Pages: 1 [2]   Go Down
Author Topic: LM2576-5.0 problem ...  (Read 7641 times)
0 Members and 1 Guest are viewing this topic.
Manchester (England England)
Online Online
Brattain Member
*****
Karma: 602
Posts: 33358
Solder is electric glue
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

While more copper is good, the layout is still poor.
That schematic is telling you to implement a star ground with pin 3, the anode of D1 and the negative leads of Cin and Cout.
You have them in a chained arrangement. What you need to do is arrange these components to join together at only one point to prevent any inductance in the ground return path.
That is the first step towards getting it right.
Logged

NZ
Offline Offline
Sr. Member
****
Karma: 0
Posts: 390
Turtle in a hard shell
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

While not exactly the same chip, it basically is - I am using the LM2596

I have created a test board, which I ended up routing wrong, actually had a component in the wrong spot, however have reworked the circuit as a trial and got it working well, so have adjusted the design and will get the next revision printed.

Here is the Top and Bottom prints of my circuit. I have had very little input from other people, so any critique is welcome before I get it printed.
NOTE: The company I get my boards printed cannot print Plated Through Holes (ie Via's), so I have to manually do them with a resistor leg soldered each side. It is a downside, however at around $10 - $15 for a single ($5 - $10 US) 80mm x 80mm board, its perfect for prototyping.

Anyway, this may be of some use to you.
Red is top layer, Blue is bottom layer. Done in Altium Designer.







As mentioned, feel free to critique. I have never built a switcher before, other than the 1st attempt.
« Last Edit: May 03, 2011, 05:04:24 am by WanaGo » Logged

Offline Offline
Full Member
***
Karma: 0
Posts: 125
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Thanks WanaGo !

So i made some changes reflecting what you have told to me.
By the way, i wanted to thank you for your help because i couldn't determine that the ground should be done as a star.

I think you will be proud of me now  smiley-mr-green ... errr ... i hope !  smiley-eek-blue

Have a look :



EDIT (2 things) :
1st : Fritzing is alcoolic so do not bother why the holes are quite full of copper
2nd : i had to let space between the diode and the LM2576 because i have to put a heat sink so i evaluated the good space as is on this PCB.

ONE MORE EDIT :
I finally found how to increase track width (by asking on Fritzing forum ...) so now the tracks around the LM2576 and the +5V / +12V / GND of my board are 80 mils !!! yeehaaaa !
« Last Edit: May 03, 2011, 05:09:03 am by simkard » Logged

Manchester (England England)
Online Online
Brattain Member
*****
Karma: 602
Posts: 33358
Solder is electric glue
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
I think you will be proud of me now
Yes I am proud of your attitude to learning, keep it up.

Unfortunately I don't think you have got the idea about star grounding. The arrangement is that all the grounds go to the one point. So let's say the star point is the ground on D1. The LM2596 should go straight to D1 no where else. Likewise the track from C2 should go directly to D1, no where else. The track from C2 should go to D1, no where else. So if you follow the paths that the ground current can take it goes from each component to that point, no where else. Any current path to the star point never has the current from two or more components flowing through it.
The star point can be anywhere, not just D1, maybe best on the LM2596 or some other point not on a component, but the principle is the same, short tracks, wide tracks, star ground.

OK on the gap between the LM2596 and the other components being due to the heatsink. Can you not turn the heatsink over and reduce that distance? It really will make it more stable. Physics is a cruel master. smiley
Logged

Offline Offline
Full Member
***
Karma: 0
Posts: 125
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Me again !!!!

And now teacher ? am i in the right direction ? is it good this time.
I understood the STAR POINT way of thinking.
I think this time is the good one  smiley-mr-green

Please tell if i've done great job  smiley-twist :



EDIT : I modified the location of the LM2576 in order to place the heat sink in the right place. Now it will fit perfectly.
So if you confirm that this will work, YEEEHAA !! let's make the PCB now !

Thanks !
« Last Edit: May 03, 2011, 11:40:17 am by simkard » Logged

Manchester (England England)
Online Online
Brattain Member
*****
Karma: 602
Posts: 33358
Solder is electric glue
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
Please tell if i've done great job
Yes good job.  smiley

Quote
So if you confirm that this will work
Hey if I could confirm that I would have been able to command a lot more money than I did.
However, it is looking good now and worth a try to see if it works. It is defiantly going to be more stable than the first layout but as I said switch mode stuff is difficult and a bit of a black art sometimes. I have had experienced engineers that don't get it right first time working for me. But at this stage it looks to me good to go.
Logged

nr Bundaberg, Australia
Offline Offline
Tesla Member
***
Karma: 126
Posts: 8471
Scattered showers my arse -- Noah, 2348BC.
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

@wanago
With my limited PSU experience that looks OK to me, as far as I can trace the appropriate paths look short and of course there's plenty of copper for the various earths. I find that once the ground planes are laid it's very difficult to see what's going on though.

One thing I just noticed, the cathodes of the diodes don't go directly to the regulator but via the inductor, it's a pretty chunky trace but still not a star arrangement.

______
Rob
Logged

Rob Gray aka the GRAYnomad www.robgray.com

Offline Offline
Full Member
***
Karma: 0
Posts: 125
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
Please tell if i've done great job
Yes good job.  smiley

Quote
So if you confirm that this will work
Hey if I could confirm that I would have been able to command a lot more money than I did.
However, it is looking good now and worth a try to see if it works. It is defiantly going to be more stable than the first layout but as I said switch mode stuff is difficult and a bit of a black art sometimes. I have had experienced engineers that don't get it right first time working for me. But at this stage it looks to me good to go.


Thanks !

The PCB is in insulating state right now, it will be ready later today and i think i will have the time to solder my components on it.
I'll keep you all informed of the situation.


Another time, thanks to all of you, i learned some precious things with you guys smiley-wink
I'm now less n00b than before =]  yeah ! but still have lot of things to learn, particularly i have to use another PCB design software because Fritzing is a pain in the A** !!! haha
« Last Edit: May 04, 2011, 07:41:24 am by simkard » Logged

NZ
Offline Offline
Sr. Member
****
Karma: 0
Posts: 390
Turtle in a hard shell
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Cheers Rob

Yep good spotting, however from what I can tell it should be fine. Everything is very close and the traces, like you said, are large.

At least this time the diodes are on the right side of the inductor - major screw up in the last design.
Logged

Offline Offline
Full Member
***
Karma: 0
Posts: 125
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Finally !!!

After some hours of drilling/piercing the PCB and desoldering components and soldering them on the new PCB, IT WORKS ! =]

I hope that it will works longer than the first time.
The first time did worked for 1 week, but the little poor inductance which by the way is clearly burned on the inside did not supported the 3A max DC current !

So i replaced the LM2576T-5.0 and the inductance by a new LM and the GOOD inductance (PulseEng PE-92108) which is far bigger than the poor little green one.


Everything is running fine for now, my big servos are moving greatly and i hope they will do so for long !


Thanks all !
If my +5V become unstable another time, i'll be back here to post  smiley-mr-green, but honestly, it's not the point that i don't like you guys, but i don't hope to be back !  smiley-mr-green
If i would, it would means that there is a big problem  smiley-twist


See you all ! and thanks for having past time to teach me good things that i will not forgot !
Logged

Pages: 1 [2]   Go Up
Jump to: