Image instability with STE booster and system memory over 512k

[DementED]

Ok, for reference here are my two 32Mhz STE boards:

First is the soldered in booster and it's the same CA4003290 board and Shifter as @DementED :


PXL_20231026_104537291.jpg


Second is the push fit booster, also a CA4003290 but a C300588-001 Shifter:


PXL_20231026_105307738.jpg


Both run fine at 32Mhz, and my 32Mhz pickup wire is taking a more direct route.

Now it's difficult to tell from your picture, but it looks like your feed from U405 is going directly to the TOS switcher. Can you confirm it's going to the booster board first please and then from IIRC the CE tap on the booster? If it goes direct to the TOS switcher, that's incorrect it should go to the booster, then that feeds the TOS switcher, but as I said, it's hard to tell from your image.

I did zoom all the way in but I couldn't tell properly, as it got a bit blurry.

Yes, the feed from U405 goes to the booster board first and then from IIRC to the CE tap on the TOS switcher.

[DementED]

@DementED

It might be a long shot, but having stuff to try out that isn't too difficult to implement is a good thing right? The creators of the TwiSTer STe booster created this guide for installing some stability fixes for the STe, to make it stable with their booster:

https://www.newtosworld.de/download/file.php?id=2942

I don't think there is any harm in trying them out. I put them on my STe which has the 32mhz Exxos booster and it works fine.

Thank you. Unfortunately it would not let me view that page. I tried registering with the site, but it kept rejecting me. Will keep trying!

[Steve]

Not sure why? It just downloads a pdf for me. Maybe you've got some security setting that stops pdfs loading.

Here I uploaded it for you:

STE-Kombi-manual-engl.pdf

(2.86 MiB) Downloaded 79 times

[rubber_jonnie]

Yes, the feed from U405 goes to the booster board first and then from IIRC to the CE tap on the TOS switcher.

OK, cool

[exxos]

Not sure why? It just downloads a pdf for me. Maybe you've got some security setting that stops pdfs loading.

Here I uploaded it for you:

I had a quick look at it but it was frankly painful to read with a lot of words cut in half due to line wrapping. While there are grounding problems, it's well known, but single wires like suggested does nothing for high frequency signals. For DC levels sure, but once you get in high frequency you are dealing with skin effect which needs surface area to get the impedance down. A large area of badly routed tracks will still have less impeedence than a single wire. The only way it would work is possibly with litz wire. That has multiple thin isolated strands which act like multiple inductors in parallel and with the increased surface area it may help. But overall, keeping the shielding in place helps with some problems as I've already documented before.

Then there is the usual noise problems with signals. Looks like a cap was added to BG. But noise is well documented on a lot of signals. Adding more caps in some cases on power rails can make ringing worse due to higher peak currents through the signals and grounds. So you have to be really careful when randomly adding capacitors that your not making the problems worse. You can end up in situations where adding caps increases currents , where then you have to combat that modification by fortifying the grounds.

[rubber_jonnie]

@rubber_jonnie at some point can you please try this program ? https://github.com/fenarinarsa/ste_dmatest/releases/ May only work in 8MHz mode.

OK, so I ran the test on my 2 STEs and my MSTE, in 8Mhz & 32Mhz on the STEs, and 8Mhz and 16Mhz on my MSTE, though I didn't bother with taking a pic of anything other than 8Mhz on any machine, though the higher speeds ran just fine.

This is the machine with the soldered in booster and same Shifter as the OP:

PXL_20231027_110221031.jpg (264.42 KiB) Viewed 2550 times

And this is the other machine with the different Shifter and plug in booster:

PXL_20231027_110552973.jpg (252.21 KiB) Viewed 2550 times

And this is the MSTE:

PXL_20231027_115501114.jpg (250.73 KiB) Viewed 2550 times

As you can see they all look pretty much the same. In addition, I heard no pops or crackles through my headphones during testing,

[exxos]

And this is the other machine with the different Shifter and plug in booster:

So that's the 588 shifter ? Odd its behaving the same as the 1712 shifter in that case... What revision board is it ?

[rubber_jonnie]

And this is the other machine with the different Shifter and plug in booster:

So that's the 588 shifter ? Odd its behaving the same as the 1712 shifter in that case... What revision board is it ?

Both boards are the same type, but there are no revision markings (My board pics are earlier in this thread) on either. A quick comparison between the boards and they look the same.
However the board with the soldered in booster and 712 shifter has a C302183 at U400, which if IIRC has MMU, GLUE and Blitter integrated into one, so U101 is an unpopulated space on the board,

The second board with the plug in booster and 588 shifter has a C300589 at U400 and U101 has a socket with Blitter chip, an MM9092V.

Other than that I am reasonably confident they are based on the same PCB, the CA4003290.

[exxos]

The second board with the plug in booster and 588 shifter has a C300589 at U400 and U101 has a socket with Blitter chip, an MM9092V.

It does pose a bit of a curve-ball if the integrated GST is a factor in all this as well as the shifter

[rubber_jonnie]

The second board with the plug in booster and 588 shifter has a C300589 at U400 and U101 has a socket with Blitter chip, an MM9092V.

It does pose a bit of a curve-ball if the integrated GST is a factor in all this as well as the shifter

Well possibly, but both of my boosters run just fine with 4MB fitted at 8 or 32Mhz, so what is different with the OPs machine?