Lexicon 480L HSP Board Repair?

[Khron]

Did you find any in Europe?

https://www.budgetronics.eu/en/ics/memory-ics/hy53c464ls-80-dram-64kx4-80ns/a-18238-10000133https://www.fabian.com.mt/en/products/webshop/10439/upd41464c-65k-x-4-bit-dynamic-ram.htm

 

[paolomunarini]

Thanks for the reply. Did you find any in Europe? Right now eBay seems like the only option. And how did you figure out you only needed to swap out 5?

I got a contact from Lexicon Italy (I am from Italy) and they sold me the last 25pcs they had in stock. They were original HY53C464LS-10.

I chose to replace one full channel at first and I tried to see if the issue was gone. And it was gone. I could probably substitute less than 5 but I decided to go full channel.

 

[michaelpnielsen]

I got a contact from Lexicon Italy (I am from Italy) and they sold me the last 25pcs they had in stock. They were original HY53C464LS-10.

I chose to replace one full channel at first and I tried to see if the issue was gone. And it was gone. I could probably substitute less than 5 but I decided to go full channel.

So either; U64,65,66,67,68 or U93,94,95,96,97? Mine has the NEC D41464C-10 does it matter if I go for that or the Hyundai?

Thanks for all the replies! Happy the thread isn't dead.

 

[paolomunarini]

So either; U64,65,66,67,68 or U93,94,95,96,97? Mine has the NEC D41464C-10 does it matter if I go for that or the Hyundai?

Thanks for all the replies! Happy the thread isn't dead.

Exactly. If I remember correctly U64-68 are the left channel RAMs and U93-97 are the right channel RAMs.

The unit I repaired had 20 total NEC chips. The chips provided by Lexicon themselves were Hyundai. I mixed and matched easily. The RAMs are JEDEC standard, so until access time and banks are equal, they should be total drop-in replacements.

 

[michaelpnielsen]

Thanks! I have someone close to my area selling 4164B-12, is it safe or might 120nsec be too slow?

 

[paolomunarini]

Thanks! I have someone close to my area selling 4164B-12, is it safe or might 120nsec be too slow?

Unfortunately I don't know... I'd suppose that it could affect the stability of the system.

I have to check both on RAM datasheet and 480 schematic if there is some spec about this.

As a safety measure though I went with 100ns...

 

[paolomunarini]

This is the job on the HSP1, if anyone interested

 

[tommia]

Another 480 with problems here. I repaired the powersupply, also found one bad tantalum on the motherboard. I replaced all the tantalums on the motherboard with electrolyrics. The 4 leds on the host go out after start up as it should, larc is intelligent. Output gives very loud noise with some kind of reverb, you can hear that if you change parameters. OVL leds stay on, the vu meters work according to the input. Self test of the motherboard give a clean dry signal but the OVL leds stay on. Extracting neither HSP doesn't help. Extacting both HSP's OVL goes out but of course unit says: not enough hsp. Any ideas?

 

[paolomunarini]

It looks like another flavour of DRAM failure. From both HSPs... I hope this is not the case because 20 DRAM chips are damn rare to find

In order to give a deeper answer I need to delve into the service manual for a bit... I'll come back to you if i find something interesting!

 

[Philbo King]

DRAM...
Reminds me of when I built a Z80 motherboard in 1979 for a personal PC. I put in 32 of 4116 DRAMs to get a 64Kx8 memory. Cost me about $1000 for those ICs.

I picked DRAM because at the time static RAM chips were being built with slightly radioactive ceramic packages, causing random intermittent bits to flip (a phenomenon called neutron single event upsets).

(Grampa ends his weird little story...)

 

[paolomunarini]

Another 480 with problems here. I repaired the powersupply, also found one bad tantalum on the motherboard. I replaced all the tantalums on the motherboard with electrolyrics. The 4 leds on the host go out after start up as it should, larc is intelligent. Output gives very loud noise with some kind of reverb, you can hear that if you change parameters. OVL leds stay on, the vu meters work according to the input. Self test of the motherboard give a clean dry signal but the OVL leds stay on. Extracting neither HSP doesn't help. Extacting both HSP's OVL goes out but of course unit says: not enough hsp. Any ideas?

The service manual says:

Check to see if 480L will pass audio in IN_OUT test. If it is still distorted try passing audio in self test, jumpers on motherboard move W1&2 to 2-3 (self test). If it still has distorted audio then the problem is likely on motherboard. If audio is clean problem is likeoly on one of the plug in boards.

This will help you discriminate a possible DAC/ADC failure from a DRAM or other HSP failure

 

[tommia]

Thanks for the help so far. But as I wrote: "Self test of the motherboard (the jumpers) give a clean dry signal but the OVL leds stay on." Is that what you suggest or do I understand you wrong?
About the DRAMs are they fragile and sensitive to static mishandling? Why I ask: I could swap one by one from Hsp Board 2 to Hsp Board 1, trying to let Hsp board 1 only to work and work from there. I know that's alot of swaps to be shure I don't swap a faulty dram by another faulty one. It' a bit guessing, I know. I can't imagine all 20 are faulty. I guess there is no easy way to test the DRAMs.

 

[radardoug]

Get the test cart available on Ebay, it will save you hours of guessing.

 

[NoisyIndividual]

I used to use a quick and dirty method to try and find the duff DRAMs by swapping them around, maybe in groups of 2 or 4, depending on how many there were. If there was a significant change in the audible artifacts, then you could then narrow it down by only swapping some instead to find those that changed to those that didn't.
I would try and swap LSB with MSB where possible to maximise any sonic differences.

I didn't really describe that very well, but hopefully you get the gist.
Worked for me anyway.

Not sure how practical this would be on a bank of 20 however ....

And they would all need to be socketed.

 

[paolomunarini]

"Self test of the motherboard (the jumpers) give a clean dry signal but the OVL leds stay on."

The self test of the motherboard is the same that you specified. Sorry for misunderstanding.

The OVL LEDs are not related to analog audio signal clipping but to digital overflow on the HSPs. So yes. The issue is most likely on the HSPs.

Regarding direct DRAM testing: a lot of folks on the internet made some NMOS DRAM tester firmwares for Arduino. I wanted to build one myself too when I was working on it, but I was concerned about the fact that the fastest clock speed of the fastest arduino on the market is nearly double the one needed for 100ns access time on the DRAMs. And this is even considering hardcore port manip on the arduino... most of the DRAM testers out there use digitalWrite() which is much much slower... I was discouraged by the fact that I wouldn't have been able to simulate correctly the access and some chips could have resulted pass intead of fail. Too much time and too much effort given the fact that I had 25 NOS spares.

Nevertheless, I bought the ZIF socket and the parts for it so I will try to build it anyways and I will try it out on the removed suspect failed chips.

Random mix and match until something clears out is surely unpractical as NoisyIndividual said...

If you can get hold of a Test Cart it would surely be the best option!

 

[NoisyIndividual]

OK, so if you even find ONE bad DRAM by swapping, you can then put it in any socket where it makes significant noise. (MSB perhaps)
Then you have a test socket where you can put each other DRAM in and test which are noisy and which aren't.

I did this successfully with typical delays/reverbs that had maybe 8 or so chips, it just may take longer with 20.

 

[NoisyIndividual]

If I had a bank of 20, I would first swap the first 4 with the last 4.
Difference would tell you something.
No difference would also tell you something.

 

[tommia]

Interesting. I will try that in the comming days. So I have to worry about the DRAMs. What will happen if I leave out the DRAMs of one channel?

 

[NoisyIndividual]

Nothing bad will happen with a DRAM not in its socket.

You could remove them all from one card and then use the other card as the testbed to again swap batches of 4 at a time to determine Difference or No Difference.

You'll get there in the end if your ears, logic, and keeping track of what came from where is good enough.

 

[Philbo King]

Interesting. I will try that in the comming days. So I have to worry about the DRAMs. What will happen if I leave out the DRAMs of one channel?

Generally you get all ones when reading from an unpopulated chip.

And it will act like a 'write-only' memory, too. You can write anything to to it, but never read it back.