Go Down

Topic: LM2576-5.0 problem ... (Read 9095 times) previous topic - next topic

simkard

Hi all !

I'm here because i'm facing a problem where i'm running a board on which i have 2 big servos (17kg/cm).
Firstly, let me tell you that i'm a noob in electronics.
Let's say, i'm beginning in switching regulators.

Until now, everything was running fine but suddenly, it seems that my +5V is unstable.
My voltage regulation is based on a "LM2576T-5.0 P+".

Symptoms :
- When i plug the +12V on my circuit, it refuses to power on everything until i disconnect the 2 servos.
- If i plug one of the servos, then everything powers off (protection mode of the LM2576 i think)
- When the system is not powering on, if i mesure the normally +5V, it is +1,95V.
- The system won't handle load correctly (if it goes further than 1A, it resets like power is cut then back on then cut then back on etc ...)

I have 2 feelings :
- The inductor (100µH) seems to be a 1/2w (you know the axial ones ; green color ...) ; i think i made a big error when buying it because it is clearly not supporting 3A !
- Could it be the LM2576T-5.0 which was running without heatsink and i don't know if it has suffered from high temperature of today as my circuit is located outside in a closed box of approx. 30cm X 20cm X 10cm. (I did a calculation of the heat sink and it seems it requires, for 3A, a 20°C/W heatsink ; but as it is not running 3A frequently, i thought it was useless to add a heatsink for now ... finally, maybe not)


Hope to have good advices ! Thanks !

RuggedCircuits

It definitely sounds like your inductor is under-rated. I'm not sure what the "1/2W" rating is -- inductors are usually rated by current (among other things). If you're planning on delivering 3A then you need a power inductor that has a saturation current rating of at least that, preferably more. I doubt an axial "green inductor" is even close.

I think you want something more like this:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=513-1674-ND

--
The Gadget Shield: accelerometer, RGB LED, IR transmit/receive, speaker, microphone, light sensor, potentiometer, pushbuttons

Grumpy_Mike

Rugged beat me to it but in addition I would say that a switching regulator is a very tricky thing to build. You need a good PCB layout and even experienced engineers often take two or three attempts at getting it right.
How is yours physically constructed?

simkard

I will go and buy new parts :
- LM2576T-5.0 (a new one because i'm not actually sure if mine is healthy)
- a GOOD inductor (toroid type this time with at least more than 3A of DC current)
- a heat sink of max. 20 C/W (and some thermal gel to conduct heat)

Hope this time everything will run OK for more than 1 week.


I think my PCB is well done because it worked fine from start.
Even when i was on the breadboard, i never had problems with heavy loads like when my 2 big servos were moving at the same time.

What i didn't know was the fact that even switching regulators deliver less heat, they do require a heat sink. Moreover, they do require an inductor that is at least of the max DC current the regulator is able to deliver... Hahaha big mistake for me here ! Shame on me ...

simkard

Here is what looks my +5V regulation part of my circuit :


Thanks !

Graynomad

Those traces look a little light to me.

(caveat, I'm far from being an expert on PSU design)

______
Rob
Rob Gray aka the GRAYnomad www.robgray.com

simkard

These traces are 48 mil thick. I used the maximum available of Fritzing (which by the way is not the easier way of designing PCBs ...)

All was working great for more than 1 week ! but i think my main 2 problems were :
- lack of heat sink on the LM2576
- lack of the good inductor which clearly doesn't support the max current of the LM2576.

siraulu

I've used LM2576 twice and never had problems with it. Very easy to get it work.
At the first time I made some tests, and with 2A it does not heat much.

Have you tested with some other load than your servos?

Grumpy_Mike

Quote
I think my PCB is well done because it worked fine from start.

That does not follow, if it stopped working that is because it was not alright.
You need to look closely at the waveforms on an oscilloscope to detect instability under a variety of loads.

The PCB layout you have is very naive, it is a poor layout. Has the data sheet got any recommended layouts?
I would expect much more copper and more attention to ground return paths.
Quote
I used the maximum available of Fritzing

No excuse. Physics doesn't care, it needs to be right.

simkard

ok, so i think i found the answer to my question "why was it working until now ?!?!" ...

i took a closer look at the very poor inductance i put on my board (axial green 100uH) and saw that it has a crack on its side.
i think this is the point why it suddenly stoped working.
definitely, this little inductance was clearly a problem from start and i didn't saw it at first (remember, i'm an electroni-noob).

i managed to find the good inductance in an internet store so now i'm going to wait for it to come and wil report back if there is still any problem.

for testing, i'l try to use the actual LM2576T-5.0 and se if it's working normaly with this new inductance, which is the recomended one based on the datashet (PulseEng PE-92108K).
of course, i bought 2 others LM2576T-5.0 to be sure mine is not in quite-dead condition.

i have done some changes on my PCB because the good inductance is very large compared to the thin cracked one owho was actualy on my board.

by the way, could you please indicate me a god PCB layout for it to run smothly ? ( do i have to fill my PCB with copper for acting like thermal heat sink ? )


thanks for your help !

simkard

Ah yes, one last thing : Fritzing doesn't go further than 48 mils for the tracks thick.

If there is a way to bypass this restriction please tell me. Thanks !

Graynomad

I've never used Fritzing but maybe you can duplicate (or triplicate) tracks right next to each other and touching.

Also duplicate traces on the other side of the board.

______
Rob
Rob Gray aka the GRAYnomad www.robgray.com

simkard

Hahaha, i did not even thought of it but it will clearly work !

I will duplicate the tracks.
What is the recommended thick for GOOD TRACKS when speaking of switching regulation and inductances ?


Thanks !

simkard

Well well wellllll ... Fritzing doesn't support multiple tracks between the same components ... Pfffff ... I should have chosen another PCB Layout design software ... Now i'm on Fritzing, i'm on fritzing !

Ok, so as i can't duplicate tracks, for the moment, i made some changes, please tell me what do you think of this new PCB :


And the wiring recommendations of the LM2576T-5.0 (bold tracks are recommended to be shorter as possible) :



I don't know how to add thick to the actual tracks in Fritzing, i think i'll post something on their forums.


Thanks !

simkard

Managed to cheat with Fritzing ... hope that will not cause some bug.

Look :

Go Up