derkom's doings

[exxos]

Here you go.

Meant the booster That bodge on cap goes to a resistor on the right of the board, what value is it ? I think it needs to be 68R for the integrated blitter... I can't check right now, but I think the cap might have to be removed as well.

[derkom]

Meant the booster That bodge on cap goes to a resistor on the right of the board, what value is it ? I think it needs to be 68R for the integrated blitter... I can't check right now, but I think the cap might have to be removed as well.

"oh"

IMG_20200608_212238.jpg (76.46 KiB) Viewed 3661 times

Looks like a 470 Ω to me. I'm certainly in no hurry here. Can wait until you've got solid information.

[PhilC]

@derkom @exxos sorry guys, real life getting in the way again which is annoying.

I'm hoping to do the reliability test tomorrow. I'm sure I have a board like yours Derkom, so will report back then.

[exxos]

Looks like a 470 Ω to me. I'm certainly in no hurry here. Can wait until you've got solid information.

Remove the cap and try 68R instead of the 470R. The 8MHz clock goes from one extreme to another depending on internal or external blitter

If the GAL has a couple spare ios ( doubt it) I may try down clocking the 32MHz to 8mhz and see if that works better as the 8MHz clock is just so bad on the STEs

[derkom]

This is definitely turning into a "two steps forward, one step back" project.

Switching out the 470R for a 68R rendered the system completely stable at 8 MHz. I realise in looking back at my posts that I'd reported it stable at 8 MHz before, but that turned out not to be true, as it did typically end up crashing after a few hours. With the resistor swap, and cap removal, the system ran through 180 GEMBench cycles and 81 YAART cycles with zero problems. Also passed diag software with flying colours (which are not red).

However, there was no change at all to the 32 MHz situation, where lately I haven't been getting a successful boot at all, and the diag software keeps reporting RAM/data line errors. But clearly the data lines are good, or this stuff wouldn't work at 8 MHz.

To see if it would improve things any, I decided to run my clock line through a via down to the underside of the board and tap the buffer chip from the bottom, but no, that didn't help, and in fact at that point I lost my ability to run reliably at 8 MHz.

Fast forward through a lot of trial and error, before I finally noticed that the extension cable I built for the PSU was quite warm. Checked VCC on the 1772 (just happened to be a handy spot) and found it to be 4.2 V. Well, that could explain a lot of my problems! I pulled out the extension cable, VCC went up to 5.2 V, and lo and behold, it boots at both 8 MHz and 32 MHz.

But...

I have very shakey video, often blanking, and it seems to be related to the presence of the 32 MHz wire (whether or not I am running the system at 32 MHz). When I removed the wire completely (and ran at 8 of course), video is fine. Install the 32 MHz wire and the video becomes unusable. I have resoldered the wire several times, and tried several routings (yes, avoiding video circuitry), and I'm always getting shaky video, and occasional reboots associated with some of the shakier moments.

I've probably overworked this wire anyway, with all the repeated reroutings and resolderings, so tomorrow I'll try installing a new wire. Maybe it would even help to use twisted pair with a ground?

[exxos]

@Steve sent me this image the other day for bests routing.. you also have to use the thin cable as well..

32mhz line.jpg (603.23 KiB) Viewed 3617 times

Though routing the wire near simms only normally causes problems with parity simms.

[derkom]

@Steve sent me this image the other day for bests routing.. you also have to use the thin cable as well..

Okay, I was wondering about going around the other side of the shifter like that, but unfortunately my current wire isn't long enough. Will attack the problem again tomorrow, with that routing in mind.

[exxos]

Okay, I was wondering about going around the other side of the shifter like that, but unfortunately my current wire isn't long enough. Will attack the problem again tomorrow, with that routing in mind.

I route the wire to the right of the socket. But generally I don't have any issues with the routing anyway.

[derkom]

I route the wire to the right of the socket. But generally I don't have any issues with the routing anyway.

And I didn't have problems initially. Back when I posted that first GEMBench screenshot, the video was rock solid, but the system eventually crashed, which I can probably attribute to the low voltage. I don't really know what would have changed except that I've soldered, resoldered, unsoldered, rerouted, soldered, unsoldered...

Edit: Just switched out the SIMMs for a 4 MB set (8 chips per SIMM, definitely non-parity) and at first the video appeared more stable, but then when I switched to medium res, I got artifacts on the desktop, and then the system froze when I tried to run GEMBench.

[exxos]

have you tried without a long lead on the psu ? You would probably have to replace the 4700uf cap on the STE as the regulation will be pretty bad..