ARP 3620 clone project

[fallout]

Thanks. I do have decent mechanical skills but most of my tools are for wood working and electronics but I don't think anything you're explaining requires very much. When you say 'extruded' aluminum plate, that's pretty vague. Can you give me a better idea of what you mean? Or do you simply mean mounting the switches on a piece of aluminum plate and attaching that to the frame? That would certainly make wiring the buses a hell of a lot easier! As for orientation of the switches, I've attached another pic. Is this what you are suggesting?

As for the switches themselves, I'm not even sure if it's worth disassembling them. I could simply just cut or remove the unused contact wires and call it a day.

Thanks for the advice!

[fallout]

*deleted*

[KSS]

Or do you simply mean mounting the switches on a piece of aluminum plate and attaching that to the frame? That would certainly make wiring the buses a hell of a lot easier!

Yes. It also makes it possible to put the switches 180 from how you've shown them here. The plate and the actuating extensions rise you above the parts that would otherwise conflict with switches in the lower portion instead of upper. Where the larger dimensions-area allow you to reduce the switch angle and so the left or right overhang may be reduced. However this might be at odds with leaving them fully assembled due to height constraints? Mounting them as a group on a flat or angle<--for stiffness- also makes the point-to-point wiring of the buses to resistors and caps more robust. I'd still recommend a PCB for those.

As for orientation of the switches, I've attached another pic. Is this what you are suggesting?

Yes.

As for the switches themselves, I'm not even sure if it's worth disassembling them. I could simply just cut or remove contact wires and call it a day.

I would disassemble them. At least one. And see if the rivets can be easily exchanged for a couple common machine screws. If not, then keeping them assd. has higher priority. But if a screw fits, you get a bunch of extra switches for a second kbd -or repairs!- and you're going to have to mount the switches somehow anyways, which means screws or glue.

I would expect they already have machine screw sized holes in addition to the rivets? Your photos all show them 90 degrees rotated from use position, so its left unclear so far. But if-since that's likely, you only have to grind away part of the 2 rivets per switch? and could keep them in place for alignment while using screws for fastening.

Edit: This also means you can add extra switch per key functions like Bob Moog did for his keyboards. As did ARP on the 2500 and Dave Rossum on the Emu modular. These KBD function switches can be used in a HUGE variety of ways! Here they're basically arriving 'for free'.

[fallout]

Yes. It also makes it possible to put the switches 180 from how you've shown them here. The plate and the actuating extensions rise you above the parts that would otherwise conflict with switches in the lower portion instead of upper. Where the larger dimensions-area allow you to reduce the switch angle and so the left or right overhang may be reduced. However this might be at odds with leaving them fully assembled due to height constraints? Mounting them as a group on a flat or angle<--for stiffness- also makes the point-to-point wiring of the buses to resistors and caps more robust. I'd still recommend a PCB for those.

Duh.. Yea, the switches would now work orientated towards the rear with a piece of aluminum clearing the conflicting parts. I think I'll go with that idea. Thanks!

I would disassemble them. At least one. And see if the rivets can be easily exchanged for a couple common machine screws. If not, then keeping them assd. has higher priority. But if a screw fits, you get a bunch of extra switches for a second kbd -or repairs!- and you're going to have to mount the switches somehow anyways, which means screws or glue.

You're correct, they're held together by the 2 gold rivets. The black one are just guides to keep the wires in place. I'm not sure how keen I am to disassemble and put 49 of these and put them back together, they're fiddly little things! I'm thinking that using glue to mount them would be the answer but I'll take one apart and see if there's any options to mount them using the holes from the rivets. I have a lot of time to think this through.

I would expect they already have machine screw sized holes in addition to the rivets? Your photos all show them 90 degrees rotated from use position, so its left unclear so far. But if-since that's likely, you only have to grind away part of the 2 rivets per switch? and could keep them in place for alignment while using screws for fastening.

With the exception of the rivets, there's really nothing that can be done with the other holes in the switches. All of these holes intersect with the wire contacts themselves.

Edit: This also means you can add extra switch per key functions like Bob Moog did for his keyboards. As did ARP on the 2500 and Dave Rossum on the Emu modular. These KBD function switches can be used in a HUGE variety of ways! Here they're basically arriving 'for free'.

...and now I enter the rabbit hole

Thanks again!

[cs1729]

I bought one of these on reverb which has the original black metal faceplate, rev 1 board + 2 other boards for the keyboard and 2 film capacitors.

Unfortunately, I didn't read through far enough in the thread and this doesn't fit into a 3604 keyboard so I can't do the project. (I just wanted to swap out my 3604 module.)

If anyone comes looking in this thread to buy one, PM me and I'll sell it for $100 + shipping. Thanks.

EDIT: SOLD

[Kipling]

Unfortunately, I didn't read through far enough in the thread and this doesn't fit into a 3604 keyboard so I can't do the project. (I just wanted to swap out my 3604 module.)

I was going to say that you can squeeze the PCB into the 3604-P case by carefully sanding it down on three or four sides (before you hit copper). The front panel would also have to be trimmed and may not look that great as the screw holes may have to be moved accordingly. At the time I designed it I didn’t have the dimensions of the 3604-P so took an inspired guess.

I do have a design which will fit but have never had any PCBs made for it as the demand is low. I need to check it over, along with the latest version of my normal (larger) PCB. If you are interested I can get a small batch of five made, or perhaps post the gerber files as it would be cheaper for you to buy five direct from a PCB fab than me have some sent to me in the UK and then post one on to the states. Similarly I’d be happy to provide a front panel drawing to fit too.

The latest version is cheaper and easier to build as it does away with all the unobtanium components and uses a much cheaper rocker switch.

Some adaptation of the case internals is required to accommodate the PCB and you probably need the 3-bus to 2-bus adapter board too, and I do have a few of those.

[cs1729]

I wouldn't have a problem sanding the pcb, but the panel was very nice so I didn't want to sand it down. I would rather pass it on to someone who will use it.

I should just fix my 3604. I have a weird problem where the cv starts dropping 1-2 cents/sec after releasing a key. I went through the electronics and they are all fine so there is something that needs isolated from the panel. I just thought I would get this to use and toss it in quick, but it's ok.

If you made another run of pcbs / panels that fit a 3604, I would be interested, but I most likely won't make one on my own. Before I found this one, I checked on OSH park or another site and it was pretty expensive to get one made. Thanks

[KSS]

I should just fix my 3604. I have a weird problem where the cv starts dropping 1-2 cents/sec after releasing a key. I went through the electronics and they are all fine so there is something that needs isolated from the panel.

Sounds like your big yellow cap has gone bad. Common problem. That or the DOA Discrete Op Amp M1 gone bad. It can be directly replaced by a modern hi Z FET input OPA.

Be sure the PCB is really clean too. Especially under the M1 -you can see the guard trace squared off- 'loop' and that area neeeds to be really clean.

Sometimes the relatively loose wiring gets pinched on re-assembly, but I'd assume you've already covered that possibility.

Keep in mind that the 3604 PCB is set up for duophonic operation. It's just missing the parts to do so. The 3604 PCB is the exact same used in all ARP2500 KBDs. Technically those are 3xxx KBDs.

[cs1729]

I should just fix my 3604. I have a weird problem where the cv starts dropping 1-2 cents/sec after releasing a key. I went through the electronics and they are all fine so there is something that needs isolated from the panel.

Sounds like your big yellow cap has gone bad. Common problem. That or the DOA Discrete Op Amp M1 gone bad. It can be directly replaced by a modern hi Z FET input OPA.

Be sure the PCB is really clean too. Especially under the M1 -you can see the guard trace squared off- 'loop' and that area neeeds to be really clean.

Sometimes the relatively loose wiring gets pinched on re-assembly, but I'd assume you've already covered that possibility.

Keep in mind that the 3604 PCB is set up for duophonic operation. It's just missing the parts to do so. The 3604 PCB is the exact same used in all ARP2500 KBDs. Technically those are 3xxx KBDs.

Thanks, I did try replacing the caps, I also replaced the module with a hi z op amp. Tried isolating the port pit from the panel which the service manual suggests. Then I realized there is a kind of layer of flux that I just can’t really get off of there on the whole board. I bought some acetone to try to clean it but I got nervous it would melt the board so I shelved it for now. Is there any other solvents you know of that would work without damaging the board or parts?

Thanks

[KSS]

90+% alcohol and a stiff toothbrush should work fine. I generally wash ARP PCBs in warm water with Simple Green or Dawn detergent as a first step after assessment and basic troubleshooting. Be careful doing this 'bathtub' wash if you've not done it before. Some parts need care to avoid making new problems. Like the sliders. I advise avoiding ultrasonic cleaners unless you REALLY know what you're doing and have the *correct* type of ultrasonic cleaner for what you;re cleaning. Hint: You won't find it at Harbor Freight!

Could of course use a commercial flux-off type product, but probably don't need to.

If you use acetone, be SURE the PCB is *vertical* rather than flat to the ground. Perpendicular to the floor. Not parallel. That way any drip stays on the PCB trace side -excepting capillary action- and you handle that by being careful with amounts and which edge -of the main four- is downward. In short, control the drip with gravity and the capillary effect with quantity applied. Work over something that won't be damaged by acetone drips.
------------------
Not too many parts in play in that circuit. If you've already eliminated the S/H -and possibly the port.- cap and OPA, the only remaining item is the FET and its drive.

[cs1729]

I got it pretty clean with acetone. I think it is some strange leak from the cv being on the panel. Service manual says to isolate the portamento pot and that helped, but still it goes out of tune after 20-30 seconds. What is a reasonable amount of time for it to hold a pitch? I

I gave up and put in a lf398 sample & hold, but even it drifts down about 1mv / 2-3 seconds. Normally I wouldn't care, bc I just jam a card between keys to hold it down, but the keys on the 3604 are not shaped to make it easy to pin it down.

My big question here is, what kind of hold times should I expect if it's working perfectly?

I am interested in the 3620 files that will fit the 3604. I may just do one slow over time. Or maybe there are other people who want one and we can group together?