• Please review our updated Terms and Rules here

HP Netserver 4d/66 LM

billdeg

billdeg

Technician
Joined
Nov 18, 2003
Messages
3,885
Location
Landenberg, PA USA
http://vintagecomputer.net/hp/Netserver_4d66LM/

Completely dead. No change with warrantied power supply installed.

I am thinking something on the motherboard has shorted. I removed peripheral boards, etc. No change. Anyone with related experience/solution other than "replace the motherboard"? I could I attempt repairs to the motherboard but that may be more trouble than it is worth to me.

UPDATE - It's probably the Dallas real-time chip, nevermind. I searched around for a battery but I did not notice it before...duh right in front of my face.

Anyway, this is one of those 486 to P/66 upgraded systems. It seems like the NetServer model had 4 versions of the same box with a different front plate (and power supply with one extra connector) but otherwise the same. I.e one could get the Netserver with a factory-installed P 66 or the version with the 486 EISA processor. I have the 486 EISA processor so this one must have been field-upgraded and the orig owner held onto the original CPU. Ordered the Dallas 1287.

Bill
 
Last edited:
Oh wow, I haven't seen a Netserver in 15 years now. I sold and serviced a ton of those... There were solid but most clients ran them too long and they died after a decade.
 
This machine has been on my list for many years. It's full of post-vintage transition hardware. FDDI comms (100mb/sec) for a file server running with a Pentium CPU would have been top of the line in 1993 given ethernet was slower then. This machine could have served files for a nicely-sized staff of users. This machine is in nice shape too, not a lot of dust at all. Not sure how long/if it was in production but even a lightly-used Dallas timer chip will die. I just hope it comes back to life after I get the new chip installed. Not sure if it's a Novell system but that's my guess. Depends if it was upgraded to Windows NT at some point. Have not been into it yet, and I will have to hack the password.

I have a few of these EISA CPU Pentium 60 or 66's. It has been a goal to study how they work and to preserve a few.
 
Last edited:
EISA server hardware's absolutely lovely. I've got a dual Pentium Pro NCR S26 and it's a superb system all around, though design-wise it's not a patch on that netserver in terms of looking cool.
 
AFAIK the motherboard in this system is the Intel xPress. most of the info I've found on it can be found in this windows .hlp file (you need XP or earlier to read the files):
http://www.elhvb.com/mboards/intel/files/help/XPRESS.HLP

also:
http://www.elhvb.com/mboards/intel/files/help/xpressLX.hlp

The later PCI version is called the ALTserver i think. There is a file in that directory for that too.

It's not just a "486 to P/66 upgraded system" although very close. It's architecture is kinda similar to the Compaq Tri-flex though not as good. IIRC the memory architecture is 32bit but faster somehow to allow for later processors, but not really a fantastic design for the eventual Pentium. Future-proofing that didn't really end up working out.

It can take a dual 60MHz Pentium board BTW. One of the strangest things in the .hlp files, is that it states something to the effect that the 32bit memory architecture doesn't really affect the memory performance. They claim it's like 99% the performance of a 64bit architecture. But at the time the Pentium 60/66 was being sold as the new "64bit Pentium" and they were claiming how the 64bit mem architecture was one of the main selling points.

here's a bit of the marketing blurb:

"The Intel-designed architecture is a strobed, asynchronous memory bus created to scale for ANY Intel 32-/64-bit processor ranging from 20 to 100 MHz. The Xpress Interface Bus allows multiple processors and cache resources to reside on the same high speed bus without replacing or removing the system baseboard. Xpress server platforms support the entire range of both write-back and write-through cache architectures while still accepting a full range of Intel486 microprocessors and Intel's 64-bit Pentium microprocessor."

Edit:
Here's the bit i was looking for:

"The Xpress Interface Bus does not incorporate a full 64-bit data path. Instead, the Xpress server architecture uses an ingenious method of time division, multiplexing 64-bit data across the dedicated 32-bit memory bus to the 64-bit DRAMs. The architecture ensures compatibility with 64-bit Pentium processor modules while optimizing the performance of the 32-bit Intel486 processor. A study found only a 1% to 6% difference between the performance of a dedicated 64-bit address/data path and the interleaved 32-bit Xpress Interface Bus. Intel simulated multiple 64-bit memory transfers using the 64-bit Pentium processor module on the 32-bit Xpress Interface Bus with a 256 KB writeback secondary cache."

Interesting system.
 
Last edited:
I need some help. I replaced both the power supply and the battery, the machine is absolutely dead. I double checked the connections. I took photos of the power supply connectors before to be sure I plugged everything in correctly. I made sure the battery pins are inserted correctly. I removed cards. Is there some sort of hidden fuse or switch to switch after the battery dies and gets replaced? I assume there is a short on the motherboard.

Suggestions? What does the right button do on the control panel:
HP_NetServer_ControlPanel.jpg
 
When you say dead, I'm assuming you mean, if you press power button nothing happens? no fans or anything? I think the right button is for cycling thru information on the LCD panel. My DEC HX6000 system, a contemporary, has a similar setup.

Does the power button connect to a PCB with the other buttons+LCD, then go thru a ribbon to the motherboard? The file I linked to indicates this, but then shows the pinout for the PSU connectors as below. There doesn't seem to be any PS_ON in that lot. Are there any other connectors for the PSU to the motherboard?

from the file I linked here is the pinout of the main PSU connectors for the motherboard:
Code: 
KJ0840 - Main Power Connector
Type:	Single Row Header Style, PC/AT standard Connector, 12 pin (Male)
Pinout:	1	PWRGOOD
	2	+5v
	3	+12v
	4	-12v
	5	GND
	6	GND     Pins 2, 10, 11, and 12 are rated for 5 amps each
	7	GND
	8	GND
	9	-5v
	10	+5v
	11	+5v
	12	+5v
KJ0850 - Aux. Power Connector
Type:	Single Row (Male); Used with 215W, 230W and 384W Power Supplies
Pinout:	1	+5v
	2	+5v
	3	+5v
	4	GND     Pins 1, 2,and 3 are rated for 5 amps each
	5	GND
	6	GND
KJ0860 - Aux. Power Connector
Type:  	Single Row Header (Male);  Used with 384W Deskside Power Supply only
Pinout:	1	+5v
	2	+5v
	3	+5v	Pins 1, 2, and 3 are rated for 5 amps each
	4	GND
	5	GND
	6	GND

Edit:

Just looked at the images again, looks like there is another purple/black wire. Have you tried shorting these together? Im guessing the black is GND.
 
Last edited:
Not to dig up a dead thread, but yes, these motherboards often have shorted tantalums that shut down the power supply. I've got a small stack of the 6-slot Intel Xpress boards to go through, all shut down a perfectly good supply, one had evidence of a previous tantalum cap explosion. There's...quite a few on the boards.

Bill, if you decide not to keep this one, be sure to let me know -- always looking for Intel Xpress parts :)
 
billdeg said:
This machine has been on my list for many years. It's full of post-vintage transition hardware. FDDI comms (100mb/sec) for a file server running with a Pentium CPU would have been top of the line in 1993 given ethernet was slower then. This machine could have served files for a nicely-sized staff of users. This machine is in nice shape too, not a lot of dust at all. Not sure how long/if it was in production but even a lightly-used Dallas timer chip will die. I just hope it comes back to life after I get the new chip installed. Not sure if it's a Novell system but that's my guess. Depends if it was upgraded to Windows NT at some point. Have not been into it yet, and I will have to hack the password.

I have a few of these EISA CPU Pentium 60 or 66's. It has been a goal to study how they work and to preserve a few.
Click to expand...


No comment on the Netserver (other then they are purtty and if I wasn't up to my gullets in massive retro 80s/90s servers I would love one) but on Dallas chip I would double check the chip and make sure it is a DS1287. The usual part for EISA systems was a DS1497. Incidentally I have found the batteries on those to last forever. My Systempro's (from 1994ish era) RTC is still running extra holding on to both EISA NVRAM config as well as date and time.

Edit: Just realized how old the OP was. DOH! :)
 
It's a DS1287 for sure, a DS12887 or GW-12887-1 will work in its place -- that's what I run.
 
You must log in or register to reply here.
Back
Top