Agilent 54624A oscilloscope not starting

[MaxFrister]

I have a 54624A oscilloscope that isn’t starting.  When powered on the fan starts, the floppy spins,  it flashes all keys, flashes the pulse-width key, lights the pattern key and stops.  The display remains blank and it is unresponsive to all keys.  I believe it should flash pulse-width, flash pattern, flash the more key and start.

All the power rails seem okay.   The cmos battery is okay.  The supervisor IC is causing a reset and the 68020 seems to be running (it continues to generate watch-dog resets and memory activity).  I’ve tried holding down a soft-key on power-up (nvram reset) with no change.

I’ve ordered and will install new SRAM but I’m not optimistic that that is the problem.

Another possibility is that the firmware is corrupt but the IC’s are of such fine pitch I'd rather not remove them unless I am pretty certain that that is the problem.

I’m running out of  ideas to figure out what the processor is waiting on.  Anyone know what is supposed to be happening while the “pattern” key is lit?  Is there a way to get a boot status log?  Anyone have a schematic/CLIP?  Anything else I can check?

Thanks.

[MaxFrister]

As I feared, replacing the SRAM did not change the symptom (booting halted with unresponsive keys).

After checking the subsystems as best I could, I repeated something I had done at the start and looked for text on the 4 obvious rs232 lines.  This time, I found activity on one of them.  After some very neat and professional work to make sure it didn't blow up my "logic analyzer" [scare quotes are intentional] I was able to read the text. 

0Could not find valid system, checking disk.
Downloading /DEV/FLOPPY/SYS5462X.JZP
Could not open SYS5462X.JZP
Downloading /DEV/FLOPPY/SYS5462X.BIN
Could not open SYS5462X.BIN
Could not find valid system on disk, checking GPIB.
No hpib module is installed!
Unable to open 'GPIB' for downloading
Could not find valid system, remaining in ROM monitor.
ROM Monitor
Enter ? for help.
->

Eureka, it finally told me what was wrong.  After a few phone calls and an hours drive, I located a computer that could still write a 3-1/4 floppy.  I copied the latest SYS6462X.JZP onto a blank diskette and tried the scope with the following result.
Quote

\0\xFF\0Could not find valid system, checking disk.
Downloading /DEV/FLOPPY/SYS5462X.JZP
Opened SYS5462X.JZP

Entering programRoms(0x20000, 1502983, 0x1ffe, 1, 1)

Entering validPart(0x20000, 16, 4, 2)
  Device: 29F160T (0x0020,0x22cc)
  readIdAdrs = 0x10
  Memory width 4 bytes
  x16 Read ID: 0x0004,0x22d6
  next part

  Device: 29F160T (0x0001,0x22d2)
  readIdAdrs = 0x10
  Memory width 4 bytes
  x16 Read ID: 0x0004,0x22d6
  next part

  Device: 29F800T (0x0001,0x22d6)
  readIdAdrs = 0x10
  Memory width 4 bytes
  x16 Read ID: 0x0004,0x22d6
  next part

  Device: 29F800T (0x0004,0x22d6)
  readIdAdrs = 0x10
  Memory width 4 bytes
  x16 Read ID: 0x0004,0x22d6

A valid part has been recognized:
  Device: 29F800T (0x0004,0x22d6)

Flash ROM is 29F800T with width=2, memorywidth=4

Flash ROM is 29F800T with width=2, memorywidth=4

    Erasing memory.
    ............

Programming from 20000 to 18ef06.
    .......................
\x11  Local build, Oct 23 2003 17:42:33
  Directory: /col/scopes/jediPlus2/projMuskrat
  Cal factors defaulted because of corruption: instSerialNumber > INST_SERIAL_NUMBER_SIZE

    error - ram volume verification error
\x11

[jedi muskrat???].

After reinstalling the case for cooling the scope was operational.  It failed self-test until a user cal was complete.  It now passes all tests and seems to be working nominally.

It still requires a few things:

1. Validate the calibration
2. Replace the long intensity knob (anyone have a spare?)
3. Figure out how to reinsert the serial number.
3. Figure out which game this scope has installed.

[Runco990]

Congratulations! 

Sounds like someone botched a system update.  Happened to me once and I thought I bricked the scope.  Fortunately this scope can be recovered by trying a floppy again, even after "bricking" it.  In my case I wrote another disk and it worked.

I love these scopes.  I use 2 of them.

[MaxFrister]

Hints for Non Starting Agilent 54621A, 54621D, 54622A, 54622D and 54624A oscilloscopes.  [Porbabably also relevant to the 54641A/D and 54642A/D]

In trying to diagnose this problem, I’ve learned a bit about these scopes. I thought I would record some notes on the off chance it might help someone in the future.

Service manual.  There is a limited service manual (https://literature.cdn.keysight.com/litweb/pdf/54622-97037.pdf)  available for these scopes.  It has some minimally useful information and things to check but no schematics.   I have not been able to locate a schematic. There is a component level service manual including schematics for the earlier generation of these scopes but they are pretty different.

Light code.  A correctly functioning scope flashes all lights and then the sequence “pulse-width”, “pattern”,  “more”, and then displays a trace.

Simple things.  If you have a scope that doesn’t start, there are some simple things to check:

• That the cmos battery is 3.0V.
• That the power supply voltage are correct without excessive ripple.  Test points are located on the bottom of the scope and shown in the “service” manual.
• Try resetting the nvram by holding down a soft-key and power cycling (but see the next note).

Soft Keys. These scopes are notorious for failing soft-keys.  To check open up the front panel and put an ohm-meter on the soft-key traces and try each key.  Resistance should be less than about 100 ohms.  If not clean the board with IPA but NOT the carbon on the rubber key.  Those can be made temporarily better with a soft carbon pencil.

Boot log.  When the scope fails to boot, it prints out diagnostic info on the serial rs-232 port on the back of the scope. The default (which can be changed) is 8 bits, no parity and 9600 baud.  Beware that rs-232 is a +/-15V (or more) protocol with active low.  In order to read it using my cheap logic analyzer I need to clamp it and invert it.

Firmware recovery.    If the firmware is missing or corrupted, the scope will attempt to recover it by loading it from a floppy or over GPIB. 

IC inventory.  I’ve attached an inventory of the major ICs that I created in trying to reverse engineer the main board. Note that in this document the “front” of the board is the part you see when you turn the scope upside down.    I was able to identify all but a few of the important ICs.  There are many assumptions and guesses in this list, so take it with a grain of salt.

[MaxFrister]

I received this question via message but I thought I'd reply in public in case others have better ideas:

    The problem for me is that all LEDs on the front panel light up without further reaction of the device. All voltages from the PSU are corect. The floppy disk drive does not initialize. Unfortunately, there is no transmission on RS232 during the start of the device. The rest of the symptoms as in the other posts, i.e. cyclical resets of the processor, etc ... I have a question, have you met with flash memory dump?
    I suspect that corrupted firmware is a problem in my oscilloscope, which, although old, I like it a lot and I'm doing my best to fix it
    Sorry for the trouble, but I'm trying to find help wherever I can...

I also really liked this scope.  I've replaced it with something more modern that takes longer to boot and I'm not sure I like as well.

Certainly sounds like a similar problem.  What equipment do you have available to you?  Another scope (you know you're not allowed to own just one old scope)?

I think the fact that it lights the panel lights means that the processor is starting the boot process.  It is just failing earlier than mine.

I think that the firmware is meant to be programmed via either the floppy or GPIB.  I don't know where the boot-loader resides....perhaps in the processor?  Unfortunately I gave the machine away so I can't check.  The only  firmware I have is the code that is intended to be loaded via floppy/GPIB.


I assume you checked the power supplies for ripple.  I'd suspect other things around the processor.  The processor itself (check address and data lines).  SRAM - U314/U315 in particular.  Supervisor U340.  RTC U350.   Failing anything obvious, I might just change the SRAM.

Best of luck and please let us know what you find.