if possible try to add a timebase to the waveforms.
How's the ripple on the power rails ? at the connector of the psu & at the other end of the board, incl voltage levels at that end.
Considering this is quite an old machine... Has it had any electrolytic capacitors replaced yet ?

I have zero experience with EISA, personally I would start by

-checking cpu/bios voltages
-monitoring BIOS data/address pins during boot

Hi! I have Proliant 1500 too not in original chassis, but I figured out how to make it work without special stuff.

Make sure you have installed correctly and set the dip switches to defaults first before starting the configuration once you get that going. Once you run the configuration and set it up, it will show you the recommended dip switches settings.

Make sure you have good battery, I used 3 cell (3x 1.5V) external battery for CMOS settings. You need to get ESR meter to check the capacitors inside the power supply. Discharge the power supply's main capacitors first with a 100 or 220 ohm resistor.

Cheers,

pentiumspeed wrote on 2022-04-03, 17:03:

Hi! I have Proliant 1500 too not in original chassis, but I figured out how to make it work without special stuff. […]
Show full quote

Hi! I have Proliant 1500 too not in original chassis, but I figured out how to make it work without special stuff.

Make sure you have installed correctly and set the dip switches to defaults first before starting the configuration once you get that going. Once you run the configuration and set it up, it will show you the recommended dip switches settings.

Make sure you have good battery, I used 3 cell (3x 1.5V) external battery for CMOS settings. You need to get ESR meter to check the capacitors inside the power supply. Discharge the power supply's main capacitors first with a 100 or 220 ohm resistor.

Cheers,

Thanks for the reply.

I have SW1 (on the backplane, near the motherboard power connectors) tested at the settings it already had while in the attic (off, on, off, off,on, off) and at the "default" settings (all off) from https://www.ultimateretro.net/motherboard/man … 15117414720.pdf howver I actually found an old post of YOURS that said those are mislabeled...? Proliant 1500 or 1500R motherboard and front panel wiring information wanted.

Also, this is very curious... I just went to double-check the sw1 switches on the processor daughterboard. They are on/on/off/on which was how it was set when it was running before, and I read that means 166mhz which should match the single P166 I have in there (but cannot find the doc that told me this setting, so looking again).

BUT

I decided to verify that the processor speed switches work by moving them back and forth then testing continuity. Switches sw1.2 (and .3 and .4) all show short / open correctly depending on position, but sw1.1 is open all the time. I need to find my reference for those switches to determine the effect of sw1.1 being open (effectively giving me off/on/off/on instead of on/on/off/on) and see if that would explain it acting weird. Maybe I can jump that with a wire to see if it gives me the correct speed. I suspect that the clock speed setting being off is going to result in unpredictable behavior relating to sync/latch/clock cycles not lining up correctly.

Going to test that and report back.

EDIT: Jumping that switch has no effect. The CLK pin (b16 on PCI and at the clk pin of the cpu)
Looking more closely, those settings for cpu speed,testing all settings and monitoring frequency and voltage:

After running through some karnaugh maps:

sw1.1 and sw1.2 have no apparent effect on frequency and amplitude. Maybe these are a multiplier?

If sw1.3 on and sw1.4 on then frequency is 33mhz, stable, and amplitude is 1.5-3v
If sw1.3 off and sw1.4 on then frequency is 33mhz, stable, and amplitude is 1.5-1.8v
If sw1.3 on and sw1.4 off then frequency is 50mhz, unstable, and amplitude is 1.6-2.1v
If sw1.3 off and sw1.4 off then frequency is 40mhz/unstable, and amplitude is 1.5-1.8v

Do that.

The switches on CPU board is first 2 for fsb or multiplier, other last 2 is for either. Forgot which was.

Compaq bios will say use 60ns memory modules if using fsb 66 MHz before proceeding. Always FPM parity modules.

Cheers,

pentiumspeed wrote on 2022-04-04, 19:29:

Do that. […]
Show full quote

Do that.

The switches on CPU board is first 2 for fsb or multiplier, other last 2 is for either. Forgot which was.

Compaq bios will say use 60ns memory modules if using fsb 66 MHz before proceeding. Always FPM parity modules.

Cheers,

(I edited my erlier post but took so long you already replied. Add the edits here ...)

Jumping sw1.1 had no discernible effect. I measured the CLK pin (b16 on PCI and at the clk pin of the cpu)
Looking more closely, those settings for cpu speed,testing all settings and monitoring frequency and voltage:

After running through some karnaugh maps:

sw1.1 and sw1.2 have no apparent effect on frequency and amplitude. Maybe these are a multiplier?

If sw1.3 on and sw1.4 on then frequency is 33mhz, stable, and amplitude is 1.5-3v
If sw1.3 off and sw1.4 on then frequency is 33mhz, stable, and amplitude is 1.5-1.8v
If sw1.3 on and sw1.4 off then frequency is 50mhz, unstable, and amplitude is 1.6-2.1v
If sw1.3 off and sw1.4 off then frequency is 40mhz/unstable, and amplitude is 1.5-1.8v

FPM vs EDO for Proliant 1500.... I've heard people arguing "Only FPM" and "only EDO ECC" and "Only EDO non-ecc". I have pairs of 60ns 72pin simms fpm and edo non-ecc, bottom two simm sockets (SIMM1 and SIMM2) but no change in behavior.

Probably multiplier locked processor but two of these switches is fsb.

No, not either EDO or ECC the service manual is explicit. Proliant 1500 is low end machine using FPM parity modules, in pairs at a time only and, two 4MB modules minimum, up to 32MB modules. Either 70ns for fsb 60MHz or 60ns for 66MHz fsb.

Cheers,

From my notes:

Update:

Dual processor board for Proliant 1500,
Investigated the dipswitch and BF0 and BF1 with multimeter. Indeed, sw 1 for BF0 and sw 2 for BF1, which can run either 100, 133, 166 and 200 using multiplier unlocked Pentium 200 processor.

Setting SW 1 or SW 2 or both "on" pulls low.

My comment: the SW3 and SW4 is can be 50MHz, 60 and 66MHz.

Cheers,

Added:

https://www.ardent-tool.com/PS2_MOST/ACROREAD … PAQ/2005018.PDF

weedeewee wrote on 2022-04-03, 13:10:

if possible try to add a timebase to the waveforms.
How's the ripple on the power rails ? at the connector of the psu & at the other end of the board, incl voltage levels at that end.
Considering this is quite an old machine... Has it had any electrolytic capacitors replaced yet ?

Thank you for your reply. I appreciate you taking the time.

More waveforms:
https://imgur.com/gallery/L6nWINT
The ripple looks pretty minor to my untrained eye. Within 1-2% and rippling with 50hz.
Timebase is at the bottom if these new ones. I captured an "address line input" versus "latch enable" relationship with timebase.

There are no electrolytic caps on the backplane or cpu daughterboard. Likely a bunch in the PSU but I don't see anything that tells me the PSU is a problem so wasn't planning to pull it apart.

pentiumspeed wrote on 2022-04-04, 20:43:

From my notes: […]
Show full quote

From my notes:

Update:

Dual processor board for Proliant 1500,
Investigated the dipswitch and BF0 and BF1 with multimeter. Indeed, sw 1 for BF0 and sw 2 for BF1, which can run either 100, 133, 166 and 200 using multiplier unlocked Pentium 200 processor.

Setting SW 1 or SW 2 or both "on" pulls low.

My comment: the SW3 and SW4 is can be 50MHz, 60 and 66MHz.

Cheers,

I am curious. The effect of not having memory installed should be to post partially, then give an POST code / beep error or similar regarding failed memory check - correct? The behavior I get (specifically the address lines being pulled high after the bus transceiver for the ISA ports) does not change regardless of what memory I put in there, or if I leave the slots empty.

I'm also trying to figure out how to measure the actual FSB at the cpu. That level of docs are not really available any more.

Make sure the processor board is correct for particular models is true. Not all processor cards is compatible even can be plugged in.
I have known working processor boards one is dual processor, used single CPU in default with 512K cache, 222925-001 and single CPU board with 16MB onboard 169174-002, these two works in this proliant 1500. None supports MMX processor due to VRM but these two does support up to Pentium 200MHz. I used their default CPU if installed on the board first to get working first.

And my notes is correct about switch configuration is incorrect according to the service manual, use the configuration diagram displayed during setup when running configuration disks. But disabling the onboard video via dipswitch is correct on the service manual. Onboard video disabled, I can use either ISA or PCI video card, cirrus logic or S3 is more compatible PCI cards and faster.

And I do confirm that I can disable the onboard SCSI via configuration software when dipswitch is correctly set as previously explained.

Floppy drive is standard as I used a generic floppy drive. I can also can use dual SATA PCI-X plugged into PCI slot or dual ATA PCI adapter card both are by 3ware.

Booting up take awhile. Set it up by defaults first and you need 4.5V battery plugged in, two battery connector is wired in parallel, use either.

Cheers,

rasz_pl wrote on 2022-04-03, 13:53:

I have zero experience with EISA, personally I would start by

-checking cpu/bios voltages
-monitoring BIOS data/address pins during boot

This is interesting...

There are 3 GAL22v10C chips on the backplane. Eproms, essentially. One is next to the bus transceivers, and has a good clock signal on pin 26.

The other two are labelled "SYS MGT ROM" and "BIOS" in the silkscreen. NEITHER of these has a clock signal on Pin 26. The signal appears to be pulled high just like the address lines between the octal bus transceivers and the isa slots.

Also noted that the SCSI IC (NCR 53C825) has a good clock signal, but the video controller (Cirrus Logic GD5420) does not. I'm going to map out which ICs are expecting a clock and which are getting it. I am wondering why the clock signal is not making it there. Also, which clock pins are just connected together and which are not.

pentiumspeed wrote on 2022-04-04, 21:42:

Make sure the processor board is correct for particular models is true. Not all processor cards is compatible even can be plug […]
Show full quote

Make sure the processor board is correct for particular models is true. Not all processor cards is compatible even can be plugged in.
I have known working processor boards one is dual processor, used single CPU in default with 512K cache, 222925-001 and single CPU board with 16MB onboard 169174-002, these two works in this proliant 1500. None supports MMX processor due to VRM but these two does support up to Pentium 200MHz. I used their default CPU if installed on the board first to get working first.

And my notes is correct about switch configuration is incorrect according to the service manual, use the configuration diagram displayed during setup when running configuration disks. But disabling the onboard video via dipswitch is correct on the service manual. Onboard video disabled, I can use either ISA or PCI video card, cirrus logic or S3 is more compatible PCI cards and faster.

And I do confirm that I can disable the onboard SCSI via configuration software when dipswitch is correctly set as previously explained.

Floppy drive is standard as I used a generic floppy drive. I can also can use dual SATA PCI-X plugged into PCI slot or dual ATA PCI adapter card both are by 3ware.

Booting up take awhile. Set it up by defaults first and you need 4.5V battery plugged in, two battery connector is wired in parallel, use either.

Cheers,

Thanks for your reply.

I have tried a trident isa card for video and swapping sw1.1 on the backplane to disable onboard video, but nothing. That does not actually surprise me given the ISA address and data lines are all messed up.

I'll obtain a 4.5v battery.

pentiumspeed wrote on 2022-04-04, 20:43:

From my notes: […]
Show full quote

From my notes:

Update:

Dual processor board for Proliant 1500,
Investigated the dipswitch and BF0 and BF1 with multimeter. Indeed, sw 1 for BF0 and sw 2 for BF1, which can run either 100, 133, 166 and 200 using multiplier unlocked Pentium 200 processor.

Setting SW 1 or SW 2 or both "on" pulls low.

My comment: the SW3 and SW4 is can be 50MHz, 60 and 66MHz.

Cheers,

Circling back to this... Thank you for your response.

My observations did not match yours. You clearly have succeeded so I'll assume my observations are incorrect, or I'm observing incorrect behavior.

So not only did I find that sw1.1 is permanently open (mentioned above I think), but also sw1.3. I traced BF0 and BF1 from the empty socket and found that only one of them was influenced by the switch. After figuring this out I soldered a bodge over sw1.1 and I get correct bus speed now. Still no post but closer.

For reference, and for anyone else looking in the future:
BF0, cpu pin Y33, goes to sw1.1 on the "processor" side of the switch. It emerges from the "memory" side of sw1.1 then goes to the back of the board (same area) and hits R33 ("51o" = 51 ohm) at the bottom. The top of that resistor is grounded.
BF1, cpu pin X34, goes to sw1.2 on the processor side. Emerges from the memory side of sw1.2 then stays on the from of the board and hits R50 ("510" = 51 ohm) on the memory side. The other end of that resistor is grounded.

sw1.3 is also permanently open, however the correct setting for this is off anyway, so that is a note I will make but no action needed.

I did note also that these settings don't affect anything on the st49c155 (clock multiplier) U43 on the backplane (next to the bios rom) so I suspect the frequency select pins A0/1/2 are locked on the correct settings intentionally and only handles backplane stuff (disks, keyboard, onboard video etc)

Instead, sw1.3 and sw1.4 (processor side) connect to pins 23 and 24 (respectively) of the ICS9158-04 (frequency generator and integrated buffer) U10. Same purpose as U43 - generates multiple frequencies from a base 14mhz input (Y1) and 3 frequency select pins (sw1.3 is pin 23, "FS1" and sw1.4 is pin 24 "FS0"). The datasheet for it shows that the output is 66.6Mhz on CLK2 out when output enable is on. FS2 high/low has no effect.

SW1.3 OFF (open) results is FS1 seeing a 1 (pullup resistor?). (Otherwise, the memory side of SW1.3 goes to ground via R36 on the back of the board.)
SW1.4 ON (short) results in FS0 seeing a 0. (Otherwise, the other side of sw1.4 goes to 5v via R149 on the back of the board)

The datasheet for it shows that FS0=0 and FS1=1 results in an output of 66.6Mhz on CLK2 out. FS2 1/0 has no effect.

This chip sends a unique clock signal to many other devices not handled by U43, like the scsi controller (via A89 on the daughterboard connector)

Still trying to connect the dots for why the bios does not see a clock signal.