Deluxe Memory Man 4x MN3005 Project

[Kekke]

Hi, first time poster here. For a long time I have had the urge to try 4x MN3005 chips in the DMM. Just to know if this is worth it or not.
Well I am trying to get it right, (I have the necessary equipment) I have been able to bias all BBD chips.
The problem I am running into is that I can't get the repeats clean. Everything functions in the pedal, I get about 900ms of delay.

I am aware of the jumper points on the pcb and I have invested hours and hours in this pedal figuring it all out.
The pcb I am working on is EC 2 002 Rev.D and originally had 4x MN3008 (now replaced by genuine MN3005).
All my Jumper points are open (as they should be). The sometimes cut trace at TP 2 is closed.

I have doubts about calibrating mostly. If I turn down both trimpot 3 and 7 all the way, the pedal responds somewhat normal but my repeats are just not clean and oscillation is just around the corner.

I increased the input impedance with about 100k, this helps but I am missing something and seem to have too much gain at trimpot 3.
Maybe someone can offer me some help or talk me through how I can properly do this calibration?
I will document every step.

This is the PCB:
It is loaded with 4x MN3005.


Let's start at the beginning:
The signal I am inserting is 250Hz at 250mV p-p, in another calibration manual from the maker, he recommends 400Hz with 3.8V p-p (because this is the max headroom of the MN3005). I think it doesn't matter so much because we will sweep through different frequencies. But if someone can verify, I would feel that I am doing this right.

[LaceSensor]

I seem to recall there being a query related the the clock / buffer being able to cope with the 3005 vs 3008

I'm not sure if that would cause the issues you are encountering

Also if the board already had 4 x 3008 why was there a need to revise a trace cut??
The jumpers etc on that revision are there so it can be loaded with either 2x3005 or 4x3008

To test the theory on the clock / buffer you could try subbing out 3005 for 3008 one by one and seeing if you get a good compromise of performance (delay time vs clarity)

[thomasha]

That is an interesting one. Have you seen a schematic or guts of a 1100 ms DMM?
I never saw one, but it would probably answer some questions.

It probably needs a buffer or another VCO to drive 4 MN3005s.

I have calibration instructions for an older DMM that uses 250 Hz 500 mV p-p at the input,
but with the level pot set to get 1.5 V p-p at the output of the compander (pin7).
Here is another one that uses 500 mV.


Can you check the bias for each chip individually, and verify the repeats? Inject the signal directly at the input pin of the BBD and see if alone it can produce a clean output. And then adjust the trimmers to have unity gain all the way.

[Kekke]

Also if the board already had 4 x 3008 why was there a need to revise a trace cut??
The jumpers etc on that revision are there so it can be loaded with either 2x3005 or 4x3008

To test the theory on the clock / buffer you could try subbing out 3005 for 3008 one by one and seeing if you get a good compromise of performance (delay time vs clarity)

The trace that is cut, disconnects the non polarized capacitor C16 (1µF) from R24(2.4k) and R25 (2.4k) in schematics with only 2x MN3005. Essentially it just cuts the extra BBD stage that is unnecessary with a 2x MN3005 setup.

See schematic:




When comparing schematics with 2x MN3005 and 4x MN3008, I found out that in the EC200 2 REV E versions there is a trace cut at that point. It basically disconnects the blue stage in the schematic above because it is not needed if you use 2x MN3005.
If someone wants to install 4x MN3008 or 4x MN3005, this cut trace should be bridged or the stages will not be able to connect in series.

[danfrank]

Yes, it will need something like a 4049 to drive the 4 X 3005 clock pins.

4047>>>4049>>>3005's

Take a look at the Current Lover build doc on the madbean website, but disregard the 4013, you will be using your 4047 instead.

BTW, I have a 1100 DMM and am not impressed with the sound. The standard DMM sounds so much better.

[Kekke]

Yes, it will need something like a 4049 to drive the 4 X 3005 clock pins.

4047>>>4049>>>3005's

The version I am using has a  HCF4047BE.  Will this be sufficient to drive the clock pins?

[Kekke]

I have another deluxe memory man that was perfectly calibrated.
Some observations, I would like to share:

Following the signal path and comparing signals through both pedals,
I have equal gain at U1A, U2B input signal is set for unity gain with the pedal's input signal (turning the level control to match).

Keeping in mind that in a factory unit with 2x MN3005, jumper points JP1 and JP2 are closed (not JP3).

When I probe and observe the signal on the O-scope at pin 7 of the first (U4) - MN3005
and compare this with a well calibrated unit that uses 2x MN3005 (PCB REV E).

I see a big difference in gain, the signal going in on U4 is much taller in amplitude when all jumper points are open
(this is with 4xMN3005 installed).

Conclusion:
So, I am thinking that there is a difference in gain structure for 2x MN3005 and 4x MN3008. There seems way more gain in the 4x MN3008 pcb configuration. I am linking this to jumper points JP1 and JP2 that are open in this situation.

The output gain of U3B U3A is controlled by trimpot 3 and also sets the gain for the second pair U6 and U7 (normally 2x MN3008, now 2x MN3005).

Problem: The output gain of U2A-B has no trimpot control to lower the input signal which is going to the first pair of BBD's, U4 and U5.


If someone has a schematic of REV_E, I would like to compare both first to confirm this. I have been looking at schematics recommended here but was not able to figure out any differences that could explain the huge amplitude at my first BBD chip apart from the opamp gain of U2 that has to be lowered.

[thomasha]

Is there a schematic for the rev D or the rev E?
Would help to understand what the jumpers are doing. Is there a gain stage between BBDs?
I would say that with 4 chips you will have one extra gain stage for each chip, which would need adjustment.

The vintage DMM has one to compensate/adjust the gain of the BBD, so that the second BBD receives the same signal level. I guess here you want something like that.

The BBD itself has a gain higher than 1(in my builds at least...), meaning more BBDs produce more signal, and at some point it will distort (without gain stages or voltage dividers to reduce the gain).

[Kekke]

Is there a schematic for the rev D or the rev E?
Would help to understand what the jumpers are doing. Is there a gain stage between BBDs?
I would say that with 4 chips you will have one extra gain stage for each chip, which would need adjustment.

The vintage DMM has one to compensate/adjust the gain of the BBD, so that the second BBD receives the same signal level. I guess here you want something like that.

The BBD itself has a gain higher than 1(in my builds at least...), meaning more BBDs produce more signal, and at some point it will distort (without gain stages or voltage dividers to reduce the gain).

The schematics don't have any information about the jumpers but my theory above about the jumpers is 100% correct.

The jumper points JP1 and JP2 bypass the extra stages which need to be connected only if you use the factory 2x MN3005 setup.
JP3 IS NEVER connected in any case at all.

If you want to use 4x MN3005, all jumper points have to be open (same as a factory 4x MN3008 setup). There is no doubt about this, but you are right that there is added gain when you use 4 chips!
Trimpot 3 (negative feedback loop of U3A) sets the gain for the second pair of BBD's (U5 and U6). I have it at the lowest setting in my case and this works out. That way the opamp U3A has no added gain at that stage (because you don't want the gain with 4x MN3005's).

What I did find out is that by adding these extra BBD stages (with 4x MN3008), like you say the gain goes up indeed, but there is enough headroom in the BBD's to take that signal).


I actually found a fix by lowering the gain of the NON-inverting opamp before the compander, I got rid of the clipping now. This opamp is set with a gain of 2 stock, it seems just a little bit too much. Still some further testing and calibration is needed now.