PDA

View Full Version : Bogus thermal problem with HDVR2



PlainBill
01-16-2004, 11:34 PM
I've hacked this HDVR2 with Sleeper's iso, and was waiting for a Netgear FA120 adapter to arrive when what claimed to be a temperature problem appeared.

I was watching TV when a blue screen popped up, saying the system had shut down because the internal temperature was too high; unplug and plug in again to restart. I do so, and hold my hand near the fan as the system restarts. There is a good air flow, but I'm a little surprised because the air is only slightly warm, and the case is slightly cool to the touch. I check the System Information screen as soon as I can; it reports the internal temperature is 43C.

The next morning I have the same message when I turn the TV on. I unplug the HDVR2 for an hour, then plug it in. Again, there is a good air flow, and now it's definitely cool. As soon as possible, I check the internal temperature and it's 35C. Within 2 more minutes, the same message pops up again! Quickly restarting the beast, I see the internal temperature is up to 37C. It's been working for the past several days, then tonight it really packed up. Part way through the 'Almost there, a few more seconds, please' screen the same "Too hot" warning pops up.

I'm convinced this is bogus; it has occured within 4 minutes of turning the system on, the cooling fan is running, neither inlet nor outlet vents are obstructed, and the case is never warm to the touch. Since the problem is now reproducible, I collected the serial output, which is attached for your information. Any advice is welcome.

Update: I've gone back to the original (unhacked) drive, and the problem persists. It's not the software, and the internal temperature isn't excessive. Does anyone know where the temperature sensor is, or how the temperature sensor software works?

PlainBill

yotool
01-30-2004, 10:05 PM
I have a similar problem except it is apparently random.

The unit isn't actually hot. I have all the thermal improvement mods installed, and the fan is definitely running, but I ocassionally get real whacko temp readings and the occassional "shutdown". I've searched everywhere for where the actual temperature sensor is located and/or what the type of sensor it is but to no avail.

I have a suspicion that the temperature sensor might the Zener Diode on the power supplyboard upstream of the fan (feeding back to some A/D on the main board via the "white ribbon cable"). This would be a logical spot for a temp sensor because if the air intakes or exhausts were blocked, or the fan died, the temp rise from the heatsinks for the TO-220 switches and dual rectifier mounted to the power supply heat sink right near the fan's output would see an immediate temp rise detectable by this strategically located zener. Just an educated guess though; could be wrong.

Any ideas anyone??? Where are the schematics????

PlainBill
01-30-2004, 11:33 PM
I have a similar problem except it is apparently random.

The unit isn't actually hot. I have all the thermal improvement mods installed, and the fan is definitely running, but I ocassionally get real whacko temp readings and the occassional "shutdown". I've searched everywhere for where the actual temperature sensor is located and/or what the type of sensor it is but to no avail.



I've resolved my problem. The thermal monitoring / control /shutdown is controlled by three modules, therm.o,fan.o (in /lib/modules) and fancontrol in /tvbin. In addition in the /proc directory is a module called therm. Typing cat /proc/therm will display the temperature.

After renaming fancontrol the system would no longer shut down due to thermal problems. Typing cat /proc/therm would sometimes give an error message 'No response', and sometimes give the temperature. Apparently fancontrol will shut down the system UNLESS it receives information from therm confirming the temperature is within safe limits.

PlainBill

yotool
01-31-2004, 12:18 AM
Interesting.

I saw the call to /proc/therm in the xPlusz code that optionally reports the hi/low temp situations. This same code making the call to /proc/therm was reporting the same bogus situation.

I still wonder then what is causing the 'No response' return from /proc/therm occassionally and tripping the unit's built in shutdown code. Could it be a random coincidence of the built in TIVO code making a call to /proc/therm at the sametime as the xPlusz call and they trip each other up?

I also now am thinking that the actual hardware based temp sensoring may be coming from circuitry internal to the IBM MPEG2 chip that runs so hot to begin with. I need some liquid freon to prove it out (the kind you can't get in the form of electronics parts cleaner anymore). I have to see if I can get some "Freeze It" that TV repairmen use for testing thermally sensitive components.

I have to get some double-sided thermal tape to apply a small heatsink to the MPEG2 chip since that can't hurt; only help.

But if what you say is true it will still trip up until I disable the fancontrol call. :confused:

PlainBill
01-31-2004, 10:15 AM
Interesting.

I saw the call to /proc/therm in the xPlusz code that optionally reports the hi/low temp situations. This same code making the call to /proc/therm was reporting the same bogus situation.

I still wonder then what is causing the 'No response' return from /proc/therm occassionally and tripping the unit's built in shutdown code. Could it be a random coincidence of the built 0in TIVO code making a call to /proc/therm at the sametime as the xPlusz call and they trip each other up?

I also now am thinking that the actual hardware based temp sensoring may be coming from circuitry internal to the IBM MPEG2 chip that runs so hot to begin with. I need some liquid freon to prove it out (the kind you can't get in the form of electronics parts cleaner anymore). I have to see if I can get some "Freeze It" that TV repairmen use for testing thermally sensitive components.

I have to get some double-sided thermal tape to apply a small heatsink to the MPEG2 chip since that can't hurt; only help.

But if what you say is true it will still trip up until I disable the fancontrol call. :confused:

I don't think it is caused by conflicting calls. There have not been a flood of people reporting this problem AND I verified the problem with the original (unhacked) OS. In my case, it reached a point where the system wouldn't boot - ever. I then renamed fancontrol and the system worked fine; therm responded often enough I was confident the system wasn't overheating. Lately therm responds every time I query it. I've even run for several hours with fancontrol enabled.

Your idea of using circuit cooler to locate the temperature sensor is a good one. It is certainly easier than looking up the specs on each of the chips on the board. Possibly you could accomplish the same thing using DRY compressed air, or CO2. I suspect it is in one of the LSI chips, but could be wrong.

Bottom line - I never saw a temperature spike, so had confidence that disabling fancontrol would not cause a problem. If you are seeing temperature spikes, I would be cautious about disabling fancontrol.

PlainBill

cali
02-01-2004, 09:14 PM
Unfortunately this one requires time without a schematic.

Search the board for a 4 0R 5 pin SOIC. It will look like a regular black transistor except it wouldnt have 3 legs...it would have more. There may be more than one on the board. Does the fan speed increase when the problem occurs?

On the series 1 the fan voltage was increased up 12V depending on the internal temp.

Hope this give you a starting pint.

Cali

yotool
03-17-2004, 07:01 PM
According to the schematics posted elsewhere on this site, there are actually 3 separate temperature sensors (P/N TC74 by Microchip). They are "serial" type addressable devices that share a common SMBus serial bus which seems to originate/terminate at the MPEG2 decoder.

The reference designators are U1000, U1001, and U47.

I have to see where thay are actually located...

The fan is PWM controlled by 2 2N2222 transistors that switch ground and the other fan pin is tied to 12V.

yotool
03-24-2004, 05:47 PM
I found U47 in between the 2 DVD Decoder ICs upstream and in line with the fan inlet.

As they are surface mount packaged devices, could the other 2 be on the bottom of the motherboard underneath some of the critical IC's?

cali
03-24-2004, 08:23 PM
I found U47 in between the 2 DVD Decoder ICs upstream and in line with the fan inlet.

As they are surface mount packaged devices, could the other 2 be on the bottom of the motherboard underneath some of the critical IC's?

Theres none on the bottom ( at least on Dtivos) so I'd suspect there are no parts on the bottom od a standalone.