Score? Maybe... SGI "granite" AT-101

[N8N]

Hello again...

Today in the mail I received an eBay impulse purchase, an SGI "Granite" board.  BIN was well under market value so I pulled the trigger.

Good news:  It looks fantastic.  It even came in its original box; had I known that I would have paid the seller extra to ship it in another box outside of that one.

Bad news: It doesn't seem to work.  Tried testing with xev and got some inconsistent results; opened a text window and I can see that sometimes it does not register keypresses at all; sometimes it recognizes a keypress but not a key release.  Let me see if I can plug it in and try it again so you can see what I'm talking about:

The quick brown fox jumped ove h lazyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyydog.

Th uick row fox jumped oer thelazy dog.

The quick brown fox jumped over the lazy dg.

The quck brown fox jmped ove th laz og.

Th quckkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmpedoer helay dog.

(where the letters started repeating, I had to hit that same letter again to get it to quit.)

I'd like to think that it's an issue with the key switches themselves and that exercising them and/or a good flushing with deoxit will get it working again.  However I'm thinking probably the controller is on the way out.  Any ideas to try?

This is going through a Belkin PS/2 to USB adapter, but I've been using the same adapter on my Model M and Wyse PS/2 board with no issues, and also used it recently to test some old Cherry POS boards, so I'm pretty sure that it's good.  I will try another one (I have several) but I am not hopeful.

[N8N]

Go figure... I am typing this on an old Ziotek adapter, *on* the SGI board.

The quick brown fox jumped over the lazy dog.

The quick brown fox jumped over the lazy dog.

The quick brown fox jumped over the lazy dog.

The (expletive deleted) quick brown fox jumped over the (expletive deleted) lazy dog.

A diferent Belkin aaaaaaaaaaaaaaaaaaadpte:

TH QUUUUUUUUUUUUUUOW         FOXJMPED OVE LAZY OG

I thinwe see what the prrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr

*ahem*

I think we see what the problem is here.  I don't know why, but this board seems to be incompatible with the Belkin F5U119-E which is one of the usually recommended adapters; but it seems to be perfectly happy with a Ziotek adapter.  I don't have the part number on that (not printed anywhere on it) but I got it from clickykeyboards.com a couple years ago when I bought my first desktop computer in ages and wasn't satisfied with the keyboard that came with it, or the SIIG Alps board that was the best I could find at Micro Center at the time.  I find this odd, this is the first time I've ever had an issue with that Belkin part number.

Edit:  Here is the Ziotek adapter that I'm using (I'm back on the SGI board through the Ziotek and it seems to be working fine)

http://www.amazon.com/Ps-Keyboard-Mouse-Usb-Adapter/dp/B000G7WYZG/ref=sr_1_3?ie=UTF8&qid=1346192797

Now I'm typing this on a Dell AT-101W through the same Belkin adapter that failed on the SGI board.

*shrug*

I really don't know what to make of this, but the combination of the Belkin adapter and the SGI Granite seems to be no bueno for some reason.

Edit2: tested with XEV, all is good.  I remember now what the "known issue" was with the Ziotek adapter that I couldn't recall the other day... Scroll Lock makes the thing freak out.  But since I mostly use Linux, well, meh.

[mich]

The problem is that PS/2 was never a solid standard - everybody just copied IBM with varying degrees of success. There are slight differences in signal timings between various keyboards and host controllers and sometimes a bad pair just won't work.

Some of this problems arise when signal rise time is longer than patience of some end of the connection. These can be fixed by adding 4,7kΩ resistors pulling CLK and DATA lines to VCC.