[SOLVED] SMPS overload shudown ?

[rohitkhanna]

Puzzle: Why does SMPS trip ?

I have put together a project which involves driving upto 8 DC motors bidirectional simultaneously. Each is driven by a uC generated variable PWM through L298's. The direction change is (almost) instantaneous but happens at different unsync'd times for each motor... as per design. The DC motors are common car wiper-motors rated for 12V ops, and typically draw 0.6 - 0.8 amps each. Going upto ~ 1.6 - 1.8 amps on heavy load only.

This whole thing I am powering from a hacked, standard SMPS power supply - the type used for pC's. Its got a 12V current spec of ~12 amps.

The problem is that the SMPS trips if a connect more than 4 motors !! I am fairly sure it is an over-current trip, but can't figure out why.

Some thoughts that occured to me --
1) The PS -to- Controller DC power lines are a thickish wire of around 3-4 feet length. I estimate this would have an inductance of around 1-2uH. Could this be a point of concern ?
2) The power caps in the controller are not very large. Maybe if i increase these to several 1000's of uF they could handle any surge currents required when the motors change direction ? The surges may be adding up & causing a spike which makes the PS trip.
3) I have not put in any 'default' load on the PS. Some PS's like to have a minimum load on their output - the +5v one. Could this absence somehow be causing trouble ?
4) If all else fails, i could try putting in another SPMS in parallel. But that has it's own issues - especially if I haven't figured out what the real problem is upfront.

Any thoughts welcome !!

 

[chuckey]

Motors take a big current on starting (Vapplied/DC resistance), this is likely to be 5 - 10 X their running current. Very large capacitor would help to relieve the PSU of this current pulse. ( Or delay the current trip?)
Frank

 

[mtwieg]

If you are trying to reverse a motor's direction quickly, then doing so will require very high peak currents. Adding more capacitance to the DC bus may help, but you may have to slow down the motor acceleration to reduce the magnitude of the current peaks.

 

[rohitkhanna]

If you are trying to reverse a motor's direction quickly, then doing so will require very high peak currents. Adding more capacitance to the DC bus may help, but you may have to slow down the motor acceleration to reduce the magnitude of the current peaks.

Motors take a big current on starting (Vapplied/DC resistance), this is likely to be 5 - 10 X their running current. Very large capacitor would help to relieve the PSU of this current pulse. ( Or delay the current trip?)
Frank

Thanks guys ... that was my first thought too. Unfortunately the system was installed before the trouble was noticed, so it'll take a bit of doing to try the additional-caps solution. All testing done in the lab was with 2 motors + 6 resistor loads only !!

I'll post results of putting in bigger caps.
cheers!

 

[rohitkhanna]

So... i went to the project site armed with solutions based on all three possible trouble secnarios - including around 26milli F of caps, low-ohm high watts resistors, and a 2nd smps), determined to sort it out once for all.

And surprise surprise !! the 1st and simplest-to-implement option of having a "default load" (my possibility #3 in my post) turned out to be all that was required !! Didn't even need bigger caps (I already had 470u on each pair of motors)

In short - i added a 10ohm/5W resistor across the +5v output. So approx a 1/2 amp steady load. In my project, the +5v is normally not used, except for special situations.
And i found that the smps now happily supports all 8 motors with no trip whatsoever regardless on the start/stop situation.

This is what I had originally envisaged - since the smps was rated for 14A @12v - capable of handling upto 23 motors steady state, and handling spikes in current when the motors switched direction.

so thanks folks... & cheers!

 

[iop95]

Problem was caused by Power Good (PG) circuit inside PC power supply, which see an overvoltage error on 5V line. Without load on 5V line, when put high load on 12V line, controller try to correct output voltage and in this case 5V line overpass maxim allowed value (5.25V). Depending of controller used, may be easy or more complicate to disconect PG circuit. Or, use dummy load, as you discovered.

 

[rohitkhanna]

Problem was caused by Power Good (PG) circuit inside PC power supply, which see an overvoltage error on 5V line. Without load on 5V line, when put high load on 12V line, controller try to correct output voltage and in this case 5V line overpass maxim allowed value (5.25V). Depending of controller used, may be easy or more complicate to disconect PG circuit. Or, use dummy load, as you discovered.

Thats an interesting and very useful & almost plausible explanation.

However one question comes to mind - are you saying that the voltage regulation circuitry is a single common one across all 4 voltages (+3.3, +5. +12, -12) ?
Why should a higher load on 12v cause the +5 to be increased ? That doesn't sound like a great design.

And conversely, how would putting even a 'light' load on +5v solve this problem ? (The +5 is rated for around 35amps, and my load was 0.5amps.)

It would seem that having the +5v lying completely open-circuit is a source of trouble. But exactly what is not still clear.

 

[iop95]

In PC power supply controller used have a single voltage feedback input; 5V line is main output voltage regulated and this is used in feedback togheter with 12V line but with different proportion, because 5V line request best tolerance (+/-5%) and 12V is not so critical; 3.3V also request small tolerance and it's derived from 5V line.
When you put small load of 0.5A, was enough to drop voltage below 5.25V and avoid PG to shut-down controller. Using a single transformer and a single controller also, designers have to choose what is most important output to use as feedback; and because tolerance restriction, 5V line is that line. A small part of 12V line is used also. Design is ok, because in a PC all time there are load on 5V line and 3.3V (same 5V) and 12V load only is not possible or will lead to a behavior like you described.

 

[rohitkhanna]

... Using a single transformer and a single controller also, designers have to choose what is most important output to use as feedback; and because tolerance restriction, 5V line is that line. ....

now that makes sense.
Thank iop95 !