Need help with circuit design/problem

[Lmandrake]

OK, here is proof positive that a little knowledge is a dangerous thing. I have a watercooled tbird system and I wanted to protect it from meltdown when it is running all by its lonesome. I attempted to do this with two sensors and a circuit that would shut the machine down and keep it shut down until it is manually restarted. I built a water flow sensor from an old dc fan (details here) and another thermal sensor with a remote thermistor mounted on my waterblock. Both are wired so that they output approximately 5 volts when either the temperature gets too high or the water flow sensor detects no water flow.

I made a simple SCR circuit to shut my computer down and keep it shut down. It is powered off the 5v standby line and and the SCR drives a 5v relay that is normally closed and is in the power good line from the power supply. The 5v outputs from either the flow sensor or the temperature sensor are fed into the trigger of the SCR. A normally closed momentary switch is used to open the scr circuit when I want to reset the machine. When I first built it, I put a 30 mf electrolytic capacitor in the trigger line to smooth out any transients and prevent false latching. A picture of the SCR board (I don't have any means to do a schematic):

picture


In spite of my ignorance, it actually worked when I first put it together. I was worried about using a 30 mf cap as a filter, but never got around to changing it out.

Now it does not work properly. Applying 5v volts to the trigger input will not latch the SCR. HOWEVER, if the thermal sensor is activated and outputs 5v and then goes inactive and the 5v goes low, the SCR will then trigger (i.e. low-high-low).

I replaced the 30 mf cap with an .1mf polyester film cap. That is in the picture and is now jumpered as it is either bad or is the wrong kind of cap. I also just added the diode to the trigger line.

I assume that the SCR has gone funky, but I really don't know. I also think using the 30 mf electrolytic cap was a bad choice.

I would appreciate any input anyone has and welcome any criticism (other than that I am a total dumb*ss). My electronics knowledge is limited to what I can pick up from rat shack pamphlets and simple circuits on the net.

I could also replace the scr circuit with 5v latching relay. I assume that if I use that properly (with a blockin diode) that would be a good choice. Is this correct?

Thanks for any help.

 

[blahblah99]

first of all, what the heck is a 'scr'?

If your goal is to shut down the computer using a thermistor AND a water flow sensor to detect when something's wrong, your circuit will not work reliably because 1) your output voltage from the thermistor will not reach the full 5V or the full 0V, and 2) you cannot just tie the outputs of the water flow sensor and the thermistor together.

But you can tie the thermistor and the sensor to two seperate comparators and feed those outputs to an OR gate. The output of the OR gate will either be 0 if there is no alarm, and a 5V if either the thermistor is too hot, or the sensor isn't detecting water flow. Feed that output into a transistor to drive your relay.

Hopes this helps!


if u need help, pm me or whatevers and i can draw up a quick schematic for you.

 

[CTho9305]

an SCR is a "silicon-controlled rectifier". 3 pins like a transistor. normally, no current (well, very little) flows across it. If you put a small current at the gate, however, it turns on, and stays on even after you remove the base current. to turn it off, you have to stop supplying current (at the anode/cathode).

I just learned that this morning. Hope its right

As blahblah99 said, you can't just tie the two outputs (from thermister and flow meter) together. However, you might be able to get by using two diodes. here is an OR gate - I'm not sure whether you need to ground it - the two diodes alone might work. You would put the "out" going to the gate of the SCR.

Also, I don't believe you actually need an SCR. If I understand your post right, in this case you're using it as a switch. As soon as it turns on, the power supply should turn off. Instantly. Since everything is off, it doesn't matter whether the SCR stays on, or a transistor returns to an off state. For that purpose, a transistor should work just as well, right?

edit: IIRC, to turn the power supply on/off, you short pin 14 to ground. couldn't you short it to ground through a transistor (saturation for the transistors I normally play with is .2V. if you put the output of the OR gate to the gate of the transistor, it will probably saturate, which will result in an instant off). I haven't found any descriptions of how to use the "power good" pin (8)

 

[Lmandrake]

Let me clarify a few things.

I am not using a thermistor by itself. I built a circuit that employs a thermistor and a comparator to turn a relay on or off depending on the temperature. As employed in my device, a relay is activated when the temperature goes to high. When the relay closes, 5v is fed to the gate of the SCR. (The resistance and impedance of the relay are needed for the circuit too work). It is essentially a thermostat using a remote thermistor.

The advice about an OR gate is well taken, I will do this. However, as a practical matter, the likelihood that either the thermal signal or the water flow would both go high at the same time is very remote. Nonetheless, I am going to build an OR gate.

I am using an SCR for a number of reasons. First, it is not correct that shutting down the power to the mainboard shuts off all power. Unless you deprive the power supply of AC current, the 5v standby is always active. If your "kill" circuit does not stay in the "disable" state when activated by an input, your system will just keep cycling on and off. Suppose the system is getting too hot. The "kill" mechanism shuts it down, everything cools a bit, the input to the transistor goes low again and the system restarts. Then it gets too hot, the kill mechanism cuts it off, etc. etc. etc. etc. Second, if I used a device that had to have power to hold the entire system in an off state (i.e. no AC in to the PS), I would have needed a second power supply just for the "kill" device.

Once turned "on," the SCR will maintain its state regardless of the state of the input to the gate. It turns off only when the current flowing through it is interrupted. In other words, it stays shut off even if the precipitating "bad" condition goes away until a user turns a switch to interrupt the current through the SCR and turn off the "kill" mechanism. An SCR or a latching relay will stay in one state until told to do otherwise.

Also my first post was inaccurate, I am not using "power good" to shut down, I am using PS-On as was suggested. Instead of using a transistor, I am using a relay wired into PS-on so that it is normally closed. If the SCR is triggered by an input to the gate, the relay is activated and the PS - on circuit is opened and and everything except the 5v standby is shut off.

 

[Haribo]

<< I have a watercooled tbird system and I wanted to protect it from meltdown when it is running all by its lonesome. >>



Ever heard of MBM and SHDN?

 

[CTho9305]

<<

<< I have a watercooled tbird system and I wanted to protect it from meltdown when it is running all by its lonesome. >>



Ever heard of MBM and SHDN? >>


the difference is that Lmandrake's setup would probably shut down in time to save it even if the heatsink fell off

 

[Lmandrake]

Certainly I have heard of MBM and other software solutions. I may be overly cautious, but I never liked the idea of software shut down programs. If you lock up, where is the protection?

BTW, I also wanted a solution that would keep my computer inert and unstartable if a cooling failure had occured as my kids would start the machine again in a heartbeat even if they smelled smoke.

 

[Superdoopercooper]

<< an SCR is a "silicon-controlled rectifier". 3 pins like a transistor. normally, no current (well, very little) flows across it. If you put a small current at the gate, however, it turns on, and stays on even after you remove the base current. to turn it off, you have to stop supplying current (at the anode/cathode). I just learned that this morning. Hope its right >>



You learn well, young grasshopper.

 

[Lmandrake]

For those who are interested, a schematic of the SCR circuit is here


A schematic of the thermostat circuit can be found here The thermostat circuit relay is wired to feed +5v to the SCR circuit trigger when an overheat condition occurs. As noted above, the water flow sensor circuit does the same thing when there is no flow. A schematic for that will have to wait for another day.


Thanks to all who have replied and shown interest.

 

[CTho9305]

<< You learn well, young grasshopper. >>


its all about 19 year old radio shack books by forrest m. mims, iii. (specifically, "Getting Started in Electronics"). I read them a long time ago, but now that I've had a freshman ECE course I now actually understand it all cool stuff. if a book is older than you are, you konw its decent

 

[dajeepster]

it actually worked when I first put it together...blah...blah..blah...Now it does not work properly.

hmm.. is it broke now?

ok... troubleshooting 101.
step one:
first... if it was working properly at first, and now is not... try repairing it to it's original state first when it worked properly. After it's working and you know that it's working...go to step two.

step two:
if step one does not work... stop and fix step one... once step one is working, then start doing modifications to your circuit. Don't make modifications to a broke circiut... why, you ask? cause it's broke. you repair broken items, you modify working items.

if it's broke and you really don't know how to fix it... try the easter egging method... replace one part at a time until it's fixed (replace with the same part) (it's called the "easter egging method" because you hunt around until you find it.) Trouble shooting isn't something that you can learn from a book, it's basically trial and error based upon knowledge that you have gained. just be careful if you are working on a live circuit.

Go forth now young grasshopper with the knowledge you have gain, and don't get stepped on by the kid in combat boots.

 

[DocBurner]

The or gate is a must. If you don't use it when you drive 5 volts and the other is at 0 volts who knows whats going to happen...

 

[CTho9305]

<< The or gate is a must. If you don't use it when you drive 5 volts and the other is at 0 volts who knows whats going to happen... >>


something is gonna get current it doesn't want, or you'll drain too much from the source at 5v and break something else

i talked to Lmandrake over AIM and he said the setup worked yesterday when he forgot to turn his pump on, but I dont know if he's still having trouble.

 

[Lmandrake]

I added an OR gate to the trigger inputs and it seems to have solved the problems.