(Solved) Repaired HP 1661C won't format hard disk (software crash)

[DIPLover]

Backstory
So I was missing my old HP 1663c logic analyzer that I sold after getting a 16700A and good cards. The 16700A is absolutely awesome but it's also quite slow to boot and cumbersome to operate through X-Window as well as loud and hot. When I need it, I need it, but often I don't.

I missed the immediacy of the 166x with their 5 seconds boot time and having every function a button press away and instantly displayed. But my 1663c had no LAN and too few channels.

I found a 1661C in great cosmetic condition, with option 15 (double RAM and LAN), it has 102 channels to my old 1663's 34. I got it for $50.

Repair
It had 2 problems : ROM self-test failed and hard-disk self-test failed.
It did boot completely and could make measurements, so I figured the boot ROM was OK, and maybe the flash had a few bitflips after 20 years, causing a bad checksum.

Surprise, the floppy was also dead when it arrived, which spiced things up a bit... (I thought they were regular PC drives, but no such luck. I was able to disassemble and repair mine though, so no biggie  )

I made fresh floppies of the OS and reflashed the ROMs. Self test passes. I ran the full performance verification suite, everything is good and all channels work  .

On to the IDE hard drive, which I thought would be the easy part...

Problem
The original drive is a 540MB Quantum from 1997 which spins up but makes a very sad sound. When I try to access it on the LA, it times out of all requests, same if I try to format it. No surprises so far. THE MACHINE DOES NOT CRASH, it just says "drive access timeout" or some such error message and happily continues along, like a proper, civilized little computer would.

I did not have another small capacity 3.5in PATA, but I had known good laptop drives and adapters so I tried that first. With a 6.4GB hitachi, the LA fails hard disk self test on boot, displays "unknown format" in the file manager and crashes when I execute a format command. The whole machine hangs and a popup displays "Software error. Please record the following numbers 0000 00000418 0020" everything is then frozen and I must power cycle the machine.

I tried to prepare the disk on a linux box by creating a new dos MBR, with a single 540MB FAT16 dos partition, same behaviour.

I next tried a compactflash adapter with a 4GB card : still hangs with the software error.

All drives spin up normally (the CF card does not spin...), all power leds light up. +5V and +12V are OK (4.98, 12.03).

Formatting FLOPPIES work, even when they are "unknown format" (apple HFS) when inserted.

Anybody ever seen something like this on HP/Agilent 166x or 167x series logic analyzers? Maybe also on 16500 from which the portables are derived?

NOTE : I had tried to format a drive before refreshing the OS in flash and had the same behaviour, which led me to restoring the flash first.

[gslick]

Try a smaller drive, 2GB or less.

I thought I replaced the hard drive of my 1660C with a CF card but just opened it up and found that it still has a hard drive. Maybe I did that on a different 1660/1670 series unit that I have.

I'm pretty sure I have a 2GB CF card in my 16500C. I'll have to open it up and verify that. I bought some 2GB InnoDisk iCF-4000 Industrial CF cards (DC1M-02GD31C1D) on eBay several years ago for this, for example: https://www.ebay.com/itm/333349100270

I didn't have to do anything special to use the CF card in my 16500C. Just boot from a system floppy, then initialize the CF drive, then copy the systems files from the floppies over to the CF drive.

[HarryDoPECC]

Yes, two things to be aware of:
1)  Use CF less than 2GB
2)  Some CF's just don't work in the CF-IDE adaptors.  I have converted my 16500C to CF and it works fine with some CF brands, but fails with others.
( I also changed the FDD for a Lotharek emulator - now my 16500C is much snappier with no spinning disks.  Still loud though!)

I wrote up the conversion with details of the brands that worked and those that did not - maybe on this forum - test your google-fu.

[Mp3]

There is a good chance, if it's really a 540mb non-SCSI hard drive, that dma mode may be needed. Sandisk 1gb/2gb cards are very reliable for that. Some more recent sd-ide adapters do dma but not atapi. I doubt atapi is needed if the drive is so small. IT is possible though.

[DIPLover]

There is a good chance, if it's really a 540mb non-SCSI hard drive, that dma mode may be needed. Sandisk 1gb/2gb cards are very reliable for that. Some more recent sd-ide adapters do dma but not atapi. I doubt atapi is needed if the drive is so small. IT is possible though.

Why would it not "really" be a non-SCSI 540Mb hard drive?

It's a 40pin connector and clearly marked as 540Mb on the drive... The whole 16500, 1660 and 1670 series use IDE drives.

[DIPLover]

Following reports on KO4BB
http://www.ko4bb.com/doku2015/doku.php?id=test_equipment:hard_drive_conversion_to_cf_for_hp16500

I have ordered a couple 1GB innodisk iCF 4000 cards.
My adapter is the SYBA SD-CF-IDE-A that was mentionned as working, so that's good.

I will report back when I get the card to test. Reports of "Software errors" on the 16500B with incompatible drives are also encouraging.

The format on 16500 drives is very weird though, I am surprised they kept it for the 1660. After all, they moved to DOS formatted floppies, why not FAT16 hard drives?

[gslick]

The format on 16500 drives is very weird though, I am surprised they kept it for the 1660. After all, they moved to DOS formatted floppies, why not FAT16 hard drives?

The LIF floppy only 16500A was new in 1988.

The LIF floppy only 1660A was new in 1993.

The 16500B was new in 1994 and added MS-DOS floppy support and the internal hard drive over the original 16500A.

The 1660C and 1670A were new in 1996 and added MS-DOS floppy support and the internal hard drive over the original 1660A.

Adding MS-DOS floppy support was a big improvement over the LIF format for exchanging files between a logic analyzer and a PC. (As long as the files were small enough to fit on a floppy. Adding Ethernet support was a much bigger improvement for exchanging large files).

As far as the hard drive format goes, if there were reasons for not using a standard MS-DOS file-system format for the 16500B there wouldn't be much reason to change it two years later for the 1660C and 1670A. Why would the implementation details of the hard drive file-system format ever matter to the logic analyzer customer?

[gslick]

I thought I had replaced the hard drive in a 1660C, 1670A, or 1670D that I have with a CF card but I opened them up and they all still have the original hard drive. Maybe I tried a CF card in a 1660C just to see if it worked and then put the original hard drive back.

I do currently have a CF card in a 16500C. It is an InnoDisk iCF 4000 2GB Industrial CF card. It is mounted in an ACS ACS-CF-IDEToCFA adapter. That particular adapter was cheap and easy to find at the time, looks like not so anymore. It is a completely passive adapter and any other passive CF to IDE adapter that is mechanically a good fit for installation should work equally as well.

[DIPLover]

As far as the hard drive format goes, if there were reasons for not using a standard MS-DOS file-system format for the 16500B there wouldn't be much reason to change it two years later for the 1660C and 1670A. Why would the implementation details of the hard drive file-system format ever matter to the logic analyzer customer?

You are right and they do not.

I am discussing this as an engineer and not a LA customer. It's a kludge. And surprising coming from HP, who knew a thing or 2 about computers and filesystems back then. Poor guy who had to retrofit a hard drive to the 16500 probably had no other choice to avoid redesigning half the thing.

It was a happy kludge however, the 2GB limit was enough to keep the platform commercially viable for another 7-8 years, and nowadays allows us to use cheap and fast CompactFlash instead of expensive and slow SCSI2SD.

[gslick]

It was a happy kludge however, the 2GB limit was enough to keep the platform commercially viable for another 7-8 years, and nowadays allows us to use cheap and fast CompactFlash instead of expensive and slow SCSI2SD.

The 1670G models kept going until at least 2004 so that would be 8 years after the 1660C and 1670A models were introduced. With the LCD 1660E, 1670E, and 1670G models they switched to using 2.5-inch laptop drives. The system software functioned with at least 3GB drives, although they still only formatted as 2GB. I have an original 3GB drive in at least one of the 1670G models that I have. I haven't experimented with larger drives or CF cards to see what the limit might be there. I should give that a try sometime just out of curiosity.

For the 16700 series with SCSI drives I have replaced missing or bad drives with other 68-pin SCSI drives. I haven't been motivated enough yet to pay for SCSI2SD adapters. I don't use the 16700 series often enough that it would really matter to me.

With the newer 16800 and 16900 series they switch to using SATA drives which makes it an easy swap to a SATA SSD.

[DIPLover]

I have an original 3GB drive in at least one of the 1670G models that I have. I haven't experimented with larger drives or CF cards to see what the limit might be there. I should give that a try sometime just out of curiosity.

I think the fact that the 1660E, 1670E and 1670G all came standard with 2GB drives point to that amount being the limit of the filesystem.

Thinking about it a bit, 2GB amounts to 64K sectors of 32K bytes. It seems there is a 16bit number limit for sectors somewhere in the firmware they inherited from the original 16500.

It may in fact be older, I have dumped the boot rom the other day and looking at it with a hex editor reveals the following string at address 0x1562 : "PSOS-6800  V4.1 Copyright 1983, 1984, 1985, 1986 SOFTWARE COMPONENTS GROUP, INC. A . H ."

Another interesting tidbit, is if you write the boot rom content to a pin compatible flash (I tried SST39F010A) chip and put it in the socket, the machine will boot in a different mode and show
PERFORMING POWER-UP SELF TEST
    emulation ROM Test
    emulation RAM Test
    passed     Interrupt Test
               Display Test

instead of the normal power-up self test Then stop. So it must ask the boot ROM for it's manufacturer ID or something.

[DIPLover]

I received the 1GB InnoDisk iCF 4000 Card today and I am happy to report that it worked perfectly on first try.