Klark Teknik DN780 DSP board repair

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

donzo93

Member
Joined
Feb 2, 2015
Messages
17
Location
Italy
Hi everyone! I've recently acquired a cheap (and broken  ::) ) dn780 reverb. It has a super noisy audio output, it gets a little bit better using program 31 (delay line) but everything else is a mess. I followed the service manual instructions and I found the culprit in the DSP board. Everything else is ok. Now, I know that this board is almost impossible to rework/fix: the pcb traces are super tiny, the DSP chip is unobtainable (and soldered in the socket, kinda strange from my point of view) and last but not least it wasn't meant to be fixed by the average tech. That said I just wanted to know if someone more experienced here attempted to fix this board and knows how it works in detail. From the schematics I roughly understand how it works, but I'm no engineer so I don't grasp everything. One thing that I noticed is that the O/Load (Overload???) led is always on apart when using program 31. This is connected to a monostable multivibrator that seems to command the write cycles to the sram, if that makes sense. As of now my intention is to start replacing all the dram/sram chip with NOS ones, hoping the problems is a broken one. If anyone has suggestions or paper/manuals to read that will be useful they're welcome. I'm also looking for the datasheet of the original dsp: the ones in my unit is a weitek wtl1010jc.

Thanks,
Marco
 
That is going to be a difficult fix. But if you're just bored then I would start by tracing the audio backwards to see where the noise begins.

But if we assume that you are correct and that it's coming right out of the DSP (have you actually verified that by looking at the output of the DAC which appears to be pin 17 of the PCM53JP chip?), then flip it over and look very carefully for cracked solder joints on one of the data pins maybe. You might get lucky.

Is the output just pure noise or is there some content present? If you put a 1kHz tone into it, do you get any 1kHz components out? If the effect is bypassed is it still noisy? Meaning can the DSP output clean signal of any kind?

If the output is not just pure noise, then there is hope in that ultimately there are some bits (as in 8 bits in a byte) that are wrong resulting in what sounds like noise but it's actually just bad data and that might be fixable. It could be that a ram chip is returning garbage as you suspect. It might be worthwhile to use a scope and check some basics of various chips. For example, if you have two probes, put on on the clock and the other on data in or data out and see if anything looks obviously wrong (possibly compared to data in / out of other similar chips).

 
Thank you swuarewave. I bought the unit to have fun repairing it and possibly learn something new. I knew it'd be a mess, so I'm not in a hurry to use it.

squarewave said:
But if we assume that you are correct and that it's coming right out of the DSP (have you actually verified that by looking at the output of the DAC which appears to be pin 17 of the PCM53JP chip?)

The instructions in the service manual are very well laid out: if you follow at 4.02 you interrupt the AD/DA clock signal coming from the DSP board and you also disconnect the data cable going to the dsp. In this way the signal does a roundtrip through the converter section. Doing this the signal is absolutely fine. So everything is ok in the converter, the problem has to be in the dsp section, or in the connection, as you said.

squarewave said:
Is the output just pure noise or is there some content present?

Sound is present at the output, even with the dsp "on". The content varies with the programs: with halls, plates and reverb in general it is very noisy, almost pure white noise. This is due to the nature of the algorithms I think, in the end they are a bunch of delays in series with a feedback network. Also the O/L led it's constantly on with these reverb programs. With delay programs the signal gets better but it's heavily distorted. The noise in program 31, is rythmic and quite musical. Also right channel is noisier than left channel: it is almost pure white noise.

I will try and probe everything as you say, but I think that's going to be a mess without a logic analyzer of some kind.


 
Change the ram, most of the time they fail. If you are in a rush, try swapping them and check that the noise change.
Also those dac are good candidate to die
Good luck
 
I'll try and report the results as soon as I've some more spare time from studying. As I said it is more of a "didactic" fix for me so I'm not in a hurry, I just wanted to hear the advices from you wise people before start messing around  ;D My biggest fear is that the PROM in the dsp board has gone bad. From the schematics it seems to control huge amount of things in DSP section, practically everything. In that case a fix does exist but it involves having the content of this PROM, which I won't have if the chip is damaged. Even a bad DSP would be a problem, but not that huge as these DSP appear regularly on ebay. DAC chips also are easily obtainable from ic brokers. Everything else is quite standard so no problems there. Now a silly question: is it normal for the dsp ic to run quite hot? Like I can still keep my finger on it, but it's very hot. Yesterday I thought it was normal under operation, after all it's a calculator, but reading the datasheet I'm not so sure anymore.
 
In the klark reverb the big one you called a dsp is indeed a 16x16 multiplier and yes it run very hot. If you would have to do the same all the time, your brain would run hot too :) I have seen some people who use some heatsink on top of it. Pretty sure the problem come from the ram... Make sure you get the same speed rating
 
Niketouille was right! One of the sram ics between 1 or 2  on dsp board (the dsp cache) is defective. I swapped 'em with ics 47,48,53,54 (ic1 to 47, ic2 to 48, ic3 to 53, ic4 to 54 ) and the unit reported "ERR 2" which means that the microprocessor can't load the program correctly on the dsp board. At this point, following the manual, I swapped ic 47,48 for 53,54 and the unit reported ERR 3. The manual itself says to just replace those two chips. So this is probably the main problem of this unit or at least I hope it. Taking a closer look on the faulty ics they do have a very nasty look: the top looks like paint has been sprayed over and then the component name was engraved in, the top surface is convex and not planar. They are probably very old, maybe the originals that came with the unit. This or they're fake. I attached a photo of the broken one. The other 4 ics were brand new instead. Already ordered a complete dram/sram set to replace everything  8) 8) 8) 8) If the unit is right after this fix I'll be doing a complete memory dump of the machine and I'll post the .hex files here. Thanks everyone for the replies.
 

Attachments

  • damaged_ic.jpg
    damaged_ic.jpg
    1.8 MB · Views: 45
I'm back! Finally the Sram ics arrived, changed 'em, aaaaaaand BANG! the unit is now functioning almost properly. Let's say 90%. The effects are here and are usable. BUT there's still that pesky rythmic noise. It is for sure coming from the dsp board: from the test written in the service manual the problem lies there, and it makes sense because the noise appears only when the dsp cable is attached and when the dsp controls the audio clock. I recorded a sample of the noise. It is HUGELY amplified, normally it is around -64/65 dbFS measured from soundcard. You can listen to it here, it is not high quality but it's ok: https://vocaroo.com/i/s1RGBOXraSi9

What do you think wise people? I have some ideas and I'll make some test soon.

 
Clearly that's not analog noise. It's quasi random "data". Just trace it backward I guess. Presumably the DAC is a separate chip but if you watch the data going into it with a scope and listen to it at the same time you might see a correlation that you can follow further upstream.
 
The noise is superimposed to the sound, and it gets effected too (in the example I was using a simple delay line so it was as if no FX was on). I think it has something to do with the multiplexer array that chose the data to send to the multiplier. Looking at the rear of the dsp pcb today I noticed someone added two capacitors from some pin of a 74f257 (ic7) to ground (?!?!???!!) and also tried to replace a broken trace. I'll try to follow the schematic to see if the arrangement is ok, but probably is not.
 
The noise is superimposed to the sound, and it gets effected too (in the example I was using a simple delay line so it was as if no FX was on). I think it has something to do with the multiplexer array that chose the data to send to the multiplier. Looking at the rear of the dsp pcb today I noticed someone added two capacitors from some pin of a 74f257 (ic7) to ground (?!?!???!!) and also tried to replace a broken trace. I'll try to follow the schematic to see if the arrangement is ok, but probably is not.
Hey Donzo, did you bring it back to life totally?
Who sells the UM2148 chips?
thanks
 
Hey Donzo, did you bring it back to life totally?
Who sells the UM2148 chips?
thanks
Hello Michinger:
I know where to get cy2148 wich are the ones installed in my unit . Did you manage to get some? did you fixed your unit? If so could you tell me what it was as im having similar problem and after replacing all of those still the same.
Thanks
 
Hello Michinger:
I know where to get cy2148 wich are the ones installed in my unit . Did you manage to get some? did you fixed your unit? If so could you tell me what it was as im having similar problem and after replacing all of those still the same.
Thanks
I had the chance to buy 2 of DN780, one defect one. But I didn´t buy it at the end. Sorry, I can´t help...
 
Back
Top