|
Vintage Computers Any vintage computer systems, calculators, video games etc., but with an emphasis on 1980s and earlier equipment. |
|
Thread Tools |
9th Apr 2021, 8:26 pm | #1 |
Heptode
Join Date: Jul 2016
Location: Portland, Dorset, UK.
Posts: 874
|
HP Power supply failure.
Saw this on CuriousMarc's channel, a HP 9825T that he has decided to have a short on one of the power transistors in the PSU sending the unregulated output of the PSU down the 5V line, so instead of 5V you got 13V!, going to be interesting to see if he can get it working again as I would guess a lot of chips have been fried and I would guess some of them are unobtanium HP custom jobs.
https://www.youtube.com/watch?v=b-eN93L6yX8 |
9th Apr 2021, 9:11 pm | #2 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,902
|
Re: HP Powersuply failure
I've had a lot of work out of a 9825 back in the day. It was a very useful machine.
I recall that it was the product one Mr Wozniak worked on. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
10th Apr 2021, 4:47 am | #3 |
Dekatron
Join Date: Jun 2015
Location: Biggin Hill, London, UK.
Posts: 5,224
|
Re: HP Powersuply failure
It was strange... The HP98x0 series machines had crowbar circuits on the +5V rail. The HP9815 and HP9825 did not. A shorted transistor in the latter can ruin your day.
There were some interesting commands in some of the option ROMs for the HP9825. One let you store a string as a program line -- from within a program. Yes, self-modifying code. Ever seen an H9831? It's the same hardware as the HP9825 (minus the internal strip printer) with BASIC rather than HPL firmware. 'Fake' ones -- HP9825s with the BASIC ROM PCB are not too uncommon. Genuine ones with the correct keycaps and the metal blanking plate in place of the printer are. |
10th Apr 2021, 9:38 am | #4 |
Octode
Join Date: Mar 2019
Location: Barry, Vale of Glamorgan, Wales, UK.
Posts: 1,364
|
Re: HP Powersuply failure
Indeed I watched that video with increasing concern about the long term fate - this is one instance where some modern short protection is worth adding to a vintage machine before that happens. I hope he can get it going - as you say if nothing else it will be an interesting journey - maybe as long as the PET thread that has been running on here.
|
10th Apr 2021, 10:06 am | #5 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: HP Powersuply failure
This is the Achilles heel of all linear regulators, the likelihood that if the main regulating device fails short-circuit from input to output it will pass the unregulated input voltage straight through to the output. The old CB 13.8V 'power packs' were especially prone to this, with the 2N3055 or whatever failing short-circuit C-E and passing on ~24V to the unfortunate recipient. Luckily, it normally only took out the audio PA IC and possibly the meter illumination bulb, if the IC did not go short-circuit quickly enough to save it.
I keep meaning to put a heavy 5.6V zener on the 5V side of my original MK14's regulator (which is mounted offboard) - no doubt I will 'keep meaning' to do it until the day the reg finally fails with predictably catastrophic results. At work we make an expensive control PCB which has regulated 3V3, 5V and 24V supplies all from 24V in. The regulators are switch-mode but they are the 'simple' type in which the switching device is between the unregulated input and regulated output, with a series coil acting as a 'flywheel' to average out the on/off/on/off voltage coming through the switcher. These are also subject to the same possibility of the switching device failing input - through - to - output so the design includes big 'just above rail voltage' zeners preceded by semiconductor fuses. Just recently I had a 6 year old PCB returned on which the 5V6 zener and its associated fuse had been reduced to slow-roasted ash. A quick inspection revealed the reason - the surface-mounted switching IC for the regulator had never been fitted, so the switcher just stayed fully on the whole time. For the whole of its working life so far this PCB had been working with its 5.0V 'supply' being provided by a 5.6V zener, the semi fuse acting as the series limiting resistor. It obviously worked well enough to get through original testing and for a surprisingly long time after that. I suppose the moral of this tale is that if you do fit a crowbar or overvoltage protection, make sure it has a fuse which can actually be blown by the crowbar. |
10th Apr 2021, 12:02 pm | #6 |
Heptode
Join Date: Jul 2016
Location: Portland, Dorset, UK.
Posts: 874
|
Re: HP Power supply failure.
If it is fixable he and his friends certinaly have the knowledge to do so, they got an Apollo flight computer fixed so this must be somewhat a walk in the park.
Will certinaly make for a good series of videos and I would guess, if necessary, some of the old custom chippery could be replaced with FPGA's, his friend Ken can decap IC's and reverse engineer them. |
10th Apr 2021, 6:18 pm | #7 | |
Octode
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,443
|
Re: HP Powersuply failure
Quote:
I did also often have to change the CB's audio PA IC, but usually due to the user blowing it up by connecting the supply backwards - sometimes on a car supply and often with no inline powerlead, after this had been chopped-out. The internal reverse-polarity 'suicide' 1A-rated shunt-diode would usually go s/c, but with no fuse to limit current, the copper track to the diode would usually burn-out instead. I've seen similar things with small SMD TVS diodes on products, that couldn't pass a sustained current for long enough to blow a fuse, and had sometimes desolder themselves? One thing I observed on their video is that they did a cold-check resistance-measurement on the 5V supply, and seemed to be using ohms law with this & supply voltage to check that supply current would be OK. However, with active IC's non-resistive load, that seems a bit flawed if DMM's resistance-range o/c voltage is < 0.4V so not enough to even bias diodes etc. |
|