DEC 3000/800 AXP boot problem - VMS

This is a discussion on DEC 3000/800 AXP boot problem - VMS ; Hi, just a quick enquiry. I have aquired a DEC 3000 Model 800 AXP and have a boot problem. The system was booting about once every ten power ups. The status LEDs don't move in the case where the system ...

+ Reply to Thread
Results 1 to 14 of 14

Thread: DEC 3000/800 AXP boot problem

  1. DEC 3000/800 AXP boot problem

    Hi, just a quick enquiry.

    I have aquired a DEC 3000 Model 800 AXP and have a boot problem. The
    system was booting about once every ten power ups. The status LEDs
    don't move in the case where the system checks do not 'kick in', they
    remain on '00'. I have cleaned the entire box and used contact cleaner
    on all connections, and re-seated the motherboard and IO module.

    Previously when the system did 'kick in' all tests completed correctly
    except the keyboard/mouse test. I am using a VT520 connected via an
    original DECconnect cable. I managed to install both OpenVMS 6.1 (the
    original disk received with the unit) and OpenVMS 8.3 (although this
    didn't boot, but probably due to the fact that I selected DECnet but
    don't have any thick/thin cables connected).

    Sorry to be so long winded. So the question is, apart from reseating/
    cleaning cables/connections, is there anything else I could check/test/
    try? I have enough competence to measure voltages and possibly even
    power ripples and the like, but I don't have a logic analyser.

    Do the SROMs age at all? It is a standard x512 EEPROM. It has a
    sticker over the programming window (assuming it has one). Is there a
    battery backed module that could be causing intermittent problems?

    Thanks for any pointers,

    Mark.


  2. Re: DEC 3000/800 AXP boot problem

    In article <1188244928.730714.166170@57g2000hsv.googlegroups.c om>,
    urbancamo wrote:

    > Hi, just a quick enquiry.
    >
    > I have aquired a DEC 3000 Model 800 AXP and have a boot problem. The
    > system was booting about once every ten power ups. The status LEDs
    > don't move in the case where the system checks do not 'kick in', they
    > remain on '00'. I have cleaned the entire box and used contact cleaner
    > on all connections, and re-seated the motherboard and IO module.


    Did the system really power on? Were the fans running? Was the LED
    idiot light lit up on the power supply?

    I've seen systems where the power supply didn't like to come on from a
    cold start. Sometimes leaving it switched on (but "dead") for 10
    minutes or so, and then cycling the power, would bring them to life.

    > Previously when the system did 'kick in' all tests completed correctly
    > except the keyboard/mouse test. I am using a VT520 connected via an
    > original DECconnect cable.


    For this configuration, you want the CONSOLE environment variable to be
    SERIAL, i.e.
    >>> SET CONSOLE SERIAL


    In this mode, missing keyboard/mouse result in "soft" test failures that
    don't block progress.

    > I managed to install both OpenVMS 6.1 (the
    > original disk received with the unit) and OpenVMS 8.3 (although this
    > didn't boot, but probably due to the fact that I selected DECnet but
    > don't have any thick/thin cables connected).


    Network cable problems would not keep VMS from booting. I can't think
    of a case where it would even keep DECnet from starting.

    VMS V8.3 should boot just fine. What are the symptoms? You should be
    getting some information on the console terminal.

    > Sorry to be so long winded. So the question is, apart from reseating/
    > cleaning cables/connections, is there anything else I could check/test/
    > try? I have enough competence to measure voltages and possibly even
    > power ripples and the like, but I don't have a logic analyser.
    >
    > Do the SROMs age at all?


    I haven't heard of the SROMs failing on these systems. If the SROM was
    bad, it's very unlikely that you could get to the SRM console prompt.
    (The >>> prompt.) The SROM code is loaded into the CPU's instruction
    cache at power-up; it contains the first instructions the CPU executes.
    Without SROM, you'd never find memory, or the console, or the firmware
    ROM.

    Firmware ROM corruption is not unheard of, but it should be detected by
    checksum and other problems. If you get console output, and no
    ROM-related error messages, I'd assume the FW ROMs (there are two
    physical ROM parts) are fine.

    > It is a standard x512 EEPROM. It has a
    > sticker over the programming window (assuming it has one). Is there a
    > battery backed module that could be causing intermittent problems?


    The NVRAM in this system is integrated into the clock module. If the
    system is retaining date and time, and the environment variables, after
    you unplug it for a while, the battery in the module is probably fine.

    You're well past the 10-year battery life of the module, but if the
    system spent most of its life powered on, the battery is probably not
    depleted.

    -- Robert

  3. Re: DEC 3000/800 AXP boot problem

    Thanks for the reply Robert :-

    > Did the system really power on? Were the fans running? Was the LED
    > idiot light lit up on the power supply?


    Yes, you here the fans wind up and then settle down each time I power
    on, and the power LED comes on. It actually seems to start better from
    cold than once I start cycling the power every 30 seconds or so. The
    failure scenario then proceeds with the two status LEDs sticking at
    '00' rather than counting down from 'FE' onwards. Nothing comes up on
    the terminal.

    > I've seen systems where the power supply didn't like to come on from a
    > cold start. Sometimes leaving it switched on (but "dead") for 10
    > minutes or so, and then cycling the power, would bring them to life.


    See above comment - it appears to be the opposite.

    > For this configuration, you want the CONSOLE environment variable to be
    > SERIAL, i.e.
    > >>> SET CONSOLE SERIAL

    >
    > In this mode, missing keyboard/mouse result in "soft" test failures that
    > don't block progress.
    >


    Yes, I do see these 'soft' failures in the output on the terminal.

    > > I managed to install both OpenVMS 6.1 (the
    > > original disk received with the unit) and OpenVMS 8.3 (although this
    > > didn't boot, but probably due to the fact that I selected DECnet but
    > > don't have any thick/thin cables connected).

    >
    > Network cable problems would not keep VMS from booting. I can't think
    > of a case where it would even keep DECnet from starting.
    >
    > VMS V8.3 should boot just fine. What are the symptoms? You should be
    > getting some information on the console terminal.
    >


    After the VMS banner, I get a DECnet informational message and then no
    more output, no disk activity, and the halt button doesn't have any
    effect.

    > I haven't heard of the SROMs failing on these systems. If the SROM was
    > bad, it's very unlikely that you could get to the SRM console prompt.
    > (The >>> prompt.) The SROM code is loaded into the CPU's instruction
    > cache at power-up; it contains the first instructions the CPU executes.
    > Without SROM, you'd never find memory, or the console, or the firmware
    > ROM.
    >
    > Firmware ROM corruption is not unheard of, but it should be detected by
    > checksum and other problems. If you get console output, and no
    > ROM-related error messages, I'd assume the FW ROMs (there are two
    > physical ROM parts) are fine.
    >
    > > It is a standard x512 EEPROM. It has a
    > > sticker over the programming window (assuming it has one). Is there a
    > > battery backed module that could be causing intermittent problems?

    >
    > The NVRAM in this system is integrated into the clock module. If the
    > system is retaining date and time, and the environment variables, after
    > you unplug it for a while, the battery in the module is probably fine.


    I don't remember having any issues with the date being wrong, except
    initially setting it when installing VMS.

    > You're well past the 10-year battery life of the module, but if the
    > system spent most of its life powered on, the battery is probably not
    > depleted.


    I failed to mention that occassionally on a 'successful boot' (in the
    sense that I get to the chevron prompt) the boot messages contain the
    following error:

    > ASIC ?? 002 0020


    > T-ERR-ASIC COREIO
    > address = f0080280
    > data read = 0
    > data exp = aaaaaa


    The status leds display 'EF'. Having looked this up in the service
    manual this appears to indicate a problem with the IO module. The
    possible solutions include reseating the system module and the IO
    module, both of which I have done.

    Even though this system only cost me 5 GBP I would love to get it
    running purely because it is such an astounding piece of engineering.
    All the fans are shock mounted, the hard drive is also shock mounted,
    when you open the motherboard side of the case there are literally
    *no* cables whatsoever, the power supply simply slots out (after
    removing power cables) and apart from the SCSI cables there are a
    couple of short power cables and that is it. A work of art I would
    have said!

    Many thanks,

    Mark.


  4. Re: DEC 3000/800 AXP boot problem

    urbancamo wrote:
    > [...snip...]
    > Even though this system only cost me 5 GBP I would love to get it
    > running purely because it is such an astounding piece of engineering.


    Mark, out of curiosity, did you get this machine from me ?

    I'll be releasing another couple of machines into the hobbyist community
    in a few weeks (3000-600 and others).

  5. Re: DEC 3000/800 AXP boot problem


    R.A.Omond wrote:

    > urbancamo wrote:
    > > [...snip...]
    > > Even though this system only cost me 5 GBP I would love to get it
    > > running purely because it is such an astounding piece of engineering.

    >
    > Mark, out of curiosity, did you get this machine from me ?


    No It came from a guy in Appleby in West Moreland.

    > I'll be releasing another couple of machines into the hobbyist community
    > in a few weeks (3000-600 and others).


    I also have a DEC 3000-600 which was bought for me by the University I
    was studying my PhD for - I managed to reclaim it last year. The '600
    is working beautifully (OpenVMS, Digital Unix and OpenBSD).

    Mark.


  6. Re: DEC 3000/800 AXP boot problem

    On Tue, 28 Aug 2007 06:29:39 -0700, urbancamo
    wrote:

    > Thanks for the reply Robert :-
    >
    >> Did the system really power on? Were the fans running? Was the LED
    >> idiot light lit up on the power supply?

    >
    > Yes, you here the fans wind up and then settle down each time I power
    > on, and the power LED comes on. It actually seems to start better from
    > cold than once I start cycling the power every 30 seconds or so. The
    > failure scenario then proceeds with the two status LEDs sticking at
    > '00' rather than counting down from 'FE' onwards. Nothing comes up on
    > the terminal.
    >
    >> I've seen systems where the power supply didn't like to come on from a
    >> cold start. Sometimes leaving it switched on (but "dead") for 10
    >> minutes or so, and then cycling the power, would bring them to life.

    >
    > See above comment - it appears to be the opposite.
    >
    >> For this configuration, you want the CONSOLE environment variable to be
    >> SERIAL, i.e.
    >> >>> SET CONSOLE SERIAL

    >>
    >> In this mode, missing keyboard/mouse result in "soft" test failures that
    >> don't block progress.
    >>

    >
    > Yes, I do see these 'soft' failures in the output on the terminal.
    >
    >> > I managed to install both OpenVMS 6.1 (the
    >> > original disk received with the unit) and OpenVMS 8.3 (although this
    >> > didn't boot, but probably due to the fact that I selected DECnet but
    >> > don't have any thick/thin cables connected).

    >>
    >> Network cable problems would not keep VMS from booting. I can't think
    >> of a case where it would even keep DECnet from starting.
    >>
    >> VMS V8.3 should boot just fine. What are the symptoms? You should be
    >> getting some information on the console terminal.
    >>

    >
    > After the VMS banner, I get a DECnet informational message and then no
    > more output, no disk activity, and the halt button doesn't have any
    > effect.
    >
    >> I haven't heard of the SROMs failing on these systems. If the SROM was
    >> bad, it's very unlikely that you could get to the SRM console prompt.
    >> (The >>> prompt.) The SROM code is loaded into the CPU's instruction
    >> cache at power-up; it contains the first instructions the CPU executes.
    >> Without SROM, you'd never find memory, or the console, or the firmware
    >> ROM.
    >>
    >> Firmware ROM corruption is not unheard of, but it should be detected by
    >> checksum and other problems. If you get console output, and no
    >> ROM-related error messages, I'd assume the FW ROMs (there are two
    >> physical ROM parts) are fine.
    >>
    >> > It is a standard x512 EEPROM. It has a
    >> > sticker over the programming window (assuming it has one). Is therea
    >> > battery backed module that could be causing intermittent problems?

    >>
    >> The NVRAM in this system is integrated into the clock module. If the
    >> system is retaining date and time, and the environment variables, after
    >> you unplug it for a while, the battery in the module is probably fine..

    >
    > I don't remember having any issues with the date being wrong, except
    > initially setting it when installing VMS.
    >
    >> You're well past the 10-year battery life of the module, but if the
    >> system spent most of its life powered on, the battery is probably not
    >> depleted.

    >
    > I failed to mention that occassionally on a 'successful boot' (in the
    > sense that I get to the chevron prompt) the boot messages contain the
    > following error:
    >
    >> ASIC ?? 002 0020

    >
    >> T-ERR-ASIC COREIO
    > > address = f0080280
    > > data read = 0
    > > data exp = aaaaaa

    >
    > The status leds display 'EF'. Having looked this up in the service
    > manual this appears to indicate a problem with the IO module. The
    > possible solutions include reseating the system module and the IO
    > module, both of which I have done.
    >
    > Even though this system only cost me 5 GBP I would love to get it
    > running purely because it is such an astounding piece of engineering.
    > All the fans are shock mounted, the hard drive is also shock mounted,
    > when you open the motherboard side of the case there are literally
    > *no* cables whatsoever, the power supply simply slots out (after
    > removing power cables) and apart from the SCSI cables there are a
    > couple of short power cables and that is it. A work of art I would
    > have said!
    >

    Try cleaning the temperature sensor.

    > Many thanks,
    >
    > Mark.
    >




    --
    PL/I for OpenVMS
    www.kednos.com

  7. Re: DEC 3000/800 AXP boot problem

    On 28 Aug, 16:50, "Tom Linden" wrote:
    > Try cleaning the temperature sensor.


    Tom, do you have any idea where it is on an '800? I tried the tech man
    but couldn't find a mention. I'm presuming it's gonna be near the
    processor?

    Cheers, Mark.


  8. Re: DEC 3000/800 AXP boot problem

    On Tue, 28 Aug 2007 12:45:50 -0700, urbancamo
    wrote:

    > On 28 Aug, 16:50, "Tom Linden" wrote:
    >> Try cleaning the temperature sensor.

    >
    > Tom, do you have any idea where it is on an '800? I tried the tech man
    > but couldn't find a mention. I'm presuming it's gonna be near the
    > processor?


    It is usually near the fan on the power supply, But I have never even seen
    an 800
    >
    > Cheers, Mark.
    >




    --
    PL/I for OpenVMS
    www.kednos.com

  9. Re: DEC 3000/800 AXP boot problem

    In article <1188330350.273447.325990@o80g2000hse.googlegroups. com>,
    urbancamo wrote:

    > On 28 Aug, 16:50, "Tom Linden" wrote:
    > > Try cleaning the temperature sensor.

    >
    > Tom, do you have any idea where it is on an '800? I tried the tech man
    > but couldn't find a mention. I'm presuming it's gonna be near the
    > processor?


    I don't believe there is a temperature sensor in these systems.

  10. Re: DEC 3000/800 AXP boot problem

    > I don't believe there is a temperature sensor in these systems.

    As the fans initially spool up and then calm down, surely there must
    be something controlling the speed. If not a temperature sensor, then
    maybe something measuring power draw?

    I found problem causing the ASIC error - there is a post-manufacturing
    hack on the IO module that involves two pins of a PLCC being lifted
    from their pads and soldered to a diode and resister in parallel. The
    other end is soldered via a wire to a pad on the motherboard. The glue
    holding the hack to the top of the PLCC has long since deteriorated
    and one of the pins was very close to its' neighbour and pad. I moved
    this clear and since then a couple of boots have now shown the error.

    What I'm noticing now however is that the system is hanging
    intermittently. My focus now is on the power supply. I have heard that
    capacitors can deteriorate in a power supply - and have personal
    experience of old capacitors blowing up. What could I check on the
    power supply to determine if that is the problem?

    Thanks for the help, Mark.


  11. Re: DEC 3000/800 AXP boot problem


    urbancamo wrote:

    > and one of the pins was very close to its' neighbour and pad. I moved
    > this clear and since then a couple of boots have now shown the error.


    That should say 'NOT shown the error'. Too early in the morning...


  12. RE: DEC 3000/800 AXP boot problem

    > -----Original Message-----
    > From: urbancamo [mailto:mark@wickensonline.co.uk]
    > Sent: August 29, 2007 3:51 AM
    > To: Info-VAX@Mvb.Saic.Com
    > Subject: Re: DEC 3000/800 AXP boot problem
    >
    > > I don't believe there is a temperature sensor in these systems.

    >
    > As the fans initially spool up and then calm down, surely there must
    > be something controlling the speed. If not a temperature sensor, then
    > maybe something measuring power draw?
    >
    > I found problem causing the ASIC error - there is a post-manufacturing
    > hack on the IO module that involves two pins of a PLCC being lifted
    > from their pads and soldered to a diode and resister in parallel. The
    > other end is soldered via a wire to a pad on the motherboard. The glue
    > holding the hack to the top of the PLCC has long since deteriorated
    > and one of the pins was very close to its' neighbour and pad. I moved
    > this clear and since then a couple of boots have now shown the error.
    >
    > What I'm noticing now however is that the system is hanging
    > intermittently. My focus now is on the power supply. I have heard that
    > capacitors can deteriorate in a power supply - and have personal
    > experience of old capacitors blowing up. What could I check on the
    > power supply to determine if that is the problem?
    >
    > Thanks for the help, Mark.


    Fwiw, I have had the occasional hang with my DEC 3000 (now running V8.3), but it has
    never hung if I do an additional INIT command after it powers up and beforeentering
    the boot command.

    Regards


    Kerry Main
    Senior Consultant
    HP Services Canada
    Voice: 613-592-4660
    Fax: 613-591-4477
    kerryDOTmainAThpDOTcom
    (remove the DOT's and AT)

    OpenVMS - the secure, multi-site OS that just works.




  13. Re: DEC 3000/800 AXP boot problem

    I put the PSU back together as is and I got the scope on the '800 this
    evening.

    Findings:

    1. The 5V line is running at 5.15V
    2. The 3.45V line is running at 3.45V
    3. The ripple on both 5V and 3.45V lines is approx. 5mv at about 30
    KHz

    I've attached the oscilloscope and meter images (nice to have a PC
    based scope!)

    Observations:

    Both 5V and 3.45V lines appear to be pretty well DC regulated.
    I wouldn't have thought that 5mv of a.c. ripple would cause any
    serious problems.
    I did a finger test on the 3.45v regulator components and couldn't
    find anything 'hot' - the large heatsink to which the regulators is
    attached gets slightly warm
    I did a finger test on the CPU heatsink and the surrounding large
    PLCCs and these get hot - almost too hot to touch.

    Conclusions:

    Maybe the fans are worn and not pulling the air through the box quick
    enough. As I mentioned, they spin up quite high and then settle down
    to a low tick over.

    Maybe the fan regulation isn't working correctly - I've not noticed
    the fans speed up once the box has been on for a while. I'm not sure
    where I would look for a temperature sensor. I can ask on the groups
    and see if anyone knows, but it'll probably be up to me to find out.
    One obvious place would be under the main CPU - do you think this
    would be integrated into the main board layout and then come out to
    one of the PSU connectors.

    I can measure the voltage on the fan circuit and quite possibly trace
    the connection back to the PSU and see if I can work out how it is
    regulated. It might be regulated on power load instead? Temperature
    would seem more sensible however.

    I noticed that with the lid off the box it did seem to be quite stable
    - I didn't really have long enough to come to any firm conclusions. I
    guess running it with a desk fan pointing at the main circuit board
    for a while might give an idea.

    What do you recon? The 5V line is a little high but I wouldn't have
    thought that in itself would cause a problem. The PSU is rated at 750W
    so it may be that fully loaded the 5V line would dip to 5V from 5.15V.

    Cheers for the help,

    Mark.


  14. Re: DEC 3000/800 AXP boot problem

    On Sep 1, 6:02 pm, urbancamo wrote:
    > I put the PSU back together as is and I got the scope on the '800 this
    > evening.
    >
    > Findings:
    >
    > 1. The 5V line is running at 5.15V
    > 2. The 3.45V line is running at 3.45V
    > 3. The ripple on both 5V and 3.45V lines is approx. 5mv at about 30
    > KHz
    >
    > I've attached the oscilloscope and meter images (nice to have a PC
    > based scope!)
    >
    > Observations:
    >
    > Both 5V and 3.45V lines appear to be pretty well DC regulated.
    > I wouldn't have thought that 5mv of a.c. ripple would cause any
    > serious problems.
    > I did a finger test on the 3.45v regulator components and couldn't
    > find anything 'hot' - the large heatsink to which the regulators is
    > attached gets slightly warm
    > I did a finger test on the CPU heatsink and the surrounding large
    > PLCCs and these get hot - almost too hot to touch.
    >
    > Conclusions:
    >
    > Maybe the fans are worn and not pulling the air through the box quick
    > enough. As I mentioned, they spin up quite high and then settle down
    > to a low tick over.
    >
    > Maybe the fan regulation isn't working correctly - I've not noticed
    > the fans speed up once the box has been on for a while. I'm not sure
    > where I would look for a temperature sensor. I can ask on the groups
    > and see if anyone knows, but it'll probably be up to me to find out.
    > One obvious place would be under the main CPU - do you think this
    > would be integrated into the main board layout and then come out to
    > one of the PSU connectors.
    >
    > I can measure the voltage on the fan circuit and quite possibly trace
    > the connection back to the PSU and see if I can work out how it is
    > regulated. It might be regulated on power load instead? Temperature
    > would seem more sensible however.
    >
    > I noticed that with the lid off the box it did seem to be quite stable
    > - I didn't really have long enough to come to any firm conclusions. I
    > guess running it with a desk fan pointing at the main circuit board
    > for a while might give an idea.
    >
    > What do you recon? The 5V line is a little high but I wouldn't have
    > thought that in itself would cause a problem. The PSU is rated at 750W
    > so it may be that fully loaded the 5V line would dip to 5V from 5.15V.
    >
    > Cheers for the help,
    >
    > Mark.




    In regards to the soft failures WRT missing keyboard/mouse, I found
    this
    old note in my personal archives:

    >
    >From: "Thomas G. Kreutz"
    >Subject: SUMMARY: Auto_start fails without a mouse!
    >Date: 1998/05/26
    >Message-ID: <896221032.8538@news.Colorado.EDU>#1/1
    >Distribution: cu
    >Approved: news@news.colorado.edu
    >Sender: daemon@lace.colorado.edu
    >Delivered-To: alpha-osf-managers@sws1.ctd.ornl.gov
    >Followup-To: poster
    >X-Note1: message-id generated by recnews
    >X-Note2: mail msgid was <9805241901.AA00838@flash.Princeton.EDU>
    >Organization: University of Colorado at Boulder
    >Newsgroups: cu.alpha.managers
    >
    >SUMMARY:
    >
    > If you remove the mouse and keyboard from a DEC 3000, it will not boot up
    >automatically after losing power unless the console variable SERVER is set to
    >ON (i.e. >>> set server on). When SERVER=OFF, the machine expects to have
    >both a mouse and keyboard upon power-up; without either, it displays a fault
    >condition and waits patiently for you to boot the system manually.
    >
    > As well documented in the manuals, you must also have set the AUTO_START
    >variable to either BOOT or RESTART. In addition, this assumes that you are
    >working on an Ralternate consoleS, with the tiny RS3S switch on the back set
    >in the down position.
    >
    > It is also possible to fake out the system by wiring a mouse/keyboard
    >loopback connector:
    >
    > 1 8
    > o o-o o o o-o o
    >
    > o o o o o o o
    > 9 15
    >
    >by connecting pins 2 and 3 as well as 6 and 7 of a male DB15
    >connector. (One person thought it was perhaps 7+8 rather than 6+7).
    >
    >Many, many thanks to all those who responded so quickly:
    >
    >



    We have several AXP 3000-800s boxes at remote sites, and found this

    SET SERVER ON

    SRM console command very very helpful :-)


    Verne


+ Reply to Thread