Hello, everybody,

I have a nice 9vx2 60VAC toroidal transformer for my device use in which there is a control board that has a mcu needing 5vdc. The following is a description of the power supply circuit arrangement.

It is in a seperated housing. Two 9v a.c secondary output lines go to two rectification parts. The first one is a full bridge diode for full bridge rectification, it goes to LM7812 and then to LM7805 to get +5Vdc for MCU use. All stages have 1000uF and 0.1 capacitors for filtering. The
second one consists of two diodes for full-wave rectification to get around 12vdc for device loading use.

The device is turned on and no loading is put on the 12vdc line. The mcu board operated. However, after running for a couples of minutes, it turns off automatically. I measure its voltage, the original +5v now goes to 3.6v. This hasn't happened before as I tested the board okay with dry
cell batteries. I add watchdog code in program, but problem still persists. I also use scope to see both +5v and +12v lines of the power supply, I didn't find any abnormaliies (no spikes or ripples) on them.

Without any components' changing, I replace the toroidal transformer with a smaller size 4.5Vx2 3VAC standard type transformer (EI type), the mcu board now works normally.

Does anyone has this experience before? Can anyone tell what 's wrong with it?

I realy don't want to add one more transformer as there is no spacing...

mcu is PIC18F452

Thanks in advance

 

What you describe is consistent with the LM78xx regulators overheating and shutting down to protect themselves. With the lower input voltage, they don't have to dissipate as much power and the +5V stays on.

Do you have heat sinks on the regulators? What package are they, TO-220, TO-92 etc? How hot are they running? Have you done the thermal math to ensure that the junction temperatures do not rise above their limits?

 

They are of TO-220 types, no heatsink added as no sensing of warm on them.

 

Hello,

Oh, I re-check again, the 7812 is hot, I missed in checking it last time. I am now trying to put a heatsink on it and thinking it maybe the cause. (why there is still 3.6V output on 7805 if 7812 is shut down due to hot?)

Many thanks JohnInTx

 

If they truly are not getting warm at all at any time, then I'm not sure what's happening. You might try a heat sink anyway. Other than that, its time to start poking around.

As for your original question, I don't know any reason that a toroid transformer would cause your problem.

EDIT: just saw your report. Good! Let us know what's happening with the sinks.

why there is still 3.6V output on 7805 if 7812 is shut down due to hot?

Because the 12V regulator will back off its output only to the point it has to to keep the temps in bounds. Its output to the 5V input is less than the 7V or so that the 7805 needs to regulate.

 

Yes, you're right, it is not the toroidal transformer, it is the cause of heatsink..., now, it works without failure. Thanks.

These regulators consumer some wattage of the power in the transformer, if i designed it marginally, it'll be no good. BTW, I remember that there are in market some sort of new regulators for directly replacement of them. They use switching technology and are made in same dimension as them. At this case, it may be also as a trial. Does anybody use them? And are they commonly used now?

 

There are some links to how to specify a heat sink in this post. You're right though. If you wind up dissipating lots of power on a heatsink you might want to reconsider your supply design or, as you are thinking, replace the linear regulators with an on-board switcher or wall wart. Even if you can keep the regulators happy, the heat has to go somewhere and the energy wasted paid for.

I've used lots of 'Simple Switchers' from TI/National Semi but I'm not familiar with any kind of drop-in switcher replacement for the LMxx. If you find one, please post a link.

Sounds like you're having fun.

 

These switchers are not without problems if you use them with a controller.

1. The A/D will have a lot of jitter. Solution: Use a seperate 78L05 for A/D reference.

2. I2C will be very disturbed. Solution: Use a small toroid as additional filter coil.

Besides that LM2576 is pretty easy to use.

Any modern electrolytic caps work + a small coil + a schottky barrier diode.

Remember even a switcher IC will require a cooler at some point.

 

These switchers are not without problems if you use them with a controller.

1. The A/D will have a lot of jitter. Solution: Use a seperate 78L05 for A/D reference.

2. I2C will be very disturbed. Solution: Use a small toroid as additional filter coil.

I am, as I type, running I2C diagnostics on a development board with 4x 18F4520 at 20mhz. Each PIC runs 4 channels of internal A/D switching between internal and external references as the channel requires. The data is rock solid as is the I2C (passing 23E6 messages since last night). Each PIC also drives a variety of peripherals including a multi channel 24-bit sigma-delta ADCs which are also fairly happy considering they are not in their enclosure. All are run by a big 24V switcher for the system (which drives other noisy things including an embedded 486) and a board level switcher for the local cards. It's not really a surprise that it works since its not the first time we've done it.

Of course, it takes careful consideration of power distribution, grounding, loop geometries and multilayer boards for the 24bit stuff. But I've also delivered many systems using 2 layer boards, simple switchers, several channels of millivolt sensing using PIC internal ADC and communications of various types with no trouble beyond using good design practices.

YMMV and for what its worth, I didn't do the hardcore, 24bit high precision part of this latest design. The guys that did do it are way good.