26th Feb 2021, 5:28 pm | #681 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
At 20ms those solid green areas on the data trace are where the sample rate and horizontal sweep are not fast enough to separate the different logic states of the data line, though it does look better for the chip select but we need to see what the data line is doing when chip select is active.
Could you show the options available for the trigger settings on your scope? On my scope, a different type, the default is to show the trigger point at the centre of the screen. It looks like yours shows trigger point at the extreme left, so when you trigger on rising edge the screen shows only the data line after the chip enable is disabled. Trigger on falling edge might be better, or move the display of the trigger point to the right, depending on your scope options. There is also my misunderstanding of the chip select behaviour on the PET. I expected a separate chip select pulse for each memory access, but thats not the case. There must be a separate output enable control on the prom that will pulse for each memory access. In the data on the 20us sweep you can start to see the logic level at different times. It shows clear high and low logic levels, but also quite a few times appears stuck at an intermediate level. This might just be times when the data line is floating, but also possible two separate devices trying to drive the data in different directions, one high and one low, which would be a problem. If you could check the data lines again after trying to change the trigger settings, we can look to see if those intermediate levels on the data are only when one prom chip select is low, or when any one prom chip select is low. |
26th Feb 2021, 6:57 pm | #682 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
The trigger options are as follows:
Sweep: Auto/Normal/Single Slope: Rise/Fall Colin. There is other open-source software (Openhantek) that I could look into if this doesn't give us what we need. |
26th Feb 2021, 7:00 pm | #683 | |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Quote:
The relevant diagram section is:- http://www.zimmers.net/anonftp/pub/c...N/320349-4.gif Or in case you are having trouble viewing that, here's a screen grab. |
|
26th Feb 2021, 7:12 pm | #684 | |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Quote:
The _CS signal (pin 20 on each PROM is an 'active low' signal which means that it is at 5V when it isn't doing whatever it does, and at 0V (low) when it is doing whatever it does, in this case... selecting a chip. A chip select pulse starts with a falling edge (going from 5V down to 0V), so it might be an idea in this case to set the trigger slope to falling edge, and that way the trace will set off on its journey across the screen at the beginning of a _CE pulse. You might also be able to manually move the trace left and right using a 'horizontal position' control if there is one. (We can't see all of the controls you can see, only the ones you happen to include in your screen shots). |
|
26th Feb 2021, 7:40 pm | #685 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
There's c.40 options for Time/Div ranging from 1ns to 5000s - so plenty to choose from.
In terms of horizontal position, when I pause the scope (which I do to take a screen shot), I can scroll the trace left for a limited time period to see some slightly different wave-forms if they exist - is that what you mean? Colin. |
26th Feb 2021, 7:52 pm | #686 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Yes, the ability to 'pan' left and right can be useful.
I'm going to let Mark go with this for a while so you don't get conflicting instructions sending you in two directions at once. In the meantime a little exercise for you: Taking your screen shot in #680, Can you say, roughly... How many milliseconds there are between the falling edge of the first _CE pulse and the falling edge of the second _CE pulse? How long (In milliseconds) is one _CE pulse? Assuming the low edges of the _CE pulses are at 0V, what voltage is the _CE line at when it is NOT low? (With the scope settings as they were, each horizontal division is 50mS: Each vertical division is 2V). Last edited by SiriusHardware; 26th Feb 2021 at 7:57 pm. |
26th Feb 2021, 7:54 pm | #687 | |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
Quote:
Try changing sweep to Normal. The trace should then not display unless it is detecting the trigger. For slope select Fall. Try one more time with these settings at 20 us per division. |
|
26th Feb 2021, 8:06 pm | #688 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
Stay with UD9-20 yellow and UE9-2 green for now.
|
26th Feb 2021, 8:33 pm | #689 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
If you still see the halfway signals on UE9-2, try connecting a 4.7 k resistor as pullup to nearby 5v. This will hopefully show if that halfway signal is floating or being driven in opposite directions by two different outputs.
|
26th Feb 2021, 8:58 pm | #690 | ||
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
Here's what I get:
https://drive.google.com/file/d/1Gog...ew?usp=sharing Colin. Quote:
|
||
26th Feb 2021, 9:04 pm | #691 | |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
So I reckon each 'tick' on the horizontal line in the middle of the display is whatever the time/div setting is, so with that set to 50ms, I can see c. 13.5 ticks = 675ms.
I reckon each pulse is 12.5 * 50 = 575ms If each dotted horizontal line is equal to the vertical voltage setting, then that's 2 x 2.0V = 4V Have I got that right? Colin. Quote:
|
|
26th Feb 2021, 9:06 pm | #692 |
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
Do you mean connect a 4.7k resistor to pin20 of UD9 to a nearby 5V (eg pin 24 of UD9)?
Colin. |
26th Feb 2021, 10:05 pm | #693 | |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Quote:
The vertical lines of dots mark the horizontal divisions. With that in mind, what is your revised figure for the time between the falling edges of the first and second _CS pulses, and approximately how long is one _CS pulse? Remember that the 'active' part of the _CS pulse is the bit where it is low, not where it is high. (again with reference to your post #680). |
|
26th Feb 2021, 10:17 pm | #694 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Mark, are you thinking that PROM _CS is held low for the entire time that the address is within the address range of the PROM? It does look that way - 4 inputs from BA12 to BA15 to the 74154 4-line-to-16 line decoder UD2. No other signal involved in the generation of the chip select. If so this is somewhat reminiscent of the situation with the MK14, you could intentionally damage something by deliberately writing to the PROM address area.
Last edited by SiriusHardware; 26th Feb 2021 at 10:22 pm. |
26th Feb 2021, 10:26 pm | #695 | |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
Quote:
Looking at the trace in #680, there is no need to try the pull up resistor yet. It looks like UE9-2 is only just reaching 2v, which might be indicating that the output of UD9 is not as strong as it should be, or that something else is loading that signal more than it should. Instead of UE9-2 use the pin for the same data line on UD9. This is just to make sure the low voltage on UE9-2 is not a poor connection in the socket of UD9. After you check that, using the same scope settings. Yellow on pin 20 of each of the proms, then green to each of the eight data lines. So what we are looking for is how each data line looks for each of the four proms. Its going to take some time, so if might want to check one prom at a time and let us review while you work on the next one. |
|
26th Feb 2021, 10:34 pm | #696 | ||
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
OK - so if each 'large' horizontal line is 50ms, then I reckon about 1.35 x 50ms = 67.5ms between falling edges.
And the active CS pulse is about 0.1 * 50ms = 5ms. Why is active low? My logic says active is high.... Colin. Quote:
|
||
26th Feb 2021, 10:37 pm | #697 | ||
Octode
Join Date: May 2012
Location: Perth, Scotland
Posts: 1,821
|
Re: Non-working Commodore PET 3016
When you say "Instead of UE9-2 use the pin for the same data line on UD9." does that mean pin 9 on UD9?
Colin. Quote:
|
||
26th Feb 2021, 10:54 pm | #698 | |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
Quote:
I have heard it was possible to damage PETs by poking to the wrong memory addresses, maybe the PROM decoding was responsible for that. It would also mean that PETs could be damaged if the software crashes, seems like a very poor design, especially with all those unused chip select pins on the proms. |
|
26th Feb 2021, 10:57 pm | #699 |
Dekatron
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,586
|
Re: Non-working Commodore PET 3016
Yes, I think if I had one working I would take those active-high CS3 pins of the PROMs to the R/_W signal line so at least they would only be enabled during a read operation.
|
26th Feb 2021, 11:08 pm | #700 |
Octode
Join Date: Mar 2020
Location: Kitchener, Ontario, Canada
Posts: 1,298
|
Re: Non-working Commodore PET 3016
|