How to determine if 7805/7812 is broken or not?

[lloydi12345]

Okay, so I have this circuit of microcontroller, transistors to control lamps, and power regulator. The input voltage is ~17Vac and when rectified it is ~24Vdc. I am feeding the dc input to 7812 and the output is connected to 7805's input. The output of 7805 then is fed to the microcontroller. There are 20 transistors controlled by 15mA. So the consumption (without including the IC) is roughly 300mA just for the transistors. The loads' supply are connected directly to 24Vdc. The circuit is fully functional on the 1st day but screwed up on the 3rd day.

(Correct me if I am wrong)Doing my approximate calculations, I came up with this:
7812:
(24-12)(0.3) = 3.6 Watts of heat

7805:
(12-5)(0.3) = 2.1 Watts of heat

As you can see the regulators obviously gets hot so I used two big heat sinks. It takes one hour and a half to make the sinks untouchable. I turned the whole circuit ON for 2 hours then turned it off for 5 mins. The sinks cooled down just like brand new. I did the same process and so on for 3 times and everything went fine. So that's a total of 8 hours on and 20 mins. off.

Then the other day I tried 4 hours and turned it off for 5 mins. and repeated the same process once again. After every 4 hours it really got too hot but the circuit is still in a good condition.

Just to check the limit of the regulators, I tried everything ON the next day for 8 hours straight without any turn off time. I took some random glances on the circuit and I noticed the problem around 6-8 hours. The circuit resets. I knew it resets because I made a routine that could only be executed once every-time the circuit is switched. The circuit resets randomly. Sometimes it resets after a minute, two, three, or more, there's no really definite time.

The question is which do you think is broken? 7812? 7805? both? How could I check the regulators if they are broken using multi-tester only?

Thank you for your time reading.

 

[Mosaic]

7805 drives the MCu, so that's the likely prob. However, does your load create inductive spikes...motor /relay? If so that could be a prob. Do u have decoupling caps on the Mcu power pins? BTW waht is the max power that is switchable by the MCu?...20 outputs is a lot.

 

[lloydi12345]

Yes, I placed decoupling caps beside mcu, caps on regulators as seen on the datasheet. Sorry but I don't know what's the max power it can produce. I used PIC18F4620. There's no inductive spikes since I used 24v tiny light bulbs. There's no other load connected to it.

 

[audioguru]

Heat sinks are designed to keep the IC from getting too hot. Yours do not. Why not?
Are your heatsinks inside a closed box? They need to radiate and conduct their heat to open air.
Thermal compound should be used between the IC and the heatsink to conduct heat better.

 

[lloydi12345]

Are your heatsinks inside a closed box? They need to radiate and conduct their heat to open air.
Thermal compound should be used between the IC and the heatsink to conduct heat better.

They are not enclosed in a box and I used a lot of thermal compound on it. So what do you think is broken? Can 7812 still be fine?

 

[MikeMl]

Sounds like your regulator(s) are right on the verge of going into (self-protection mode) thermal shutdown.

Set up a fan. Blow room air on the heatsinks while everything is on and running. If no shut down, your heatsinking without forced air movement is marginal.

Get a hair dryer. Blow hot air on the heatsinks while everything is on. If that induces a shutdown more quickly, that also proves that the heatsinks are marginal...

 

[lloydi12345]

Sounds like your regulator(s) are right on the verge of going into (self-protection mode) thermal shutdown.

Set up a fan. Blow room air on the heatsinks while everything is on and running. If no shut down, your heatsinking without forced air movement is marginal.

Get a hair dryer. Blow hot air on the heatsinks while everything is on. If that induces a shutdown more quickly, that also proves that the heatsinks are marginal...

So you're saying that the regulator still works fine? The strange thing here is that, even when the heat sinks are at room temperature, the circuit still resets randomly, making the regulators not get any hotter.

 

[Mike odom]

Neither one is probably broken, but I'd say the 7812 is going into thermal shutdown first, since it is dissapating the most wattage.
the 78xx family of regulators have a thermal shutdown, so they turn themselves off before they melt.

I have the same thing, a 7812 being fed from a +24VDC supply, supplying 320mA (.32A) max (16 4-20mA circuits). I use a 3 watt resistor to pull some of the heat out of the 7812 regulator. The resistor gets too hot to touch, but it runs nonstop.

To calculate the resistor value, set the regulator input to a minimum of 16v at 400mA. 24v - 16v = 8v. 8v/.4mA = 20 ohms.
I*I*r = .4 * .4 * 20 = 3.2W, so use 2 10 ohm 3W resistors in series. You can work the formula backwards if you want to pull out half the heat... 1.6W/(.4A*.4A) = 10 ohm, or use a 10 ohm 3W resistor to pull half the heat out of the regulator.

You can figure out if the regulator is going into thermal shutdown by it's dissipation. What is the part number of the heatsink you are using, and I am assuming you are using the regulator in a TO220 package? The TO220 has a thermal resistance junction to case of 5°C/W, so with 3.6W the max case temp would be 107°C, so at 30°C ambient, you would need a heatsink with a thermal resistance less than 21°C/W.

In air, the TO220 has a thermal resistance to air of 65°C/W, which means the highest temperature you can dissipate 3.6W without a heatsink is -109°C ambient, or the most power you can dissipate at 30°C would be 1.46W.

View attachment 12V regulator.PDF

 

[Mike odom]

So you're saying that the regulator still works fine? The strange thing here is that, even when the heat sinks are at room temperature, the circuit still resets randomly, making the regulators not get any hotter.

Even though the heatsinks are cool, the case of the regulators may still be too hot.

 

[lloydi12345]

Neither one is probably broken, but I'd say the 7812 is going into thermal shutdown first, since it is dissapating the most wattage.
the 78xx family of regulators have a thermal shutdown, so they turn themselves off before they melt.

I have the same thing, a 7812 being fed from a +24VDC supply, supplying 320mA (.32A) max (16 4-20mA circuits). I use a 3 watt resistor to pull some of the heat out of the 7812 regulator. The resistor gets too hot to touch, but it runs nonstop.

To calculate the resistor value, set the regulator input to a minimum of 16v at 400mA. 24v - 16v = 8v. 8v/.4mA = 20 ohms.
I*I*r = .4 * .4 * 20 = 3.2W, so use 2 10 ohm 3W resistors in series. You can work the formula backwards if you want to pull out half the heat... 1.6W/(.4A*.4A) = 10 ohm, or use a 10 ohm 3W resistor to pull half the heat out of the regulator.

You can figure out if the regulator is going into thermal shutdown by it's dissipation. What is the part number of the heatsink you are using, and I am assuming you are using the regulator in a TO220 package? The TO220 has a thermal resistance junction to case of 5°C/W, so with 3.6W the max case temp would be 107°C, so at 30°C ambient, you would need a heatsink with a thermal resistance less than 21°C/W.

In air, the TO220 has a thermal resistance to air of 65°C/W, which means the highest temperature you can dissipate 3.6W without a heatsink is -109°C ambient, or the most power you can dissipate at 30°C would be 1.46W.

View attachment 69260

The heatsink I used is like this one **broken link removed**

So last solution is to add the resistor. I guess?

Even though the heatsinks are cool, the case of the regulators may still be too hot.

No, I started them both cool. Both regulators are cool and both sinks are cool.

 

[dr pepper]

Could you use a bigger sink and run the 7805 direct from the dc rail?

Have you thought about using a lm2575-5 and a lm2575-12 switching regs, for the added complexity of a diode and an inductor you'll get hardly any heat and a whole load of extra efficiency.

 

[Mike odom]

for the added complexity of a diode and an inductor you'll get hardly any heat and a whole load of extra efficiency.

not to mention 4 very low esr capacitors (2 for each regulator). Panasonic FK series recommended.

 

[Mike odom]

Okay, so I have this circuit of microcontroller, transistors to control lamps, ... There are 20 transistors controlled by 15mA. So the consumption (without including the IC) is roughly 300mA just for the transistors. The loads' supply are connected directly to 24Vdc. The circuit is fully functional on the 1st day but screwed up on the 3rd day.

300 mA is a lot to pass through a micro... even though most allow 20mA on a port pin (40mA sink on the ATMEGA16), the overall per port and the overall per device is limited to somewhat less than that. The ATMEGA8 can sink/source 300mA as it's absolute max, while the PIC12F683 is around 95mA. Read your micro's electrical specifications.

 

[Mike odom]

Yes, I placed decoupling caps beside mcu, caps on regulators as seen on the datasheet. Sorry but I don't know what's the max power it can produce. I used PIC18F4620.

here's what I do... go to digikey.com, type in PIC18F4620, bring up the part, click on the datasheet to open it up, click on sec 26, electrical specifications, and READ... (if digikey.com doesn't have the part, then microchip.com should have).
Amazing, if you are sinking the transistor gate current, you are borderline (which I doubt, since the transistors are hooked up to a 24V source/load). Since you are probably sourcing, you exceed max current by 50mA. Make that exceeded by 100mA, as there is a note "max current by all IO ports = 200mA." And by max current, I mean you can't have so much as an additonal pullup connected other than your transistor outputs.

This is probably why you are getting random failures... once you exceed the max current, all bets are off. You've probably damaged the micro... read grey box at bottom of attachment.

There's no inductive spikes since I used 24v tiny light bulbs. There's no other load connected to it.

light bulbs, lamps are inductive loads. Big Capacitor and resistor in front of regulator would take care of any spikes, and see my D2 on the power supply I attached in the earlier post, that would take care of negative spikes on the ground also.

 

[Mike odom]

Have you thought about using a lm2575-5 and a lm2575-12 switching regs, for the added complexity of a diode and an inductor you'll get hardly any heat and a whole load of extra efficiency.

and I agree with the DR. : switchers are the way to go. My next revision on my board will be to replace the linear supply with a switcher. Although this would clear up your heatsink problem, it will do nothing to solve your current in/out of your micro.

I mean, 78xx regulators are nice, and simple, and CHEAP... but when the power becomes an issue, adding resistors/heatsinks, etc... then the cost of a switcher is more equal. Also, you probably only need the one (5V), unless you have some other 12V load you didn't mention. Input is rated at 40V, so you can go directly from 24V to 5V.

 

[lloydi12345]

Hi, thanks for the replies mike, yeah I already considered about using switching regulators. As a matter of fact, I already built the mc34063 circuit last time but it was used for another project. I am just utilizing whatever components are left here so I'm just left with 7812's and 7805's.

I think I have made bad calculations for the output current for microcontroller. I should have increased the resistance. I guess I played safe too much with opening the transistor. It was supposedly to be controlled with just 2mA my bad.

 

[shokjok]

Are the heatsinks not large enough to fit the circuit area? No matter how much thermal grease you use, the fins will get hot and the devices may go into thermal shutdown mode. Have you tried the 7805K and 7812K TO-3 versions? Both devices should not get as warm as the TO-220 devices, and you may not need a fan to circulate the warmer air. A taller heatsink should also work with the TO-220 devices.

 

[ronv]

I think the regulators are ok.. Hot but ok..
The datasheet for your heat sink says 17.9C per watt so the 7812 may be 98C or so if it's warm outside.
Maybe the micro.

 

[Mike odom]

They are not enclosed in a box and I used a lot of thermal compound on it. So what do you think is broken? Can 7812 still be fine?

Also, you DON'T want to use A LOT of thermal grease. You want to put the thinnest layer on it that you can. It is used to "fill in the cracks" to make sure you have 100% contact between device and heatsink. Too much and it actually becomes an insulator.

 

[atferrari]

Are you sure that they do not oscillate sometimes? Seen that before.