exxos blog - random goings on

[stephen_usher]

It's well known that the UltraSATAN pushes the boundaries of timing beyond the specification to increase performance.

[exxos]

Removed the "Good DMA", fitted a socket, put the "Bad DMA" in in, and no change

IMG_0964.JPG (462.45 KiB) Viewed 1265 times

Though there may be some light at the end of the tunnel..

I noticed that this machine has a IMP blitter in it, and we all know the history of IMP don't we ? So I took it out and plugged in a "normal" National blitter. The blitter itself is actually turned off on the desktop anyway. I have not noticed any problems between having it on or off in all my tests. the Wadley got down to 1861 files left to delete before it crashed. This time with 11 bombs.

IMG_0965.JPG (87.49 KiB) Viewed 1265 times

Though I seem to have developed a odd problem now since changing the blitter...I don't know if this blitter is good or there is a bad connection yet..

IMG_0966.JPG (48.55 KiB) Viewed 1265 times

IMG_0967.JPG (33.29 KiB) Viewed 1265 times

Putting the IMP blitter in is fine again. So found another National blitter and plugged that in. Now fine.

So I will redo my SD card and repeat the tests again.

[exxos]

Well well well..

IMG_0968.JPG (78.46 KiB) Viewed 1254 times

So it seems the IMP blitter is causing issues. I know I have had all sorts of clock problems on the STE. So possible that and/or the bus grant signals are a bit wonky somewhere. I will get the scope on it see if I can see anything amiss.

[exxos]

I scoped out all the busgrant and clock so far..

My notes on pins tested.

Code: Select all

NATIONAL - IMP

0-6 8mhz (34)
1-7 BR (30)
2-8 BG (28)
3-9 BGI (37)
4-10 BGACK (29)
5-11 BGACK (40)

0-6.png (14.9 KiB) Viewed 1248 times

1-7.png (12.97 KiB) Viewed 1248 times

2-8.png (13.02 KiB) Viewed 1248 times

3-9.png (12.34 KiB) Viewed 1248 times

4-10.png (12.4 KiB) Viewed 1248 times

5-11.png (12.93 KiB) Viewed 1248 times

There is a fair bit of noise on both BR low voltages. I am pushed really see much difference in anything else other than the spike negative on IMP BGACK with low noise also.

BR even has 1.2K resistor on it and still takes like 200ns to pull high

EDIT:

I've dumped X1 probe on pin 40 (BGACK) and will just let it run see if anything changes. Signal looks mint at the moment!

IMG_0969.JPG (380.1 KiB) Viewed 1246 times

[exxos]

So with the scope probe on X1 it worked. So to sanity check, I removed the probe, copied all my files back to C:, re-tested, and it's still bloody working!!

So basically nothing has changed other than "re-seating" the blitter

I will do another full test to make sure. I can't believe all these problems were simply a bad connection on the blitter!

[exxos]

Plugged the pico drive in and it crashed. So that noise is still a possibility...

I've put the national blitter back in to continue deleting my test files as I had been using ultrasatan before.

[stephen_usher]

If you look at the second set of traces from your ‘scope. The second one goes low far earlier than the first.

Is that the bus request line?

[exxos]

If you look at the second set of traces from your ‘scope. The second one goes low far earlier than the first.

Is that the bus request line?

It's BR but I wouldn't read into timings to much as I was having to zoom in and out signals to capture them for best image.

[exxos]

Test passed with the Pico drive with the National blitter. So its the IMP blitter which is not happy. So I will chuck some small caps on those 2 noisey lines and try again with the IMP blitter.

[exxos]

33pF bodge caps added. Will try my tests again with the IMP DMA, see if it makes any odds.

IMG_0975.JPG (406.51 KiB) Viewed 1230 times

NewFile12.png (6.49 KiB) Viewed 1230 times

NewFile13.png (6.28 KiB) Viewed 1230 times

EDIT:

First pass crossing the finishing line!

IMG_0976.JPG (55.53 KiB) Viewed 1229 times

I have also ordered a X100 probe. I am not overly convinced that a X10 probe isn't affecting the signals somewhat.