

[CheapPedalCollector]

Yes can confirm it is SAD512D, sorry I didn't know there was a distinction between the two.

So what would the frequencies be in KHz and what voltage swing to the clock input?

I was reading the 512D datasheet and it says 5V or so, which seems like quite a lot, but I'm not sure how the inverter connected nand gates output voltage.

I tried to sim, but getting part libraries for LTSpice working is way too much hassle for me, especially with such old parts.

[anotherjim]

Must admit the SAD512D is completely "new" to me. But it's only 8pin and the single-phase clock input is TTL compatible so 0-5V swing and then it must divide the clock frequency by 2 to obtain the 2 phases to operate the BBD section.

[duck_arse]

> how to calculate the operating frequency  ...  I can't seem to find the circuit on the web, any of my books or datasheets.

The new kidz @ TI have totally gerfarkled the classic datasheets. Fortunately Futurlec has the 2000 sheet archived. Page 13 has this plan with minor transposition. But no theory.

https://www.ti.com/lit/an/snoa653/snoa653.pdf

page 26 of the LM3900 apps sheet has all the guff. that circuit used to be in every dual/quad datasheet.

[anotherjim]

I think this is the VCO in question...

The clock here runs off +15v, so I suppose 5v is a minimum.
I don't know how it works apart from what looks like a variable current sink sets the timing. I suspect there are narrow pulses out as this BBD won't care about duty cycle.

[ElectricDruid]

I suspect that that VCO design is fairly variable. The logic thresholds will vary with supply, and probably between different versions of the 4001 too, so I'd expect only a fairly loose relationship between any equation and reality.

[anotherjim]

And could be a bit different again if the original used the old unbuffered CD4001A or CD4001UB devices and was cloned with the current buffered CD4001B. Some old "clever" circuits don't behave well with buffered CMOS. However, all of the gates are only used as inverters so 4/6th of a modern CD4069UB hex inverter chip could serve.

[Rob Strand]

I was reading the 512D datasheet and it says 5V or so, which seems like quite a lot, but I'm not sure how the inverter connected nand gates output voltage.

In the SAD512D datasheet it says the clock input is 5V *min* but in the datasheet text it says "or higher".  Since the chip has an internal clock generator the device is less fussy about the clock details.   Also the clock doesn't need to drive a high input capacitance like on the old SAD devices.

The clock generator divides the clock down by 2.  My initial delay estimates need to be increased by 2, 

    Td =  2*1.8ms to 2*3.6ms = 3.6ms to 7.2ms.             ;I've patched the previous post with that fix.

The middle of that is 5.4ms pretty typical for a chorus pedal.  Just OK for a 512 stage BBD.  However, it still doesn't agree with Craig Anderton's article which is around 20ms.  The DOD brochure only gives a single delay time of 21ms.   Not sure what's going here.   My calculations make sense for the chip and Craig Anderton's article matches the the brochure.

The SAD512D was an 8 pin dip, smaller than the SAD512/SAD1024.   Kind of the first step at modernizing and shrinking the devices - perhaps to compete with the Panasonic devices.  After that came Reticon's R5106 (IIRC used in MXR microchorus? - maybe the microchorus was SAD512D and the later pedals were R5106) which was a better SAD512D .

I tried to sim, but getting part libraries for LTSpice working is way too much hassle for me, especially with such old parts.

Yes it can take a bit to get this stuff up and running.

So what would the frequencies be in KHz and what voltage swing to the clock input?

I'd assume everything is running from 15V, it will work.  There's no other supply rails on the schematic, so that's a good assumption.

The frequencies at the output of the VCO are the reciprocal of the period,

        f = 1 / T

So take my estimates for the clock period  T = 7us to 14us, then  f = 71.4kHz to 143 kHz.

The hanging issue is the discrepancy in the delay times.   It could mean those clock frequencies are wrong.   I'm only seeing a clock divide by 2 in the SAD512D, not a clock divide by 4 which would double the delays again.

suspect that that VCO design is fairly variable. The logic thresholds will vary with supply, and probably between different versions of the 4001 too, so I'd expect only a fairly loose relationship between any equation and reality.

It's not too bad at 15V.   You might only see 20% variation.   As for the calculations the uncertainties of device speed, chip capacitance, the effect of the diode, gate output resistance all blur the estimates - but that's nothing to do with the performance of built units (and you can always make the calculations more complicated).

[CheapPedalCollector]

And could be a bit different again if the original used the old unbuffered CD4001A or CD4001UB devices and was cloned with the current buffered CD4001B. Some old "clever" circuits don't behave well with buffered CMOS. However, all of the gates are only used as inverters so 4/6th of a modern CD4069UB hex inverter chip could serve.

They use CD4001BCN, so does the Flanger 670. I'm not sure what the N means, C is ceramic package yes?

The older version of it with SAD1024 seems to use a CD4013BCN as well.

In the SAD512D datasheet it says the clock input is 5V *min* but in the datasheet text it says "or higher".  Since the chip has an internal clock generator the device is less fussy about the clock details.   Also the clock doesn't need to drive a high input capacitance like on the old SAD devices.

Ok, that's good to know. I assembled the circuit on a breadboard, and it reads about 100mv on my meter, so I think my meter doesn't handle AC at that high of a frequency very well. I really need a scope badly... soon.

I'd assume everything is running from 15V, it will work.  There's no other supply rails on the schematic, so that's a good assumption.

The frequencies at the output of the VCO are the reciprocal of the period,

        f = 1 / T

So take my estimates for the clock period  T = 7us to 14us, then  f = 71.4kHz to 143 kHz.

The hanging issue is the discrepancy in the delay times.   It could mean those clock frequencies are wrong.   I'm only seeing a clock divide by 2 in the SAD512D, not a clock divide by 4 which would double the delays again.

Ok that makes sense, thank you.

It could just be part variation too, CTS pots are notorious for being out of spec by a large tolerance (I've seen up to 30% off), and also these thing had 0 cherry picking so I gather he designed it to be somewhere in the middle so that part tolerances would still produce a working effect even if not ideal. So Craig my have been assuming based on measurements of frequency without checking part values.

David did design things to be as cheap to produce as possible. Might explain why the MXR effects are more long lived, as they have better overall designs.

The point of all my asking is I hope to design some daughter boards for these pedals to repair them, and also to make some new ones with XVIVE/Coolaudio Panasonic reproductions. I've been collecting the different models and circuit variations for many years now, in both working and non-working order. I don't see a lot of people who want to preserve these things very much and I love them so I want to keep them all working (and working again) as long as possible. I just lacked the knowledge to do so, but I've been correcting that as much as I can.

You all rock, thank you so much.

[Rob Strand]

I'm not sure what the N means

Usually means plastic.

Ok, that's good to know. I assembled the circuit on a breadboard, and it reads about 100mv on my meter, so I think my meter doesn't handle AC at that high of a frequency very well. I really need a scope badly... soon.

Yes a generic multimeter can't be trusted over about 1kHz, 10kHz at a stretch.  Higher end DMM's might do 100kHz to 200kHz max.   I've got a very unique bench meter which goes out to about 20MHz.  It's about the size of a 4 litre ice-cream container, only does voltage, no ohms or current.   That only gives you voltage anyway what you really want is frequency.

You can go some way if your multimeter has a frequency range.   When the frequency is modulated it's not 100% reliable but you can set the Width pot to zero and measure a fixed frequency as a check.  Beyond that you can input you own fixed DC voltages into the VCO input.   Old-school "analog based" DMM's only go upto 200kHz on frequency, not quite making the full range of a BBD.    You can of course build you own frequency meter adaptor for your multimeter, even a purpose built one for this project!

It could just be part variation too, CTS pots are notorious for being out of spec by a large tolerance (I've seen up to 30% off), and also these thing had 0 cherry picking so I gather he designed it to be somewhere in the middle so that part tolerances would still produce a working effect even if not ideal. So Craig my have been assuming based on measurements of frequency without checking part values.

I know what you are saying I've seen it too.  In this circuit it's not the pot since the frequency points are at the pot extremes and the pot's value doesn't affect things too much.   You can get some offset on the opamp used for the voltage to current converter.   At the low end the VCO input voltage is only 50mV and the opamp offset could be up to 7mV but even that's not going to double the period.

My feeling is the specs aren't correct.  I wouldn't even raise an eyebrow if that was the case!  And yes, maybe Craig went off the specs.   Without knowing the specs or seeing Craig A's article I would probably accept my calculated values - with the caveat that they could easily have 20% longer period (and delay), ie. 20% lower frequency.

The point of all my asking is I hope to design some daughter boards for these pedals to repair

It's a good idea.  I suspect SAD512D's are going to be extremely hard to find and/or be expensive.


Something I forgot to say was we are assuming the DOD schematic is correct.  It's not unreasonable to think it's not 100% correct.    For example the thing could cause a difference is the VCO cap value.  We don't really know what value that part is.

I did a search for the schematic for the DOD 670 Flanger and found a similar circuit with different part values here and there!   The Flanger spec is 1.5ms to 15ms.  I have not redone the calculations to see how the values of that one pan out.

And lets not forget different versions of the DOD pedals!   How many versions of these pedals where there?

[CheapPedalCollector]

Yes a generic multimeter can't be trusted over about 1kHz, 10kHz at a stretch.  Higher end DMM's might do 100kHz to 200kHz max.   I've got a very unique bench meter which goes out to about 20MHz.  It's about the size of a 4 litre ice-cream container, only does voltage, no ohms or current.   That only gives you voltage anyway what you really want is frequency.

You can go some way if your multimeter has a frequency range.   When the frequency is modulated it's not 100% reliable but you can set the Width pot to zero and measure a fixed frequency as a check.  Beyond that you can input you own fixed DC voltages into the VCO input.   Old-school "analog based" DMM's only go upto 200kHz on frequency, not quite making the full range of a BBD.    You can of course build you own frequency meter adaptor for your multimeter, even a purpose built one for this project!

Cool, I'd be interested to do such a thing until I have the money to splurge on a scope. My meter is an old Fluke 87-III

It's a good idea.  I suspect SAD512D's are going to be extremely hard to find and/or be expensive.

Something I forgot to say was we are assuming the DOD schematic is correct.  It's not unreasonable to think it's not 100% correct.    For example the thing could cause a difference is the VCO cap value.  We don't really know what value that part is.

I did a search for the schematic for the DOD 670 Flanger and found a similar circuit with different part values here and there!   The Flanger spec is 1.5ms to 15ms.  I have not redone the calculations to see how the values of that one pan out.

And lets not forget different versions of the DOD pedals!   How many versions of these pedals where there?

The cap is indeed 25pf, I have one that doesn't work currently that I just got in the mail. I've had two 670's, I repaired and sold one, and the other one is a working unit. I also have a 680 delay, that the SAD4096 is toast, but I posted about that and the 670's already.

Afaik there is only 1 version of the 670, the older flanger was the 640 and used SAD1024. There are two versions each of the delay and chorus that I've seen. One delay uses SAD4096, and the other uses R5101. One chorus uses SAD512D, and the other uses SAD1024. I wouldn't doubt if there is also a 670 that uses a SAD1024, and I also wouldn't be surprised if there were versions of both the chorus and flanger that use R5106 and a 12 volt power supply.

So far I think it's quite possible to make boards for these with very minimum parts needed, and a single capacitor swap if the circuit is how you think it is. That's good news. I would think just a CD4013 and the appropriate BBD device, and doubling the clock cap, while using the VDD/VGG reversing trick they used in other devices. What do you think?

[anotherjim]

I think the 25p cap would be halved to double frequency. But the math is messed up with small caps as the stray & input capacitances become more significant. There will be at least 3pF stray and the input capacitance of 4001 is double as both gate inputs are tied together. So the 25pF circuit may actually see 30pF with 5pF being stray. If the cap changes to 12pF then it now sees 17pF instead of 15pF. A 10pF cap could actually get closer to halving the timing capacitance. These values are guesses to illustrate and reality depends on part tolerances and board layout.
Anyway, if these DODs have a signature sound, it would probably be important to make sure the linearity (or not) of the original clock VCO sweep is maintained.
If you don't want to buy a scope or other frequency measurement, you could rig up binary counters to divide the clock down to a range a DMM frequency counter can handle. An 8-bit counter dividing by 256 will render 256kHz at 1kHz.
However, you would probably want to freeze the sweep at min and max depths to give the reading a chance to resolve, otherwise, you have a moving target. This probably applies even if you have a method that can directly read the clock frequency (or pulse period).

[Rob Strand]

Cool, I'd be interested to do such a thing until I have the money to splurge on a scope. My meter is an old Fluke 87-III

Those meters can measure frequency.   IIRC the frequency measurement is digital and quite accurate.   I don't know how well they work over 200kHz.

The cap is indeed 25pf, I have one that doesn't work currently that I just got in the mail. I've had two 670's, I repaired and sold one, and the other one is a working unit. I also have a 680 delay, that the SAD4096 is toast, but I posted about that and the 670's already.

OK cool, good to know.

I redid the calculations for the DOD 670 Flanger.   As a side note the schematic I have has a whole lot of value changes marked in by hand.   Then it has additional notes talking about old units which imply the hand marked changes are not for old/new!  To the best of my abilities and after doing a lot of calculations I concluded the hand-marked changes are likely to be "more correct".   As far as the LFO design goes.  It's really hanging on by a thread.



Anyway to cut a long story short:
- I started with the LFO output swing at the width pot at 1.4V to 13.5V, essentially the same range
  as the Depth control pot.
- That feeds a 470k + 33k (hand marked values) divider which then goes to a similar LFO circuit to the DOD 690 Chorus.
   That sets the range of VCO_in to 0.092V to 0.888V
- The Flanger circuit uses a 120pF cap instead of a 25pF cap.
- SAD512D delay chip

I got the follow estimates:
- clock period range 4us to 35us
- clock frequency range 28.6kHz to 250kHz

Based on the same assumptions as the chorus:
the SAD512D is 512 stages and has a divide by 2 on the clock, so

   Td = 2*(512/2) * T

That gives an estimated delay range of 2ms to 18ms.
The brochure for the Dod 670 Flanger states 1.5ms to 15ms.

Unfortunately the fact the Flanger schematic has been marked up puts some doubt on what the production units really are. And the fact old units were mentioned.    Despite that the best estimates are in close agreement with the brochure.   

That being the case it makes me think the chorus calculations are OK and the calculated delays are correct.

I think the 25p cap would be halved to double frequency. But the math is messed up with small caps as the stray & input capacitances become more significant. There will be at least 3pF stray and the input capacitance of 4001 is double as both gate inputs are tied together. So the 25pF circuit may actually see 30pF with 5pF being stray. If the cap changes to 12pF then it now sees 17pF instead of 15pF. A 10pF cap could actually get closer to halving the timing capacitance. These values are guesses to illustrate and reality depends on part tolerances and board layout

In the scheme of things small errors.   You might see 20% increase in periods.  The thing is I'm estimating 3.6ms to 7.2ms and the brochure is 21ms.   The finer points can't fix that.    Moreover 21ms from a 512 stage delay is really pushing your luck.  It's going to be noisy.   We could check the filter frequencies but often they aren't technically correct on a lot of guitar pedals.

Anyway, if these DODs have a signature sound, it would probably be important to make sure the linearity (or not) of the original clock VCO sweep is maintained.

No doubt because the frequency of the VCO is proportional to the LFO waveform.

So far I think it's quite possible to make boards for these with very minimum parts needed, and a single capacitor swap if the circuit is how you think it is. That's good news. I would think just a CD4013 and the appropriate BBD device, and doubling the clock cap, while using the VDD/VGG reversing trick they used in other devices. What do you think?

It has to be possible because there's already units out there.  I guess what you don't want to do is mess with how the VCO works as that would change the character of the unit.   In this case you don't care what the delay time or frequency is you just copy the existing ckt.

If you were to rejig the whole VCO that's going to come at a risk.   I wouldn't use any calculations to ball-park what the circuit is doing.  You would need to build one.   Then you would need to check your new circuit produced the exact clock frequencies over the whole range of VCO input voltages (at fairly fine steps).

The filter cap at the input to the VCO also has an effect and you would need to mimic that.

[CheapPedalCollector]

In the other thread for the 670 I posted a schematic that I cleaned up and verified from the one you have with both the units I had. That schematic also has an error in it at the output of the SAD512D, it does *not* connect to both +/- inputs of the opamp. It is otherwise accurate to two actual units. That schematic has part values for R5106, but doesn't say anything about changing the VREG to a 12 volt one, that would blow the chip pretty quickly I would think.

I have found the schematics for the 670 and 680 to also not be accurate, to my units at least. In fact, I found most DOD schematics are not accurate to production models and often contain fatal errors.

[/quote]
Those meters can measure frequency.   IIRC the frequency measurement is digital and quite accurate.   I don't know how well they work over 200kHz.
[/quote]

Oh good, that means I found the issue with both pedals as the clocks are very very low, the delay is only about 31khz. The AC reading I'm getting from them is also about 110mv so that can't be right can it? The chips are new, I'm not sure what would cause that behavior. The op amps have been tested and are fine, I tested the transistors too but I guess they may be failing under load. I really dislike having to pull parts to do this. I bought an ESR meter recommended by EEVBlog but it sucks and can't seemingly measure anything so I just use my cheap cap meter.

It has to be possible because there's already units out there.  I guess what you don't want to do is mess with how the VCO works as that would change the character of the unit.   In this case you don't care what the delay time or frequency is you just copy the existing ckt.

If you were to rejig the whole VCO that's going to come at a risk.   I wouldn't use any calculations to ball-park what the circuit is doing.  You would need to build one.   Then you would need to check your new circuit produced the exact clock frequencies over the whole range of VCO input voltages (at fairly fine steps).

The filter cap at the input to the VCO also has an effect and you would need to mimic that.

Yeah I would definitely have to build the circuits and do some comparisons against working units to make sure any modifications I come up with produce desired results. I have found that theory doesn't always work as expected in reality.

[Rob Strand]

In the other thread for the 670 I posted a schematic that I cleaned up and verified from the one you have with both the units I had. That schematic also has an error in it at the output of the SAD512D, it does *not* connect to both +/- inputs of the opamp. It is otherwise accurate to two actual units. That schematic has part values for R5106, but doesn't say anything about changing the VREG to a 12 volt one, that would blow the chip pretty quickly I would think.

I have found the schematics for the 670 and 680 to also not be accurate, to my units at least. In fact, I found most DOD schematics are not accurate to production models and often contain fatal errors.

The one you posted was the R670 and it has quite a few differences to the plain vanilla 670, but also similarities.

And yes, I've got little trust in DOD schematics.

Oh good, that means I found the issue with both pedals as the clocks are very very low, the delay is only about 31khz. The AC reading I'm getting from them is also about 110mv so that can't be right can it? The chips are new, I'm not sure what would cause that behavior. The op amps have been tested and are fine, I tested the transistors too but I guess they may be failing under load.  I really dislike having to pull parts to do this.

Well, the low voltage is expected.  The waveform isn't square it is a *very* narrow pulse.  It's probably like 200ns wide with a period in the order of 10us (100kHz) so the width is 1/50 th the period.  That means a very small AC voltage.

What pot settings did you get the 31khz on?   

FWIW,  the very narrow pulses in this circuit can screw up frequency meters!   Yes, more complications.  Some of this stuff is a losing battle.

I bought an ESR meter recommended by EEVBlog but it sucks and can't seemingly measure anything so I just use my cheap cap meter.

Cap meters vary all over the place.   The LCR types are the most reliable as they use sinewaves but some pulse types are OK provided you are only measuring a capacitor.   Most meters will not measure correctly in circuit although the LCR types can do a better job.

Note also the job of an ESR meter is to measure ESR.  Capacitance is often thrown in but the measurements aren't always trustworthy or accurate.

Yeah I would definitely have to build the circuits and do some comparisons against working units to make sure any modifications I come up with produce desired results. I have found that theory doesn't always work as expected in reality.

Theory does work but precise theory can get overly complicated.   This is where simulators like LTSpice take over.

However in this case the limiting factor is not having LTspice models for the CMOS gates.   In a circuit like this which pushes the capabilities of the chip and also depends on things like output resistance and input capacitance you need a fairly accurate model.   Such a model is difficult to construct.  It would be a project in itself to come up with a good model.

I've done a few sims with unverified models I've created myself and the results are in the ball park of the calculations.   The uncertainties like the input capacitance have a noticeable effect on the chorus but not so much effect on the flanger.  More complications are: In the circuit the gate has two inputs tied together.  Inside the CMOS chips there are protection diodes (which have capacitance) and protection resistors.   It all complicates getting the simulation right down to the small decimal points.

When you design things like this the simple theory is fine to ball-park circuit values.  You know there are uncertainties.  When you build a circuit you can make a few tweaks to compensate.   In the bad old days you could get caught out as not all brands of CMOS gates worked the same when pushed to the limit.  Typically you would either specify known brands that worked, or one day a production run would have problems and you had to make some adjustments for different brand chips.   Back in those days you couldn't buy faster chips you had to make do and squeeze the last drop out using whatever tricks or tweaks you could.

[CheapPedalCollector]

I figured out my LCR meter and now I can use it reliably, it's pretty sensitive to what frequency it's set to. It's the DE5000. I'm getting accurate values now and verified all the caps I've replaced are way out of spec and have silly dissipation factors compared to the new caps.

OK so the low AC reading is OK, that's good to know, maybe I've got this thing working now. I've been troubleshooting with the SAD512D chip out of it and also trying to understand the circuit absolutely so I don't make any wrong assumptions about it.

Looks like the best chip replacement for it would be MN3004, but that would be hard to fit into a small pedal like an MXR Micro Chorus or Micro Flanger, or the DOD 460 mini chorus. I don't think there is any 8 pin equivalents unfortunately.

I think I'll continue to learn the theory as best I can with my limited math knowledge, but always build stuff and tweak it as needed like I always have. This has been really informative and I learned a lot. I understand what's going on in most analog delay based pedals now. Of course the ones using the specific clock chips are easier to understand, but knowing these old ways are also pretty interesting. Using CMOS in analog ways is fascinating and I want to experiment and see what else can be done with it besides ring modulators, bit crushers and the MXR envelope filter.

[Rob Strand]

It's the DE5000.

They are very good units.  I've got one myself.  I got it after I left my last job since I missed the cool LCR meters they had in the lab there.

I came across this page at one point.  You can see some of the cheaper LCR don't perform so well,
https://www.edn.com/review-continuing-the-search-for-the-ultimate-lcr-meter/

Looks like the best chip replacement for it would be MN3004, but that would be hard to fit into a small pedal like an MXR Micro Chorus or Micro Flanger, or the DOD 460 mini chorus. I don't think there is any 8 pin equivalents unfortunately.

Yes, it going to be a pain whatever way you go.   

If you want to make a board that plugs into *any* pedal then it has to look like a SAD512D from outside, get the LFO/VCO from the existing board, be compatible with existing input output circuits.

Something that occurred to me was to use more modern devices but it seems they are only available in 256 stage or 1024 stage.    Probably need to operate the BBD at 5V or 9V.   You would need more circuit but the idea would be to use SMD to keep the size down.   Unfortunately the modern BBDs are still through-hole.   Two 256 stages is going to be bigger than a MN3004.   If you used a 256 stage divider the clock down by it would work but the quality wouldn't be great.  You could add more filtering but that may limit the circuit you can plug in to - and the filtering will change the sound.   Going for a 1024 and doubling the clock is going to be a headache since you would need to put in a PLL to double the clock.   Other options would be to make mods to the VCO but that going to be pedal specific and doubling the frequency is unlikely to keep the original behaviour.   All messy and intrusive options.