exxos blog - random goings on

[Icky]

EDIT:

I gave the MB a hard calibration, and now only getting this... so must be bad connection somewhere...

So how far from the floor did you drop it

[exxos]

So how far from the floor did you drop it

Dropping stuff is for girls.. smacking it with your first is a mans calibration..

Oddly it just started working fine again

[stephen_usher]

So how far from the floor did you drop it

Dropping stuff is for girls.. smacking it with your first is a mans calibration..

Oddly it just started working fine again

Percussive maintenance is the key?

[exxos]

Percussive maintenance is the key?

Must be!

[exxos]

Its still being a pain for some odd reason... I did check the shifter output vs the 32mhz and wonder if that is the issue.. as we basically are inverted. The original 16Mhz mod had a inverter in there... so maybe that is the issue

00.png (8.59 KiB) Viewed 4801 times

000.png (8.28 KiB) Viewed 4801 times

[exxos]

OK so it didn't really make any difference adding delays on the 32MHz line to the MMU. I had to 90 and 180deg out and doesn't seem to change anything. Though with it being a sod to reset, its hard to really tell 100%.

Been messing with this for ages.. its odd because it seems to go from a row of flickering bombs, to going to the desktop and refusing to access the floppy.. to working perfectly...

I mean I ran GB6 all tests fine.. Powered off for 10 seconds, powered back on, and the floppy drive isn't working again

I'm really wondering if theres a bad connection on one of the PLCC chips somewhere but cant find one

[exxos]

I have re-wired the mod board on the motherboard, and put jumper links and cables in , so I can quickly remove the mod board and resort back to stock speeds..

My odd issues are persisting though even that stock speeds I am back to the symptoms that I do not get any video other than a white screen so something is wrong with my board somewhere, but at least thankfully it is not relating to the speed boost board...

IMG_4300.JPG (299.56 KiB) Viewed 4774 times

[exxos]

So I have at last found out why the shifter isn't doing anything intermittent times.. I suspected that either DE and 2MHz was going out of whack, one of them has to be for the AND gate not to work and enable shifter data..

So measuring the mod board 2MHz to the GLUE 2MHz... And there is indeed different wakeup states but actually on the clock itself


The top waveform is the 2MHz clock which is generated from the 161.. The bottom waveform is the GLUE's 2MHz output..

The images below are when the video is working...

wok.png (9.37 KiB) Viewed 4762 times

wok2.png (9.43 KiB) Viewed 4762 times

wok3.png (9.41 KiB) Viewed 4762 times

This image is when the video is not working..

wxxx.png (9.39 KiB) Viewed 4762 times

So indeed the 2MHz output from the GLUE shifts in phase as is not a direct clock down from the 8MHz line

[exxos]

8MHz goes out of whack as well


Video working...

xok.png (9.29 KiB) Viewed 4756 times

Video not working...

xf.png (9.38 KiB) Viewed 4756 times

I am assuming that there are some internal FF's in the MMU doing the clock divisions which start up at the random state, I know there are "wake up states" relating to the shifter etc but this is a PITA now

[ijor]

I am assuming that there are some internal FF's in the MMU doing the clock divisions which start up at the random state, I know there are "wake up states" relating to the shifter etc but this is a PITA now

Correct. There are clock divisions at SHIFTER, MMU & GLUE, and all of them involve wake states. I told you in the other thread that if you bypass the chipset clock division you will create some subtle wakeup issues.

So indeed the 2MHz output from the GLUE shifts in phase as is not a direct clock down from the 8MHz line

It is, of course, derived from the 8 MHz. But the initial phase is non deterministic. And the relation with this clock and MMU RAM process, is exactly what produces the GLUE/MMU wake states.