Hi everyone!
Myself and @DigitalRampage have finally restored the DECTalk DTC01 to a working state!
Restoration
In a prior thread, I posted the capacitors for the power supply. After Adrian reworked all the capacitors, we tried again, but no luck, the machine still wouldn’t boot. Adrian found a video that suggested the problem could have been faulty ROMs, which I didn’t consider at the time, but seemed reasonable, given the ROMs were EPROMs manufactured in the 1980s.
What I was then able to do is read back all the ROMs and compare their MD5 hashes against known working copies from MAME. Sure enough, 3 ROMs were faulty. Two read back different MD5 hashes, and one read back a completely random checksum each time, which was interesting (write only memory? lol). I used a Minipro programmer with the Linux software to dump the ROMs.
I then produced the following table to match the recovered ROMs against known good copies use by MAME:
| Our label | Chip marking | MAME Hash Match? | MAME Chip Match? | Notes |
|---|---|---|---|---|
| DECTALK1 | 098E5 | Yes | Yes | Corrupted on first read, now seems okay? |
| DECTALK2 | 097E5 | Yes | Yes | |
| DECTALK3 | 096E5 | Yes | Yes | |
| DECTALK4 | 095E5 | Yes | Yes | |
| DECTALK5 | 122E5 | No | No | Reads back the same bad value each time |
| DECTALK6 | 121E5 | Yes | Yes | |
| DECTALK7 | 120E5 | Yes | Yes | |
| DECTALK8 | 119E5 | Yes | Yes | |
| DECTALKA | 106E5 | Yes | Yes | |
| DECTALKB | 105E5 | Yes | Yes | |
| DECTALKC | 104E5 | Yes | Yes | |
| DECTALKD | 103E5 | Yes | Yes | |
| DECTALKE | 126E5 | No | No | Reads back a different hash each time, totally busted |
| DECTALKF | 125E5 | Yes | Yes | |
| DECTALKG | 124E5 | No | No | Reads back the same bad value each time |
| DECTALKH | 123E5 | Yes | Yes |
Match this against the image below:
As you can see, 3 ROMs were completely busted. DECTALK1 is still a bit sketchy, but has been cooperating thus far.
Interestingly enough, porting the faulty ROMs into the MAME DECtalk DTC01 emulator actually reproduced the exact behaviour we were seeing IRL: a speaker pop, followed by 0xD2 on the LEDs, but it not booting. If you swapped any of one these bad copies we had out with good copies, but not all, the speaker pop would not occur and the LEDs would go into some sort of infinite loop. I think this goes to show how accurate the MAME emulator is.
Anyways, after identifying the faulty ROMs, we were able to borrow 3 working ROMs from Adrian’s personal DECtalk, which read back correctly.
We were able to power it up, and sure enough, it worked! The DECtalk announced itself loud and clear. However, we weren’t able to get it connected to the DEC serial terminal, due to both cabling problems and then later a serial baud rate difference. The user can program in a custom baud rate into the NVRAM of the DECtalk, which we weren’t able to figure out, apparently it wasn’t any of the standard baud rates, which was super weird but whatever.
However, all was not lost. Adrian had the excellent idea of actually calling up the DECtalk by connecting it to a landline. I was then able to call up the DECtalk, and it presented a debug and diagnostic menu! Using that, I could then press the ‘*’ key to reset to factory settings. Sure enough, then, it just worked! Many celebrations 
Assets
Recovered ROMs (incl. faulty ROMs. DECTALK_122E5, DECTALK_124E5 and DECTALK_126E5 are good copies of the faulty ones): https://files.mlyoung.cool/acms/dectalk/dectalk_recovered_roms.zip
Multiple reads of the bad ROMs (in case you want to analyse the errors): https://files.mlyoung.cool/acms/dectalk/dectalk_bad_roms_test.zip
Data spreadsheet: https://files.mlyoung.cool/acms/dectalk/dectalk_roms.ods
Future plans
The plan now is to connect the DECtalk to a 1800 phone number that will be active on Saturdays. You will be able to call the phone number, and the DECtalk will tell you the time, the weather in Croydon, maybe a fun fact, and sing you a song. This will be driven by a small Raspberry Pi with a USB<->RS-232 adapter. The code for this will live here: mlyoung101/acms_dectalk: Code that runs the DECTalk at the Australian Computer Museum Society - Codeberg.org
We also plan to replace the borrowed ROMs with a blank ROMs that have been sourced from Tasmania kindly by Adrian.
Conclusion
This was the first restoration I’ve worked on here at the ACMS, and it’s been super fun! Thank you so much again to Adrian @DigitalRampage, whom without this work absolutely wouldn’t have been possible; you’ve been such a big help along the way.
