Epson V700 scanner repair. A comedy of errors.

[bernard_L]

I bought my V700 maybe 15 years ago, second hand. I use it almost exclusively for scanning MF, and also 35mm when not after best quality. I noticed that one of the two Cold Cathode Fluorescent lamps (CCFL) used for reflective scanning (opaque documents) was not working.
Is it the lamp? An acquaintance gave me a CCFL tube from a dismantled scanner. I made the substitution. No joy. So...
First error. Should have swapped the connections between the two CCFL tubes, to find out if the fault lay in the tube or the electronics.
Is it the inverter? If it is not the lamp it must be the electronics (called an inverter) supplying the high voltage (1100V AC p-p for starting and 500V in operation). Fortunately there is this video showing how to dismantle the V700:
And a service manual on https://elektrotanya.com/ (recommended). Except that for some obscure reason Epson gave all schematics (including trivial ones with two push-buttons and three LEDs) except the inverter. Never mind, the inverter seems to follow a standard schematic. There are in fact two inverters, one for each CCFL. One gives no output. Measurements with digital voltmeter to find out if the faulty inverter receives 24VDC, and
Second error. Made a short with the probe between 24V and Gnd. General failure of the scanner. The DC fuse (1.3A, fast) is a surface mount component 0805 size (2mmx1.25mm). Not found at the hobbyist store. And pro stores have a minimum order and/or substantial shipping charges.
Third error. As I had to order a pedestrian glass envelope fuse with the proper rating (1.25A) I might as well order replacement transistors. Because clearly the inverter failed, and clearly the only active components are the pair of transistors. As already mentioned no schematic, and no type number for the original part, so I order a pair of BUT12AF transistors, with generous voltage and current specs 100V 8A.
Inverter "repair". After bypassing the blown SMD fuse with a classical one --luckily the holder just fits under the main board cover-- I replace the two transistors on the "failed" inverter. No joy.
Fourth error. Should have tested as-is before embarking into time-consuming substitutions.

End of story. The root cause was on the main board. For whatever reason, the +24V supplies for the two inverters are switched on/off independently on the main board following logical signals from the µProc: LMP and LMP2. And LMP never went on. I was able to do a wired-OR between two open collectors, allowing the LMP2 signal to control both +24V supplies. Restored the original transistors on the "faulty" inverter. Everything works!!

Bypassing the blown SMD fuse with a classical fuse

A wired-OR in the switching logic for the 24V supply of inverters

Finally, success.

 

[Kino]

Great job!

I couldn't have done it.

 

[Don_ih]

Such frustration that ultimately results in triumph is uplifting. Good job!

 

[koraks]

Not found at the hobbyist store.

They can be had from e.g. AliExpress, but in Europe, TME also sells them to private individuals. The fuse is probably a resettable polyfuse.

The mystery is how come LMP doesn't turn on while LMP2 does. No output on the GPIO on the uC that drives the LMP switch?

Nice troubleshooting process and I recognize the "errors". Either you do the 'right thing' and do a proper root cause analysis with measurements etc. You end up realizing you just should have replaced that leaky cap in the first place...so next time you strategically replace some usual suspects only to realize that it would have been more efficient overall to do the root cause thing...you never win!?

 

[bernard_L]

They can be had from e.g. AliExpress, but in Europe, TME also sells them to private individuals. The fuse is probably a resettable polyfuse.

The mystery is how come LMP doesn't turn on while LMP2 does. No output on the GPIO on the uC that drives the LMP switch?

Nice troubleshooting process and I recognize the "errors". Either you do the 'right thing' and do a proper root cause analysis with measurements etc. You end up realizing you just should have replaced that leaky cap in the first place...so next time you strategically replace some usual suspects only to realize that it would have been more efficient overall to do the root cause thing...you never win!?

The fuse is FCC20132ADTP. A "real" fuse.

https://www.mouser.fr/datasheet/2/210/kamaya_products_FCC__english_20170207_094544-1842064.pdf

On Aliexpress I found only resettable fuses; I have the impression that these exhibit a wide ratio between the sustainable current, and the current that shuts off in, say 0.5s.
But, thank you for mentioning TME; it will replace Farnell for my occasional needs. And they even have a "real" fuse in 0805 format:

https://www.tme.eu/fr/katalog/fusibles-smd-0805_112999/?params=77:1439582_courant-nominal:1-5a

There are, on the schematic, separate pins on the µP labeled LMP and LMP2. LMP just is not asserted during power-on, like LMP2 is. Strange things in Japanese design. There are also two CCFL tubes for transmission scanning, and their inverter(s) are fed from a common (logically switched) +24V line. I suspect that the two scanning modes have been assigned to different engineers. Who would like, anyway, to turn on just one of the tubes for reflective scanning? This just makes the paper wrinkles and creases apparent. And implementing a non-resetting fuse in SMD! The smallest SKU in the Epson maintenance manual if the fuse blows is the Main Board!!

My mistakes were not just unfortunate choices in either-or decisions. Just plain stupid. After I installed the new fuse, I should have re-started the diagnostic where I had left it: checking the availability of DC power. Instead I rushed to do the "clever" thing: changing the transistors.