CPU Over-Temp Detector / PSU Shutdown help

[xy]

i just made a CPU Over-Temp Detector / PSU Shutdown as a fail safe in case my pump fails according to the instructions on the procooling site.

here is my lay out for the board:

and here is the finished board:

the board sorta works when the green LED lights the Power-on signal passes through and when the red LED lights the Power-on signal is cut off. BUT the circuit won't fire and latch pins 2+3, 8+9 (cut circuit and light red LED) unless l push the reset switch, in which case no matter where i turn the pot. it will not reset and latch pins 3+4, 7+8 (connects power-on signal and lights green LED). what am i doing wrong

here is a list of the components that i used (scroll all the way down)

[BladeRunner]

Far be it for me to pour cold water on your efforts but is there a reason why you couldn't use MBM5 in conjunction with SHDN?. It's what I'm using and if I get 40°C CPU temp my system auto shuts down in the normal way.

Thinking about it there are 3 areas your unit would be better>

1. If you are using win 2K in "standard PC mode" or an older PC type so you have to press the power button to turn it off

2. If you don't have CPU bios temp reporting from your mobo

3. If Windows just happens to lock-up and or crash in some way when shuting down, which is always a possibility with Windows.

So far the two apps in the links below have saved my system once, when the pump fuse blew for no apparent reason.

full details on MBM5 with SHDN and how to use them HERE

Do make sure it works once installed by setting a shutdown temp below your max, and seeing that it will actually shutdown the PC. It wont work if you have to press the power button to turn off your PC, (like you do if you are using Win2K in "standard PC" mode).

It does amaze me we don't have user selectable bios shutoff temp settings with the AMD mobos just like my old Pentium systems used to.?

[xy]

i thought about using software but because i already made the circuit board i would like to use it if at all possible, if not then i guess i will go with MBM5 and SHDN.

[Joe]

Blade you just dont understand the principle behind a stand alone shutdown system do you???


if your HSF falls off your CPU will lock up almost instantly... and when that happens your lil shutdown feature in MBM will do jack shit.

In most everycase that you want to shut the system down cause its over heating there is a risk the core will lock far before it gets shut down.

[BladeRunner]

No I do understand the principal, and I don't consider heatsinks falling off myself, but you are correct. However I was commenting in the context of this thread and forum we are in, namely Liquid cooling.

Surely something as sever as a heatsink falling off will lead to core death, (and VGA card death), before the standalone can do anything. How and where does it detect the CPU temp?

In reflection it is a very good second back up but I still think it's wise to have the software as well anyway, (especially if you have nothing else). Like I said before, is there a valid reason why it's not done in the bios now with AMD's?. My Pentium's had a setting shutdown if CPU reaches X temp.............

[xy]

Originally posted by BladeRunner:
How and where does it detect the CPU temp

i was planning on drilling a small hole on the maze 2 and insert a temp probe in there.

[BladeRunner]

Yeah that would be ok but I was really asking the question in Joe's "heatsink context", as with your probe placement in the w/b it still wouldn't help save the CPU if the waterblock fell off.

This to me is possible but very unlikely and to be honest if a heatsink or waterblock does fall off then a dead CPU is probably deserved . I think what you and I are trying to prevent is Pump failure death. In my system if waterflow stops, the block and coolant in it heats up fairly slowly and SHDN works fine. Lack of coolant in the block could be borderline though so I will look into a separate shutdown controller it is a good idea. I will however have reservoir low level warning and pump flow warning via the VGA smart doctor when I get around to it.

[xy]

ok the problem was the pin placement
the pin out is:
1 2 3 4 5 6 7
14 13 12 11 10 9 8
but on the board the pins were like so:
7 6 5 4 3 2 1
8 9 10 11 12 13 14

[GuyBFF]

Another way to make a reliable shutdown controller is with a DigiDoc 5 and a relay to one of the fan outputs. This will obviously cost more, but I like the DigiDoc 5 for other reasons.

Wire the relay to be normally closed and connect between the power good to the motherboard. This way you have a temperature setable shutdown.

I also use MBM5 and SHDN like bladerunner suggested, but if my CPU is running at 1700+ it will lock if it warms to much before shutting down, so I have a backup.

Another thing the DigiDoc can do is control fan speed to a preset if you use a LM317T, one resistor and a pot. Basically you can set a low speed, then when the DigiDoc kicks in it will increase voltage to the full 12V. Because the LM317T will see a higher signal on the ADJ it won't add any more power, and won't allow any sort of current drain. The reason I use a voltage regulator is they will supply current without dropping voltage like a rheostat or resistor at fan start (where DC fans can draw 5 times the current, droping voltage to a point where the fan won't sometimes start).

[ck42]

xy:

Sorry about the confusion with the relay. I just now ran across this thread.

The ommision of the pinout for the relay was my fault. I should have included it in the original article but didn't realize I forgot it until I started getting emails asking me how to hook-up the #$%#$&&*% thing.

Is it working ok now that you have the pinout correct?

[xy]

Let me clarify the ping out that I tried to switch was the LM339N, that didn’t work so I switched the relay now when I hit the reset switch the green will light and the Power-on will pass through. However no matter how I set the potentiometer the relay will not trigger

[Butcher]

Um, I may be being dense here but where exactly on that board does the temp probe connect? I've looked but there doesn't seem to be a connector for it anywhere...

[xy]

if you look at the top right corner on the first image of the first post you will see 2 purple blocks labeled thermostat. that is where the temp probe connects

[ck42]

xy:

What is the value of the pot you're using?

Are you using the exact relay that I specified in the article? or did you buy one somewhere else?

I'll get back with you shortly to try and troubleshoot what is happening. It's a fairly simple circuit...don't worry, we'll get it figured out.

edit:

Just noticed something. From the looks of the transistor, you have it in backwards. Make sure that it's the Emitter that is going to the ground line and the Collector that's going to the realy Pin1.

In your pix, it looks like you have the transistor turned so that the exact opposite is happening which would explain why turning the pot NEVER activates the realy.

For this type of transistor, the Base is pretty much always the middle pin. The Emitter and Collector are the left and right pins. Double check and make sure.

[xy]

thanks for the hint my transistor is connected the wrong way. i used the exact components that were specified in the article, i think i killed the transistor and the LM339 when i was switching everything around , so I’m going out and getting some new replacements. right now the relay is turned 180 deg. so the line is facing the bottom. i think it will work once i get the new components because i can manually trigger the relay by connecting pin 1 of the relay and the ground then reset the circuit by hitting the reset switch.

[ck42]

xy,

Ok...hope the new components fix you up.

You are correct that you should be able to falsely trigger the circuit by connecting Pin1 of the relay to ground. This is exactly what the circuit is doing under normal circumstances when the high temp limit is reached. So, if by doing that, the circuit behaves normally, then everything else must be hooked up correctly

Sounds like you're ALMOST there.

[hielko]

Originally posted by xy:
thanks for the hint my transistor is connected the wrong way. i used the exact components that were specified in the article, i think i killed the transistor and the LM339 when i was switching everything around , so I’m going out and getting some new replacements. right now the relay is turned 180 deg. so the line is facing the bottom. i think it will work once i get the new components because i can manually trigger the relay by connecting pin 1 of the relay and the ground then reset the circuit by hitting the reset switch.

Sounds familair to me . I also did something like that... But after replacing the LM chip and the transistor with a 2N 2222 everything worked perfect.

[xy]

i replaced the transistor and the LM339 chip, now everything works!!

thanks for helping me everyone!