29th Mar 2021, 5:49 pm | #1261 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
|
Re: Non-working Commodore PET 3016
You could check the refresh address to the rams and the multiplexers driving the ram address.
Probably best to use the nop circuit for this. There should be a square wave on each of the following, with frequency divided by two on each successive pin. UE3-10 UE4-6 UE4-10 UE5-6 UE5-10 UE6-6 UE3-6 Then check activity on the following multiplexer outputs. UE3-7 UE3-9 UE4-7 UE4-9 UE5-7 UE5-9 UE6-7 Is you could let us know what that activity looks like then we can try checking the data outputs of the ram. |
29th Mar 2021, 7:23 pm | #1262 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
|
Re: Non-working Commodore PET 3016
After #1261 and still running NOPs.
For each of the ram chips J2-9 and I2-9. One scope probe on pin 15, second probe on pin 2 or 14. Set scope to trigger on falling edge of pin 15, trigger level 1.4v. Vertical at 2v per division, horizontal at 2 us per division. The output data should be random, but we should see pin 2/14 output driving at good high or low logic levels while pin 15 is low. This might indicate suspect ram chips, but probably best wait until ram test software can be tried before attempting any replacement. |
29th Mar 2021, 8:41 pm | #1263 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
First set of results from 1261:
Square waves present. UE3-10 - 500 khz UE4-6 - 250 khz UE4-10 - 125 khz UE5-6 - 62.5 khz UE5-10 - 31.25 khz UE6-6 - 15.63 khz UE3-6 - 7.813 khz Colin. Quote:
|
|
29th Mar 2021, 8:50 pm | #1264 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
Looking good, carry on.
|
29th Mar 2021, 8:52 pm | #1265 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
|
Re: Non-working Commodore PET 3016
#1263 looks correct, I expected it would as you had looked at the outputs of the timing chain earlier.
Next set should verify if the multiplexers for dram addressing looks ok. |
29th Mar 2021, 8:53 pm | #1266 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
Second part of the results zipped into one file.
Colin. Quote:
|
|
29th Mar 2021, 9:02 pm | #1267 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
I get consistent results on all RAM chips with this test.
I attach a link to a video showing what I get as it's difficult to capture in a screenshot. Pin 15 is Yellow/Channel1, Pin 14 is Green/Channel 2. https://drive.google.com/file/d/1zf1...ew?usp=sharing Colin. Quote:
|
|
29th Mar 2021, 9:10 pm | #1268 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
When the NOP test is in place the CPU sweeps the whole memory range from 0000 to FFFF, but only one quarter of that range 0000-3FFF is occupied by system RAM. So what you are seeing on that video is the system RAM being inactive for three quarters of the time and active for one quarter of the time.
The sweep through the system RAM area coincides with the time period during which the BANK SEL signal is high, so maybe we need captures with the rising edge of BANK SEL at the left hand side on the upper trace and each RAM data output in turn on the lower trace. Set the horizontal sweep time to get about 8-12 memory reads just after the rising edge of BANK SEL? Mark? Edit: The falling edge of A15 would also serve as a trigger but the high BANK SEL pulse nicely brackets the system memory area sweep, so whenever BANK SEL is high we know we are in the middle of a system memory sweep. |
29th Mar 2021, 9:19 pm | #1269 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
So if each chip gives those results (which they do), is there hope that the RAM chips may not be faulty?
Colin. Quote:
|
|
29th Mar 2021, 9:39 pm | #1270 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
Too early to say, unfortunately. Ideally we need captures of the RAM data output lines only when the RAM is being accessed because the 75% period of inactivity is a distraction. I'll wait and see if Mark has any specific follow up checks he wants done on the back of your results so far.
|
29th Mar 2021, 9:57 pm | #1271 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
I'm due to be doing a bit of chip programming some time in the next day or two, so while I have the programmer set up I will program a couple of test 'PROMs', one with Slothie's test code which he kindly posted in the thread a while back and one from another source - each of them may have features or show up problems that the other does not - assuming that the system is now at the point of being able to run test code.
We are reasonably sure the system is running some of the code in its own PROMs (that's how the screen gets cleared, when you have the 6502 directly fitted). The test code is designed so that it does not need to use RAM itself, and so it can run without working RAM but at the same time it can try to write to and read back from RAM to see if that does work. It does need working video RAM and video circuitry but as you know, you successfully fixed that part a while back. |
29th Mar 2021, 10:59 pm | #1272 | |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
|
Re: Non-working Commodore PET 3016
Quote:
If you could capture the data output on the other memory chips in the same way, maybe by switching the scope to hold or try the trigger set to normal instead of auto. That would then eliminate the possibility of failed output drivers on any of the other ram chips. I didn’t get chance to check the output of the multiplexers, as I’m not able to open zip files on the ipad, I’ll take a look at those later on the laptop. We still need to verify the write signal and that we read back the same data thats written, but both of these are going to be better done with the ram test software. |
|
29th Mar 2021, 11:06 pm | #1273 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
I'll try to program the ICs up tomorrow (Tuesday). if I can do that then hopefully I can put them + an adaptor in the post in time to get there on Thursday or if not then, Easter Saturday at least. (I don't imagine there will be any post on Good Friday).
|
30th Mar 2021, 11:59 am | #1274 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
That's awesome - thanks very much for your help again.
Colin. |
30th Mar 2021, 12:00 pm | #1275 | ||
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
The video is consistent for all RAM chips. I'll re-test with the settings below
Colin. Quote:
|
||
30th Mar 2021, 3:03 pm | #1276 | ||
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,762
|
Re: Non-working Commodore PET 3016
Random RAM chip chosen (UI5) as they all give the same waveforms.
Image attached. Colin. Quote:
|
||
30th Mar 2021, 9:05 pm | #1277 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
You obviously feel there aren't any spectacular differences between any one of those lines and the rest so we'll have to say that also seems inconclusive (but we did not know that until you looked, so it was worth looking).
The test 'PROMs' are now programmed. Both of the test code files I have are 2K in length so I have programmed them into 2 x 2716 EPROMs. These can't be dropped straight into a PET 2001N PROM socket without a slight tweak to the pin wiring, so I've made a simple adaptor from two stacked sockets. I've fitted both of the test EPROMs into the adaptor and read them as though they were 2532 EPROMs which we know are directly drop-in compatible with the PET's PROM sockets - the 2532 is a 4K device so, with A11 of the PROM socket disconnected by the adaptor, the content of these smaller (2K) devices just appears twice when read as though they are a 2532. Slothie's code is written to run in the UD9 (Kernel PROM) position - when placed there the 2K of code will appear at both F000-F7FF and F800-FFFF, the second being the 'true' image. The other test code, written by Daver2 over on VCFED, is written to run in the UD8 (edit PROM) position and that will appear at E000-E7FF and E800-EFFF, the latter being the 'true' image as far as I know. I'll try, circumstances allowing, to get these into the post in time to arrive on Saturday, if not earlier. My only concern is that depending on the type of IC sockets the PROMs are fitted in, the pins on the adaptor may not reach far enough into the socket holes. If so, that could probably be got around by plugging a conventional 24-pin IC socket into the mainboard PROM socket and then plugging the adaptor into that. |
30th Mar 2021, 10:20 pm | #1278 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
|
Re: Non-working Commodore PET 3016
Um, just noticed a potential problem with the second bit of test code which lives in the UD8 range - reading Slothie's test code source it seems the peripheral ICs are mapped into the upper half of the E000-EFFF block, so I can't let the EPROM activate if the address is in the range E800-EFFF.
I'm actually wondering if this code will work in a 2716 plugged straight into the UD8 socket. Then, Pin 21 (2716 VPP) = constantly held high, as needs to be in read mode Pin 20 (2716 OE, connected to UD8 CS signal Pin 18 (2716 CE, connected to A11, so output of IC is enabled only when A11 is low. (E000-E7FF). How does that sound? |
31st Mar 2021, 12:07 am | #1279 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,264
|
Re: Non-working Commodore PET 3016
That looks right to me.
Does that mean the 2716 is a drop in for the 6316 after all? It might cause slightly higher current drain, but with all the TTL in the PET it probably won’t even be noticed, and it is only a debug tool. |
31st Mar 2021, 2:22 am | #1280 |
Octode
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
|
Re: Non-working Commodore PET 3016
Yes, Using this instead may be a bit better:
Pin 18 (2716 CE, connected to UD8 CS signal Pin 20 (2716 OE, connected to A11, so output of IC is enabled only when A11 is low. (E000-E7FF). keeping current down a bit further - as otherwise 2716 is enabled whenever A10 is Low and only need to disable outputs when upper-half of it is being accessed, to avoid conflicts. Although knowing how warm some Commodore-MOS ROM's could run, and with most EPROM's often seem to run reasonable cool, then it may draw a bit less power whichever of these two methods are used |