Aeroflex 2975 CPU Problems

This forum exists for the purposes for discussing service monitors (This includes but is not limited to Motorola, HP, Aeroflex, GD, etc). Additional topics allowed include test procedures, interpretation of test results, where to find information about specific tests, antenna VSWR, return loss testing, duplexer and filter alignment, etc.

Moderator: Queue Moderator

Post Reply
k7dmk
New User
Posts: 13
Joined: Mon Oct 15, 2012 12:24 pm

Aeroflex 2975 CPU Problems

Post by k7dmk »

I'm working on an Aeroflex 2975 that is crashing after warmup. The LCD suddenly displays a blank white background and sometimes reboots and sometimes just hangs.

When I plug a keyboard and hit F2 to enter the bios, I'm getting some readings in the hardware monitor section that are questionable. It's showing the core voltage to be about 0.91 to 0.93 volts. That doesn't seem right for a 400 MHz Celeron CPU, if the bios is reading it correctly. The CPU temp starts out at about 52 degrees C and goes up to about 65 to 67 C after a few minutes and them the box crashes. I checked the thermal pads visually and they seem OK. The heat sink is getting hot.

Would someone with a 2975 please plug in a keyboard, hit F2 and see what they are getting in the "Hardware Monitor" for core voltafe and temp? I have been unable to find any service info on this unit and don't know if this core voltage reading is correct or a red herring.

This particular CPU board has the processor chip and memory on a mezzanine board that then plugs into the main CPU pcb. The serial no. range is 281X for this unit.

The service monitor was working normally until very recently. A new CMOS battery was installed and a few days later the problem started.

Any advise would be appreciated.

Dave, K7DMK
Al
Posts: 1045
Joined: Tue Sep 04, 2001 4:00 pm

Re: Aeroflex 2975 CPU Problems

Post by Al »

I have 2975, s/n 598X. On checking the bios' hardware monitor, I see a core voltage of .91V, and CPU temp of 58C/136F on startup and 64C/147F after the unit runs for 10 minutes or so. So it doesn't appear that you have a CPU fault. I don't believe that the CPU in my unit is a mezzannine style because of the later s/n, but I've never had the unit's skins off. I would start by looking very closely at the supply voltages to the CPU for switching noise over a period of time if you haven't already done so. If I didn't see anything abnormal, I'd start focusing on memory faults/intermittents.
k7dmk
New User
Posts: 13
Joined: Mon Oct 15, 2012 12:24 pm

Re: Aeroflex 2975 CPU Problems

Post by k7dmk »

Al: Thanks you very much for taking the time to get the bios hardware monitor parameters for me.

I pulled the CPU mezzanine board off of the main carrier PCB and examined the the thermal pads and also found the stepping number of the precessor. It's SL6SE and it is a ultra-low voltage CPU. The spec I found showed a core voltage in the range of about 0.95V. I didn't realize that Intel made older Celeron's that were only 400 MHz and had such low core voltage. The CPU is rated at only 4.5W thermal power.

The voltages all look good. There are several LED's on the CPU main board that show the various rails. I also have measured the rails and they are very close to nominal values. The main power supply delivers +5 volts, and +/- 15 volts. There is a 3.3V regulator and the CPU card and a led for that. The power supply was pulled and tested outside the unit with a partial load and it seems OK. When the unit crashes, all the led's are still on and the voltages read correctly. An interesting side note on the power supply; it has some sort of "watchdog" timer when disconnected from the main board shuts it off after about 6 minutes. I think perhaps the processor must sent a "keep me on" signal to the PS once it's booted.

I am getting a bad vibe from those mezzanine connectors.The problem started when the CPU board was removed to replace the CMOS battery and also to make backup's of the internal hard drive and to test those backups. So the board was in and out several times and was flexed, etc.

Some things that I'm seeing are partial boots where the unit will get half-way through the boot process and hang, random crashes and hangs while running, and other weird behavior.

Intermittent mezzanine connectors would be one possible explaination. They must have stopped using them for some reason. I wonder if they have been a problem as they age and perhaps the contact tension decreases or they otherwise degrade. They carry hundeds of connections.

I wonder if these connectors have a history of causing problems in these units? I also wonder it a new style CPU, without the separate processor board, would work in my unit and whether these are available for less than a king's ransom?

Thanks, Dave K7DMK
Al
Posts: 1045
Joined: Tue Sep 04, 2001 4:00 pm

Re: Aeroflex 2975 CPU Problems

Post by Al »

As I understand it, Dave, Aeroflex stopped supporting the 2975 around last December, and judging from others who've tried to get some support from Cobham(Aeroflex) they're told that there's no help from Cobham. It's possible that David Hagood(wowbagger on here) could put you on to a source for a replacement CPU - you could check with him for ideas because he was intimately involved with the 2975 software development. He also might be able to better point you in the right direction for troubleshooting your existing your CPU card.
k7dmk
New User
Posts: 13
Joined: Mon Oct 15, 2012 12:24 pm

Re: Aeroflex 2975 CPU Problems

Post by k7dmk »

Al: Thanks again for your comments and time. Intermittent problems like this are the worst to troubleshoot. And yes, I heard that Cobham has ceased support. We are on a Ham club budget and have more time then money, so I guess we will have to keep trying. It's pretty strange. Once it finally boots it can run for a few minutes to a few hours before it crashes. The fan in the back suddenly stops and the backlight stays on with a blank bright white screen.

I did clean and inspect the mezzanine connectors under a stereo microscope and they look OK. They were not really dirty or corroded in any way. Also applied some Caig Labs ProGold very lightly to the contacts for good measure. Things are still abnormal with a few more crashes since this was done so it's not fixed. I'm going to try carefully monitoring the 3 supply rails over time to be sure nothing is going on there.

Dave, K7DMK
k7dmk
New User
Posts: 13
Joined: Mon Oct 15, 2012 12:24 pm

Re: Aeroflex 2975 CPU Problems

Post by k7dmk »

Also, I forgot to mention that it may be that the little CPU is still overheating. With the covers off, booting into the bios and starting the hardware monitor brings the processor temp into the 70-73 degree C range after several minutes. Then the unit shuts down. When I place a desk fan over the CPU board, the temp stays in the low to mid 60's and it keeps running. Granted, other components besides the CPU are being cooled by the external fan. With the covers on and the internal fan running the processor may be cooler, but there is no way to measure it. There is no temp shut-down setting in the bios only CPU throttling at 100 degrees C. I wonder if there is some hidden temp shutdown threshold being reached?

The CPU cooling system design is pretty marginal. There is a thick thermal pad that bonds a small piece of copper to an aluminum plate. It looks like a lot of thermal resistance to the small flip chip surface area. There is about 1/2mm gap between the top of the cpu and the copper pad that has another fairly thick thermal pad. Bad setup. It's a low power processor but it's still getting pretty darned hot. I've had monster 100W 6 core CPU's run much cooler with a proper heatsink.

Dave, K7DMK
ljohn6
New User
Posts: 37
Joined: Sat Jan 08, 2011 12:47 pm
What radios do you own?: Kenwood,Icom,Hytera

Re: Aeroflex 2975 CPU Problems

Post by ljohn6 »

Sorry to hear of the problems Dave, I have the same board but mine is a Pentium 200, I could get the temps but mine is a dead unit as my hard drive failed and will no longer boot beyond the loading symbols, my temps would be during start up and not a running temp. Good luck with the fix hope you find a solution.
John
Post Reply

Return to “Test Equipment & RF Equipment Alignment”