• Please review our updated Terms and Rules here

CBM PET 3032 STRANGE BOOT

Now I look at the video again it looks like it is the 'letters' that are changing but the 'numbers' remain static. Is that true or not?

Dave
 
Interesting. Look at the top line, it should contain the static text "32K DRAM MEMORY TEST.". This text should occur directly after the VERT DRIVE signal.

Oscilloscope channel #1 measuring G10 pin 11. Trigger your oscilloscope on this signal.

Oscilloscope channel #2 measuring E11 pin 9.

Set your oscilloscope to CHOP mode and adjust the timebase to observe some detail. You should observe the video signal going up and down. Some of the ups and downs should be the static video signal of black and white pixels that should be there, but some of the ups and downs should come and go. This latter signal is the problem we are trying to identify. See if you can see it on the oscilloscope trace...

Dave
 
daver2 said:
Interesting. Look at the top line, it should contain the static text "32K DRAM MEMORY TEST.". This text should occur directly after the VERT DRIVE signal.

Oscilloscope channel #1 measuring G10 pin 11. Trigger your oscilloscope on this signal.

Oscilloscope channel #2 measuring E11 pin 9.

Set your oscilloscope to CHOP mode and adjust the timebase to observe some detail. You should observe the video signal going up and down. Some of the ups and downs should be the static video signal of black and white pixels that should be there, but some of the ups and downs should come and go. This latter signal is the problem we are trying to identify. See if you can see it on the oscilloscope trace...

Dave
Click to expand...
Must i scope when i observe the flickering chars?
 
daver2 said:
Interesting. Look at the top line, it should contain the static text "32K DRAM MEMORY TEST.". This text should occur directly after the VERT DRIVE signal.

Oscilloscope channel #1 measuring G10 pin 11. Trigger your oscilloscope on this signal.

Oscilloscope channel #2 measuring E11 pin 9.

Set your oscilloscope to CHOP mode and adjust the timebase to observe some detail. You should observe the video signal going up and down. Some of the ups and downs should be the static video signal of black and white pixels that should be there, but some of the ups and downs should come and go. This latter signal is the problem we are trying to identify. See if you can see it on the oscilloscope trace...

Dave
Click to expand...
I am desperate!! I can't see nothing :(
 
>>> Must i scope when i observe the flickering chars?

I would have thought that was obvious... If it isn't flickering you won't be able to observe it with the oscilloscope will you?!

>>> I am desperate!! I can't see nothing

Well, just looking at the oscilloscope doesn't really help does it? How do you know whether you have selected a suitable timebase or not? Try adjusting the timebase (probably faster) one click by one click at a time clockwise and see if anything appears from the 'mush'...

If you can not see anything untoward, perhaps the problem is logic after the point we are measuring? I think not at this point in time though...

Dave
 
This could be a tricky one for desperado's scope. For this I would use a scope (like the Hitachi V509) that has a delay timebase. Lock the sweep onto channel one of the scope,on the V drive pulse and look at the signal with channel 2 with the delay B timebase which can be adjusted on the vernier to see any number of scan lines after the the V sync and adjust that until the 32k DRAM MEMORY TEST data appeared for example.

The other way, with no delay timebase available, is to fill the screen with the same character (if the computer will run a short program), and use the two channel method with channel 1 sync'd to the V drive pulse, it should be easy to see the letters changing then using scope channel 2 probe to go back through the system. In other words like Daver2 suggested, but with more characters on the screen to make it easier for he scope to find and display them.
 
I think the timebase is much too slow in the video at either 2 or 5 ms/div to see the detail we are looking for.

Dave
 
The fact that the numbers are not affected (nor is the period symbol) and that letters are suggests this is related to b6 or b7 of the symbol as output from video RAM. Given it does flash the correct answer we know that the value in Video RAM is good.

So... Suspect the connections for the bit 6 from the RAM through the latch to the Character ROM.

I would slightly suspect the Socket Character ROM pin 22, the Character ROM or possibly the latch at F9.

What you could do it boot it into BASIC and then write a small program to flood the screen with say 0xFF characters (255 decimal) and see what you actually see.

(I think moving the B1 jumper from OFF to ON should work for you)
 
Desperado, your board is the same Dynamic PET as mine, on this board it is very easy to disable the DRAM memory reads.

In this article I photographically captured over 20 scope recordings of the circuitry that supports the 4116 DRAM's. If you get to fault finding them, those recordings might help you for comparison purposes.

I was kind of hoping somebody else would do more scope recordings of other parts of the Dynamic PET's circuitry too, so we could make a library of "normal waveforms" so as to help everyone with repairs and even web publish it, as a book or folder. At least we have a diagram of the 4 state machine's timing too.

At the end of the article there is some very simple firmware to put in a 2532JL ROM, to test out the DRAM's, which works if BASIC is running, if not it requires the hardware adapter described there. But all this doesn't do nearly as much machine testing as Nivag's great system. My system was just dedicated for the task of isolating defective DRAM chips and/or finding defects in the DRAM support circuitry.

 
Nivag Swerdna said:
The more I think about this the more I think.... dodgy drive from the latch F9... (or the Character ROM)...
Click to expand...

Nivag,

It may be worth you reading the group of posts starting at #284. This was an interesting test we did with the video RAM and CPU removed and 'forcing' specific characters to be displayed on the screen. See if the results align with (or disagree with) your thinking.

Dave
 
Why the sad face, just turn the timebase knob clockwise a few clicks and see if you see something. If not, keep turning it one click at a time and see if you can see anything at each click. It is not too difficult...

You can work out (in microseconds) how long a line takes to draw on the screen. You can even measure the frequency of the HDRIVE signal using your oscilloscope and determine the time from that. Each character line on the screen is comprised of a number of video lines. Multiply the time taken to draw one line by the number of video lines to draw each line of characters and use that as your approximate oscilloscope timebase setting. Then tweak accordingly based upon what you see....

Dave

Dave
 
daver2 said:
Why the sad face, just turn the timebase knob clockwise a few clicks and see if you see something. If not, keep turning it one click at a time and see if you can see anything at each click. It is not too difficult...

You can work out (in microseconds) how long a line takes to draw on the screen. You can even measure the frequency of the HDRIVE signal using your oscilloscope and determine the time from that. Each character line on the screen is comprised of a number of video lines. Multiply the time taken to draw one line by the number of video lines to draw each line of characters and use that as your approximate oscilloscope timebase setting. Then tweak accordingly based upon what you see....

Dave

Dave
Click to expand...
I am desperate, i can't see nothing :(

 
You must log in or register to reply here.
Back
Top