• Please review our updated Terms and Rules here

APF Imagination Machine - 4108 Replacement and some Newbie Questions

D

Dubis7

Experienced Member
Joined
Mar 20, 2018
Messages
105
Hi all,

Emboldened by the realization that APF graced the Imagination Machine with socketed chips, I've decided to try and get mine working. Currently, the M-1000 console is fully functional, and when I plug in my RAM cartridge I can see the title screen no problem. However, upon boot up, I'm presented with a screen full of @ symbols and then it restarts.

I suspect a RAM issue, and am hoping to identify the bad chip and replace it. However, it looks like the 4108 chips used in the Imagination Machine are no longer readily available. Is there a substitute?

Additionally, while I've done some repairs before, I've never quite understood how to identify a bad RAM chip. I've got a test probe, and I've successfully used it with guidance, but is there a trick to being able to tell which pins should be high, low or pulsing? is it literally just a matter of identifying where the power is coming in and out, checking that those are high, and then identifying the data lines and ensuring those are pulsing? Could I use a chip data sheet to tell me which pins should be doing what? Or am I completely off the mark?
 
The 4108 DRAMs were from the days when the yield of 16K DRAMs was pretty low. So TI and Intel and Mostek sold "half-bad" 4116s/2116s as 4108 or 2108 8K units. Recall that at the time there was a shortage of DRAM. Read the TMS4108 data sheet and you'll see that there were two versions--one where A0=0 and another where A0=1 when asserting CAS.
The bottom line is that you can use 4116 DRAMs in place of the 4108s with no bad effects.
 
Got it, Thanks! I've gone ahead and ordered a batch of new 4116s. Not a bad idea to have some spares, and that way I can go ahead and swap the whole batch, then work backwards to identify the bad chips.
 
Last I looked Anchor Electronics in Sunnyvale actually had some 4108s on their price list (they have lots of weird things), but there's really no reason to go that way unless you want to try to make all look original. And as Chuck noted, you'd have to make sure you got the right "gender" of 4108 to match the rest of the bank.
 
Funny thing about those half-bad 8Kb DRAMs. My first (S100) 64KB DRAM board was populated with them--got them for free from the local Intel sales guy. They all performed well as 16Kb DRAMs. I probably still have a few of them in my hellbox.
 
Okay, so a progress update on this, and hopefully seeking a little help. I replaced all the RAM chips in the imagination machine with new 4116s. It gets past the title screen to the cursor now, but crashes when I press a key. Most of the time it just locks up, but sometimes it displays garbage on the screen or resets entirely.

It's not proper debugging, but since most of the chips were socketed I decided to pick up a known good spare for each and swap it in. No change. I figured it had to be one of the support chips soldered directly to the board, and recognizing that the bottom RF shield was a bit of a pain to get off I decided to go the extra step and socket all 8 of them. That process went smoothly and the entire thing is essentially running on new chips now.

Despite that, it still crashes when I try to type anything. Literally no change. I didn't break anything, but I didn't fix it either.

I realized afterwards that I missed one chip, something called a Belfuse 0447-0250-01. I thought it was some sort of fuse, but apparently it's something called a delay line. Might that be my issue, and if so does anyone know how I might go about checking it? If it is broken, it looks like it might be a hard to find part.
 
Dubis7 said:
I realized afterwards that I missed one chip, something called a Belfuse 0447-0250-01. I thought it was some sort of fuse, but apparently it's something called a delay line. Might that be my issue, and if so does anyone know how I might go about checking it?
Click to expand...
See post #60 at [here]. You will need either a logic analyser, or an oscilloscope with at least two channels.

( A logic probe will not show the delay, but it would reveal whether or not pulses are exiting on the 'tap delay' pins. )

Dubis7 said:
Might that be my issue,
Click to expand...
Known to be used in support of RAM, providing the delay between the RAS and CAS signals sent to the RAM chips.
See [here] for an example.

Dubis7 said:
If it is broken, it looks like it might be a hard to find part.
Click to expand...
See [here]. There were at least four manufacturers making these things.
 
I might be able to help you in a round about way.

I have had some run-ins with defective delay lines on vintage computer pcb's, not the same as your (Processor technology).

I tackled a faulty delay line problem. I disassembled the delay lines to find out how they worked, why they had failed over time and created ways of making a replica.

In general they are prone to failure primarily because the solder flux inside ultimately corrodes through the fine copper wire, or possibly cracks in the terminating resistance.

Typically the delay line consists of a TTL Hex inverter logic IC and an L-C delay analogous to a transmission line, driven at one end and terminated by a resistor (see page 5)

I had a go at re-manufacturing one with new IC and hand wound inductors and surface mount capacitors. In the end I used some delay IC's from Dallas and Maxim.

You should be able to create a new delay line (if you need one) to suit, using these parts to replace the original unit, if it is defective. The information is in this article:

 
Okay, I'm raising the white flag here. I need some guidance beyond my expertise.

Here's where things stand: As far as I can tell, the MP1000 console works perfectly. Rocket patrol runs, as do cartridges. It successfully interfaces with the Imagination Machine keyboard as well, at least as far as cartridges go.

When running Basic, it will show the splash screen, and pressing the EN key on the controller shows a prompt. Then things break. Pressing keys either locks the system, resets it or generates garbage characters. I haven't found a pattern on that yet.

Just to confirm, I checked the voltages coming out of the power supply board. They're all within tolerance, so this doesn't appear to be a power problem.

Here's what I've done: There's a new PIA chip in the MP1000, which was not originally playing rocket patrol. That fixed that problem. The main board on the Imagination Machine is completely rebuilt. All new chips, all socketed. I replaced the delay line, which was originally a Bel Fuse. The new one is by Dallas Semiconductor. Of note is that the original delay line had 16 total pins, while the new one has 14. That threw me off for awhile, but as far as I've been able to tell pins 8 and 9 on the original weren't connected to anything and the pinout is the same. It seems that modern equivalents omit those last two pins. The system isn't working any worse than it did before I made the swap so I'm keen to presume that I interpreted things correctly, but if anyone notices anything there by all means let me know.

I made two videos - one to highlight how the system is operating, and the other to show my work on the board.

Does anyone know where I should be looking next? I've got a logic probe handy to start working out where things are wrong. Now I just need a map. I'll also pass along the service manual I've been working from in case that helps.
 
You must log in or register to reply here.
Back
Top