It’s not often I start a post with a picture, but this is exactly what I’ll do. This is the first of a series of posts, which could have only come about due to good fortune, and the generosity of an OCAU member who has so kind as to alert me to his father’s house move and thus clean-out of old gear.
One of the things he cleaned out was the above – a Netcomm NC-35C modem. The copyright dates this to 1985, and the V2123 naming suggests a V.21/V.23 capable modem – i.e. 300bps to 1200bps. It was a smart-modem for sure, interfacing directly to the telephone line with status LED provisions in the front. The black box near the top – that’s the line relay, the big yellow thing is the hybrid transformer for the phone line. If the printing is to be believed, that part was built in week 21 of 1974! Next to that and a little below is the timing crystal, and there’s also a speaker on the board as well.
Near where the DB-25 connector should be is two silver topped chips – these are the RS-232 line-drivers. There’s a few assorted discrete logic gates, an EPROM with the firmware on it (labelled V.0.90, suggesting this was actually an engineering pre-release), and next to that, the AMD World-Chip FSK modem itself. A set of DIP switches set certain options for interfacing between different computers with different requirements.
Unfortunately, the power socket wasn’t populated, instead bare wires were provided. Four of them. By tracing the circuits, one was Vcc for the logic chips (5v), one was Ground (0v). The other two provided the high and low voltage for the RS-232 line drivers – i.e. +9v and -9v (or +12v and -12v if strictly adhering to standards). With four connections, power sequencing could be an issue, otherwise you could have voltages flowing to and from different places in ways which the chips were not designed to do.
The data connection was adapted to a DIN-like socket, unfortunately, nothing that I knew of. I took care of that by desoldering the wires, and soldering in wires to a DB-9 cable so I could attach it to a modern computer.
The bottom of the PCB shows a simplicity not found in modern circuitry. Just two layers for the PCB, all through hole components. The foil traces were found to have a bubbled appearance, as if delaminating from the substrate – this seems to be a common trait of most PCBs from the 80’s.
Unfortunately, on applying power, things got bad. The AMD modem and the PROM got really warm – a sign that it may be either static damaged or damaged by polarity reversal at some stage. Maybe I had cocked up a connection or the power sequencing. In short, the modem was dead.
I had high hopes in resurrecting the modem, to make it dial a few more calls. You can always get a modern modem to talk in older modes using the AT+MS command (Rockwell Chipset, Modulation Select), or though tweaking some S-registers, but it’s not quite the same feeling as getting the “real deal”.
But it wasn’t all a loss. In its death, we get to see something pretty that we don’t see nowadays – an EPROM chip itself. Erasable Programmable ROM chips often came in ceramic DIP packages which contain a quartz glass window over the die, which, when exposed to ultraviolet light for about 30 minutes, erases the data stored inside so that it can be “reburned” (i.e. reprogrammed). Depending on the storage capacity and age of the chip, the dies are of varying sizes, but they all are pretty. Very pretty.
You can even see the Motorola logo in the corner of the die at the top right of the image. (Image taken with an LED torch for illumination and a Canon SX110is point-and-shoot.)
As a bonus, here’s a much later STMicroelectronics EPROM. Notice that the die area is much smaller.