Image Map Image Map
Results 1 to 4 of 4

Thread: Help fixing socket 5 upgrade module - Evergreen Spectra 333 - Vout does not adjust

  1. #1

    Default Help fixing socket 5 upgrade module - Evergreen Spectra 333 - Vout does not adjust

    I have an Evergreen Spectra 333 upgrade module which I use in a dual socket 5 NEC Proserva V Plus along with a Kingston upgrade module. Both modules are set to 2.8 V and are running the Pentium 233 MMX. They were working fine until I was lifting up a CD-ROM (while powered up) to check the model number, at which point there must have been a short circuit detection because the PSU shut off completely. Upon powering up again, I was getting BSOD's in NT4 and W2K related to NTKRNL-something. If I set both upgrade modules to 3.0x instead of 3.5x, so running at 200 MHz, I could use the system normally again.

    I started to think that the brief short must have affected one of the upgrade modules. I checked the Vcc2 voltage on the Kingston and it was 2.8 V as it should be, but on the Evergreen, it is at 2.2 V, even though it was jumpered for 2.8 V. The Pentium MMX is designed to run at 2.8 V, so I've concluded, for now, that the issue is with the Evergreen upgrade module. When measuring the Vcc2 voltage output on the Evergreen module, the system starts out booting with 2.4 V, but with load, it levels off around 2.21 V. Once in Windows 2000, at idle, it still sits at 2.2 V, which is far too low.

    For voltage regulation, the Evergreen uses an LP2951C, which is a low-dropout regulator in the SOIC-8 package. But its output current is limited to, I think, 200 mA, which isn't nearly enough for the Pentium MMX 233 (around 6 A). To make-up for the low output current, the Evergreen modules also uses a 10 A PNP power transistor with part number TIP110A. Both modules have this separated P on it, which indicates the brand. Perhaps an older Pericom or Power Integration logo? Knowing this would help finding the proper datasheet.

    It is likely that one or both of the LDO and PNP have gone bad. I've performed a basic out of circuit PNP test, but it didn't find a short. On the attached jumper sheet, on the bottom right, I show which tests I did. But passing this test doesn't necessarily mean the PNP is performing within specs.

    The simplest route is just to replace the LDO and the PNP with modern equivalents from digikey. As I cannot seem to find the old datasheets that contain the separated P logo, I can't see the exact specs. Does anyone know which modern equivalents will work on this upgrade? A concern is that too much variation in IC specs could alter the set voltages and I was hoping to avoid swapping out the SMD resistors to get the output voltages back to 2.1, 2.8. 2.9, and 3.2 V (jumper dependent).

    I've enclosed photos of the Evergreen Spectra 333 unit. I have desoldered the molex power connector, the PNP, and heatsink. Also shown is the jumper table and a schematic diagram of how the LDO, resistors, jumpers, and PNP are connected to form Evergreen's VRM circuit.

    EDIT: One other point I wanted to make is that when things were working properly, I did not measure Vcc2 to see if the Evergreen was ever outputting 2.8 V. The Spectra-333 normally runs and AMD K6-2 at 2.2 V. Its possible that the module only runs at 2.2 V, but then why would there be jumpers and silk screened settings for the other voltage. Following the circuit diagram, the jumpers do change a resistance at Vcc2, so I assume it should work at the other specified voltages.

    Thanks a lot for your help.
    Attached Images Attached Images
    Last edited by feipoa; February 26th, 2020 at 04:50 PM.

  2. #2
    Join Date
    Apr 2015
    Location
    Austin, Texas
    Posts
    1,925

    Default

    Before blaming the Evergreen interposer, did you try swapping them around in opposite sockets? It could be a motherboard issue.

  3. #3

    Default

    Yes, tried that before posting. The interposer takes its power from the 5 V molex, so I don't think the socket would matter much with respect to seeing 2.2 V at Vcc2. The interposers are outputting to Vcc2.

    EDIT: I am considering:

    BDX34C for the PNP
    https://www.digikey.ca/product-detai...mSeq=318860896

    And either TI or ON Semi for LP2951CM,
    https://www.digikey.ca/product-detai...mSeq=318860914
    https://www.digikey.ca/product-detai...mSeq=318860911

  4. #4

    Default

    I replaced the power transistor with the STMicroelectronics BDX34C and now the Evergreen Spectra module outputs the proper voltages. I didn't need to change the LP2951C voltage regulator chip. Looks like passing various PNP measurements between the base, emitter, and collector for the standard transistor test did not help in this situation. I have one extra BDX34C and a few extra LP2951C that I will keep in the Evergreen Spectra's box for safe keeping.

    The jumper sheet for this module had to be measured as I couldn't find a complete listing anywhere. For your convenience, it is provided here:

    CPU MULTIPLIER

    MUL J3 J2 J1
    2.5x 1-2 1-2 O
    3.0x 1-2 O O
    3.5x O O O
    4.0x O 1-2 1-2
    4.5x 1-2 1-2 1-2
    5.0x 1-2 O 1-2
    5.5x O O 1-2
    6.0x O 1-2 O


    CPU CORE VOLTAGE VCC2

    VCC2 J5 J4
    2.20V O 1-2 (supposed to be 2.1V)
    2.82V O 2-3
    2.90V 1-2 O
    3.10V 2-3 O (supposed to be 3.2V)
    3.70V O O
    1.86V 1-2 1-2
    2.05V 1-2 2-3
    1.89V 2-3 1-2
    2.10V 2-3 2-3

    I suppose if you have a 1.6 or 2.0 V K6-2/3+ chip, you'd want to use the 2.05 V setting.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •