General Stuff(tm)

[stephen_usher]

All the PiStorm stuff is on hold today as it's server upgrade time.

I didn't notice that the distro release I am using went out of support three weeks ago so I won't get the PHP fix. So, time to bite the bullet and do an upgrade to the next version. Full back-up to external disk first though, then I have the recreate the upgrade script I used to use at work. *sigh*

Oh, the new PiStorm board arrived yesterday and... it has similar issues to my original one.. seemingly too slow.

[rubber_jonnie]

All the PiStorm stuff is on hold today as it's server upgrade time.

I didn't notice that the distro release I am using went out of support three weeks ago so I won't get the PHP fix. So, time to bite the bullet and do an upgrade to the next version. Full back-up to external disk first though, then I have the recreate the upgrade script I used to use at work. *sigh*

Oh, the new PiStorm board arrived yesterday and... it has similar issues to my original one.. seemingly too slow.

Good luck with the upgrade.

[stephen_usher]

Good luck with the upgrade.

Thanks!

[stephen_usher]

And finished...

That was "fun". I had to use a few symlinks to point to new library versions for some of my local executables. I can rebuild them later. Fail2ban refused to run with a cryptic error until I found that there was no default ban action, which didn't seem to worry the previous version. Also somehow the mysql database got corrupted even though I'd shut down the server before I did the back-up or anything. Thankfully I have a nightly dump so I could start again with a new database and reload the dump.

*phew*

It's working well enough now... and PHP is up to date with a version built yesterday.

[stephen_usher]

Re: PiStorm

Well, well, well.... I took my two PiStorm boards over to @dad664npc and he tested them out. They worked fine for him.

This evening I got some time to play with the new one, which was giving fewer memory test errors.

When I put my finger on a specific area of the board next to the CPLD near CPU pins 15 and 16 the number of memory errors dropped massively. If I wet my finger they would almost go away entirely. Seeing as R4 wasn't installed I wondered if that was the issue, so I followed that red herring for a bit.

Then I put my probe onto the non-ground pad of the R4 position, which gave me the 8MHz clock. I happened to run the memory test whilst I had the probe on there and... no memory errors. I then tried the same on pin15 (CLK) and again, no errors.

So, starting at the smallest ceramic cap I had I went up until finally no more errors, 47pF.

IMG_5481.jpeg (231.07 KiB) Viewed 27 times

Looking at the system clock on the motherboard I could see a harmonic. It only had a small amplitude but it made the rising edge of the clock saw toothed. The 47pF cap removed this.

It seems that some PiStorm board CPLDs, but not all of them, saw the saw toothed rising pulse as an extra, very short, CPU clock pulse which upset the timing.

I may try adding another 47pF cap on the motherboard connection on my relocator board to try filtering at source as it should help all CPUs.

[exxos]

I've had to do basically the same on on my stuff as well because of basically noise and bad ringing on the clocks etc. It all drove me nuts on my STE boosters !

May just need some series resistance rather than a cap ? I guess it depends how fussy other things are on the clock as well.. It all ends up being trial and error at the end of the day.

[stephen_usher]

I've had to do basically the same on on my stuff as well because of basically noise and bad ringing on the clocks etc. It all drove me nuts on my STE boosters !

May just need some series resistance rather than a cap ? I guess it depends how fussy other things are on the clock as well.. It all ends up being trial and error at the end of the day.

It looked to be a harmonic, so a cap to filter it is the most appropriate. There's a small amount of ringing but it's the edges which are important.

[exxos]

It looked to be a harmonic, so a cap to filter it is the most appropriate. There's a small amount of ringing but it's the edges which are important.

Mine was on a STE with some values.
viewtopic.php?t=20#p41514

It changed depending on if the machine had a internal or external blitter ! So looks like I settled on 47pF eventually as well

[PhilC]

Any chance of posting a pic of the waveform before and after please?

[stephen_usher]

Any chance of posting a pic of the waveform before and after please?

I was going to try moving the cap this morning so I'll see what I can do.