LLOYD L9000 PROGRAMMER REPAIR

[bondarevy]

Hello, can you tell me where l can download the manual and software for the LLOYD M9000

[exxos]

Hello, can you tell me where l can download the manual and software for the LLOYD M9000

Afraid I don't know. Probably look around on the Internet / archive.org or something.

EDIT:

First result the manual on google.

https://www.google.com/search?q=LLOYD+M ... e&ie=UTF-8

[exxos]

So after what can only be described as the most exhausting couple of days of fault finding in living memory, I finally got my batch of chips programmed. I'm writing this partly as a repair log and partly as therapy. Make yourself a cup of tea. This is going to take a while.

It started simply enough — the L9000 was showing a CE fault. No programming voltages. Easy, right? Famous last words.

First, The Voltages

Checked all the rails. Found the 25v rail sitting at 27v — a preset had drifted somehow. Adjusted it back. Chips verified blank. Self test passing. Voltages all correct. Prodded every chip on the board. Pressed on every capacitor. Pushed down hard on every IC with the meter probe. Flexed the board in every direction I could manage. Bashed it around the bench. Nothing failing. Absolutely rock solid. Clearly fixed then.

Reassembly #1 → CE fault again.

Took it apart. Again.

Found RV2 — which should be sitting at 6v — reading zero. Sprayed IPA over the area and things started behaving erratically — voltages jumping around seemingly at random. Strong suspicion that there was some conductive contamination somewhere, possibly a blob of old flux or dirt that was partially conducting and upsetting things. The IPA appeared to wash it away or temporarily change its behaviour because after working it in and letting things settle, the voltage came back to the correct 6v. Whether that was actually the cause was never definitively established — much like everything else on this repair, it was more educated guess than proven diagnosis.

While poking around in calibration mode one of the OTHER presets decided to start smoking. At the correct voltage. Just because it felt like it. Why it did this was never established either. Turned it off sharpish and decided not to think about it too hard.

There are two op amps in the voltage regulator circuit. Compared voltages between the working rails and the non-working one — nothing significantly different that pointed to an obvious fault. Resoldered the op amp sockets and sprayed IPA over the op amps and sockets themselves on the basis that it might be a contact or contamination issue. That appeared to solve it — all three voltages correct, self test passing, chips verifying blank.

It is worth noting that all the "problem sockets" on this board had already been replaced during a previous repair session last year, so oxidisation wasn't the issue — more likely general contamination or the act of reseating things that made the difference. Or possibly something else entirely. The L9000 was not in a particularly forthcoming mood about explaining itself.

Then the usual pre-reassembly ritual — prodded every single component on the board, pushed down hard on every IC and op amp with the meter probe, pressed on every capacitor, flexed and twisted the board every way possible, bashed it repeatedly, wiggled every chip in every socket. Not a single failure. Completely solid. Right then.

Reassembly #2 → CE fault. Again.

Naturally.

Took it apart. Again.

It should be noted at this point that the voltage presets were probably not actually at fault at all. The more likely explanation is that the faulty programming voltage rail was somehow affecting the master power rail and causing other voltages to shift as a side effect — or possibly ground bounce, or some other interaction that was never fully traced or understood. The voltages appeared to drift and then stabilise seemingly at random, and the actual root cause of that behaviour was never definitively pinned down. It just came along for the ride while the real fault lurked elsewhere entirely.

Poked and prodded every component on the board. Checked every voltage. Flexed the board, bashed it, pressed on every IC, wiggled everything that could be wiggled. All fine. Everything perfect. Rock solid on the bench. As usual. Beginning to wonder if the L9000 simply didn't want to be in its case.

Reassembly #3 → CE fault. Still.

You couldn't make it up. And yet here we are.

Took it apart. Again.

Did some baseplate experiments this time — removing just the single baseplate screw seemed to cure it. Refitting the baseplate and it still worked. Puzzling. Pressed down firmly on the PCB in every location including directly over the screw hole. Bent the board around the mounting points. Pushed hard on every component near the mounting area. No problems whatsoever. Put the baseplate back on — still working. Prodded everything again just to be thoroughly sure. Still nothing failing.

Reassembly #4 → "Device Reversed" error. With absolutely no chips inserted anywhere.

I cannot stress this enough. There were no chips in any of the sockets. Not one. In any direction. The programmer was completely convinced otherwise and there was absolutely nothing I could do on the bench to make it misbehave. The board was apparently saving its special tricks exclusively for the moment the lid went back on.

Tried unplugging and replugging the gang adapter every possible way. No change. Removed the baseplate. No change. Put it back. No change. Wiggled everything. No change. Checked voltages — 25v had drifted again. Of course it had.

Took it apart. Again.

Now The RAM Joins The Party

Because why have one fault when you can have two.

With the voltages finally stable and everything apparently working, ran the self test — now getting intermittent user RAM fail. A whole new fault to deal with on top of everything else. Brilliant.

The L9000 has two types of RAM — a bank of 8 chips making up the user RAM buffer which holds your device data, and two smaller chips for the CPU. Unplugging one chip from the bank made the failure consistent rather than intermittent, confirming it was definitely the user RAM bank at fault. Not entirely surprising as I'd had problems with this bank before — previously replaced six of the eight chips, repaired corroded tracks, and replaced all the sockets. The remaining two chips are soldered directly to the board and I really didn't fancy desoldering them given the board's history of heavy corrosion damage and fragile tracks.

Spotted a loose piece of solder sitting between two of the RAM pins. Removed it carefully. Sprayed IPA over the RAM area and worked it in — more to shift any contamination and exercise the contacts than anything else given the sockets were already replaced last year. Then did what I always do before reassembly — prodded every component, pushed hard on every IC with the meter probe, pressed on every capacitor, flexed the board in every possible direction, bashed it repeatedly on the bench, wiggled every chip in its socket. Not a single failure. Solid as a rock.

Reassembly #5 → RAM fail. Immediately.

Took it apart. Again. Obviously.

More IPA, More Hope

Sprayed more IPA over the RAM area and worked it in more thoroughly this time. Powered up with chips partially removed — 2mb working. Added more chips — 4mb working. Tried 6mb — failed, but turns out 6mb isn't even a valid configuration so that's irrelevant. Full 8mb — working perfectly. Self test passing consistently.

Then the usual pre-reassembly ritual. Bashed the board around. Flexed it every which way. Prodded every single chip. Pushed down hard on every IC with the meter probe. Pressed on every capacitor. Wiggled everything in its socket. Poked at every repaired track. Ran the self test again. Still passing. Completely solid. Practically begging it to fail at this point. It refused.

Reassembly #6 → CE fault. The voltages again. Naturally.

At this point the screws were getting visibly worn and I was developing a Pavlovian twitch every time I picked up a screwdriver. The L9000 had clearly decided that functioning inside its own case was beneath it.

Took it apart. Again.

Checked voltages — drifted again for reasons that remained stubbornly unclear, quite possibly still a symptom of the underlying fault affecting everything else rather than any actual problem with the presets themselves. Adjusted back. Found a broken CAS line in the RAM bank while I was in there — repaired it. Everything working on the bench. Again. Ran the full pre-reassembly ritual — every chip prodded, every IC pushed down hard with the meter probe, every capacitor pressed, board flexed and bashed in every conceivable direction, every socket wiggled. Self test passing perfectly. Nothing failing. As always.

Reassembly #7 → CE fault. Still.

I may have said some words at this point that aren't suitable for a family forum.

Took it apart. Again.

Finding The Actual Root Cause — Eventually, And Why It Wasn't Simple

Decided enough was enough and properly traced the 6v programming voltage rail rather than just adjusting things and hoping. The circuit uses two op amps, a BC337 driver transistor, and a power transistor. Compared voltages across every pin against the working rails — nothing significantly different that pointed to an obvious smoking gun anywhere.

BC337 base and emitter sitting at identical voltages — no forward bias, completely off. Spent considerable time tracing what was driving what and measuring every pin in the chain. Eventually fed 10v via a 10k resistor directly to the power transistor base as an experiment — got 5.6v on the output rail immediately, proving the output circuit itself was perfectly fine.

Now, at this point I had already physically pushed down hard on the power transistor body with my meter probe during every single bench test session. Pressed it firmly. Pushed every pin down individually. Pushed down on the transistor body itself. Nothing. It just sat there working perfectly as it always did on the bench. So a bad connection on the transistor had already been mentally dismissed as a possible cause.

Which made what happened next all the more infuriating.

Lifted the power transistor — which bolts to the case as a heatsink — and found the base track had completely fractured away from the solder pad. Clean break. Utterly invisible until the transistor was physically lifted from the board. Pushing down on it from above, even quite firmly, did absolutely nothing because the break was on the underside of the pad and only opened when the transistor was pulled upward and away from the board by the tension of the bolt during assembly.

And there it was. The actual root cause of absolutely everything.

That transistor bolts directly to the case for heatsinking. On the bench with the case open the transistor sits relaxed, the cracked pad makes just enough contact to work, and no amount of probing, flexing, bashing, pressing, poking, or pushing can replicate what happens when that bolt pulls the transistor tight against the case during assembly. The precise mechanical stress of that one bolt on that one pad in that one specific direction is simply impossible to reproduce by hand. Every bench test passed perfectly every single time because the crack was closed. Every reassembly failed because the crack opened. The only way to find it was to physically lift the transistor — which only becomes obvious in hindsight, as these things always do.

Over the past year that pad had been slowly cracking under the cumulative stress of repeated assembly and disassembly cycles, getting progressively more marginal until finally giving up completely. The "works perfectly on the bench, fails every single time it goes in the box" mystery that had been driving me absolutely insane for hours — and had apparently been causing intermittent grief for the past year — was one hairline crack on one solder pad. That was it.

Soldered a bodge wire across the break with plenty of slack so it can never be stressed again. Voltage came straight up immediately.

Whether the broken programming voltage rail had also been contributing to the intermittent RAM failures via ground bounce or supply noise is pure speculation — it's possible, it would explain some of the weirdness, but it was never proven. There was also a broken CAS line in the RAM bank found and repaired separately, which was a real and definite fault. Beyond that, exactly how much of the RAM oddness was genuine RAM trouble versus collateral damage from the power rail will never be known for certain.

Reassembly #8 — It Actually Worked. Mostly.

Ran the first batch — 32 chips programmed successfully without a single error. Progress at last! Resisted the overwhelming urge to immediately start poking at things to see what would break.

Then during the next batch the programmer had one odd mid-operation reset, seemingly jumping randomly to a different part of its own program mid-cycle. Suspected the watchdog and mains cycle detector circuit which I'd previously had to patch just to get the machine to power up at all — probably still marginal. Restarted, carried on. Another 32 chips done successfully.

Last batch of 8 chips caused repeatable failures mid-programming. Erased the whole lot and tried again — same result every time, blank check fine beforehand but programming falling apart in the same place. Clearly something wrong with that particular batch of chips rather than the programmer. Binned them — not worth an hour of chip-by-chip testing for a handful of EPROMs.

Final 8 chips — programmed perfectly. Order complete. The L9000 was returned to its case where it sat looking entirely innocent, as if none of this had ever happened.

Summary

Number of full disassembly and reassembly cycles: well into double figures. Every single time it went back together something new decided to fail, and half the time taking it apart again mysteriously fixed whatever had just broken without any obvious reason why. Every single bench session involved thorough prodding of every chip, every capacitor, pushing down hard on every component with the meter probe, flexing and bashing the board in every direction, wiggling every contact — and every single time it passed with flying colours right up until the moment the lid went back on.

Definite faults found and fixed:

• Cracked track on programming voltage power transistor pad — the real villain all along, responsible for the "works on bench, fails in box" insanity throughout, and completely undetectable by any conventional bench test short of physically lifting the transistor
• Broken CAS line in the RAM bank

Things that happened but were never fully explained:

• Voltage presets drifting intermittently — most likely a symptom of the broken power rail affecting the overall supply rather than the presets themselves, but never definitively proven
• One preset smoking at the correct voltage for no apparent reason
• Various RAM failures of uncertain origin — possibly related to the power rail, possibly genuine, possibly both, possibly neither

The moral of the story: if your L9000 works perfectly on the bench but fails every time you box it up, physically lift every bolted component and inspect the pads underneath before you do anything else. No amount of bench testing, prodding, flexing, bashing, or probing will find a crack that only opens under the specific mechanical stress of assembly. Check every track near every mounting point. Mechanical stress and thermal cycling will find every marginal repair eventually, and they will wait until the most inconvenient possible moment to do so.

And keep a large can of IPA handy. It won't fix a cracked track but it'll keep you busy while you're hunting for the real fault.

Also, consider taking up a less stressful hobby. Like bomb disposal.

IMG_5096.JPG

IMG_5095.JPG

IMG_5094.JPG

IMG_5092.JPG

IMG_5091.JPG

IMG_5089.JPG

The last Puzzle.. Why was a single green gang LED lite up for no reason.. Well.. it was 2am.. people can figure that out themselves ;)

IMG_5097.JPG