Resolved Mac Pro 1,1 (2006) Apple Service Diagnostic error: FBDIMM voltage sensor reading below low point

[Soba]

I was recently given several 2006 Mac Pros and one of them is intermittently failing 2 tests that are part of the Apple Service Diagnostic suite. There is an error on the voltage sensors of both DIMM risers that reads as follows:

Voltage Sensor - FBDIMM Riser A + 12V rail and VRs test 1 - Sensor Reads Within Operating Range
-Check to ensure that sensor reads within operating range
ERROR - Sensor reading below low point
- T E S T F A I L E D -

Voltage Sensor - FBDIMM Riser B + 12V rail and VRs test 1 - Sensor Reads Within Operating Range
-Check to ensure that sensor reads within operating range
ERROR - Sensor reading below low point
- T E S T F A I L E D -

I've attached a photo showing this, as well.

I have run all tests dozens of times. These DIMM riser sensor tests fail intermittently, but not always. All other tests have passed every time.

Apple Hardware Test shows no problems with the system. The system also seems to operate normally, at least during my testing so far; macOS installs fine, software installs and runs, and so on, but this is no indication that the system will work reliably, long term.

The system has 4GB of RAM installed on (8) 512MB DIMMs. I removed both DIMM risers and cleaned them, and also unplugged and reseated all DIMMs, but as expected, this had no impact on the error.

I have two other identical Mac Pros and they do not exhibit this error.

Any suggestions for where the problem lies and how I might fix (if possible)? Many thanks.

 

[b0fh666]

if the computer works fine and reliably, the problem might be in the sensor itself. i’d just live with it.

 

[Soba]

if the computer works fine and reliably, the problem might be in the sensor itself. i’d just live with it.

I'm still testing reliability of the machine with some RAM tests and will report back in a day or two. I'm not sure how reliable the system is at this point, and I'm planning to give these systems to someone else, so it's important that they have no problems.

 

[konqerror]

I'm not sure how reliable the system is at this point, and I'm planning to give these systems to someone else, so it's important that they have no problems.

It's a 14 year old system with no manufacturer support and no availability of spares... if reliability is important, the only answer is to junk it and get something newer.

 

[Soba]

It's a 14 year old system with no manufacturer support and no availability of spares... if reliability is important, the only answer is to junk it and get something newer.

Spare parts are readily available from countless sources, and in the case of questions I don't have the answers to, there are forums. Throwing away equipment that is still serviceable is not one of my habits, hence the desire to fix it, if that's feasible (or even determined to be necessary after further testing).

 

[konqerror]

Spare parts are readily available from countless sources, and in the case of questions I don't have the answers to, there are forums. Throwing away equipment that is still serviceable is not one of my habits, hence the desire to fix it, if that's feasible (or even determined to be necessary after further testing).

Untested used parts. If you were able to find a tested, warranted FBDIMM at this point, it would cost more than the whole computer.

I'm not doubting the computer can be made to be useful, I'm saying your need for reliability is not realistic. It is a simple fact that all machines have lifetimes. Polymers degrade, solder joints crack under repeated thermal cycling, transistors degrade via electromigration...

You also should realize the environmental impact of the excess energy use of a workstation-class machine of that era. A single FBDIMM AMB sitting idle burns 6 watts, more than the entire idle power consumption of a faster, modern laptop with screen on. The idle power of that system is 171 watts, just about the power of two fully-loaded Macbook Pros at max blast.

I would bet if you do a lifecycle analysis, if you're doing real work, throwing away that computer and replacing it with a modern system will be better for the environment.

 

[Soba]

Untested used parts. If you were able to find a tested, warranted FBDIMM at this point, it would cost more than the whole computer.

I'm not doubting the computer can be made to be useful, I'm saying your need for reliability is not realistic. It is a simple fact that all machines have lifetimes. Polymers degrade, solder joints crack under repeated thermal cycling, transistors degrade via electromigration...

You also should realize the environmental impact of the excess energy use of a workstation-class machine of that era. A single FBDIMM AMB sitting idle burns 6 watts, more than the entire idle power consumption of a faster, modern laptop with screen on. The idle power of that system is 171 watts, just about the power of two fully-loaded Macbook Pros at max blast.

I would bet if you do a lifecycle analysis, if you're doing real work, throwing away that computer and replacing it with a modern system will be better for the environment.

Thank you for your posts.

 

[Soba]

if the computer works fine and reliably, the problem might be in the sensor itself. i’d just live with it.

Many thanks for your input.

To close the loop on this, after several days of diagnostics and stress testing the machine, I have been unable to get it to crash or even to display any problem whatsoever. That being the case, it does indeed look like just a faulty sensor.

For the record, I tried swapping out all the DIMMs and even installed a different set of DIMM riser cards pulled from an identical system, but the error remained. (If there really is an issue somewhere, it's in the power supply or the logic board, but is apparently so inconsequential that it doesn't affect the functioning of the machine.)

I'm going to assume it's just a faulty sensor and call it a day.

Update: I changed the RAM in this system to a matched set of 32GB DIMMs from Data Memory Systems and the error went away completely. Even after 20+ runs of the diagnostic tests, the error did not show up again. It looks like the RAM that was installed had some kind of incompatibility that was causing the error; installing RAM from Data Memory Systems (a company that I highly recommend, by the way, because everything is guaranteed compatible) solved the problem.