exxos's Atari, Amiga & retro forum

PaulJ wrote: 19 Apr 2020 15:02 Don't know @rubber_jonnie , I'd be hard pressed to use a mega 4 as a donor. I would keep fixing even if it wanted to die.

Yep, that's why the CO70789 is gonna be in the frame for the next H4. I just cannot bring myself to kill the Mega, and why I'm going to get one of exxos' RAM upgrades. It will live again!!!

Well, the MMU/SHIFTER upgrade was ordered from Harry this morning (Amongst other things), so hopefully next weekend I can put this whole Mega 4 thing to bed!

I'll keep the V1.1C IDE/TOS adapter and fit the CF card I'd already built whilst it was working. If it stays dead, well that's another matter, and I will put the upgrade into the Mega 1 instead.

I guess we'll see once it arrives.

Whilst I'm waiting for the RAM upgrade to arrive from Harry, I thought I'd do a little more prodding around with the Mega4.

On my H4, also 4MB, all the MAD lines have a pulse at the MMU, and all the data lines have a pulse too.

The Mega 4 on the other hand doesn't seem to have a pulse at D2 and D4 at the MMU, which may be why things are going bad. D2 & D4 have pulses at the CPU.

TBH I have only checked so far with my cheap ass logic probe (Must get a better one), so next up is the scope, see what is going on there, plus another look at the schematics.

MMU has been tested in the H4 so I know it's good, as is the GLUE, DMA Shifter, YM, and the various logic chips.

Be interesting to see what I can find out, because if I can get it running again, I may give the 4MB upgrade that's on order to the Mega 1.

It's a fun way to fill all this free time IMHO :)

Out of interest, I ended up using the Meanwell PSU from the Mega to power the H4, because when I used a standard STFM PSU on the H4, any time I put the logic probe anywhere near the MMU, which is very close to the PSU, all the lights lit up on the probe, clearly a result of electromagnetic interference. Clean as a whistle with the Meanwell instead.

Sounds like a step in the right direction @rubber_jonnie

PhilC wrote: 26 Apr 2020 12:31 Sounds like a step in the right direction @rubber_jonnie

@PhilC Turns out it was exactly the right direction.

I decided the oscilloscope wasn't necessary until I'd actually buzzed out the data lines back to the CPU.

Turns out the fault was literally just outside the socket:


Buzzing out the lines from the socket to the VIAs literally the next hop along, and there was no continuity. That is about 10mm from socket pin to the VIA as indicated by the red lines.

The VIA to the MMU buzzed out just fine, so bodge wires added as follows:


And now she runs just fine again.

Now bearing in mind we've talked about manhandling boards a few times on here, and how something mechanical can be a problem, well this is a prime example.

I removed a Beta IDE board and installed the CPU in the socket, plugged in a Megabus IDE adapter for testing, and plugged in a 6 ROM set so I could test the Megabus IDE adapter.

All that activity near the CPU socket was enough to kill the machine with some sort of mechanical force that broke the D2 and D4 connections from the socket to the board.

I tried reflowing the joints first but nothing until I physically added the bodge wires.

After that, booted right away, either with the compact IDE board or the original CPU and a 6 ROM set.

Turns out the RAM wasn't faulty at all, even though the error message said RAM first, it was actually the data line part of the following message:

I1 RAM/data line error (D15)0000000000000000(D0) (D15)0000010100000000(D0) 000008

Which was relevant.

What is interesting is how far you can get with minimal tools once you start to get a feel for just what makes these machines tick.

All I used to diagnose this in the end was:

A logic probe (A cheap ass one at that!)
A multi-meter in continuity testing mode (Buzz out)

Also goes to show, like in the post @Icky made earlier about just how much longer these boards will last. It seems pretty clear that the answer is "Probably not much longer!"

I'd also better go clear up the flux from some earlier reflowing I reckon!

Congrats @rubber_jonnie ! I bet you are well pleased after all your work investigating that! should we take bears as to how long it will last again now ? :lol:

I think the boards are just so fragile that simply plugging in and out a upgrade is enough to break the thing, which I think is basically what happened to you ?

I had a bad Via on my MEGA as well... it was /AS from the CPU to MMU... There was a break in it, as to how this machine ever worked in that state is beyond me.

Via's are evil :twisted: :twisted:

exxos wrote: 26 Apr 2020 18:05 Congrats @rubber_jonnie ! I bet you are well pleased after all your work investigating that! should we take bets as to how long it will last again now ? :lol:

I think the boards are just so fragile that simply plugging in and out a upgrade is enough to break the thing, which I think is basically what happened to you ?

@exxos Please, no bets, I just want to finish building this machine and close the lid!!

And yes, simply plugging & unplugging an upgrade was just enough to break it.

If it's any consolation, one of my Megas has developed a ram fault today. Bit 13 is faulty, so out with the scope in a bit.

Mind you, I'm lucky, mine will boot still

PhilC wrote: 26 Apr 2020 18:44 If it's any consolation, one of my Megas has developed a ram fault today. Bit 13 is faulty, so out with the scope in a bit.

Mind you, I'm lucky, mine will boot still

After all my fun and games, I wish you good luck! At least I can say I'm learning, which is nice.

Booting is good, I'm bloody lucky the diag cart at least gave me some help via the serial port.

Let us know how you get on, I'm rooting for you!

Well, it's fixed and I am amused at the root cause, see below.
That pin was touching ground.