• Please review our updated Terms and Rules here

Kaypro 4 '83 - Turns on but doesn't boot. Pop sound and smoke.

Another update. I pulled every single chip from the motherboard and re-seated them. Bent one or two pins a bit, but managed to straighten them up ok and they went back into the sockets correctly as far as I can tell.

I also soldered a new leg onto the memory IC that got snapped off before.

PXL_20240806_095550791.jpg

PXL_20240806_100811717.jpg

PXL_20240806_101226944.jpg

Used a breakaway male header from a breadboard kit which was a pretty good fit as a replacement.

Unfortunately I'm not seeing an improvement so I'll have to wait for some EEPROM's, an Arduino Mega and a DB9 to DB25 cable to arrive which would allow me to program a diagnostic ROM and attempt to get a serial connection.

I'm also curious to see if perhaps powering the board from an ATX power supply I have lying around would change anything. I require some sort of harness though so I'm trying to work out the best way to get something clean and safe that won't require chopping up a good power supply. Looks like ATX headers are pretty easy to buy, but I have no idea what sort of connectors to get for the motherboard power. I saw elsewhere an AT power plug (Molex 90331) might work? Unfortunately I don't have any AT power supplies lying around or the crimping tools required to make a plug manually so might need to resort to using alligator clips or something and just being very careful not to touch the connectors together so I can power it on long enough to test it.

Something weird that happened after I put all the components back in was I got the repeating 'N pattern again, but then when turning it off and turning it back on again I got the scrambled characters which got me thinking perhaps it's power related. I did measure stable voltages at the motherboard power header but maybe elsewhere on the board the power isn't what's required.
 
Just something I have noticed about IC's with very rusty pins, in my experience, they appear to have a high incidence of being defective too.

I'm not 100% sure of the mechanism of this, it could possibly be that rust and corrosion migrates down the pin surface in the package and into the interconnects connecting the pin to the die. This is known to happen with epoxy package transistors too, not just with rust either, but ones that had silver plated pins, which were common on many early TTL & CMOS IC's.

Even though you have re-soldered a new pin onto the stump of the remaining one, I would recommend testing every RAM chip in a tester (unless you have a good number of known good replacements) because all of the pins on those RAM IC's looked pretty rusty.

Also if you are buying 4164 ram chips, I would avoid sellers where a lot of refurbished parts of unknown history appear with bright shiny looking pins. Buy genuine looking new old stock from a reputable seller. Japanese brand chips such as Mitsubishi, Hitachi etc are generally very good. Avoid used parts, and look for 80's era date codes to help to avoid fakes.
 
Last edited:
Ok, well who knows if there's multiple other issues (I'm sure there must be since I was getting a different screen before re-seating every chip), but according to this DRAM tester I built with an Arduino Uno R3 - 2 DRAM chips failed a memory test.

Say hello to this monstrosity:

PXL_20240807_135131828.jpg

The very first chip I tested I got a red LED and I was wondering if it was even working correctly.

PXL_20240807_152155733.jpg

But then I went through a bunch more and kept getting green LED's so clearly it was working.

PXL_20240807_153001692.jpg

So in the end these are the 2 casualties. One chip from '83 and one from '82.

PXL_20240807_154140212.jpg

Funny enough the chip I had to solder a leg onto came back with a pass, so it's nice to know the damage wasn't something I caused.

I used the guide here to build it:


Unrelated to the DRAM issue (???), it seems there has to be some weird power thing going on as well, because I moved the system downstairs so I could pull the memory chips out for testing and figured I'd just turn it on first to see what sort of curse it has today and I was greeted with this:

PXL_20240807_151206855.jpg

Most of the characters were flashing. Reset button didn't make any change to screen. Turned it off and turned it back on again and then back to the matrix with the same character pattern as previously

PXL_20240807_151302984.jpg

So that would be the second time this has happened when the power has been off for a while. When I turn on initially I'll get a bunch of random characters, often many repeating characters, and after turning it off and back on I then get the matrix screen. The characters on the matrix screen are almost exactly the same as previous pictures, however if you look closely there are actually some differences. Eg, line 20 at the end of the line is a "K-" but previously it had been "C-" which is interesting. Perhaps a change caused by moving RAM around.

It would seem that the matrix effect happens after the machine is "warmed up". Can anyone guess as to what could cause this? Capacitors storing up energy after being off for hours and that's causing the weird behaviour? - Note that before I got the repeating N' pattern over and over before re-seating everything. I'll see what happens when I replace the bad chips and then I might need to re-seat everything again.

It's nearly 3am now, so tomorrow I'll try hunting down some 4164 DRAM. It would make life easier if eBay actually enforced the "List Australian sellers only" option. Seems many Australian's are able to somehow ship packages for only $3 but the estimated arrival date is 2-3 weeks away...
 
Last edited:
I had a look around on ebay. If I were buying parts, one advantage being in Australia, is that many new genuine old stock IC's are coming out of Europe now, places like Germany, Hungary etc and shipping to here is cheap & easy. There are some good one in the USA, but also there are a lot of refurb'd parts about too, reprocessed board pulls etc.

I would probably buy these ones, because, they are new old stock, with a package with physical characteristics that you don't see with fake IC's. The Japanese chip makers were super fastidious, and the failure rates are really low. These ones are in the USA as it happens and it usually takes about 10 days by air post to get them to AU and you will have plenty of spares and an amazing price for 25 pcs:

www.ebay.com.au

25pcs National MN4164P-15A NMOS 65,536 X 1 BIT DYNAMIC RAM DIP-16 Chip Japan MIJ | eBay

Find many great new & used options and get the best deals for 25pcs National MN4164P-15A NMOS 65,536 X 1 BIT DYNAMIC RAM DIP-16 Chip Japan MIJ at the best online prices at eBay!
www.ebay.com.au

One type of IC to watch out for and avoid are ones with laser engraved markings, these didn't generally exist in the era of these IC's, and are re-labelled parts.

Of note, DRAM IC's can have failure modes where they work over specific time frames because of their internal refresh arrangement. If the chip is good, it will hold its current data within the refresh interval. As chips age and fail, they can develop corrupt data in the time frame between normal refreshes, or alternatively there can be defects in the external refresh circuitry in the computer. A rapid DRAM test protocol not only can miss DRAM IC's with these defects, but even a normal DRAM chip can appear to fail in the computer if the support refresh circuitry is defective. DRAM IC are highly complex inside. compared to SRAM.

The strange behavior of your computer makes me wonder about DRAM refresh issues, which noted can reside inside or outside the chips.

There is a way to get around this, but the DRAM needs testing in the computer to confirm that it holds it data correctly over extended time frames, and a method devised to thoroughly check the refresh support circuitry. It pays to write data patterns to the memory , then re-write to adjacent bytes (there can be cross interference). One way to check the DRAM support circuits, is to make the computer independent of the DRAM completely for its OS only, by electrically substituting in SRAM to support its operations, then examining the DRAM refresh circuits with the scope. These are active even when the computer is doing nothing. I did these things to evaluate 4116 DRAM IC's in the PET, because I found for the reasons above, unique to DRAM chips, the ebay DRAM testers had limited utility:

www.worldphaco.com/uploads/DRAM%20MEMORY%20TEST%20SYSTEM%20FOR%20THE%20DYNAMIC%20PET%20COMPUTER.pdf
 
Last edited:
Pyroteq said:
So that would be the second time this has happened when the power has been off for a while. When I turn on initially I'll get a bunch of random characters, often many repeating characters, and after turning it off and back on I then get the matrix screen. The characters on the matrix screen are almost exactly the same as previous pictures, however if you look closely there are actually some differences. Eg, line 20 at the end of the line is a "K-" but previously it had been "C-" which is interesting. Perhaps a change caused by moving RAM around.

It would seem that the matrix effect happens after the machine is "warmed up". Can anyone guess as to what could cause this? Capacitors storing up energy after being off for hours and that's causing the weird behaviour? - Note that before I got the repeating N' pattern over and over before re-seating everything. I'll see what happens when I replace the bad chips and then I might need to re-seat everything again.

It's nearly 3am now, so tomorrow I'll try hunting down some 4164 DRAM. It would make life easier if eBay actually enforced the "List Australian sellers only" option. Seems many Australian's are able to somehow ship packages for only $3 but the estimated arrival date is 2-3 weeks away...
Click to expand...

I think your RAM is just showing bad data.

Which state are you in? Didn't know you were from Australia - I have some 4164's you can have in Perth you can pick up if you're local?

Until you fix the DRAM, you won't get any improvement in the troublehsooting. :(

Cool work you did on the tester though !

p.s. Just checked your time stamp. I see the post from 12.48am so if you're nearly at 3am, I guess that puts you over East.
 
Hugo Holden said:
I would probably buy these ones, because, they are new old stock, with a package with physical characteristics that you don't see with fake IC's. The Japanese chip makers were super fastidious, and the failure rates are really low. These ones are in the USA as it happens and it usually takes about 10 days by air post to get them to AU and you will have plenty of spares and an amazing price for 25 pcs:
Click to expand...

YOLO, ordered that one, I'm trusting your judgement.

cj7hawk said:
I think your RAM is just showing bad data.

Which state are you in? Didn't know you were from Australia - I have some 4164's you can have in Perth you can pick up if you're local?

Until you fix the DRAM, you won't get any improvement in the troublehsooting. :(

Cool work you did on the tester though !

p.s. Just checked your time stamp. I see the post from 12.48am so if you're nearly at 3am, I guess that puts you over East.
Click to expand...

Yeah, I'm in Sydney. Thanks for the offer, I already ordered the 25pc Hugo linked. I figure I'll either sell the rest and break even or I'll keep them around and maybe I can use it to build an 8 bit breadboard computer with the kids.

Also ordered a few AT28C16 EEPROM's. That will allow me to program the diagnostic ROMs if the DRAM doesn't fix the issue.

Now just gotta wait a couple of weeks for the parts to arrive. Thanks everyone for your help so far. =)
 
You must log in or register to reply here.
Back
Top