ARP 3620 clone project

[Fernandozarone]

THANK YOU SO MUCH.
I'll check all of the steps you addressed, then will be back for a detailed report.
Grazie!

[Fernandozarone]

OK,
I checked the rails and found +15.01 and ... -15.65 V. After calibrating the PSU (It had been replaced with a Power One HBB15 - 1.5 AG) the juice is correct and stable now . Also pressing keys, on the rails no voltage change. I did not go with the scope yet...
But I have checked the V at the pins of 4011 in both positions (OFF and ON of Portamento SW) and after keypressing. Hope this helps defining a "zone" where to pay more attention. PIN 1 = OFF 0, ON 0 - PIN 2 = OFF 0 > 14,37 V after keypressing; ON= ditto - PIN 3 = OFF 15 V > 14,99 V after keypressing; ON = ditto - PIN 4 = OFF 15 V > 14,99 V after keypressing; ON = 15 V > 0 V after keypressing - PIN 5 = OFF 0 V > 14,38 V after keypressing; ON = ditto - PIN 6 = OFF -0.58 V ; ON= 11 V > 13.36 after keypressing - PIN 7 = 0 V. PIN 8 = OFF 0 V > 12.45 V after keypressing : ON= ditto - PIN 9 = OFF 15 V > 14.99 V after keypressing; ON = ditto - PIN 10 = OFF 15 V > 0V after keypressing; ON = ditto - PIN 11 = see PIN 9 (interconnected) - PIN 12 = OFF 0 V ; ON = ditto - PIN 13 and PIN 14 (interconnected) = 15 V always .
At the level of PIN 3 something doesn't seem so right... Service manual , paragraph Memory control : "When the switch is on pin 6 of Z5 (CD4011BE) should be held at +15v by R66 (10k)". I found: PIN 3 = OFF 15 V > 14,99 V after keypressing; ON = ditto... Am I wrong?
As to R66, that is involved in the game, operating on CD 4011 (that I socketed and replaced, naturally) something is strange. In the schematics it is indicated as 10 K; in the board drawing, it is marked as 100k... and that is the one I find on my board. Any clue? ...

[Fernandozarone]

Naturally, I am going on with further checks

[Fernandozarone]

Q3 is ok, swapped with Q 12, vibrato does still work, but the problem is still there...

[Fernandozarone]

My bad, 100K is R65... it's ok, low resolution of the drawing.

[Fernandozarone]

OK I'm back, after many tests.
All of the TP TESTS reported in the service manual are ok. The problem remains.
I have a question for you experts and for the owners of the former 3620: If I play the keyboard with the upper voice CV jack connected, I say, to OSC 2 CV in, (OSC 1 and 2 in the mixer) with Portamento switch OFF I should hear the two different notes played (and this is in my kbd). But if I left my patch cord in, connected to OSC 2 CV in, and switch the portamento SW OFF, the second voice should be disabled... it shouldn't be heard, should it? If this is correct please let me know... Thanks
F

[Fernandozarone]

I correct... "switch the portamento SW ON!, the second voice should be disabled."
Apologize for my bad typing....

[Fernandozarone]

Hi
I beg your pardon for not being up to the task of this problem solving. First of all, I am not able to fix the problem because I am not a professional (just an amateur...) and apologize for this matter. Secondly, because I am really involved in the post-COVID "reprise"at my Dept (excuse my quotation marks... do not ask me why... let's be positive and fingers crossed !) and my time for my beloved, analogue stuff hobby is becoming shorter and shorter...
In any case, I have tested all of the TPs reported in the 3620 Service manual , also with the help of VERY talented friends, and the keyboard DOES work, at least according to those parameters. Tested all of the GATE / TRIGGER / PULSE and they are perfectly OK ( as to instrumental data - but not in a FAST PLAY situation !)
Going beyond the tests, when I play it in Portamento ON > OK
But Portamento SW OFF > BAD news !!!
playing in a rapid sequence 4/5 notes, the first time it works (all of the notes play as they should) but when I replay the sequence , it becomes shorter and shorter. I mean, the second time the sequence stops at the fourth note (but when I replay the fifth it comes out), the third becoming shorter and shorter and so on.
BTW I used the KBD with my OB SEM, it gives the same problem, so it does not seem to be due to a 2600 main box issue!)
I am now giving up. Not a drama, but... if you can give me a clue... very welcome! otherwise, I shall just use it in the Portamento ON SW position. No problem. The thing is, as a physician, ... I would understand what the matter is!!! You know, I would like to find an explanation of everything in the science world...
I also worried about the key contacts. I tried to change the position of the J wires. Closer to the bar, farther ... always with CV touching before Gate. NOTHING changes. All the contacts thoroughly cleaned up, obviuosly.
All of the problems are shown in the Portamento SW OFF position.
BTW, both the polycarbonate capacitors replaced, with similar, tested ones. All of the JFETs are ok. The tests' data are perfect.
I do not want to bother you, if you have any idea to guide me, otherwise ... thanks as well!
Grazie ...
Fernando

[mcbinc]

(I'm the person who was helping you elsewhere with this 'problem'. Suggestions follow...)

There might be people here who can point you to places where 2600/3620 people talk about playing and quirks. They have to be hiding out somewhere. Speak up, folks!

Learn the instrument. Go through the 3620 User Manual examples. Every one. Think about what is going on, make modifications to each example, understand the limits and possibilities. Features in the keyboard are not independent of one another: they interact and constrain. Playing technique matters.

Adjust your expectations. The 3620 is 1970s technology. It uses mechanical contacts which bounce and behave inconsistently and which can cause inconsistent triggering and incorrect CV 'memorization'. It simply cannot deliver digital speed and precision. If you demand digital speed and precision, the path is to use a MIDI keyboard with CV converter. In my case, a previous owner threw the keyboard action away and grafted an MPU-101 to the 3620 electronics. That said, performance tuning is possible but you are going to need a good, dedicated tech or learn some deeper diagnostic skills.

As a physician, I hope you wouldn't just replace human parts until your patients started working better. Same rule should apply to rare synths. :-)

[KSS]

Excellent points, mcbinc!

[Fernandozarone]

:-)
Actually, with the help of a very skilled engineer on Skype we have tested all of the TPs with the help of the scope, all of them are ok. Moreover, we double checked the kbd behavior with the help of the Symetrix simulator (he kindly designed the whole schematics into the program...) and it matches perfectly what we have seen testing the kbd. And his conclusions are that the delay is due to the process itself of sampling the CVs, so, in agreement with mcbinc, it is the "old" design of the analogue 3620. The problem is, it is totally unplayable in the PORTAMENTO OFF position. Another friend has an orange 2600 with the same 3620 (more recent) and it plays perfectly.
Did something change over the years in the schematics?

[mcbinc]

Having another 3620 available helps. What it may show here is a design or possibly components on the edge of failure. Portamento OFF means CV sample-and-hold is ON and this requires a correct trigger signal (correct in value and timing with respect to CV change) and not false triggers. (One issue shown by your keyboard is that it was re-sampling the CV bus on key release. This shouldn't happen but could be caused by switch bounce or marginal component.) Here's where I would use at least three channels on a storage oscilloscope and monitor CV bus, gate, and trigger (internal preferred over external) and look for missing, mis-timed, and extra trigger pulses. The trigger generating circuit is simple but not perfect: one diode, one transistor, one capacitor, and two resistors (plus a third providing a path to ground). Faults found in the trigger signal generation may suggest how to adjust that circuit (or replace and/or substitute components).

[mcbinc]

Here's an experiment about design limitations. This may or may not work. It's based on a read of the schematic. Let's see... Play slow bell chords (sustained arpeggios): one note down, two, three, up to 10. Ascending using upper voice CV. Descending using normalized lower voice CV. I suspect triggering may become unreliable towards the end of that chord.

[Fernandozarone]

Unfortunately I only have an old scope, TEKTRONICS, 4 channels but no memory. In any case, the trigger values (timing and mV) seem to be all fine.
mcbinc, your suggestion is probably right. Some clues could be taken from this experiment.
- Portamento SW ON, upper voice CV cable connected to VCO2: Low C pressed down (plays steadily) , the other fingers play the following keys, in ascending sequence, keeping the fingers down > the scale plays upwards regularly
- Portamento SW ON, upper voice CV cable connected to VCO2: High C pressed down (plays steadily), the other fingers play the following keys, in descending sequence, keeping the fingers down > the scale plays downwards regularly
- Portamento SW OFF, upper voice CV cable connected to VCO2: Low C pressed down (plays steadily), the other fingers play the following keys, in ascending sequence, keeping the fingers down > only the first TWO keys play in sequence (Low C stays ON). Then, from the third finger on, the CV stays on the second note, no matter how many other keys you press. The scale stops there: LOW C plus the second finger key pressed.
- Portamento SW OFF, upper voice CV cable connected to VCO2: High C pressed down (plays steadily), the other fingers play the following keys, in descending sequence, keeping the fingers down > only the first TWO keys play in sequence (High C stays ON). Then, from the third finger downward, the CV stays fixed on the second note. The descending scale stops there: HIGH C kept pressed plus the second finger of the other hand key pressed down.
This could help... Thanks mcbinc!

[mcbinc]

Wow. That is much worse than I expected. If you hold a note with one hand and play with too much legato in the other, triggering would be unreliable. (I.e. technique matters a great deal.) Asking your friend to repeat this experiment with his 3620 might be revealing.

That Tektronics scope should be good enough. Channel for CV bus, GATE bus, trigger test point. Single sweep, trigger when GATE bus drops in voltage when a second (or third) key is pressed. Then play and watch. Look for timing/level differences between good captures and bad captures.

[Fernandozarone]

Will be done!
Thanks!
F

[Fernandozarone]

(un)fortunately,
the test was reproduced on the "sound", perfectly working my friends' 3620 (orange style, later than mine, that's the grey one) and it shows the same, identical result: in Portamento On , one low key down, the other hand plays all of the higher notes in sequence maintaining the keys pressed, all of them sound. Same in descending, reverse style. In Portamento OFF mode, only two of the upper voices are played, because from the third on the upper voice stays still on the second pressed (third considering the first, left hand voice...)
Everything seems to indicate that it is ok.
Do not know what to think. Not that it makes me mad or changes my life, but...I wanted to find an explanation...

[janvanvolt]

Made some progress with mine, awaiting the punch tool to make the wires between both boards. The case however, is shining in its beauty:

[solar1]

Hello Janavolt,
impressive project...I must compliment the keyboard case housing....
what is it made from? cant see a seam anywhere.
beautiful work!

[Fernandozarone]

As to my "haunted" 3620, this is the updated reckoning:
- all of the tests at the TP reported in the manual are absolutely OK, using a Fluke DVM and a Tektronix scope.
- triggers and gates are OK for timing and value, as far as tested in the rules of the manual
- all of the J-wire contacts have been cleaned up shiny, gold is still present on the contacts
The keyboard is perfect as far as I use it with the portamento switch on; it's unplayable in off. Unplayable means that also pressing not too fast the keys, CV does not reach the last note, that needs to be pressed again to get the correct note.
In the service manual it is reported "previous pith or incorrect pitch memorized on key depression" = CV contact meeting bus too late or Gate too soon
I took into account this, that could explain the problem, but
1) IMHO it should happen in both Porta sw positions... shouldn't it?
2) I have tried to change the bending angulation of the J wires, but the problem still remains.
I am seriously thinking of giving up...
If any other suggestions , very welcome!

[mcbinc]

1) IMHO it should happen in both Porta sw positions... shouldn't it?

No. This is why I say go through the User Manual and learn the keyboard. It does not work the way you are assuming. In the case of Portamento ON, the CV values are not sampled-and-held the way you assume. I.e. they are not controlled by the internal trigger.

2) I have tried to change the bending angulation of the J wires, but the problem still remains.
I am seriously thinking of giving up...
If any other suggestions , very welcome!

The Service Manual provides tests for mostly static/steady-state conditions. You are trying to improve a dynamic condition. You will need to set up a suitable diagnostic for this. The manual does not supply this but it does have sufficient information for a tech to set one up. (I gave you some hints above.)

[mcbinc]

Made some progress with mine, awaiting the punch tool to make the wires between both boards. The case however, is shining in its beauty:

That's the Fatar plus converter board combination? That is lovely!

[Fernandozarone]

I know, the static tests are not ok in this situation...
I ask myself if I could send the kbd alone to some tech not so far...

[janvanvolt]

Hello Janavolt,
impressive project...I must compliment the keyboard case housing....
what is it made from? cant see a seam anywhere.
beautiful work!

Thank you - i forward the compliments to the guy. He's my studio tech and also doing 3rd party work. It is plywood, but laminated, laquered and clearcoat. He suprised me with the blue inner color which complements very much the black.

I am starting with punching wires today.

[Fernandozarone]

To whom it may concern,
especially to those who tried to help me solving the problem...
Eventually, after weeks of positive tests … I found the reason why the response of the 3620 was that … lazy.
My kid, Mattia, (I made a gift to him, a basic Arduino course) 13 yrs old, when I showed the 3620 board (opened up on my desktop) in order to see what the electronic components look like, said: “Papi, why that diode seems to be fatter than the others?” Well . It was not a 1N4148, but… after desoldering, it clearly showed the scipt : PH4148, erroneously replaced by … somebody,.
That explains it all.
And now it works well
…
Grazie! And forgive my ignorant approach,.. but believe me nobody could imagine such a crazy replacement!



Fernando