13th Mar 2021, 7:33 pm | #1021 | |
Octode
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,440
|
Re: Non-working Commodore PET 3016
Quote:
Although I've never tried it, always liking to keep the IC intact, especially for blanket removal and socketing of RAM IC's where it's uncertain which are at faulty but a quicker approach to be able to just swap with good ones / test them in something else. I used to do this with a desolder sucker + manually freeing up each pin in the hole, but I've recently been using a heat / paint-stripper gun - although a higher-end variable-temperature / airflow Steinel etc type. If the IC's are particularly-expensive / hard to obtain types, then you may be able to re-use it if it wasn't actually faulty, by soldering it back on top of an IC socket. I've sometimes done this with IC's that the lower-leg has got snapped off / suffered pin-rot, by pushing some tinned copper wire into missing pin slots of an IC socket, and soldering against the upper-shoulder of the IC. |
|
13th Mar 2021, 7:57 pm | #1022 | ||
Dekatron
Join Date: May 2008
Location: Derby, UK.
Posts: 7,735
|
Re: Non-working Commodore PET 3016
Quote:
To make any character code you like, you will need to earth (via a suitable resistor) whichever data lines should be a 0 and leave any that should be a 1 unconnected. For instance, connecting pins 6, 15, 9 and 12 of UE9 to 0V should produce a screen full of capital O (= binary 00001111 = hex 0F); and disconnecting pin 12 should change them all to inverse. As long as touching the resistor on each pin caused the display to change, though, I think we have more or less already exonerated everything downstream of UF9. If you can't get the proper video RAM chips for positions UF7 and UF8, a 6264 could be bodged in as a temporary substitute -- it's a bit wiring-intensive, and it's wasting 7/8 of the chip (which just goes to show how cheap memory of all flavours has become), but it would not take too long to draw up a PCB for a more permanent job. Quote:
__________________
If I have seen further than others, it is because I was standing on a pile of failed experiments. |
||
13th Mar 2021, 8:04 pm | #1023 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,577
|
Re: Non-working Commodore PET 3016
Colin and aj between them appear to have located some reasonably priced replacement ICs (post #1005) - yes, we have now gone past the 1000-post mark in this thread!
|
13th Mar 2021, 8:10 pm | #1024 |
Octode
Join Date: Mar 2019
Location: Barry, Vale of Glamorgan, Wales, UK.
Posts: 1,364
|
Re: Non-working Commodore PET 3016
In fact your post was exactly a real 1K - it needs to double its views to beat the two MK14 ones even though it has beaten them in replies...
|
13th Mar 2021, 8:21 pm | #1025 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,577
|
Re: Non-working Commodore PET 3016
Looking back, it makes for very heavy reading at the start because we spent so much time chasing the missing CLK1 signal which was never, in fact, missing. That was all down to that quirky frequency meter feature on one of Colin's meters. We've made much better progress since Colin acquired his scope - I doubt we could have made it this far without it, or without borrowing one at least.
|
13th Mar 2021, 8:23 pm | #1026 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,817
|
Re: Non-working Commodore PET 3016
OK - here's where we stand on the orders. The ebay RAM chips are on their way with an indeterminate deliver time - it's either in the next 3 days or by 23rd March; somewhat difficult to decipher.
I have ordered 2x18 pin turned sockets from Cricklewood today - I expect they should be with me Tuesday. So unless there's anything more I can test, I thin kwe stand down and have some well earned beer/wine/soft drink of your choice and I'll let you know when the deliveries have come. Thanks all so far. I'm determined to get this PET running. Colin. |
13th Mar 2021, 8:29 pm | #1027 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,817
|
Re: Non-working Commodore PET 3016
To be fair, I was (am) a novice at all of this and made some simple time-consuming mistakes at the beginning too, but I'm learning a great deal now thanks to everyone.
If anyone wanted to know, this is the USB scope I have been using for all of the screenprints/testing - £33 when I bought it. Currently unavailable from Amazon, but I'm guessing that it should be available elsewhere. Lots of components/electronic tools seem to have gone unavailable or on long lead times recently. Good job the PET's so old... https://www.amazon.co.uk/gp/product/B088JX46VQ Colin. Quote:
|
|
13th Mar 2021, 9:10 pm | #1028 | |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Non-working Commodore PET 3016
Quote:
Alan |
|
13th Mar 2021, 9:39 pm | #1029 |
Octode
Join Date: Mar 2019
Location: Barry, Vale of Glamorgan, Wales, UK.
Posts: 1,364
|
Re: Non-working Commodore PET 3016
I have already had a few beers thanks Colin I would echo the praise for you dogged determination to get this running. In the end this is probably the best computer science education we can all get. Sirius deserves a medal mind for ignoring all our interjections and sticking to his logic.
I have followed with great interest and learnt a lot about a machine I only dimly remember from a rich friend who's dad had one BITD. This thread has reminded me at times of playing chess by mail... another pastime that is probably no more. |
13th Mar 2021, 10:08 pm | #1030 |
Tetrode
Join Date: Mar 2018
Location: Nottingham, Nottinghamshire, UK.
Posts: 88
|
Re: Non-working Commodore PET 3016
That oscilloscope appears to be a clone of a Hantek!
The 16 channel logic analyser could come in handy soon. |
13th Mar 2021, 10:34 pm | #1031 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,817
|
Re: Non-working Commodore PET 3016
The one I have from Amazon is a Hantek - it has some rough edges; the way it works on Windows 10 is a bit interesting and the device drivers cause blue-screens more often than I would like, but it does the job for a good price.
Logic analysers are a completely new field for me - I may/will need more help there too. Colin. |
13th Mar 2021, 10:38 pm | #1032 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,817
|
Re: Non-working Commodore PET 3016
|
13th Mar 2021, 10:41 pm | #1033 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,577
|
Re: Non-working Commodore PET 3016
I think Ian is suggesting that yours is not a 'real' Hantek, rather, a copy (although it comes with a copy of the Hantek software). The real thing tends to cost at least 2-3 times as much as yours did.
No matter though - it has more than paid for itself already, and will no doubt continue to do so. It is ideal for what you've been using it for. I actually want to make it clear that every single contribution to this thread has been appreciated by me, we haven't always taken my initially preferred route through the icebergs because there have been many alternative ideas which were more sensible than mine. Even when we have gone my way it has often only been when two or three of us concurred on a particular course of action. As we get closer to the end, continued input especially from those who actually do know the 6502 and the PET will be ever more vital, so please stick with it. We are getting slightly ahead though, as it isn't working yet and the evidence so far suggests that there is very likely to be more defective silicon to find. If (as hoped) replacing the video RAMs finally allows the screen RAM content to be rendered on-screen then we'll see if we can get Slothie's test code - another really important contribution - into something which Colin can just plug in. In the meantime, let's just park the thread until the next batch of parts arrives. Last edited by SiriusHardware; 13th Mar 2021 at 10:59 pm. |
13th Mar 2021, 11:29 pm | #1034 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,817
|
Re: Non-working Commodore PET 3016
Mine is a Hantek; box and scope are labelled as such - Amazon seem to have changed their description.
Colin. Quote:
|
|
14th Mar 2021, 12:08 pm | #1035 | |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Non-working Commodore PET 3016
Quote:
By chance I happened across the attached Commodore document tucked away in a quiet corner of Bo Zimmerman's site. Others may already have seen it. It's interesting from two points of view: 1. It describes how the video RAMs operate in relation to the character generator and provides some troubleshooting tips if we find ouselves in a situation whereby the new chips don't work as hoped. 2. It gives a good overview of the different dynamic system RAM chips fitted to the PET 3016 plus more fault finding tips which could be helpful in the quite likely event of Colin encountering more faulty silicon. One thing worth noting is that Colin's PET is fitted with sixteen TI 4108 dynamic RAM chips (see post #8) whereas the circuit diagram we've been working from has sixteen 4116 chips ie, as in the PET 3032. This is significant because I can't find anywhere currently selling legacy 4108 chips. On the other hand NOS 4116s are still available, possibly because some versions of the Spectrum used them. In fact there's one vendor offering eight 4116s for £10 plus £1.60 p&p. In theory and with changes to jumper settings it would appear that eight 4116s could be used as direct replacements for the sixteen 4108s in Colin's machine. Food for thought. Alan |
|
14th Mar 2021, 12:21 pm | #1036 | |
Octode
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,440
|
Re: Non-working Commodore PET 3016
Quote:
Open-source Software & Hardware design info for testing 2114's on an Arduino Mega is here: http://theoddys.com/acorn/Developmen...14_tester.html - based on the original Arduino Uno design, here: https://github.com/Danjovic/2114-Tester So you could build either of these / maybe just use flying lead to a socket on veroboard / breadboard.. BTW, I bought an Uno-R3 (smaller USB-B connector, CH340 USB-UART IC and SMD AT-Mega328 etc.) for only £2.80 delivered via Aliexpress last year. Although that was before it seems you now have to pay VAT on everything from outside UK with no min. threshold. It may be possible to make a kind of loopback plug & mod. code to test the assembled tester. But it's unlike to pass IC's at least, if there is a wiring-fault on all RAM IC's. It should also be possible to get it to put some more helpful diagnostic messages out on serial-USB interface to PC, rather than just a simple Go / No go LED indication |
|
14th Mar 2021, 12:29 pm | #1037 |
Nonode
Join Date: May 2004
Location: Halifax, West Yorkshire, UK.
Posts: 2,587
|
Re: Non-working Commodore PET 3016
Really don't see the point in spending time and effort testing the old video RAM chips when NOS ones are available at four for six pounds. Just the way I happen to look at life but each to their own.
Alan |
14th Mar 2021, 1:29 pm | #1038 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,577
|
Re: Non-working Commodore PET 3016
I'm surprised aj, every technician I know always has an inherent craving to want to know not only what went wrong, but how / in what way it went wrong, so I will very often check / test parts which I already know, from substitution, are dud. UF9, the video latch, would be another interesting case to look at. I agree, though, that for the price of the video RAM we can probably dispense with my usual morbid curiosity.
In this particular case I am also very conscious that we are spending someone else's money and time - Colin's in this case, every time we make a decision to replace something, so I am always inclined to try to think of ways to test suspect parts using equipment he already has or substitute them in some way before replacing them - in part because I prefer to know, through diagnostic or other means, that a part really is dud before we change it. Going to the other extreme, we could just ask Colin to price up every chip on the board and completely re-chip it and it would very possibly work straight away, but we would never know what the faults had been - which would be a source of annoyance to me personally, but might not bother others at all. When it comes to the system RAM which is not in sockets we could adopt the 'test in situ' approach using Slothie's test routines - either that or hack all the original ICs out and replace them either with 'legacy' substitutes such as the ones you have indicated or with a modern RAM replacement board similar to the one you have in your older PET. I'd really prefer to keep it original, or as original as it can be, if possible. Thanks for the pointer to the document regarding operation of the video circuitry, which I will peruse while I am downloading my lunch. I suggest you also study it carefully yourself in case I have a nervous breakdown before we get to the end of this project... RRaaaaack! (sound of handbrake being pulled on again...) |
14th Mar 2021, 1:56 pm | #1039 | |
Octode
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,440
|
Re: Non-working Commodore PET 3016
Quote:
I usually always re-use IC's which I've desoldered, that I wasn't definitely sure were faulty but I'd just removed to make fault finding easier (Like DRAM's on Spectrum's where if they haven't found they may well do at some point, so I tended to just blanket socket these - especially the very vulnerable 3-rail 4116's), as I don't like wasting good components. But I have also socketed virtually all IC's on smaller microcomputers, to also make it into a bit of a useful IC-testing board. I rarely use turned-pin sockets, unless I need to solder to top-side tracks to repair damaged thru-plating etc. as I like sockets to match those originally used / I haven't wanted to spend a lot in total on many of these (Until old computers acquired retro status / prices, it often wasn't worth spending too much on repairing them, with so many others being thrown out). So I tended to only use them for larger IC's on something I wanted to be very reliable. However, even using non turned-pin socket, it was usually a good idea to clean-up the pins of removed IC's - either for pins to actually fit into small holes in the top on types with 'Y' style contacts inserted in the underside, or to not catch and damage the socket upon future removal on the type of sockets with a diagonal sprung part against a vertical metal edge. I first try to remove all the solder, using good quality / small size desolder braid. In the absence of this, I have sometimes using a desolder pump, but have to be careful for the sudden recoil action not to bend pins - And really need to hold the IC in a small vice. If the IC was pulled-out with just hot-air heat gun (and fresh flux applied) on the rear, then there is usually a lot more solder on the pins than if they were removed with a desolder-pump etc. that has already removed most (although I use to apply extra fresh solder, as didn't have flux, to aid solder flow and removal with desolder pumps) With virtually all solder off the pins, I then ensure they are all dead-straight, using long-nose pliers across all the pins - On some boards, the manufacturer has bent some to hold in place, and if pulled out with hot-air, then you don't tend to straighten them first (that's usually a good idea when desoldering individually). I then use a flat micro fine-pitch 'Needle' file (or edge of stiff coarse emery-cloth) across both side's outside and inside edges, to remove last traces of solder / ensure there's no spikes to catch in sockets. As with brand-new IC's, before inserting into sockets, it's also often useful to place each side of pins against a solid flat surface and bend them all slightly-inwards, to ensure they are perpendicular, if they still slightly-splayed apart (as they tend to be when supplied originally and IC insertion equipment compresses during insertion into PCB resulting in a bit of sprung retention). This ensures they just push straight into socket, with less chance of a pin not going in and getting bent. |
|
14th Mar 2021, 2:24 pm | #1040 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,577
|
Re: Non-working Commodore PET 3016
(Thanks for the link to the adapted-for-Mega version of Danjovic's 2114 tester code - I already had that about 80% done myself, but now I don't have to).
Next up may be an Arduino or Raspi based 28C16 EEPROM programmer so that we have the option of posting bits of test code for Colin to try running. |