exxos blog - random goings on

[mrbombermillzy]

@Rustynutt Is that a Sony BVM/PVM in the corner of the first picture?

This one?
Bought this around 2014. Finally getting my cables in order to connect the Falcon.

Wow! That must be a BVM then. Arguably the best 15Khz monitor to use with the Falcon (despite people saying they are too good! lol)

I had around 5 PVM20s. The connections were not quite as plentiful as yours...

P1030235.JPG (65.89 KiB) Viewed 1527 times

P1030293.JPG (94.11 KiB) Viewed 1527 times

Oh and while we are on the subject of 'how messy was your computer room' here's some memories of around 3 years ago:

P1030461.JPG (108.47 KiB) Viewed 1527 times

P1030409.JPG (142.85 KiB) Viewed 1527 times

P1030410.JPG (131.07 KiB) Viewed 1527 times

P1030458.JPG (149.93 KiB) Viewed 1527 times

Although I have slimmed it down now (around 40% Id say). Had to fit the darn kids in somewhere!

[Rustynutt]

Yes, it's a BVM 24".
Main reason it sets where it does is it weighs so damn much!
I bought this from a retired Sony Tech. His names Patrick. It only has 50 hours on it. He got it from a friend who worked on the set of the 1st Avatar film. It was a "spare" unit
He actually loaded it up and drove it 200 miles so not to risk shipping.

[exxos]

Finally a bit of progress with the 1MB 30 pin simms new production run

simm.jpg (34.51 KiB) Viewed 1491 times

This image is from the factory ( I'm not assembling these myself this time).

[JezC]

Finally a bit of progress with the 1MB 30 pin simms new production run

Great, there's an STE to upgrade from 1M to 4M waiting...then the Dual fast ROM board & accelerator will go in it

[Rustynutt]

Finally a bit of progress with the 1MB 30 pin simms new production run


simm.jpg


This image is from the factory ( I'm not assembling these myself this time).

What would a "short" run of 72 pin 32MB EDO 50ns DRAM run?
I could go for 12 minimum, with 2-3 other users probably bumping that quantity to 18.
I've hand sorted 60ns simms, out of around 25, have 2 pair that work at 46MHz, both Micron.

Maybe Deja Vu, seems I've asked this before?

Exxos, haven't seen a package yet from you. Think there's 3 Falcon owners in Oregon, maybe one is the postman?

[exxos]

What would a "short" run of 72 pin 32MB EDO 50ns DRAM run?
I could go for 12 minimum, with 2-3 other users probably bumping that quantity to 18.
I've hand sorted 60ns simms, out of around 25, have 2 pair that work at 46MHz, both Micron.

Maybe Deja Vu, seems I've asked this before?

Exxos, haven't seen a package yet from you. Think there's 3 Falcon owners in Oregon, maybe one is the postman?

It seems to have got lost then along with the one I sent to @Steve who is in the UK like me. So not sure what's going on there .

As for Simms. I'm affraid my PCB routing days are over. I find it hard to even use a PC these days unfortunately.

[exxos]

Cute

IMG_0557.JPG (95.91 KiB) Viewed 1258 times

[exxos]

Been doing some testing with a SD hard drive and my V1 STE 32MHz booster only to run into some bizarre issues that I don't have with ultrasatan.

It has been driving me nuts for the past few days. But long story short it seems that the RW signal after a DMA cycle "takes ages" to go high again.

YELLOW = RW
BLUE = BGACK

rwbgack.png (5.89 KiB) Viewed 1181 times

Then with 1K on RW (it has the 1.2K resistor pack so in total its more like 600R) This is the RP100 pullup thing again.

rw2.png (6.34 KiB) Viewed 1181 times

I referenced with BGACK because I am using it as the switch between 8/32MHz. It only seemed to work with RW in that equation which is why I started looking into that signal more. I think with the speed of the CPU running at 32MHz is just causing a clash with RW somehow screwing DMA access up. Which does not make much sense by itself. But the "problem" is that ultrasatan does not seem to be effected by this issue.

With just a original CPU.. Without and with the pull-up resistor..

rw3.png (5.71 KiB) Viewed 1181 times

rw4.png (6.14 KiB) Viewed 1181 times

The rise time is still relatively long but it does not have the "bootleg" type curve to it like it does with the HC CPU.

I think this problem will only manifest itself with a 32MHz booster. I shouldn't imagine it would ever be a problem on a stock machine with a stock or HC CPU anyway. Though I may release a note that this problem exists and may cause malfunctions with some hard drives. The only other alternative is to release different firmware which ones fractionally slower in the booster. Though it is still a mystery why the some hard drives behave differently in the first place.

Anyway, I thought I would just dump this info here because I will probably forget by tomorrow anyway , and it might be important

[exxos]

Another oddity is that the RW signal ends up delayed going to the DMA port because of that weird latch buffering thing going on with the STE. It seems rather erratic. Write can be delayed by about 184ns to 276ns.

I wonder if this could be potential issue with write failures on drives But probably not that simple because a lot of the other signals are also delayed.

Capture.PNG (245.11 KiB) Viewed 1134 times

I was mostly trying to see if there was any correlation between the RW times on the CPU vs the DMA port.

[exxos]

I think I figured out what is going on with the DMA write corruption. At least for the most part.

The HC CPU in the STE works fine as normal. However in this particular machine when the booster is added, even when running at 8MHz, it causes write failures on the hard drive.

Basically when copying my test floppy to the C: drive, it would intermittently lock up with " data on the drive may be damaged" . This is the hard drive not the floppy drive! It would consistently do this pretty much exactly the same files every time. There was also odd read problems as well in the mix.

I remember having no end of trouble with the clocks on the STE in early booster work. In that when the original CPU is in there, it just loads the clock slightly different enough for things to work properly.

The STE I am testing is the one with the internal blitter with a -38 DMA. I also remember documenting the clocks with different amounts of noise on it between the internal and external blitters. This resulted in lots of weird and wonderful mods originally on the STE boosters clock depending on motherboard revision. Looking back it seems the external blitter and more problematic with the clock than the internal blitter motherboards.

So the clock on the DMA is basically this..

IMG_0576.JPG (64.23 KiB) Viewed 1148 times

8.5V P-P! No wonder the poor DMA is unhappy!


So I did some hackery to add a 100R resistor in series with the DMA clock line..

IMG_0579.JPG (317.41 KiB) Viewed 1148 times

IMG_0578.JPG (60.03 KiB) Viewed 1148 times

And now the hard drive is a lot more stable and there does not appear to be any more write issues.

So while the HC CPU tends to solve one set of DMA related problems, it is perfectly possible it could actually create problems on the DMA. However this is not limited to simply swapping to the HC CPU either. it just as a more dramatic effect as opposed to changing between brands of NMOS CPU.. which is also generally enough to make or break "DMA problems".