Nelma Data Corporation - Persona 100 Build Log

[chillywaffle]

 

I know that this isn’t typically the type of build log that gets posted here, but I found this weird computer when I was browsing around Kijiji.  At first glance, it looked like an IBM 5150 but the disk drives were on the wrong side.

 

According to the seller, it was a Nelma Persona 100 and it was made in Canada.  It even had an Ontario Hydro certification sticker on the back. It was listed as not working and it did not include any boot media.  

 

A quick Google search revealed practically nothing about the company or the computer.  I did find a reference to it on a CP/M collectors blog and an article from 1983 in Computerworld magazine.  The only facts I had about the computer was that it ran CP/M, it was powered by a Zilog Z80 microprocessor and it was originally released in 1983.  Because I am terrible with money, I bought it.

 

When I want to pick it up, the seller had it set up and powered on.  The system had power but the monitor was displaying a bunch of random characters.  The system came with a 13” green phosphor monitor and a keyboard.

 

As I was unloading it from the car, I noticed a weird rattling sound inside of the computer case and saw a couple of sunflower seeds fall out of it.  I quickly got it into the house and opened the case. It was packed with sunflower seeds and shells. At some point, something had been living in this thing.  

 

Unfortunately, I did not get before pictures of the seed infested case since I was worried that something was still living in it but here are some pictures of the computer after a good vacuuming.

 

Ancient 17” laptop for scale

 

Something heavy was sitting on that front bezel for a bit

 

The inside of the computer with the disk drives and fan removed

 

The 13” green phosphor display and the keyboard.  The data corruption on the screen is the fault with the machine.

 

My original plan was to clean up the case and build something more modern into it but I really love the look of that green phosphor monitor.  It looks like something out of Fallout.  So instead I am going to try and restore this thing back to its original 1983 condition and document the process for you guys.  

 

There are several big obstacles for this project.  The biggest one is that I have actually never tried something like this before. I have played around with a couple of arduinos in the past but my knowledge is more geared towards the IT/Devops side of the fence.  So it will be a learning experience.

 

I am also missing several key pieces of information.  Nelma Data Corporation seems to have only existed from 1983 to 1987 and I can find almost no information about them online.  I have no documentation for the computer and no boot disks which is challenge since the BIOS portion of CP/M is machine specify.  

 

On the long shot that you guys know anything, let me know.  I’ll keep this updated as I dive into it. Thanks for checking this out!

 

 

 

[TrigrH]

[chillywaffle]

@TrigrH

That's pretty cool, I looked up the address on street view and it looks like it's an aquatic fitness center now.  

Based on that ad, it looks like I just have the standard build.  One thing that is weird on this board is that there is space for 4 expansion card slots but the power and card slot sockets where not soldered in.  I wonder if this unit was a further cost reduced option that was made after this ad was printed.  

[chillywaffle]

OK, so a quick update.  I have pulled the board out of the machine and have had a chance to look it over.  Based on the symptoms, there could be several things wrong with it.  I have attached a picture highlighting the chips that could be causing the data corruption on screen.  Sorry about the blurry shot :-(. 

I know this looks like I have basically painted the entire board, but I swear there is a good reason to suspect any of these chips.  

Let's start with the column of 8 ICs highlighted in green on the left hand side of the board.  This is 64k of system memory.  If I have a bad DRAM chip, it could be causing a bunch of random data to be outputted to the screen.  The other reason that I suspect it could be a RAM issue is that the disk controller is not sending the signal to kill the disk drive motors when no disk is detected.  

I don't think I will be able to test the memory while it is in the board.  I will probably have to remove these and test them one at a time.  

The second culprit could be the video memory on the right side of the board highlighted in blue.  Those are 6 SRAM chips that hold character data before it is outputted to the display.  They will also need to be removed to be tested. 

Below those chips are the two CRT controllers.  They have built in video memory as well.  I actually haven't got the faintest idea on how I am going to test these.  I also think that one of them regulates power to the monitor as the monitor is powered right off of the display port on the motherboard. 

The chips highlighted in purple are the ROM chips.  I actually don't know how much of the included software is stored on these.  I have removed them from the board and found they make no difference except that the distinguishable characters all disappear.  One of them must hold the character map.  

The orange chip at the top of the board is the Zilog Z80A CPU.  Variants of this CPU are actually still being made and there is a ton of information available for it.  Hopefully I can come up with an easy way to test it.  

The last thing it could be is a fault with one of the glue logic chips that are highlighted in teal.  This is honestly where the problem probably lies.  Fortunately, it is mostly made up of 74 series TTL chips which are readily available.  Unfortunately, there is a pile of them to test.  I don't have a schematic of the board so I will probably need to test all of the chips individually.  

I will need to find a way to test all of those TTL chips in circuit because I really do not want to have to remove them just to test them.  

Anyway, that is all I have for now.  I want to get a start on the TTL chips this week and I will update when I figure out a good way to test them.  

[chillywaffle]

Adventures in Testing TTL Logic in Circuit

 

I have had a chance to look over the problem of testing the TTL glue logic without removing the chips from the board.  Unfortunately, I have to do this instead of just tracking the issue down with the schematic as I do not have a schematic for this board.  The silver lining is that by doing the glue logic first, I can follow the traces off of each chip and try and map out the whole board in kicad. 

 

At first I tried using the 5V and ground pins on an Arduino to drive the chips individually.  This ended up being a terrible idea because suppling power to a single 5V pin while it is still wired into the circuit supplies 5 volts to all of the chips on that 5V trace. This pulled too much current through my Arduino triggering the circuit protector.  I really should have taken physics in high school. 

 

I ended up having to use the power supply that came with the machine and this is what I set up: 

 

(10 points if you spot all the fire hazards) 

 

So basically what I am doing is running jumpers from the power switch to the AC terminals on the power supply with some alligator clips wrapped in electrical tape.  From there, I have the main motherboard connector plugged into the board and two multimeter probes jammed into a molex connector to supple +5V and ground to the bread board.  

 

The bread board is going to act as a power distribution system to supply power for my logic probes and to act as logical 1 and 0 for testing purposes.  

 

To make sure that there is no activity on the board, I grounded out both of the crystals and shorted out the reset line to the Z80A CPU so all of the address lines are held in an impedance state.  

 

 

To connect to the chips, I found these cool little grabbers on amazon that can clip onto chip leads without shorting everything out and starting a fire.  

 

 

So now all of the pieces are in place to do some testing.  

 

The first step to testing 74 series logic chips is to pull up the data sheet of the chip that you are going to test.  Let's start with these 74LS00 chips since they are pretty simple.  

 

 

When you pull up the Datasheet for the chip, the first two things you want to look for are the connection diagram and the truth or function table.  The connection diagram shows what each pin is connected too and the truth table shows the logical operation of the circuit.  

 

This is the connection diagram and truth table for the 74LS00 IC from Fairchild Semiconductor: 

 

So, we can see from the diagram that the 74LS00 contains 4 two input NAND gates and we are going to test this chip by applying the signals in the input columns to the inputs of the chip and measuring the output.  If each gate matches the truth table, then the IC is fully functional. 

 

The first pins that I test are pins 7 and 14 to make sure that the chip is being powered.  Pin 14 should be high and pin 7 should be low.  Just touch the chip leads with the logic probe.  

 

 

 

Now it is time to test each of the NAND gates.  To do this, simply attach the clips to the 2 input chip leads and run them back to the breadboard.  The +5V from the molex connector will act as logical high while the ground lead from the molex connector will act as logical low. Since those are connected to the breadboard, just move the wires between the two rails to create every combination on the truth table while testing the output with the logic probe.  

 

(I actually lost the picture of me testing the 74LS00 chip, but this picture should give you an idea) 

 

Now, not all 74 series chips are as easy to test as the 74LS00 but it is a good place to start.  Some chips require a clock signal to function correctly as they are edge triggered (The chip is triggered when the voltage shifts from low to high or high to low).  I usually just fake it by touching the chip lead with a 5V jumper wire.  

 

Some chips like line drivers are better tested with a logic pulser/probe combo.  This way you can see the signal pass through the chip successfully.  

 

Just remember if you get stuck, I found that the Datasheet + Google will usually get you out of a bind.  

 

In the end, I found about 9 chips that had anomalous results.  This is mostly likely due to my less then perfect testing method.  I'll likely just order them and replace them one at a time since they are super cheap.  

 

Now I just need to figure out how to test DRAM and SRAM chips.