Deluxe Electric Mistress Clock Frequency

[miketbass]

Hello, I have an interesting finding that I'd like to run by everyone. About 6 months ago I purchased a DEM v2 and set about cleaning and calibrating it. It was in great condition and the only bumble along the way was getting the clock in range. The trim pots appeared to have been untouched but the clock frequency was around 60Khz! Bringing it down to about 40Khz (spec per EH is 35Khz) was the lowest it would trim to, so I chalked it up to shoddy EH factory calibration. This board was a revision C and had some trace cuts and a jumper across the clock trim from the factory.

Flash forward and I happened across a DEM needing a new SAD1024 chip - this one is dead mint and came with the OEM box and paperwork too! I could tell that the back had never been off by the way the screws loosened - that slight crack of 40 years of being untouched.

This board is slightly later, a rev D but still with the silver/black graphic. Popped a new Reticon chip in and went about checking calibration - needless to say everything was a hair off of perfection. But the clock! Measured at 58Khz at pin 1 and 5 of the 4013. Head scratcher huh? All measurements are done with a Fluke 87 and my scope confirms it...

Side by side it is noticeable that the new rev D with the higher clock sounds different. Of course it sweeps much higher and the turn-around in the sweep is near impossible to dial out, where as the older DEM sounds much more in line with any of my other 70s flangers.

Has anyone ever encountered this? All calibration steps and measurements are gleaned from a factory EH doc for post '79 DEMs. I have double checked that my measurements are accurate and at this point my only explanation is that the service document in circulation for this circuit and the actual clock frequency of units leaving the factory conflict - the OEM calibration sets the sweep much higher. Thoughts?

[Scruffie]

Don't think I've ever had a problem setting one to 35kHz...

Got some photos of the component side of the boards? Seeing if there was any circuit changes might shed some light on matters.

[miketbass]

Not exactly world class photography, but theres really nothing of note going on with the board to my eye. This would have been the last board revision made for this model and it doesnt have any of the tagged on components on the backside - which was common to EH. Ill snap some pics of my rev C board here in a minute. Note the fixed resistor in place of the gain trim. Both of my boards have this and I suspect EH decided they could do without the additional trim pot sometime in 1978.

This one has a pots that date to 1978 but is likely a 1979 unit and would have been made right before they switched to the black/green graphics. Adjusting the clock trim gets me down to 39Khz but no further, identical behavior of the slightly older rev C.

[miketbass]

Heres some pics of the rev C board, and a side by side of the two. Note the tagged on components on the older board. Both of these exhibit the same behavior during calibration - a seemingly factory setup around 60Khz and a minimum trim range of 40Khz or so, measured at pin 1 of the 4013 with a Fluke 87 and confirmed with a Tek 453.

[Scruffie]

I was hoping to check the component values and don't think I can with those.

Admittedly it's one of the EHX pedals I get the least of in for repairs but I did a V2 about 6 months back and I definitely set it to 35kHz because I wrote the sweep readings down afterwards as I'd been meaning to check them.

You're definitely doing the set up procedure correctly (Range Fully CW in filter matrix mode, I've misread those docs before)? And have you checked supply voltages etc? Could be some bad component drift which does happen although the odds of it happening twice in a row...

Edit: Hang on, next to your clock trimmer, above the 311, it looks like the resistor is an 8k2? In the schematic that's 3k9 and I have seen that as 2k7 in a couple of units. Maybe they put what they had on hand in and... left it as is.

[miketbass]

Yeah the photos arent all that great - Ive been trying to handle this unit extremely carefully to avoid any damage. Im doing the calibration procedures by the letter, filter matrix on to freeze the LFO and color and rate set CCW. I played some more with the clock trim and checked range, heres what I have:

Rev D: 58Khz - 515Khz
Rev C: 38.5Khz - 290Khz

These are at the extremes of the range knob rotation. I can play with the trim on both to get similar ranges, although the rev D sweeps wider. Im curious as to the sweep cap value now. It seems that setting the trim anywhere near the factory spec makes flanger fall way short of what I would expect the short delay times to be. My MXR and Ross flangers (also SAD1024 based) sweep a much more expected 32Khz to 500Khz or so. That is what I would expect to see here. Im going to pull the board on the Rev C as its on my bench right now and examine the sweep cap and clock/LFO. I expect to find a 47pf value but we shall see....

[Scruffie]

See the above edit I made.

Is that just sweeping the range knob or in auto mode? If it's auto the Rev C is abysmal, in fact both are way off.

The DEM I measured ran 35-680kHz in swept mode.

For the sweep of the range pot in filter matrix, the clock cap isn't the only important value, the divider chain from the LFO sets the range too, so check the voltages on it.

Also worth checking the 1uF cap off pin 2 of the LM311 hasn't gone dry... in fact that would probably be my bet as to the problem considering the age of the caps.

[miketbass]

My measurements were in filter matrix mode. I'm gonna focus on the rev D board for now, clicking over to sweep and Im getting 60Khz to 700Khz or so. Heres some better pics of the Rev D board.  Both boards when inspected have a 47pf range cap in place. Something of note, when the trim pot is adjusted it does not move in a linear fashion on either unit. The lowest frequency is somewhere the middle, where it "bottoms out", and then climbs up to over 100Khz on either extreme of the rotation. This is not typical behavior in my experience but both units do this. Strange?

[Scruffie]

Again, I modified my message above so I don't keep flooding you with posts, quite late and my brain is on auto pilot

But on the photos, see where the diode, clock trimmer and LM311 are? Just to the right (of the top) of the LM311 and above the clock trimmer is a resistor, what value is it? Could be an 8k2 or 1k2 (it's 3k9 on the schematic) Anyway, it was obviously adjusted at the factory.

The trimmer thing is normal, they wired it in a weird way earlier on. Is that a jumper I see across it? If you remove that, it'll probably be easier to set.

I suspect the issue is the voltage range you're getting at pin 2 of the LM311 though and as I said above, I would suspect the 1uF cap, tantalums dry out quicker than regular electros.

[miketbass]

Edit: Hang on, next to your clock trimmer, above the 311, it looks like the resistor is an 8k2? In the schematic that's 3k9 and I have seen that as 2k7 in a couple of units. Maybe they put what they had on hand in and... left it as is.

That's definitely a 8k2 resistor, but unrelated to the clock trim. The 3k9 tied to the trim is there under the big cap kinda hidden. I lifted one leg of the 1uf cap in question and it tests good - measures 1.08uf and charges up and down with a meter measuring resistance across it. The bridge across the trim is now etched into the pcb. Im not really sure what to make of all of this... If I didnt have a second unit to compare to I'd be going crazy! I guess now my mind is at that I have two vintage DEMs and neither is dialled in optimally.

[Scruffie]

I said it was late But interestingly, that is a 2k7 on the schematic and 3k3 in some units, that's actually the pull up on the 311, that's quite a large value change.

I'm looking at a clearer image I have of a DEM track side, it doesn't have the bridge, hence the confusion.

I think I'll get some sleep and reread this thread tomorrow rather than just confuse matters further but yeah, something ain't right with these.

In the meantime, test what voltages you get at pin 7 of the 324 with the range full cw & ccw in filter matrix mode.

[miketbass]

Man, thank you for your help Scruffie. It helps just having someone else to confirm that something that strikes me as bizarre truly is, and you seem more than knowledgeable with this circuit. Get some rest and if anything jumps out at you tomorrow let me know.

I first would like to ease some confusion, the frequency measurements taken early in this thread were ALL with filter matrix mode engaged. With the pedal still cracked open face down, I can flick that switch and get a nice slow and wide sweep to measure full clock frequency spectrum. Pretty nice to be able measure it easily too and my new Fluke meter allows me to (mostly) catch the top end of the sweep, whereas in the past its been a game of visual ping pong trying to watch it on the scope. Anyways...

With the sweep engaged and the clock trimmed all the way to its lowest possible point of 40Khhz, I get a sweep up over 650Khz and then my meter catch quite catch it. Hey, thats way beyond spec from Fluke so I'm pleased to see that. Further it indicates that I am only missing the ability to trim lower frequencies and the upper end of the clock sweep would seem to be fine.

I took measure of pin 7 of the LM324, filter matrix on, with range CCW=1.73v and CW=6.17v. I'm a little out of my expertise here to make sense of these figures being in the ballpark of correct so I defer to the experts. For now I am going to study the schem and see if I can reason why my trim range will not drop low enough.

[DrAlx]

I don't have that exact circuit but I am very familiar with that LFO/VCO combo.

Tolerances on pots (and I am refering to both the range pot and trim pot here) can vary 10% easily.
You can add in 5% for the clock cap, plus some variation in the gain of the PNP transistor charging the clock cap.
It all adds up.  So in filter matrix mode with pot at maximum, two different EMs can be sending different control voltage to pin 2 of the LM311.

When you take you clock frequency measurements in filter matrix mode, take note of the voltage on pin 2 of the LM311.  That's the "control voltage" (CV).  The PNP transistor charges the 47pF clock capacitor and when the cap voltage reaches CV the LM311 will quickly discharge the cap through the diode, and the process repeats.  The discharge creates the VCO "tick".

Larger CV on pin 2 means the cap will take longer to reach the CV level, so clock frequency goes down.
Similarly a clock cap that is 10% larger in value takes 10% longer to reach the CV level, so the clock frequency goes down.

[Scruffie]

You're mentioning of 5% for the clock cap made me realise something, usually they would specify a 5% tolerance cap for clock caps on the schematics (and follow through in the pedals) in this circuit they haven't so it could be 20% if it's a bog standard ceramic, what's the story in the actual pedal?

I looked at some more gut shots and some units would modify the clock trimmer giving you only 50k range and some don't. I also found a photo of another one with the 8k2 for the 311 pull up.

Usually they would modify a circuit if something was off but I could see the 40kHz leaving the factory as is and it's possible this 60kHz unit was made at the end of the day on a Friday.

[ElectricDruid]

What's the typical effect of ageing on the clock caps in these things? Could it be that they all age in one direction? (capacitance decreases with age?)
I know that's the case with electrolytics, but I've not heard such a thing claimed for ceramics.

Cheap +/-20% caps are pretty much a given in gear of that era, I'd have thought. EH were famous for chucking whatever they could get in things - witness the million variations of the BMP. "Does noise come out?", "Ok, it'll do" seems to have been the testing methodology for component substitution!

[miketbass]

So I measured both my DEMs last night at the center lug of the range pot, both swing from about 1.7-6v at the extremes. Looking at some reference material, it seems that the high end of the voltage should swing to 8+v or so. What would be the cause of this?  Parts tolerance? The clock cap looks to be a standard 47pf ceramic job, Ill look closer when I get home.

[Scruffie]

What's the typical effect of ageing on the clock caps in these things? Could it be that they all age in one direction? (capacitance decreases with age?)
I know that's the case with electrolytics, but I've not heard such a thing claimed for ceramics.

Cheap +/-20% caps are pretty much a given in gear of that era, I'd have thought. EH were famous for chucking whatever they could get in things - witness the million variations of the BMP. "Does noise come out?", "Ok, it'll do" seems to have been the testing methodology for component substitution!

Yeah, ceramics can lose capacitance over time after long periods of storage and by the description in the first post, this would fit that criteria so it's certainly a possibility, especially if it was already on the low value side of 20%

That's a bit of a misconception from everything I've seen, they designed around a lot of parts they could just get their hands on cheaply but they were consistent with those designs. They really did use 5% tolerance caps for clock caps and 10% for filter caps in a lot of stuff when called for, but the schematic on this doesn't call for any particular tolerance where it would usually be noted.

[miketbass]

Well I'm at my wits end here guys. Playing with the clock trim has yielded best case results, but still fall short of specs. Both of the units, when calibrated just so and measured in auto sweep mode yield:

Rev C : 35Khz-450Khz
Rev D: 41Khz-650Khz

These numbers are slightly rounded off but these are what I can get. I checked both 1uf caps in the Rev D LFO and both test good. Tagged a 100k resistor in parallel across the trim to bring it to 50k and it raised the floor of what I was able to trim in to 44Khz or so. I know there is much at play here determining the voltages but I suspect that my CV is not going high enough, or swinging wide enough. I can get a range of roughly 1.7-6v measured at pin 2 of the LM311. This seems to be a narrow range with 6v being much too low compared to what I have read.

Any suggestions, or should I box it all up and live with it?

[DrAlx]

Depends on how much you are willing to modify the circuit. 
I would say the CV range is the problem.  There are a few ways you can widen it.
E.g. Look at the schematic and near the rate pot there is a 10k and a 30k resistor. The ratio of those two values sets the width of the triangle wave. So putting some resistance (e.g. 200k) in parallel with one of those resistors will narrow or widen the triangle wave amplitude (depending on which resistor you are effectively modding).  The triangle wave voltage can't go too low or you will get dropouts in the sweep. Try that mod using a large trim pot (instead of the 200k value in my example) and see if you can get the triangle wave closer to spec.

EDIT: Read this old post of mine regarding improving sweep ratio ..
https://www.diystompboxes.com/smfforum/index.php?topic=91981.msg965363#msg965363

Also read the corrected post immediately after it.


EDIT2:  So in your case, you can improve the sweep ratio by lowering the 30k value by putting a large resistor in parallel to it.  200k will effectively lower it to 26k which will probably be too wide a sweep.  You might need larger (e.g. 470k in parallel to it) to give about 28k.

I also noticed that the midpoint of you triangle wave seems much lower than the waves in the pictures.  Could you report the voltage on pin 10 of the LFO IC (LM324), and also the min and max square wave voltages measured on pin 1 of that IC.
To get a symmetrical wave, the voltage on pin 10 should be half-way between the min and max square wave voltages.

[Scruffie]

Depends on how much you are willing to modify the circuit. 
I would say the CV range is the problem.  There are a few ways you can widen it.
E.g. Look at the schematic and near the rate pot there is a 10k and a 30k resistor. The ratio of those two values sets the width of the triangle wave. So putting some resistance (e.g. 200k) in parallel with one of those resistors will narrow or widen the triangle wave amplitude (depending on which resistor you are effectively modding).  The triangle wave voltage can't go too low or you will get dropouts in the sweep. Try that mod using a large trim pot (instead of the 200k value in my example) and see if you can get the triangle wave closer to spec.

EDIT: Read this old post of mine regarding improving sweep ratio ..
https://www.diystompboxes.com/smfforum/index.php?topic=91981.msg965363#msg965363

Also read the corrected post immediately after it.


EDIT2:  So in your case, you can improve the sweep ratio by lowering the 30k value by putting a large resistor in parallel to it.  200k will effectively lower it to 26k which will probably be too wide a sweep.  You might need larger (e.g. 470k in parallel to it) to give about 28k.

I also noticed that the midpoint of you triangle wave seems much lower than the waves in the pictures.  Could you report the voltage on pin 10 of the LFO IC (LM324), and also the min and max square wave voltages measured on pin 1 of that IC.
To get a symmetrical wave, the voltage on pin 10 should be half-way between the min and max square wave voltages.

In this case as the filter matrix range is quite small too wouldn't a more appropriate place be the 47k gain setting resistor coming from pin 6?