exxos's Atari, Amiga & retro forum

I have an STE, which experienced various weird stability issues, for example, trashing floppy disks, nonworking cartridge port (NetUSBee, also Test Cart not working, yes the fuse was OK, dumping the test cart contents revealed some data trashing, actually), so I ordered a bunch of parts for it, to apply the fixes from the "mandatory fixes" topic.

First I replaced the CPU to a 68HC000, that worked fine. Then I replaced the bus resistors to 2.2K and 1.2K ones (ordered from exxos) respectively, and this is where the problems started. The machine no longer booted, just black screen. Then I realized, I managed to lift two traces (D8/D9) near resistor pack P104 going towards the CPU, which then I patched on the back side of the board. Which is when I got to the state of non boot. Either I get a black screen in NTSC frequency (I guess the ROM hasn't got to video init state at all), or a PAL screen with white border and garbled output, or an entirely white screen. It's totally random and I could not find any consistency in it. The CPU is halted after this point, so runs into a double fault (HLT pin is low), verified on a scope. Actually tried with both the HC and the original CPU. I also tried multiple RAM modules, both 2MB and 4MB combinations.

I checked through the entire board for more broken traces or signals, on both sides of the resistor packs, but I found nothing. All address/data lines, and all control signals going through P100. I also verified multiple times, that the resistor packs don't have shorted pins to each other, and that all pins have the right resistance towards the common pin. So now I'm out of ideas, and decided to ask here, maybe someone has an idea.

A weird thing, that might be related: I noticed that between ground and +5V VCC across the entire board, I get only a low Ohm resistance, usually in the 50-150 Ohm range, slowly increasing when measured. It which is so low, that my multimeter often just indicates a short, most of the time. I'm not sure if this is normal or not, sadly I have no reference value from before changing the resistor packs.

Side note, but ironically while measuring all this, I found out that A15 was broken on the cartridge port, probably right below the connector (the connector's joint is not broken, it's the line that got corroded probably), which probably resulted my cartridge issues in the first place. (I can trace A15 just fine to the left edge of the board, then in the final section where it goes "up" towards the back of the machine and the cart port, it loses continuity.)

Oh, one thing I forgot to mention before - the PSU has been recapped before and was working, currently set to 4.95 volts. Also the big axial cap has been changed on the board to the set from exxos (this has been a previous round of changes, and was working OK before, sans the issues mentioned). Also, the mainboard doesn't boot with my bench PSU, set to 5V and 3A, but the power draw tops around 1.25A, with the HC CPU, so at least the power draw doesn't indicate a major short. Before the resistor pack changes, the board did boot with the bench PSU as well.

You probably have more broken tracks or vias. It's not simple to diagnose.

Have a look at these 2 threads as they were track damage also.

https://www.exxosforum.co.uk/forum/viewt ... =23&t=4867

https://www.exxosforum.co.uk/forum/viewt ... =18&t=5047

Resistor pack the wrong way round?

Thanks for the replies. I try to be quite methodical, and as I wrote I checked through for lifted pads already (at least 3 times, I was looking at this board for a few days already, before posting), and I verified continuity in both directions of the resistor packs. The STE remake PCB in KiCad is a great help, so I can see where traces are going, so I can verify continuity. It could be I still missed something. In trace D8 and D9, one can clearly see 1-2mm the lifted traces near the pads. There's nothing like that on any of the other traces, not during a visual inspection, also not according to the multimeter.

So my question was more like, what one could verify to make sure the ROM access of the CPU is OK, can I remove some chips to simplify the equation, in case something is pulling the bus the wrong way, etc. I do have a fairly OK RIGOL scope, although I'm not very proficient in using it. But I have the feeling I'm about to be... :)

Resistor packs the wrong way around - not as far as I can tell. If the silkscreen markings are anything to go by. The "dot" on the resistor packs has to line up with the marking on the silk screen? Because that's how they're placed right now. I will try to make some pictures to back up what I wrote, but it might take a bit. Sunday, and family stuff...

Ok, so just to wrap this up - with family Sunday done, I managed to get the machine back to life. Apparently I have some contact issues either with the ROM socket (I have an Exxos Fast TOS adapter with 1.62/2.06), and/or (more likely) the CPU PLCC socket. So I cleaned/readjusted the ROM/CPU socket pins before going forward with the diagnostics, and my STE sprung to life! I will eventually replace both the CPU and the ROM sockets, but for now it's working.

So I proceeded, and also did the 1772 10K pull up resistor mod, which fixed my "drive is write protected" issues. Then I patched the A15 line, so the test cart was running too! I was over the Moon, but then the machine was still bombing kinda randomly, and my NetUSBee still refused to work. Bleh. More precisely, the USB part seemed to totally pass now, but not the network part. Which isn't funny, because that would be my main use case for the thing. So I kept checking lines and just scratching my head over it, then I decided to just run the machine once more from my bench PSU...

And just like that, the NetUSBee's network sprung to life! I haven't touched anything else, just replaced the PSU. So apparently the PSU does something, that the NetUSBee really hates, be it either some ground noise, power noise, or something else. Now, before I order a new PSU online, or decide to make a PicoPSU hack or something, I will also do the reset cap fix, just in case. Maybe it's related. ... (Again, the PSU was already reworked once. So at this point I'm just annoyed, and will probably get rid of it anyway. We'll see.)

Anyway, thanks for the opportunity for some rubber ducking here. :)

:bravo:

Yep the ROM sockets are generally bad on the STE.

IIRC theres been posts about NetUSBee being very fickle for a while now. Nobody knows why or a solution to this. It does seem very sensitive to what voltages it will work properly at. The builders of NetUSBee really need to resolve that problem.

Is it an sr-98 PSU? If so they have a 5v adjustment potentiometer. It could need tuning. Sometimes if mine isn't quite right the Netusbee Usb ports won't work on mine, always seems to be the 5v rail needing adjustment.

That is a really good hint, and it is an SR-98 PSU indeed. However, the NetUSBee problem doesn't seem to be related to the voltage of the PSU, I tried various voltages from 4.5 to 5.3 volts, and it never works with the stock PSU. Sometimes the USB also stops working, but the Ethernet part never works with the stock PSU. However, on my bench PSU it always works, regardless of the voltage (again, tried a range of voltages).

If I can speculate, it might be related to the startup of the PSU. Perhaps the 8019AS reset is somehow completed before the power stabilizes on the cart port, or there are some weird power spikes or bus glitches, and this puts the chip into some weird state. I don't know. Sadly, there's no reset line on the cart port, so any further reset won't affect this initial state of the chip.

The NetUSBee is a very finicky device.

On one of my STs it (ethernet only) will work flawlessly with the unrecapped stock psu but not at all with the Exxos PSU. On the TT I can sometimes read the Ethernet chip ROM but only sometimes and this is consistent during a usage session, but will change the next day.