Hi, my AdvHepax was having some contact problems and in the end stopped making contact with the Calc at all (It was not detected in Cat 2, but sometimes with some gentle pressure in the right direction it would be). I decided to take a look inside and was able to resolder one of the loose joints so now it works perfectly again (I know, die Depperten ham immer des Glueck...)
As a side effect, there are a few pictures from the open module and I thought it might be interesting to some. I had to reduce them as I don't have a place to host the full res ones. VM also put something over the chips so that one can not read the numbers. I tried to make them visible but with only some success.

Anyway, enjoy.

Cheers

Peter

Top side of the PCB 

Top Side of the PCB 

Magnified 1st chip of the bottom side 

Very strange. I also have a defective Hepax module... Have a look at this picture:

Matthias

Thanks for posting the pictures.

The Fujitsu MB671209U is an ASIC, which acts as the controller for the memory parts.

Microchip 27C256-25/L is a 32K*8 OTP EPROM with 250 ns access time. That's where the HEPAX ROM code lives.

The other chip must be the SRAM, and it is most likely a common 32K*8 CMOS part, which was made by quite a few different vendors. A typical example would be a Hitachi HM62256, though I don't necessarily think that's the actual part in your HEPAX. The "-12LL" suffix indicates 120 ns access time and extra-low standby current.

The EPROM and SRAM could easily be replaced if they failed, though one would have to have the code to program into the EPROM, and know how it is organized. If the ASIC failed, the HEPAX would be toast, as the only source for a replacement would be another HEPAX.

Hi,

some of the solder joints between the pcb and the connector strip look as if they could use some upgrade;-)

Maybe a short-circuit?

Also, it looks as if the PCB already experienced some heat in the resoldering process. Toasted?

Raymond

Hi Raymond

No, it's not due to a resoltering experiment but it's a sort of caoutchouc that protected the contacts.
Anyhow, the board of Peter looks better than mine... ;(

It realy looks like there has been some repair on Matthias' Hepax module. Maybe the connector has been replaced, and the result glued or protected with some silicone paste. I would say that there is a good possibility of reviving this module by replacing the connector (again) and checking all soldering joints. Also check the PCB for cracks.

Meindert

Quote:

---

Also check the PCB for cracks.

---

EPROMs don't hold their data indefinitely though. I don't think any of them are guaranteed more than ten years, although they usually go much longer.

Edited: 11 May 2008, 3:07 p.m.

Great info and detective work Eric, very impressive. From a couple of other pictures, the writing on the 2nd chip could be something like

M884 S-6 12LL

Japan 3/34 G55

Maybe it gives you some more clues...

Cheers

Peter

Edited: 11 May 2008, 6:46 p.m.

Most likely Fujitsu MB84256-12LL.

The SRAM is a 32kbyte part that can hold 16kwords. My hypothesis is that the same part was used in both the Standard (8k) and Advanced (16k) modules, probably one pin of the ASIC selects the RAM size. If I'm correct, it *may* be possible to turn a Standard module into an Extended one.

You can find an interesting page (in French) about the Hepax, with pictures of an open module here.

J-F

Hi there,

Thanks a lot for the pictures Peter.

I do agree with your appreciation Eric, MB84256 is the part. And certainly will also agree with Garth in that cracking tracks are very difficult to find in circuits not working under severe stress conditions. However, the observation regarding FR-4 PCB substrate may well be not that good... to my mind (and eyes) it's a ceramic substrate.

May Peter post a super-macro pic of one of its edges to determine... :-)

Regarding data retention, while early UV erasable EPROMS ranges in the "decades", later OTP's usually goes near the "century". To be more precise, this particular part (27C256) from Microchip is entilted to >200 years acording to manufacturer specs.

Cheers from the Canaries.

Diego.

Edited: 12 May 2008, 7:41 p.m.