PAK 68/3 in STE

[Badwolf]

If @Badwolf has some clues. it would be great as his adapter is really good with other configurations.

Sorry, I've only had half an eye on this thread.

Have I got that right that Exxos' adapter (with 4k7 pull ups) works but mine (with 3k3 pull ups) doesn't?

Obvious differences would be those pull ups (measure the combined resistance of any address or data line to VCC with the machine off). Perhaps the main board is one of those with the lower initial value (rather than 10k)?

Next one would be the clock line -- Exxos' adapter has a Schmitt trigger on the clock line IIRC. I forwent this to reduce clock skew, but it could be the path is a bit complicated and that's causing a cumulation of reflections and other messiness. Scoping the clock line as it enters the PAK & comparing the two would be the only way to look for that.

Last one would be power lines -- perhaps my adapter can't pass through enough power and Exxos' can? Would seem the least likely as I think both are four layer boards with a dedicated VCC and GND layer, but easy to measure: what's VCC on the PAK with both installed and running?

Cheers,

BW

[exxos]

If @Badwolf has some clues. it would be great as his adapter is really good with other configurations.

Sorry, I've only had half an eye on this thread.

Have I got that right that Exxos' adapter (with 4k7 pull ups) works but mine (with 3k3 pull ups) doesn't?

I am not following many threads currently..

But if it is 3.3K AND the original (which can be 4.7K) then in parallel they will give 2K which means its in untested waters. I know going lower starts to cause problems at some point. For example, IIRC, 1.2K causes issues.

The schmitt buffer is there because there is bad ringing on the 8Mhz clock. It causes glitching with my accelerators. I added it onto the ST536 because of such issues. But of course I cannot speak for anyone else's accelerators.

[HigashiJun]

Have I got that right that Exxos' adapter (with 4k7 pull ups) works but mine (with 3k3 pull ups) doesn't?

Yes, both adapters (yours and exxos' one) work great with other expansions and exxos' one also works with my PAK68 while yours doesn't seem to like it...

Motherboard SILs are as per exxos' recommendation, i.e. 2.2k, so I have a total of 1.32k with your adapter and 1.50k with exxos' one.

I will try to solder 4.7k SILs on your stuff and see what the result is.

Cheers.

[HigashiJun]

But if it is 3.3K AND the original (which can be 4.7K) then in parallel they will give 2K which means its in untested waters. I know going lower starts to cause problems at some point. For example, IIRC, 1.2K causes issues.

So 1.32k might be to low for the PAK, but everything seems to be rock solid at 1.50k.

[Badwolf]

Yes, both adapters (yours and exxos' one) work great with other expansions and exxos' one also works with my PAK68 while yours doesn't seem to like it...
Motherboard SILs are as per exxos' recommendation, i.e. 2.2k, so I have a total of 1.32k with your adapter and 1.50k with exxos' one.
I will try to solder 4.7k SILs on your stuff and see what the result is.

Oh gosh no! If your motherboard resistors are already 2k2, don't put any SIL resistor packs on the adapter board.

The packs on the adapter are for motherboards with the original 10k pull-ups still in place.

BW

[HigashiJun]

Oh gosh no! If your motherboard resistors are already 2k2, don't put any SIL resistor packs on the adapter board.

The packs on the adapter are for motherboards with the original 10k pull-ups still in place.

BW

Yes, i know (it's even written on your adapter PCB).

But as it works with exxos' adapter, I wanted to give it a try.

My first version of your adapter has no SILs at all and it doesn'r work either, so that's why I'd like to test with 4.7k.

[Badwolf]

But as it works with exxos' adapter, I wanted to give it a try.

My first version of your adapter has no SILs at all and it doesn'r work either, so that's why I'd like to test with 4.7k.

Ah, that's interesting.

I'll wager it's the clock line, then. The Schmitt trigger is probably cleaning it up for the PAK -- perhaps it has a high load or is of a slightly sensitive disposition.

BW

[Rustynutt]

@Darklord

Yes, NVDI is really good.

@Rustynutt

Centek bench shows the right speed, but Sysinfo has a quirk.

Something else and may be specific to my Falcon setup with Afterburner.
Sysinfo, all menus work fine, except when going to Environment Application Information. After reaching this page, the Falcon locks up and must be reset.

[HigashiJun]

Ah, that's interesting.

I'll wager it's the clock line, then. The Schmitt trigger is probably cleaning it up for the PAK -- perhaps it has a high load or is of a slightly sensitive disposition.

BW

Hope to tell you that soon.

Cheers.

[HigashiJun]

Something else and may be specific to my Falcon setup with Afterburner.
Sysinfo, all menus work fine, except when going to Environment Application Information. After reaching this page, the Falcon locks up and must be reset.

That's weird...

Apart the CPU speed issue, everything else works good here.