Rohde & Schwarz FSL Windows XP trouble

[ivaylo]

So I have the opportunity to trade a Rohde & Schwartz FSL Spectrum Analyzer which has a few intermittent problems:

• It would occasionally at random times fail to boot (then a few hours later would have no problem and run for hours). During boot time winlogon.exe would throw an "application error" - "The instruction at XXX referenced memory at 0x00000000. The memory could not be read", and when you click OK it would go straight to the blue screen (see attached images).
• It would occasionally freeze after it's been running for a while. Then only a reboot takes it out of that state. I strongly suspect these two problems are one and the same.

The unit is running Windows XP embedded. A picture of the motherboard is attached. The Toshiba part is the flash disk I believe, RAM is soldered to the board as far as I can tell. So my questions are:

• What could be causing this? From my Windows times I remember getting things like these, but was it in the context of bad RAM? I remember having to replace or move SIMM-s around in the 90s and such, but here they are all soldered. Or could it be indication of the flash going bad or something?
• Should I proceed with the trade? IMO right now it is favorable to me and if I can deal with this issue I'll be happy.

Again, happens rarely, the unit works fine most of the time, passes all possible embedded tests and diagnostics, etc. Appreciate your help...

[marshallh]

Can you run memtest86, or some other memory tester?


The ram is easily replaceable with the right tools, just time-consuming. Coudl also be power supply.

[unitedatoms]

The googling for this fault address, error being intermittent and fact that the device has changed hands makes it most likely being a network problem. DNS/DHCP and may be even authentication (why lab instrument needs a network authentication ? this is too much).

I'd first try to see if there is any way to see what network IP stack looks like. What is cached as dynamic IP left from previous user, and was previous user in a big or small organization: the larger the organization the more abusive is the network settings management for devices like this.

If there is any way to get to console, locally or remotely: these commands may help

ipconfig /all
ipconfig /renew
ipconfig /flushdns
ipconfig /registerdns
Again: ipconfig /all

[ivaylo]

Can you run memtest86, or some other memory tester?

OK, I'll try that. Thank you!

[ivaylo]

The googling for this fault address, error being intermittent and fact that the device has changed hands makes it most likely being a network problem. DNS/DHCP and may be even authentication (why lab instrument needs a network authentication ? this is too much).

I'd first try to see if there is any way to see what network IP stack looks like. What is cached as dynamic IP left from previous user, and was previous user in a big or small organization: the larger the organization the more abusive is the network settings management for devices like this.

If there is any way to get to console, locally or remotely: these commands may help

ipconfig /all
ipconfig /renew
ipconfig /flushdns
ipconfig /registerdns
Again: ipconfig /all

Ha, it never occurred to me that stale network could cause this. I did all of the above with no problem. What should be the result of it? Should it show me errors or flush something or what? After it crashed and then recovered yesterday I did independently (trying to connect to it with the R&S software) check if DHCP is working and all seemed fine, but I didn't do it before it crashed. Also why would time cure this? A few hours after it was crashing it just came back. Well, it was also with an open case at that point. Thank you for your help!

[unitedatoms]

It will get rid of older stale IP address, NETBIOS names of nonexistent servers to which this device was talking back ago in different network. There are idle timers in Windows like every 6000 minutes to send a cryptic Kerberos heartbit to renew useless token to update the state of network drive \\SOMESERVER\C$ which was used 180 days ago to move a kitten picture to background, or nonexistent network printer etc.

This dead connections are intermittently being renewed by windows with very slow timers

There is another cleaner for NETBIOS release refresh: nbtstat -RR

https://technet.microsoft.com/en-us/library/Cc784285(v=WS.10).aspx

[gowf67]

This very much sounds like RAM issues. It will run fine until it tries to reference what the OS originally thought was a good address. Agree that you need a more exhaustive RAM test. Any possible way of booting another kernel (linux) via USB etc..??

[ivaylo]

This very much sounds like RAM issues. It will run fine until it tries to reference what the OS originally thought was a good address. Agree that you need a more exhaustive RAM test. Any possible way of booting another kernel (linux) via USB etc..??

I can try that. What test do you recommend under Linux?

[nctnico]

I guess this is a pretty standard PC based platform. If yes then you can create a bootable compact flash card with memtest86. Memtest86 is a self contained bootable image so setting it up is pretty easy.

[Wuerstchenhund]

So I have the opportunity to trade a Rohde & Schwartz

It's "Rohde & Schwarz" (no 't') 

FSL Spectrum Analyzer which has a few intermittent problems:

• It would occasionally at random times fail to boot (then a few hours later would have no problem and run for hours). During boot time winlogon.exe would throw an "application error" - "The instruction at XXX referenced memory at 0x00000000. The memory could not be read", and when you click OK it would go straight to the blue screen (see attached images).
• It would occasionally freeze after it's been running for a while. Then only a reboot takes it out of that state. I strongly suspect these two problems are one and the same.

It's certainly not a network problem as suggested by someone else, this error points to a hardware issue. I've seen this error a few times on R&S Windows XP based spectrum analyzers, and in my experience it's usually caused by one of the following:

• Defective RAM - relatively easy to check with tools like memtest86+ which can be booted from USB stick
• Defective flash storage on analyzers like the FSL which use a CF card instead of a hard drive or SSD. For some reason R&S had decided that Windows XP on a standard CF card is a good idea, which it really isn't unless you use a RAM disk which redirects all writes to RAM instead of flash. However, that isn't the case on these R&S instruments, which means the cards will wear out over time as they're not made for the amount of writes.
• PCB connectors worn out/defective, although that's very rare

For this case my guess would be on #2. I'd remove the CF card, create a backup and then replace it with a new one from a reputable brand like Sandisk (and from a reputable source to make sure it's no fake). The cards are cheap and easy to replace so that's a possible quick fix if that's the problem.

If it's RAM, well, then on the FSL that means replacing ICs as the RAM isn't socketed as in other R&S spectrum analyzers. In this case I hope the thing is cheap, as investing a lot of money into a FSL3 isn't worth it when SA's with much better specs like the FSP go for not that much money in working condition.

[ivaylo]

For this case my guess would be on #2.

Appreciate your response! How can I positively diagnose that it's #2?

[gowf67]

I might try something like this:

http://www.pendrivelinux.com/testing-your-system-for-usb-boot-compatibility/#more-231

Includes Memtest86+

[ivaylo]

OK, this is what I did so far:

memtest86 does not find any errors.

Ran chkdsk both locally from the instrument as well as mounted on another PC, result from there looks like:

C:\Documents and Settings\iiliev>chkdsk /x e:
The type of the file system is NTFS.
Volume label is DRIVE_C.

CHKDSK is verifying files (stage 1 of 3)...
File verification completed.
CHKDSK is verifying indexes (stage 2 of 3)...
Index verification completed.
CHKDSK is verifying security descriptors (stage 3 of 3)...
Security descriptor verification completed.
Windows has checked the file system and found no problems.

    993887 KB total disk space.
    391657 KB in 4059 files.
      1268 KB in 583 indexes.
         0 KB in bad sectors.
     12686 KB in use by the system.
      7024 KB occupied by the log file.
    588276 KB available on disk.

       512 bytes in each allocation unit.
   1987775 total allocation units on disk.
   1176553 allocation units available on disk.

Anything else I might be missing?

[Muxr]

Power supply fluctuations can also case problems like this. I'd inspect you power rails make sure they are providing clean power. Perhaps you got some electrolytic caps that are drying out.

[nctnico]

Seems like it could be any Windows related problem. I'd just re-install Windows -if possible-.
http://discussions.virtualdr.com/showthread.php?243986-RESOLVED-winlogon.exe-application-error

[Wuerstchenhund]

OK, this is what I did so far:

memtest86 does not find any errors.

That's good, although it's no guarantee that the RAM is fine.

Ran chkdsk both locally from the instrument as well as mounted on another PC,

Chkdsk doesn't help much as might show no problems even if the flash storage is defective. The only way to be sure is to replace the card with a new one.

However, as Muxr said power supply fluctuations can easily produce errors like this as well, so the first thing I'd do is to check the various voltages with a scope, and also check the caps (the FSL Series has been made over a very long time and some of them could very well be affected by the capacitor plague).

[amyk]

Try running benchmarks that stress more than just the memory, like Prime95 and LinX. Be sure the CPU heatsink is clean and has good thermal contact.

[ivaylo]

Thanks guys, all good advice... I'll try some more to replicate (hasn't happened since the first time, btw). But a general question here. Is this how you keep these older PC based instruments running? I have a couple of oscilloscopes and a SMU but nothing with a hard disk that I am aware of. Wuerstchenhund (spelling! ) suggested I get a FSP model, but isn't that an even older beast also running on XP (has a floppy disk on it's face for God's sake...). Or are the specs that dramatically different so constantly nursing the ancient PC (and probably hard disk) inside to health is worth it? You can tell I have very little experience with these, and was just hoping I get something for the lab a bit better than a Rigol (or is the R&S FSL even better than the 3GHz Rigol?)...

[Wuerstchenhund]

But a general question here. Is this how you keep these older PC based instruments running? I have a couple of oscilloscopes and a SMU but nothing with a hard disk that I am aware of.

PC based test instruments are not worse in terms of reliability or repair than other complex instruments. Hard drives are often a weak point but at least for Windows based instruments they can usually be replaced easily (these days with an SSD).

On the other side, having Windows on the instrument brings quite a few advantages.

Wuerstchenhund (spelling! ) suggested I get a FSP model, but isn't that an even older beast also running on XP (has a floppy disk on it's face for God's sake...).

Don't get mislead by the floppy drive, even later R&S analyzers came with them.

As to the FSP: there are two versions of the FSP, the first generation one (1093.4495.xx) which came out in 2000 and which runs Windows NT and comes with PS/2 mouse port, and the later generation (1164.4391.xx, came out around 2005) which runs Windows XP and comes with a faster processor and USB ports. R&S has kept the floppy drive in the later model as standard equipment unless you ordered the optional CF card reader.

The FSL came out in 2007 if I remember right, however the FSP has been replaced by the Windows 7 based FSV a around three years ago while the FSL is still sold.

Or are the specs that dramatically different so constantly nursing the ancient PC (and probably hard disk) inside to health is worth it? You can tell I have very little experience with these, and was just hoping I get something for the lab a bit better than a Rigol (or is the R&S FSL even better than the 3GHz Rigol?)...

My point was simply that investing a lot of money to get an FSL (which is a low-end spectrum analyzer, although a much better one than the Rigol) working may not be worth it when fully working second generation FSPs are often sold for pretty reasonable money. Of course it depends on what's wrong with the device and how much you'd have to pay to for the instrument plus to get it going again.

In terms of performance, R&S considers the FSP a mid-range analyzer but it's performance is much better than the FSL and even bests the HP 8566/8568 Series (which by many is considered the gold standard in spectrum analyzers).

[ivaylo]

Thank you, sir! This is the type of advice I was looking for. The fact that most companies don't list when a product was introduced and how long it was produced is maddening...

[carver]

I apologize for new reg and up the old topic.

someone can share the image dump of CF card(DD like), from any the FSL device ?

thanks in advance.

[ivaylo]

Sorry, I got rid of the device I was referring to in this topic long time ago. Haven’t owned an FSL since. Good luck...

[Eugene1970]

Hi.
I also need cf dump.
Have you got it?

[hamilton_qrn]

Hello,

I solved this the same problem with a Rohde & Schwarz FSQ26 Signal Analyzer doing the following steps:

1- During start-up press DEL to enter in BIOS CMOS Setup screen
2- Go to "Load Optimized Defaults" and press Enter
3- Press F10 to Save and Exit

After that, you need to set the time and date on WIN XP and that's it.