Subject: Re: Two dead AXPpxi33 boards?
To: None <>
From: Berndt Josef Wulf <>
List: port-alpha
Date: 10/18/2001 20:08:56

Whatever you do, don't test the working CPU in the two apparent dead
boards as you risk to fry it.

I would first take a multimeter and checkout the power supply
voltages. Assuming that you've one of the motherboards connected,
measure the following voltages on J33 in reference to 
Pin 5, 6, 7 or 8 (Power Ground):

	Pin 2		+5V
	Pin 3		+12V
	Pin 4		-12V
	Pin 9		-5
	Pin 10		+5

The bright shining LED may be due for several reasons, one being that
you've used an ordinary LED and connected it straight between ground
and one of the output pins without a serial resistors - they don't
like to be connected directly to +5V!!! Try an LED with a 470Ohm
resistor in series. 

Once you're sure that the power supply is operational, you may want to
measure the output voltage of the regulator that supplies the CPU. I
don't know what voltage is expected at the output, but it should be
substantially lower that +5V (appr. 3V). You may find some information
on the Internet.

Perhaps not relevant but for completeness, the reference manunal makes
mentioning about connecting output ground and the safety ground. But
if you have mounted the board in a housing using metal screws, this
should be ok.

After this place CPU and RAM onto the board. It would be unlikely
to have two faulty CPU's, but I would spent some time testing the memory.
I assume that the memory you're using is the correct type. If
available try different memory modules that are known to work.
Remember, you need a pair of 36 bit wide RAM.

Are any cache chips installed? If so, remove them temporarily and see
if this helps. Also check the Cache Speed Select jumper for correct
configuration appropriate for the type of cache installed.

(J17  no jumper = 15ns, with jumper in = 20ns)

Check CPU Clock Frequencey jumper J7

Speed  Processor    P1-2   P3-4  P5-6
66     21068        Out    Out   Out
100    21068A       Out    Out   In
166    21066        Out    In    In
233    21066A       In     Out   In

Check SROM select jumpers J28 and J29

Function                Jumper   Pin
Boot from Flash         J28      1-3
Boot with BCache        J28      3-5
Boot without BCache     J28      2-4
Boot from loader floppy J29      1-3

In any case, try to connect a serial console as this will
be the best way of telling whether or not the system is
alive. You won't be able to see much activity booting to
the SRM console.

Hope this helps

cheerio Bernd
Name    : Berndt Josef Wulf            | +++ With BSD on Packet Radio +++
E-Mail  :             |    tfkiss, tnt, dpbox, wampes
ICQ     : 18196098                     |  VK5ABN, Nairne, South Australia 
URL     : | MBOX :
Sysinfo : DEC AXPpci33+, NetBSD-1.5    | BBS  : vk5abn.#lmr.#sa.aus.oc wrote
> I think I have a pair of dead AXPpci33 boards here. I'd like a second
> opinion, preferably useful but sympathetic will do. I'd swear this board
> worked before.
> I have the board sitting on it's cardboard box sitting on top of the
> case I'm testing it with. The power supply is good as far as I know,
> since it worked fine with a 486 up until I took the 486 board out.
> I have two 8 meg SIMMs in the lower two slots; the SIMMs were taken
> out of my AS 200 4/233 where they've been working fine since 1996.
> No video card I have works, including the PCI card out of my working
> AS 200. I can't test the serial ports until I get some adaptors, so
> I'm ordering some next week. 
> Hooking up a keyboard doesn't do anything to affect the video card. A
> keyboard (PS/2-style) that I hooked up had all three lights 
> (num,caps,scroll) on, and no key affected their status. 
> A speaker connected to the board does nothing. 
> My case happens to have a power led with wires to two 1-pin cables, so
> I'm able to test the diagnostic connector (J6, not the manufacturing
> one). It appears the pins are 1-16 with 1 being the lowest and farthest
> back, and it appears that the pins zig zag. 
> Having said that, connecting the LED's white wire to pin 15 and the green
> wire to pin 1 results in a LED shining waaaaay too bright (pin 1 is +5V).
> Moving the green wire to what I'm pretty sure is pin 7 or pin 9 results
> in a normally shining LED (7 and 9 are LED enable and clock). None of 
> the lines D0-D7 ever do anything. In particular, D7 should light up
> while any test is running, but it doesn't. I get nothing. So, it appears
> the board doesn't even try to self-test.
> Now, I have a working AXPpci33 sitting here running, and I figured I'd
> drop that processor on these two problem boards. Problem is that the
> running board's processor is too hot to touch for some period of time
> after power-down, and I don't like having the machine down for very
> long (it's my secondary DNS server, a test MySQL server, and the primary
> Kerberos server). So, I don't know if the boards are dead or the 
> processor is dead (I have two boards to test but only one processor).
> Now, provided the processor works fine then are these boards dead?
> Or have I missed something? I've got two deadweights, right?
> You know, I hate to throw out a good board if I've missed something 
> important. I need a working board. What do PC64 or 164LX boards go for
> nowadays?
> Thanks for any assistance.
> -- 
> Kevin P. Neal                      
> Seen on bottom of IBM part number 1887724: