Hi

If your 9383 is running on a >= 15 year old GPS module, that could be 
a source of trouble. Most of the modules from that era handle the
GPS rollover problem poorly. 

Bob

> On Feb 16, 2021, at 10:00 AM, John Miller via time-nuts 
> <time-nuts@lists.febo.com> wrote:
> 
> Hello Everyone!
> 
> TL;DR: If anyone with a Spectracom 9383 would be willing to help me recover 
> mine, I would be eternally grateful!
> 
> Early last year I stumbled upon a local Spectracom NetClock 9383 for a very 
> reasonable price. It worked just fine out of the gate, but because I am 
> incapable of leaving well enough alone, I popped the hood, took an image of 
> the CF card, plugged in a monitor and keyboard and instantly started to take 
> a look at the BIOS and the OS. Curiosity satisfied, I put it back together 
> and resumed using it as intended, but I noticed it was a bit unstable. (For 
> what it's worth, it may have been unstable before I poked around inside of 
> it, but I never ran it for long enough before opening it to know...) When I 
> say that it was unstable, I don't mean the clock, but the PC hardware itself 
> - it was crashing and locking up.
> 
> So if figured, if it's not entirely broken, fix it 'till it is!
> 
> I tried two things first: I put the image from the original CF card onto a 
> similar linear CF card and I swapped out the RAM SODIMM for another, though 
> those changes didn't improve the situation. So, despite the fact that I know 
> the Spectracom engineers configured the SOM modules specifically to be 
> reliable in the configuration in which they were shipped, I decided to 
> re-flash the BIOS on mine to a newer version. This was, I believe, the 
> pivotal mistake. The Advantech SOM-4455 PC module was shipped in the 9383 
> with BIOS v1.14, and Advantech had a v1.15 available on their website. I 
> downloaded this updated BIOS image, the flashing tool (or at least what I 
> thought was the right flashing tool...), and I made a FreeDOS-booting CF 
> card. The 9383 booted off of this right away, and I used the utility to make 
> a backup of the original BIOS and configuration, then wrote the new BIOS to 
> the chip - and then it crashed and stalled.
> 
> I rebooted the machine, but sure enough the BIOS was corrupted and didn't 
> boot. I went out and bought a TL866II EEPROM programmer and PLCC adapter and 
> confirmed that, sure enough, the BIOS on the chip was trashed. To make 
> matters worse, the backup that the utility took of the chip was similarly 
> trash. I was able to use the EEPROM programmer to burn the v1.15 BIOS to the 
> chip and get it booting again, and while I can get into the BIOS and start to 
> boot the OS, it won't boot completely - it hangs when it loads one of the 
> custom Spectracom kernel modules. Additionally, any BIOS settings that may 
> have been there are now gone.
> 
> While digging through configuration files and scripts on the disk image, I 
> was able to find names of a few engineers who had worked on the 9383, and 
> when I contacted them they were willing to talk about it. They confirmed that 
> the only ran the 9383 with the stock BIOS  from Advantech, they didn't make 
> any custom modifications to it, but they probably did make some changes to 
> the default settings, though they weren't able to confirm for sure.
> 
> I suspect that one or more things is happening here:
> 1) BIOS v1.15 breaks something that their custom hardware kernel modules 
> relied on
> 2) Losing a setting in the BIOS is preventing said module from loading 
> correctly (I have tried iterating through BIOS settings hundreds of times)
> 3) The interface on either the planar or the SOM-4455 module is damaged (I 
> have inspected everything as closely as I can and it all looks fine...)
> 4) The SOM-4455 has been damaged
> 5) The planar has been damaged
> 6) The universe has determined that it is not meant to be.
> 
> I have tried contacting both Spectracom and Advantech, but Spectracom has no 
> interest in helping me fix something that A) is 15 years old, B)has  been 
> tinkered with in massively unsupported ways, C) is not even in the hands of 
> the original buyer. Advantech hasn't responded to me at all, and I have been 
> able to find BIOS v1.14 on their website at all. Best I can determine the 
> file should be called "4455v114.bin" but I have not been able to turn it up 
> anywhere. Go figure they're not interested in supporting 15 year old hardware.
> 
> I have looked into getting a SOM-compatible backplane to see if my module 
> works in other circumstances, but they tend to be very expensive when they do 
> show up. Similarly I've thought about simply getting another SOM-4455, but 
> all of those that I have found has been outside of my price range.
> 
> At the very least what I hope that someone here with a 9383 can help me with 
> is restore the proper BIOS configuration settings. If anyone is willing to 
> crack theirs open, attach a VGA monitor and PS/2 keyboard, and simply take 
> pictures of each of the configuration screens I can try and match my v1.15 
> BIOS to what it is *supposed* to be. Even better would be if someone could 
> pop out their BIOS chip and get an EEPROM dump of v1.14 + the configuration. 
> If you don't have an EEPROM reader, I'd be happy to loan you mine, or if you 
> want to send me the chip I can dump it and mail it back. I would also be 
> happy to mail my SOM-4455 to someone to see if it works in an otherwise-fine 
> 9383.
> 
> I know that this isn't the best piece of equipment in the world (mine isn't 
> even an Ru or OCXO model), but it deserves to live and I have a strong 
> feeling that it's not dead! I just can't bare to see it get trashed.
> 
> 
> Regards,
> John
> KK4YWH
> 
> 
> PS - Failing all of that.. if anyone is looking for a spare 9383 for parts...
> 
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe, go to 
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.


_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe, go to 
http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
and follow the instructions there.

Reply via email to