Project: HDMI/DVI out for STFM

[rubber_jonnie]

DVI to HDMI adapter just arrived, which will give me additional testing options.

[rubber_jonnie]

Ok, so I quickly ran some tests in low/med/hi res on a 1040STFM running TOS 1.02 (I may have mistakenly said 1.04 earlier) including Blitter, with the DVI board set to jumper B and no vsync cable, and the outcome was as follows:

1. Throughout the tests, even when the screen was blank, the LED on the HDMI adaptor remained on at all resolutions. At no time did it extinguish.

2. No matter what resolution, a blank black screen could be provoked by simply moving the mouse pointer between the TOS pull down menus.

3. At low/med resolution, the screen would blank and become all black, and come back as it was after a second or two.

4. At hi res, the screen would sometimes come back ok, and sometimes would have vertical white bars and what appeared to be shifted columns. It wouldn't recover, but the machine was not hung, and menus were available, and the mouse pointer would move. Changing between the TOS pull down menus would still cause the blanking.

Whilst navigating the menus I noted that the Blitter was enabled, so wondered what would happen if I turned it off, and this is where it gets interesting.

With the Blitter turned off, I was unable to reproduce the "blanking" at any of the screen resolutions. They were all 100% fine.

What was also interesting was, that when I ran a short Blitter test using the diagnostic cartridge, the test caused a blank black screen and when the display came back to normal, the test was marked as failed.

I then tried the long Blitter test and it passed no problem.

It would seem then that the problem is Blitter related. However I will reserve judgement until I have tested with another couple of STs, both with and without Blitter.

I'm also trying to get a vid uploaded, but it is taking an age to upload from my phone for some reason, but it will show the basic TOS low/med/hi res tests with and without Blitter. I still need to get a vid of the diagnostic cart tests, but right now, sleep is looking like the more appealing option!

[Smonson]

That's very interesting and strange, because the blitter doesn't interact with the shifter. And surely nothing here can tell where the mouse pointer is... right!? Unless moving the mouse to a certain spot causes the ST or its associated wires to move. Question: does it still happen when you use the keyboard to move the pointer with ALT+arrows?

I'm theorising that a brief loss of sync, not long enough to see on the LEDs, could cause the monitor to go blank for a couple of seconds while it searches for a stable sync again.

My only other theory right now is that power supply variations could cause a problem, some issue that is not present/as bad when the blitter is disabled.

[Smonson]

Another question (which is not as easily answered): if you had an original shifter installed, would it still fail the blitter short test?

[Icky]

Ok, so I quickly ran some tests in low/med/hi res on a 1040STFM running TOS 1.02 (I may have mistakenly said 1.04 earlier) including Blitter, with the DVI board set to jumper B and no vsync cable, and the outcome was as follows:

RJ do you get the same responses with the vsync connected?

When I get time tonight I will run the same tests to see if I can reproduce.

[rubber_jonnie]

So in answer to the questions you've raised:

1. Unless moving the mouse to a certain spot causes the ST or its associated wires to move. Question: does it still happen when you use the keyboard to move the pointer with ALT+arrows? The ST doesn't move when I'm moving the mouse pointer, the only cable that does move is that of the mouse.. I will try ALT+Arrow keys and report back.

2. I'm theorising that a brief loss of sync, not long enough to see on the LEDs, could cause the monitor to go blank for a couple of seconds while it searches for a stable sync again.Whatever happens, it's faster than I can see, the LED doesn't even appear to blink.

3. My only other theory right now is that power supply variations could cause a problem, some issue that is not present/as bad when the blitter is disabled.The power supply is good, recapped with one of exxos' kits.

4. Another question (which is not as easily answered): if you had an original shifter installed, would it still fail the blitter short test? I couldn't say, but it is easy enough to test. I'll let you know the outcome.

5. RJ do you get the same responses with the vsync connected? I've not yet tried with vsync connected, I will do and will report back.

In addition to the above, I will try other boards and see what the result is. I have a mix of blitters installed by me and at the factory.

Incidentally, all of my PSUs bar one, have been recapped. The machine that has that PSU won't be a part of the testing because it doesn't have a blitter.

Here's a link to the video, which includes testing at all resolutions, with and without the blitter.

https://youtu.be/JbsaOFb2zgE

I didn't get a video of the diag cart tests as I thought to try it later on, and forgot to take my phone with me when I went back upstairs. I'll get a video of that at some pint and upload it.

[exxos]

I was going to suggest trying the remake board

Mine has the blitter and I did not see any issues. I think I only tried with Vsync wire connected though.

It may be you need to change the address and databus pullups to 2.2K. With the bliter enabled, the extra load on the bus could be causing it to trip up and the shifter issues are the outcome...

I guess you could try running YAART to see if the RAM itself trips up but may not be a fair test..

[rubber_jonnie]

I was going to suggest trying the remake board

Can't do that until I finish the board

In the meantime I will try some other STFMs, provided I can find one or two with the Shifter that is socketed.

I'll also try with the vsync connected. Which pin of the Glue does this need to be connected to?

[Smonson]

Great video - thanks for going to the trouble to post it. Those vertical bars are definitely not data line issues - that's the so-called "every other 16 pixels background" problem which manifests as a blank bar of 16 pixels every 80 pixels in mono. It's purely to do with the timing of when data is loaded into the shifter, compared with the shifter's internal counter. And it probably means the shifter misread one of the signals lines for a moment: load, rw, cs, or de.

With the bliter enabled, the extra load on the bus could be causing it to trip up and the shifter issues are the outcome...

Agreed! By the way, I didn't necessarily mean a problem with your power supply, but maybe the power supply just to the mod board - it draws a lot of current through just one pin each of the shifter socket, 5v and GND, which is generally not a great idea electrically, but very convenient... Do you have a machined-pin IC socket for your shifter, or the normal cheaper type?

[exxos]

Agreed! By the way, I didn't necessarily mean a problem with your power supply, but maybe the power supply just to the mod board - it draws a lot of current through just one pin each of the shifter socket, 5v and GND, which is generally not a great idea electrically, but very convenient... Do you have a machined-pin IC socket for your shifter, or the normal cheaper type?

All my sockets 40 pins are the "cheaper" type. Though to be fair, I have had a lot less trouble with those than the turned pin ones.

If its a possible "miss-read" on a line, if RJ places his scope on x1 on each shifter pin one by one, normally its enough to solve such issues, assuming its a noise related issue, which 99% of the time things are.