Help: Roland SDE 3000 delay runs 20 minutes then dies

[tho_dk]

Hi
I have a Roland SDE 3000 Digital Delay. Works and sounds great except for one thing: It is only able to run about 20 minutes at a time. After 20 minutes or so it dies and has to rest about an hour or two before being able to power up again. I seen in other threads that people (Fletcher for one) have had problems with these units overheating. Could this be the problem here?
Does anybody have experience with this problem and/or any ideas how to solve it?
By the way it occurs even if the unit is out of the rack so its not due to being in an overheated closed rack.

Hope very much someone is able to help... Thanks in advance!

Best regards
Thomas

[tho_dk]

Really... nobody?

[John Lance]

I don't have a schematic. You need a schematic.

Take the top off of the unit first, then power the unit on. Cover the unit with something non-metallic and within reason so that it will heat up and fail.

Start diagnosis after the unit has failed. Begin checking the power supply voltages. Be aware that one slip with meter probes and you'll be dealing with more problems than you started with.

There are sooooooooo many things that can fail causing your problem. It is easier to take it to a technician that has all appropriate equipment, schematics, know-how, and has access to Roland for parts and support.

[conleycd]

I don't know maybe a bad capictor or power regulator?

CC

[nofi]

yeah... if you just want to chuck new parts at it, I'd be looking at the regulator... probably the 5V digital supply regulator seeing as the whole unit just dies.

but also, look into the main bridge rectifier, and the bulk caps after that. could be something funny going on there, too.

that will probably fix the problem.


but really - the SDE-3000 is a very nice delay unit, and totally worth having working in your studio - so definitely worth taking it to a tech and getting it sorted out properly!

[John Eppstein]

After you've had it on till it fails you might try unplugging it and checking to see if anything is hotter than it should be.

It also might be worthwhile pulling it apart and inspecting all the solder joints, perhaps touching them up. Look for joints that show a "ring" pattern in the solder that could be a sign of a cracked joint. Often cracked joints will open up due to thermal expansion. Use a magnifying headband for this inspection - you can get one for only $5 plus shipping from All Electronics

All Electronics | Electronic and Electro-Mechanical Parts and Supplies at Discount Prices

It might be a good idea to touch up the solder around any parts that get warm on general principle, even if you don't see anything.

Also if any chips are socketed, make sure they're seated properly.

You could also try using freeze spray to see if you can make it come back on.

[Lasermonkey]

My SRV-330 kept crashing/re-starting. I eventually found it was one of the bridge rectifiers. One of the diodes had failed. Once replaced, it worked fine.
From what I can gather, the PSUs in Roland units from this era can be a little problematic.

[kb_cheapmods]

On Sunday I picked up a dead SDE-330 from CL for $5. Had the classic 'powers up but no function, only slight backlight'.

I opened it up to find a lot of crud surrounding the power supply caps on the top of the board - I thought it might have been glue to hold them in place, but I'd read other forums where the power supply on the Sxx-330 had issues. The regulator looked OK as did all the solder joints.

So, I replaced all the caps in the power supply with same or next-higher-standard-value uF values and same-or-next-higher-voltage ratings (mostly because that's what I had on hand).

Hooked it back up to mains and it booted perfectly. I had it running the night before last for about 1hr playing with it (my first rack delay, so a lot of newbie fun!) and it seems fine.

So I recommend recapping the power supply. It's just a good idea for any electronic equipment approaching the drinking age!

[auddoc]

The power supply is definitely the first place to check. measure all the main points, +5vdc and the bipolar +/-15 volts (may be 12, I forget) when first up. Let it fail. repeat measurements. Also take note of the amount of ripple (AC) on the 5 rail. If the voltage drops or the ripple increases in failure mode, either the regulator or the electrolytic caps are suspect. Roland and Yamaha both use high speed switching diodes that can get funky- check those as well. Low speed diodes won't work so direct replacement is called for.
One good trick, is to use cold spray on individual components that may have gone thermal - slowly, one at a time and let return to normal operating temp- I've had sucess locating thermal parts that way. Other than that, check the reset line on the CPU is not pulled high. The power supply can probably be done without a schematic, and some experience, but the rest is pretty dense, lot's of luck. I believe that unit has some proprietary gate arrays (big-ass chip) that are, more of the replace and cross your fingers situation.

[auddoc]

As John Lance suggests, a steady hand is required to not accidently short components and make it worse. Or Voila!- paperweight! Make all your measurements at the connectors where possible. Use an alligator clip cable.to make a seperate solid ground connection first.... that way you only have one lead to deal with....and lay off the caffiene, for a surgeon-steady hand. I also put my DMM test probes on a grinder to get an acupuncture-sharp tip happening- then you can pierce wires and conformal coatings with sharp tips during measurement.

[Dissdnt_penguin]

I know this is an old thread, but I don't know if you found a solution to your problem.

It sounds more like overheating to me.

Do you have it in a well ventilated place? This things get extremely hot, and if you don't have enough room to dissipate the heat, it could really damage your unit.

I have an open slot right under it in my rack, and sometimes place a fan under it. There are some cheap multimeters equipped with a thermal probe. Try measuring the temp on the IC's and transistors once it shuts down. That will probably give you an idea on how hot it is getting.

Hope this helps.
DP

[Dissdnt_penguin]

Originally Posted by John Eppstein

You could also try using freeze spray to see if you can make it come back on.

That could really damage the unit. Violent temperature changes have very bad effects on electronics, and some components spec's recommend not to get below 0 degrees, sometimes more. Also solder joints can crack.

DP

[John Lance]

Originally Posted by Dissdnt_penguin

That could really damage the unit. Violent temperature changes have very bad effects on electronics, and some components spec's recommend not to get below 0 degrees, sometimes more. Also solder joints can crack.

DP

In experienced hands, a judicious touch of freeze spray can be just perfect for quickly isolating overly temperature sensitive defective components, and this persons particular case would be a good candidate.

A little reading up on component specs before use is good.

A little perusing of a schematic before use is good too. You could be violently altering bias points in a circuit designed to be at an overall temperature that is more stable. A case in point would be the output section biasing circuitry in a power amp. So what does happen to a component and the surrounding circuit when the temperature drops precipitously on that one part and the circuit is active? What kind of reaction should you expect? You should know before using the stuff. Most of the signal level and digital processing equipment is pretty safe to use freeze spray in though.

I would rather take care of a solder joint that would crack (I've never had one do this) now, than to have the joint fail later, which it would do so eventually if freeze spray is going to be messing with it.

[Dissdnt_penguin]

Originally Posted by John Lance

In experienced hands, a judicious touch of freeze spray can be just perfect for quickly isolating overly temperature sensitive defective components, and this persons particular case would be a good candidate.

Good to know. Thanks!
DP