Ressurecting HP3478A

[grumpydoc]

Not quite in Mike's league for reverse engineering but...

A while ago I bought a pair of HP3478As - one working and one not.

The one that didn't had a scrambled display and broken switches - I suspect that the 3478 itself was OK and the problem was broken pins on the connector. However the display uses a slightly odd connector which just plugs into the driver PCB. The broken pins got stuck in the board and all attempts to retrieve them failed - in fact they fell into the display. Unfortunately in taking the display apart to try to retrieve them I think I put too much pressure on the LCD itself as it ended up looking like this:



Ooops 

I haven't, so far, come across a suitable donor unit or display so I thought about trying to replace the display with a PIC to decode the original signals and drive a graphical LCD, or perhaps a VFD as I happen to have a few 20x2 alphanumeric ones which are about the right size.

However the controllers are just bare dies epoxied to the board:



so no clues as to what they are.

It's probably not going to be too difficult to figure it out by having a look at the signals on a 'scope - especially as I have a working one to compare and the service manual but I thought I'd see if anyone could help guess what the controllers are.

It's probably similar to the D7225, similar vintage and the display takes the bias voltages from the DMM rather than generating them. The service manual shows one side of the connection as follows:



I was hoping that the signal names might jog someone's memory and match a standard controller of 1980's vintage. OS1 and OS2 are the oscillator connections (OS1 has a 470 pF cap to ground) and SYNC is presumably the start of serial data signal but I can't quite figure out what PWO/DPWO/IWA/IWB/ISB might be. These don't seem to map to the 7225 signals so I guess that's not the controller.

I've tried googling the signal names but that doesn't hit anything useful. Googling variants of "LCD controller/driver" only (so far) hits modern stuff, either dot matrix drivers or how to drive displays from microcontrollers.

So,.... do the signal names mean anything to anyone?

[JackOfVA]

Have you posted this question on the Agilent group forum? Can't recall seeing it there.

Also, if you have a GPIB adapter, you can exercise the defective 3478A and see if it functions. All else having failed, one could just use it via the GPIB interface with a bit of suitable code.

[grumpydoc]

Also, if you have a GPIB adapter,

I don't.

Or rather I do but its an old ISA bus one and I don't have a working PC with an ISA slot to put it in.

[lowimpedance]

 I cannot help with the devices used on the LCD, but I can suggest a possible alternative source besides another 3478a and that is the HP3488a switch/control unit. I found quite a few on ebay at good prices (probably much cheaper than the 3478a etc).
 I recently scrapped one and noticed the LCD looked familiar and indeed it was the same part number used in the 3478a.

[c4757p]

Hmm... I may have to throw the logic analyzer at my 3468A (same LCD) when I have time. I don't really like how hard it is to source replacements, especially after all the fun I've had with bad LCDs in Fluke meters... I don't suppose it will be all that complicated.

[BravoV]

Hmm... I may have to throw the logic analyzer at my 3468A (same LCD) when I have time. I don't really like how hard it is to source replacements, especially after all the fun I've had with bad LCDs in Fluke meters... I don't suppose it will be all that complicated.

Hey c4757p, share it please once you managed to decode it sucessfully, it would be nice to be able to replace them with LED based segment display, really hate those dim non back-lighted LCD especially at not very bright ambient light. 

[grumpydoc]

a possible alternative source besides another 3478a and that is the HP3488a

OK, thanks - I don't recall seeing on of those on sale in the UK working or not. It is also used in the 3468A but I haven't seen a non-working one of those either.

Hmm... I may have to throw the logic analyzer at my 3468A (same LCD) when I have time. I don't really like how hard it is to source replacements, especially after all the fun I've had with bad LCDs in Fluke meters... I don't suppose it will be all that complicated.

Hey c4757p, share it please once you managed to decode it sucessfully, it would be nice to be able to replace them with LED based segment display, really hate those dim non back-lighted LCD especially at not very bright ambient light. 

I have had the offer of a replacement so we'll see how that goes but I agree it would be worth reverse engineering these as replacements are hard to find and the display itself isn't great anyway.

I've had a look around for a suitable display - the original is a 12 digit 16-segment ("Union Jack") display with a 12.5cm x 2cm viewable area in a 13x3cm module.



So far I haven't found any dot-matrix display with the right dimensions to fit.

I do, however, have some 20x2 alphanumeric VFD displays which have the right viewable area and which might fit with a small modification or two (for a start the connector will need changing from a right angle one that sticks out of the side to a straight one sticking out of the back).

[BravoV]

Hell, even the LED based replacement is smaller, I still want it.

Also if with smaller led, the spare unused area will be also suitable for individual leds for those custom 12 indicators.

[free_electron]

This thing is called a starburst display. The interface is serial. It uses a 6 bit (64 chars) code.

The special voltages or for the plane control.

You should be able to 'sniff' the protocol using a little logic sampler like a salae or usbee.

[c4757p]

I will share the protocol if I get it. Quite busy with school right now, though - not much time to go poking around with that sort of thing.

[grumpydoc]

This thing is called a starburst display.

Transatlantic bias, I think This side of that particular divide I've only heard them called 14 or 16-segment or "Union Jack"  displays.

The interface is serial.

Yes I expect so.

It uses a 6 bit (64 chars) code.

That depends on the controller I would have thought - unless you speak with particular knowledge of the one used in the HP 3478A.

The D7225G, for instance, can drive 14-segment displays but uses an 8 bit serial display.

Some older "controllers" are little more than a shift register and the plane drive circuitry, however, I've only seen those do up to a couple of 7 segment digits which is 16 bits of data to shift in on the serial interface. To do that with this display would be 204 bits (12x16 for the displays plus the 12 custom annunciators) which would be way too unwieldy.

The special voltages or for the plane control.

Yep - modern controllers tend to generate them internally.

You should be able to 'sniff' the protocol using a little logic sampler like a salae or usbee.

Indeed. I don't have one but do have a 4-channel DSO with reasonably deep memories which should do the trick if I have time.

I'm less sanguine about the meter working. I just powered it up and there were no nice relay clicking noises as I'm sure it did last time.

Looks like a longer project than I thought