• Please review our updated Terms and Rules here

Neglected PET needing some love

This is the 'noddy' RAM test that is run first prior to using Page 0 and Page 1 (stack) RAM during the full memory test.

Give me a few minutes to send some work emails off and I will have a look.

In the meantime you can look at the 'g' and 'b' letters with the knowledge that I am storing $00 in address $0000, $01 in address $0001, ... , $FF in address $00FF. Repeat for addresses $0100 through $01FF. You are looking for 'stuck' data patterns or 'stuck' address patterns.

The first 256 bytes on the screen consist of a g or b status (good or bad) for the first 256 bytes of memory (page 0) and the next 256 bytes indicate the character corresponding to the data that was read. The character codes (when converted back from PETSCII to hex) would give you the data pattern that was read. The next 2 * 256 bytes is duplicated for addresses $0100 to $01FF (page 1).

See if you can make some sense of it...

Dave
 
Last edited:
Okay so what this is showing is all within the first ram chip, ua4 in this case?

If so then we can assume that at least that chip is bad but potentially others (at least others in the upper ram block).
 
My initial assessment would be that data bit 0 is stuck LOW (UA4), data bit 1 is stuck HIGH (UA6) and data bit 5 (UA14) is (possibly) stuck LOW.

I would suggest replacing data bit 0 (UA4) first and then retesting. You should observe more 'goods' and less 'bads' as you work through the problem...

Then post a new photograph so I can see...

I can explain my thought process if that would be helpful for your information later.

Dave
 
daver2 said:
My initial assessment would be that data bit 0 is stuck LOW (UA4), data bit 1 is stuck HIGH (UA6) and data bit 5 (UA14) is (possibly) stuck LOW.

I would suggest replacing data bit 0 (UA4) first and then retesting. You should observe more 'goods' and less 'bads' as you work through the problem...

Then post a new photograph so I can see...

I can explain my thought process if that would be helpful for your information later.

Dave
Click to expand...
oh that's because any given byte is spanned across multiple chips, right? Okay so sounds like I need to order some ram chips.
 
Yes, Data bit 0 is UA4, Data bit 1 is UA6, etc.

I assume that there could be multiple data bit faults - so more than one faulty device can contribute to a 'bad' status indication.

I then look for patterns. I see that we get 'bbgbbbgbbbg" and then this repeats. I also observe a whole block of 'bbbbb...'.

Address $0000 (data $00) is BAD. So is address $0001 (data $01).

But address $0002 (data $02) is GOOD.

So the fastest repeating pattern is actually 'bbgb'.

I have then looked up the data that is returned for these memory tests - being 'bb.b' (the '.' indicating the good test value). It is a little confusing having 'b' indicating a 'bad' test and the data value returned as 'b' also. You have to count very carefully so that you correctly lock onto the change between the 'good/bad' memory cell status indications and the data related to the tests.

Address $0000 Data $00 should give $00 (@) but gives 'b' which is $02 = 0000 0010.
Address $0001 Data $01 should give $01 (a) but gives 'b' which is $02 = 0000 0010.
Address $0002 Data $02 should give $02 (b) and is correct.
Address $0003 Data $03 should give $03 (c) but gives 'b' which is $02 = 0000 0010.

The red text indicates the bits that are in error.

This is how I deduced that bit 0 is stuck at 0 and bit 1 is stuck at 1.

A then looked at the large block of 'b' and guessed that this relates to data bit 5. This would give us an approximate $20 (32) string of 'bad' if this bit was stuck. As we get a nice pattern when data bit 5 is LOW and a permanent 'bad' when data bit 5 should be HIGH I am assuming it is stuck LOW.

You know as much as I do until we get some 4116 DRAMs replaced now...

It's a logic puzzle. My wife does Sodoku puzzles - I do DRAM puzzles :) ...

Dave
 
Last edited:
Divarin said:
I thought maybe these keyboards are just supposed to operate with some higher resistance level so in order to get a connection strong enough to make the multi-meter buzzer go off I need to press really hard?
Click to expand...
Yes, in theory hard presses should not be necessary. The PET circuit uses a 10K Ohm pull up package (UB11). So if my math is correct, the keyboard contact resistance has to be only less than 417 Ohms for the input of the PIA to see a good logic zero of 0.2V. I've always thought that a simple solution to the PET keyboard contact problem would be to replace UB11 with a 100K Ohm package which would allow the resistance to go up to 4K Ohm. However I think that might lead to a RC time constant issue and perhaps to double input characters.
-dave_m
 
dave_m said:
Yes, in theory hard presses should not be necessary. The PET circuit uses a 10K Ohm pull up package (UB11). So if my math is correct, the keyboard contact resistance has to be only less than 417 Ohms for the input of the PIA to see a good logic zero of 0.2V. I've always thought that a simple solution to the PET keyboard contact problem would be to replace UB11 with a 100K Ohm package which would allow the resistance to go up to 4K Ohm. However I think that might lead to a RC time constant issue and perhaps to double input characters.
-dave_m
Click to expand...
Ah okay good to know. My copper foil fix didn't work too well. there's a small number of keys working and some working if I press really hard but the majority don't work at all. I went ahead and ordered some conductive paint but if copper foil can't make a strong enough connection I don't have much hope that conductive paint would work better, probably worse.
 
oh man I got the wrong kind then.
Well anyway I've been poking around the contacts today and focused in on one in particular that only makes a connection if I press really hard on it, what I'm doing is laying a small strip of copper foil over it, then pressing down with a pencil eraser. while holding multimeter probes to another point where they make contact:

1666449829176.png

Some just a light press with the eraser works but most (like this one) you have to really press. Even if the multimeter is set to 2000 ohms or 20k, to get any reading at all you have to press down super hard. I also tried bridging the connection with a small bit of solder (solid, I didn't melt it, just using it like a wire) I'd lay the solder over the connection and press down with my finger. I have to push really hard.

There's good continuity from either end of the contact to anywhere else on the board including the cardedge but having a really hard time connecting the two halves. I'm wondering if they're too worn to let any flat surface connect them.

I'm thinking of trying to flow a layer of solder over these, but also sort of worried that might cause more problems.
 
Please don't use solder, you will regret it...

Is it possible that the surface of the copper has oxidised (or otherwise become high resistance). This is what happens to bare copper PCBs if left for a period of time. Roller tinning is better. Gold us best...

Please do some reading on the other VCFED threads (Commodore in particular) related to this topic to see how other people have resolved their issues. Some of them have been difficult also. Other people have already walked this way for you...

I think @dave_m has been involved more in this subject?

Dave
 
Last edited:
alright I'll read up more, stange that I have good continuity on either side all the way to the card edge just can't connect the two halves very well.
 
I cleaned the plungers by pulling a clean sheet of folded paper across them. I also cleaned the PCB with IPA and a light rub. I had breaks in the tracks I had to cope with too, but it seems you don't have them.

I bought some conductive paint but never used it.

Good luck.

Colin.
 
Yes, I would think a little isopropyl alcohol should do it. Use a high grade of 98+ pure.
The Deoxit products of contact cleaners should also be good. Do not use the Ruby pink eraser trick as it is too abrasive and will rub off some of the metal.
 
It's the rubberised plunger pads that need cleaning. Colin's 'paper method' seems to work well.

Alan
 
One other thing to check. There are a lot of tiny screws that hold the keyboard PCB to the keyboard frame. Make sure none are missing. You want no flexing or bowing of the PCB.
 
I am not familiar with this keyboard but there are two things that would probably fix it for sure.

Where the Gold plated fingers are, clean that with a cotton cue tip and contact Cleaner. CO contact cleaner is good.

Although the Gold looks shiny and metallic as it ages it gets a transparent insulating film on it that you cannot see with the naked eye, it is very deceptive, it looks clean but it is not.

This problem affects Gold contacts in switches where the operating voltages and currents are low and there are no sliding motions or friction on the Gold surfaces. Other places the effect turns up where there are touching contacts are in scope attenuators, ZIF IC sockets, & some brands of DIP switches and on the Target electrodes of single tube color video cameras, but that is not a full list.

This problem causes chaos & intermittent connections and connections exhibiting high resistances in aged systems, even though we are taught that Gold is the go to metal for low corrosion and good conductivity, if there is no sliding friction, the effect will appear after the decades roll by. Switches with sliding contacts don't have the trouble as it breaks through the thin transparent film. But, any system of Gold touching Gold, or Gold touching another conductor, the effect will ultimately occur.

Then, check the surface resistance of the rubber plungers. If they are not low after cleaning you could replace them. Conductive rubber discs of different diameters are readily available on ebay, if you search under calculator parts & repairs. I bought some very good 10mm diameter ones while doing some keyboard experiments, they had very low range resistances.

Don't apply anything significantly abrasive to the pcb. The most abrasive effect I would ever apply to the pcb fingers would be A4 paper soaked in contact clear, never anything more abrasive than that, not a rubber and certainly not any abrasive papers. Most of the time the cotton cue tip with CRC's CO contact cleaner is all that is required to remove the invisible film.
 
It looks like the surface resistance of the rubber plungers is really high, I have to set my multimeter to 20K ohm to get a reading at all, at 2000 ohms I get nothing, at 20k I get ... something, numbers just jump around a lot but start out around 5.something

anyway when I first opened this keyboard up to clean it about a month ago I used isoprop. alcohol on the circuit board contacts (using q-tips) and I tried cleaning the rubber plungers by gently sliding the pad across a piece of paper (a post-it note) Normally when I do this there's a light black steak left on the paper which tells me that there's still carbon on plunger. With these there's nothing on the paper. So I think they've been too aggressively cleaned in the past.

As for the copper foil I used I didn't glue them they just have a light adhesive on one side, just (barely) enough to hold them on so it shouldn't leave a glue residue at least nothing that another light slide across a sheet of paper won't remove.

Luckily my Amazon order for the (wrong) conductive paint didn't go through due to issues with me switching from an about-to-expire debit card to a new debit card so that gave me a chance to cancel that order and buy the silver stuff instead.

As for the contacts on the circuit board they seem fine. With my multi-meter I can measure good continuity (no measurable resistance) between any part of the "fingers" to the card-edge and I don't have to give any pressure on the probes. I sort of expected if I move the probe around on the fingers I'd find "dead spots" that needed to be cleaned but nope it's all good. I just can't bridge the two halves of the contact with these plungers and for some reason the copper foil isn't doing it either even though that foil has no measurable resistance either.

But I'll wait until the paint comes and try just painting one rubber plunger and let it dry for 24 hours, if it gets good contact I'll do that with the rest of them, if not I may just see if I can get replacement plungers.
 
Ah, so the switch pads on the circuit board are fine then.

I thought you were having problems with those as well.

So, some new pads or 'proper' paint should solve your problems...

Hopefully, you will also have some replacement 4116 DRAM by then and we can get this sucker working properly (apart from your monitor line 'tearing' issue that we still need to address).

Dave
 
yeah I hope so, it's hit or miss with ebay sellers though, about 30% of the time they buy the shipping label but never drop it off at the post office and I end up having to get a refund from ebay.

these ram chips still show as "usps awaits item" but that order was placed just before the weekend so that's not so unexpected.
 
You must log in or register to reply here.
Back
Top