Time Electronics 9814 Programmable Voltage Calibrator

I left school having just turned 17 after my 'A' levels in 1980, needing to fill a year before going to University, and ended up working at Time Electronics in Tonbridge.  

The company made electronic calibration lab instruments, and I worked there for a year before University, and then in most vacations subsequently until I started my first post-graduation job at DEC in 1984. More reminiscences about this in a future post perhaps.

The biggest project I was involved with was a new range of programmable instrumentation, with remote operation via IEEE-488/GPIB and local front panel controls. This range included voltage and current sources, a programmable resistance, multi-way switch, and screwdriver *(1).  Later on, a multi-function calibrator was produced, combining several functions in one unit.

Since I retired from actual paid employment a few years ago, I've been slowly collecting working versions of things I worked on during my career.  For example, using simh, I can run ALL-IN-1 email and DECnet-VAX Phase V WAN devices (see earlier posts).  It's been fun to revisit these things and keep my modest technical skills ticking over.

A while ago I was idly googling for Time Electronics and came across this article https://xdevs.com/fix/te9823/ by Illya Tsemenko, in which he debugs a mildly broken version of the later 9823 multi-function calibrator, from 1988.  He seems quite complimentary about various aspects of the design, and this page contains links to the manuals he acquired, as well as a firmware dump.

In the back of my filing cabinet at home, I've got a source listing of a version of the related 9822 code that I wrote, dating from 1983, so I thought it would be interesting to compare the two.   A bit of work with f9dasm from github, and I had a disassembled version to look at. It's very similar to my listing, but with a couple of less useful features removed, and an upgrade to higher resolution operation, using later revisions of the analogue boards, with a different DAC chip.

These instruments are so old now that they rarely seem to come up second hand, but I did find a listing for a 9814 Voltage Calibrator on Ebay. It was marked as ex-MoD and "spares or repair", seemed to be basically working, but was also too expensive. The guy wouldn't negotiate on price, so I left it. A few months later he had reduced the price to a level I was prepared to consider (especially given there weren't any other ones about), and so I bought it.

It seems to have been dropped on both corners, which wasn't mentioned in the advert !

Although the eBay seller had pictures of it working, that was 6 months earlier, so I thought it'd be sensible to check the power supply before actually powering the unit up. This is easy, as it just slides out of the back of the cage, and I was able to power it up in isolation and check with a meter that the digital and analogue supplies looked OK.

The seller mentioned that some of the switches were sticking, and I knew from experience that this was probably just them fouling on the front panel (these panels were precision-engineered back in the day by a bloke with a fly press). So I took the panel off, adjusted the mounting screws a bit, and now the switches are fine.

1980s UI technology

The big Intersil chip is the 7-segment LED driver, there are two octal flip-flops to drive the switch LEDs, a keyboard encoder to allow the switches to be read, and an address decoder for all those.

I also pulled out the CPU and IEEE boards to check they looked OK. 

CPU "Complex"

This must be a fairly early unit because the board doesn't have the optional 1K SRAM. So what we have is a very simple setup of a 6802 microprocessor, two 6821 PIAs, a TMS 2532 EPROM, a 12 bit timer that provides an NMI clock interrupt, an address decoder and a few gates worth of glue logic.

Without the external memory, the firmware only has access to 128 bytes of SRAM inside the 6802, so it's going to be quite limited in features.

One of the pullup resistors looks like it's been lifted at some point and then not put it back properly.

After all that, I straightened the bent bits, reassembled, and it does basically work. It must be a long way out of calibration though. At this point, which is more accurate, the once expensive calibration instrument or my cheap-ass multimeter ?

After playing with it for a few minutes, I found a couple of UI glitches in the way the switches and modes work that I couldn't imagine I'd have let go if I'd been responsible for it. I know that the listing I've got was a rewrite I did in Summer 1983, of code that was produced by a temporary employee, who'd been at the company while I was away during University term time.  I think the rewrite was started pretty much as soon as this other guy had left the building.

So I got curious about how different this earlier software was to what I'd later rewritten, although I knew that some of the original basic framework was mine, as it was reused in other instruments.

So, the only thing to do was dump the EPROM, disassemble it and have a look !

The cheapest way to do this was to use an Arduino, and write a simple program to dump the EPROM contents to the serial port.

https://github.com/trevorwarwick/arduinopromreader

I'd never played with an Arduino before, but it was really simple to do what I wanted.

With the dump completed, I spent a while with the disassembler adding some notes and comments to see how much of the code was recognisable in my later version. 

This version turns out to be quite poorly structured and messy, but I did manage to decode it to a reasonable extent, at least enough to see that my later version is definitely much better.  I obviously re-used a few subroutines almost verbatim as these were also recognisable in the original.

Disassembled 9814 code

I might now investigate connecting up the IEEE-488 interface and controlling it directly from a computer. 

Footnotes:

1) I don't think they sold (m)any of the programmable screwdrivers,  it was essentially a trimpot adusting screwdriver on the end of a flexible drive, that could be driven in either direction under program control, and must have been a very expensive solution for what it was.