Trouble with new Exxos Integrated Falcon RAM Board

[exxos]

I tried 40R, 80R, 220R. You basically took 220R and X1 scope probe to snub out all the crap.

IMG_0125.JPG

IMG_0128.JPG

On 10X it was like this... But not on every cycle.

IMG_0127.JPG

This is with 220R and 100pF to GND. :roll:

IMG_0132.JPG

Pretty mad really. I will try a 470R but its getting onto a crowbar type of thing to clamp this out now :roll:

Thing is though.. what else on the motherboard is prone to failure with all these spikes :roll:

EDIT:

Didn't boot with 440R. It could be the scope probe picking up noise which isn't helping. Though its hard to figure out what resistor / cap combo works best.

EDIT2:

I guess the RAM board only goes back to the Videl.. So will try a pull down instead.

[exxos]

The only way to solve it is to use something like 2x 40R in series and the middle point a 100pF to GND. A "T" filter type of thing. So both ends of the filter look like this.

IMG_0134.JPG

IMG_0135.JPG

[forty2]

It's weird though that you're seeing those spikes, and I'm not - those first few images led me to think incorrect probe compensation, but then the address lines look OK as you said. I just did another pass at a bunch of pins on the RAM board, and they all kind of look like this, both on my Digital Agilent & on my Analog Tek, with a 100MHz probe. Baaah, my attachment quota has been reached, but my Vmin was -160mV and Vmax was 5.12V.

Was older RAM just more resilient, or are people going to have these problems with their original RAM boards soon too? A 'T' filter on every data line seems bonkers, right?

I was gonna go ahead and buy the SIMM adapter instead - if you want to credit me for the RAM board when I place that order, that'd be swell. Thanks @exxos !

Thanks @dml - I have... a lot of Tek scopes (this one might be the newest!) My instagram handle is toomanyoscilloscopes :D

[exxos]

It's weird though that you're seeing those spikes, and I'm not - those first few images led me to think incorrect probe compensation, but then the address lines look OK as you said. I just did another pass at a bunch of pins on the RAM board, and they all kind of look like this, both on my Digital Agilent & on my Analog Tek, with a 100MHz probe. Baaah, my attachment quota has been reached, but my Vmin was -160mV and Vmax was 5.12V.

Was older RAM just more resilient, or are people going to have these problems with their original RAM boards soon too? A 'T' filter on every data line seems bonkers, right?

A lot of original RAM boards have failed. I've had falcons in for repair with failed RAM boards and even sent boards to repair. Some chips may be more resilient than others to abuse. I tried many brands during testing. But some didn't like being in the falcon at all. I was also limited to what I could actually still obtain.

I don't see how you can not see the spikes. At the speeds the signals are at and the long traces from videl to RAM, they simply cannot not be there. I did not see the issues with X1 prove but the probe has to be more than 100pf in that mode. So either your probes are "bad" or you have a bandwidth limit or noise filter set on your scope or something. I really don't know. You would really have to send me your motherboard to test myself but not really worth the hassle.

Can you also PM me your order number? The SIMM based board was designed with RAM failure in mind. It's easy to change a simm if it dies.

[forty2]

Bandwidth. Limit. :facepalm: Rookie move... Yeah, I'm seeing Vmax of 5.9v & Vmin of -800mv now (sorry).

[exxos]

Bandwidth. Limit. :facepalm: Rookie move... Yeah, I'm seeing Vmax of 5.9v & Vmin of -800mv now (sorry).

Easily done. Thanks for letting us know anyway :)

If you take an image and upload, the quota problem should be fixed now.

EDIT:

I have taken this over to random posts in my blog now. https://exxosforum.co.uk/forum/viewtopi ... 087#p93087

[forty2]

Got 'er open again this morning just to get a few more shots
you can clearly see the excursions and ringing at 50ns

IMG_5826.jpg

took a few with the analog scope too - results are similar

IMG_5828.jpg