I do have some updates.After backing up the HDD, replacing the HDD and fixing a number of sectors with chkdsk the unit went back into XP, but then stops at the splash screen.
The checkIt Utility doesn't come up with any issues, so I assume the P4 board to be good.
I disassembled the unit and found a dead BIOS battery, and 5 blown caps. The battery was replaced and the caps are my job for tonight -
although I doubt it will have any noticable effect.
SW wise all drivers seem to be there. I am not sure however if I am missing the PowerPCs boot image, because
c:/vxBoot does not exist.
All references to this image come from various TDS7000 threads though.
So I read the service manual and followed the troubleshooting flowgraph.
check voltages
check Application
check Acquisition board
check PowerPC board
1: Check voltagesAt first I was quite happy with the manual, but that soon changed.
I checked the voltages as per table 4-7 in the manual
The two
6V signals on J3 D1/D11 are 4.25 V and the 12V on J1 are way off. //while reading this line just before posting I realize I should maybe check these signals with a scope, not a DVM. It might be pulsed.
After some investigation I am relatively sure that the manual is wrong. The two 4.25V signals come from an open drain output of two comparators which judge the 2.6V of the acq frontend as good.
Measuring with a DVM leads to the same result.
The description of J1 seems to be completely wrong. ABC 9,10 and 11 should be 12V, but are measured 5V GND and 5V, which makes perfect sense when following the traces.
My conclusion is that the power rails are all good.
2. Check the ApplicationI did not find any version number in the software, only a reference to TDS7000 software, which I found strange.
Since I had the backup I simply removed the original and installed the most recent Application SW from the Tek Website, just to find that
the reference to TDS7000 series seems to be correct. Now I know the version, but Application is still not starting properly.
3. Check the Acquisition boardChecking the ACQ board is supposed to be done in the Utilities Menu of the Scope SW
4. Check the PowerPC boardThe 7 segment display shows "
8." - power seems to be there, and according to the manual this means MPC740 initialisation and MPC106 initialisation
Again strongly misleading information in the manual.
There is no MPC106 on the board - It is a MPC107 - both DRAM controllers, but why would you not just write the correct information?
Same with the MPC740 - after long hours of datasheet and PowerPC History research, the only match between Tek Clock rate claims of the PPC and the
BGA package found on the PPC board that I could find is an MPC750CXe. But the only datahseet that comes up is for the MPC750CX.
I am relatively confident that the only noticable difference is the maximum clock rate - I hope at least.
At this point I startet to give the service manual as much credibility as conspiracy theories about the moonlandings.
5. Reverse engineering the PPC BoardAs the manual didn't really help a lot, I started poking at the board.
I was quickly annoyed by the fact that I couldn't reach the bottom side of the pcb, so I investigated if I could use the ATX connector on the board, which originally is used to power the P4
board the other direction to power the PPC from a standart ATX powersupply. Turns out it works like a charm. Fortunately the board indicators are a lot better than the manual, and putting a jumper onto the
2 pin header J612 labeled with "FORCE PWR ON" compfortably switches on the ATX PS under test conditions.
All Oscillators seem to be good.
PLL U441 seems to be fine (33.3MHz) as well
clock input to PPC chip 83.3MHz - no idea where that comes from - is there, and seems to match quite well to the sales claim of 583MHz PowerPC when an internal PLL factor of 7 is used.
2 of the 4 PLL config pins however seem to be the only 2 from roughly 300 pins which are not accessible by a via. The remaining 2 accessible signals are at matching levels to set a factor of 7.
The NVRAM carries a date code of 1st week 09 - dangerously old for an expected lifetime of 10 years when it comes to my taste.
I unsoldered the package, shortened a 40pin socket to match the 32pins, backed up the content and put the device into the newly installed socket.
Since I did this I received a replacement but I guess there are some issues with the content so I don't bother replacing it for now.
In the data I do find multiple records of the units serial number and model, as well as what looks like a license key, that is significantly longer than the
one printed on the back. I hope that means more options than the expected optionless unit.
The boot image location I expected to find after having read a few threads about 7000 series scope repairs however was nowhere to be found.
Large blocks of seemingly random information are in the image.
Also it seems as if several blocks of informations repeat inside the memory. A backup for older settings or calibrations maybe?
So I it might very well be that part of the memory is corrupted.
As the PPC chip doesn't seem to do a lot I checked the flash pins while switching on - nothing, all lines high.
I will repeat the same measurement with the nvram, but I expect it to be the same, as according to the wanky service manual it is not used as early in the POST.
I checked the 2 serial ports, but nothing there as well - they are also said to come up later in the POST
Later I found why - the
#HRESET (Hard Reset) pin of the PPC chip is pulled low permanently (only when powered on, otherwise HIGH-Z).
The SpartenXL seems to be configured successfully. Configuration in Master Serial Mode takes place, although I am not sure if I checked the DONE signal - I should check it again.
It has a XC17S01 config memory attached to it - I have nothing to read it out, so I will order one of these cheap logicanalysers in the hope to be able to get a binary image from it.
At 1.25MHz roughly it shouldn't need a proper tool for the job.
By the way, I am really mad at myself because I was so stupid to pay 600EUR for the 8 logic channels option on my DSO-X 2024A - every time I want to use it does not do the job.
I think Auto Scale is the only button I use less.
The Spartan2 is situated close to what could be a JTAG interface, so I quickly hooked up the Jtagulator and found a set of pins. Connecting with a Xilinx USB Platform cable to the pins
recognised earlier, I found an XC18V01 - which failed most of the communication efforts. Turns out it was an old version of the chip, for which the support was removed 2009 in Impact.
The idcode of the chip was changed, and simply correcting the XC18V01 bsd file in impact to accept an "X" instead of a "1" in one position of the idcode made it work.
The FPGA however does not start up properly I found later. The
#PROGRAM pin keeps being pulled low by some part of the reset logic which I have not reverse engineered sufficiently, yet.
I do not care to much about the DRAM or the DRAM controller at this stage of the POST as well.
Does anybody have a
datasheet of the T3W53AF Microprocessor on the board, I could not find one?
It seems to be called "Mite" from National Semi. I never heared from these, and i fear this one might be in control of starting up the rest.
I did find some connections to the reset logic, but no Idea if input or output, yet.
So for now I keep poking at the board to find the schematic of the reset section. Not very quick though with 12 layers.
Luckily the vias are not covered by solder mask - it would make the job close to impossible.
Any input is highly appreciated.