exxos's Atari, Amiga & retro forum

@rubber_jonnie Still can't get this PAK to work :roll:

Been experimenting with the "Z-seam" issue... seems cooling is the only thing which helps with this.. I had been messing around with the retract settings all the time and that does not seem to do anything :roll:

In any case, the one on the left was the first one without any fans running, the one on the right was with fans running...

exxos wrote: 05 Jun 2020 19:21 @rubber_jonnie Still can't get this PAK to work :roll:


IMG_5266.JPG


IMG_5267.JPG

Guess I'd better try mine.

I've been looking at these SMT stencil machines...

www.amazon.com/product-FRAMELESS-Stenci ... B07WT112Z7


Feking expensive :roll: There is a UK firm who do a similar thing, but its getting to the price of getting a small SMT paste printer :roll: There are some cheaper things, but they look pretty terrible in all respects.

So I am thinking of cannibalizing my old I3 printer for parts and making my own version. Or if bars and bearings are cheap, I will just order a set. I can 3D print various parts I need to make the thing.

This came about as I need to paste 75 PSU boards. Putting the SMT parts on is bad enough without messing about with pasting of it all. It kinda goes back to me needing a pick and place machine again as well :roll:

I started designing my own jig a few days ago and just printed out the first test parts.

The little dots are supports for the "overhang" parts of the bars. I'm surprised how well these worked and how easily they come off!

I managed to order some bars and stuff ( as used on 3-D printers) yesterday as well, so I am hoping they will arrive tomorrow so I can see better how this will ultimately all fit together.

I've been experimenting with the STE booster code again. Basically bypassing the 8mhz signal as its so far seems to be causing various odd issues. So the 32mhz gets down clocked to 8mhz and that is used instead.

But geese. Could I get the clock divider to work. Its a PITA in a gal as there is only 1 global clock input. Even so, long story short, for some bizzare reason the reset line which switches the clock into 8MHZ during reset was causing the gal generated 8mhz clock to not work at all!? Even more odd, if I output on a spare in pin , it works , but when I connect to back into the 8mhz gal input, it dies again. The MB clock is isolated so there is no short. So I removed the reset signal from a different unrelated part of the code, and it now works fine. I may fiddle with complier optimisations tomorrow as something is well messes up somewhere.

Anyway, the booster is now using a 8mhz gal generated clock.the tests in GB6 have changed slightly. Integer division has gone up from 355% to 364%, and some other tests have dropped a few percent. Its probably to be expected as there would be a slight phase shift in the 8mhz clock so it would explain the slight differences in the scores.

I may send this Jed to @PhilC to try out as well..

Just a quick post to show the original STE booster using the 8MHz clock, and the second screenshot is using 32MHz / 4 to generate 8MHz.

Noticeably integer division has gone from 355% to 364%... But on most other tests the schools are generally around 1% lower. Nothing wrong with this of course, it is just how that slight clock skew is effecting the results.

As said earlier, test was to stop using the buggy 8MHz system clock.


8MHz system clock (orignal)

32MHz / 4 8Mhz generated (test firmware)

GB6 just ticked over 100 passes. This is with 68R on board with 100R in series with the wire. So 168R seems to work well on my board. I think it really depends on tolerances of all the hardware and the GAL.

exxos wrote: 12 Jun 2020 22:22 GB6 just ticked over 100 passes. This is with 68R on board with 100R in series with the wire. So 168R seems to work well on my board. I think it really depends on tolerances of all the hardware and the GAL.

We are almost in sync. My GB6 just started pass 99, with only the 68R on the board. Might as well leave it running overnight just to see. :)

derkom wrote: 12 Jun 2020 22:47 We are almost in sync. My GB6 just started pass 99, with only the 68R on the board. Might as well leave it running overnight just to see. :)

68R worked fine with every other booster, just not this one on test currently. :shrug: