Getting 5.35V from a LM2675-5

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
I am getting 5.35V (loaded) from a LM2675-5 from the PCB I just created. My Pic simply will not tolerate the over voltage.

I bread boarded the same circuit and I am getting 4.97V from the bread boarded circuit. I placed the chip in the PCB in the bread board and it gives me the 4.97V too.

Any ideas what could cause this over voltage?
 

Attachments

GopherT

Joined Nov 23, 2012
8,009
I am getting 5.35V (loaded) from a LM2675-5 from the PCB I just created. My Pic simply will not tolerate the over voltage.

I bread boarded the same circuit and I am getting 4.97V from the bread boarded circuit. I placed the chip in the PCB in the bread board and it gives me the 4.97V too.

Any ideas what could cause this over voltage?
This chip seems to need a 20 mA minimum load. Are you consuming enough? Try adding a 100 ohm or 220 ohm resistor to the load (in parallel to your existing load).
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
This chip seems to need a 20 mA minimum load. Are you consuming enough? Try adding a 100 ohm or 220 ohm resistor to the load (in parallel to your existing load).
Well as I mentioned work fine on the breadboard and that is no load (not the module but a separate complete circuit). When the new module is plugged in, it is powering a LCD with backlight and a pic so it must be consuming at least 20 ma.

I thought maybe the feedback was not connected but I just measured 6.3V there.
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
Nope. Stuck the chip into the breadboard and it works like a champ. Bad diode maybe? Or error in designing the board but I have been over it and I do not think so. If it was the coil I should think it would not work at all. There really is no much that can go wrong.

This is what it says about the diode. Not sure if that has anything to go with my issue.

upload_2015-12-30_0-56-0.png
 

GopherT

Joined Nov 23, 2012
8,009
Nope. Stuck the chip into the breadboard and it works like a champ. Bad diode maybe? Or error in designing the board but I have been over it and I do not think so. If it was the coil I should think it would not work at all. There really is no much that can go wrong.

This is what it says about the diode. Not sure if that has anything to go with my issue.

View attachment 97500
Which diode are you using? Is the diode fast enough for that regulator?
 

MrAl

Joined Jun 17, 2014
11,389
Hi,

What inductor are you using on each board? (Maybe a link)
If an inductor saturates for a little part of the cycle it can cause all kinds of problems.

Also, check the ground paths from input to chip and from chip to output.

Is there any way you can post a picture of the PC Board layout?
You do have to be a little careful about the placement of the feedback trace relative to the inductor mounting.

Since you have a test bed that is working, you can try swapping inductors and/or diodes. If swapping the inductors for example produces the opposite effect in each circuit then you know one of the inductors is not quite right. If there is no change for any swap then it must be the layout.
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
That must be it! I have the output going right under the inductor! What is the best way to fix this? (Other than move the trace from out under the inductor).

How would I accomplish this?

"ground plane construction or single point grounding."


I need to learn to read the whole datasheet. :oops:

upload_2015-12-30_6-49-56.png


They provide a sample below using surface mount. Just follow it as close as possible? I am not sure what I am looking at in the first image. Not really sure of the second. Which are the traces? Why are the caps so far from the chip (I realize it is 4x)?


upload_2015-12-30_7-2-7.png
 

Attachments

ISB123

Joined May 21, 2014
1,236
Shield it and connect the shield with GND.

FEEDBACK
This is the input to a two-stage high gain amplifier, which drives the PWM controller. It is necessary to connect
pin 6 to the actual output of the power supply to set the dc output voltage. For the fixed output devices (3.3V, 5V
and 12V outputs), a direct wire connection to the output is all that is required as internal gain setting resistors are
provided inside the LM2676. For the adjustable output version two external resistors are required to set the dc
output voltage. For stable operation of the power supply it is important to prevent coupling of any inductor flux to
the feedback input.
 

ISB123

Joined May 21, 2014
1,236
Those don't need to be connected to the ground since they are encased in ferrite. You could try making a shield from alu. foil and then just bolt it down to GND.
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
Those don't need to be connected to the ground since they are encased in ferrite. You could try making a shield from alu. foil and then just bolt it down to GND.
I think I might just bite the bullet and buy those. Foil would look a bit funky. Darn I was going to do that when I purchased the parts for this project but figured it was over kill. Thanks for the help.
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
I redesigned the PCB. I isolated the coil off to the bottom left, widened the traces and reduced the space taken by that diode. I added a pad to ground near the coil in case I want to add a makeshift shield. The connector is flipped from the last one because I had to put the pins on the trace side of the board.


Image is increased in size to 429%.

upload_2015-12-30_15-59-13.png
 

ISB123

Joined May 21, 2014
1,236
Looks fine although I would move GND trace completely on other side so its not even remotely close to the inductor.
 

Thread Starter

spinnaker

Joined Oct 29, 2009
7,830
I was thinking about that myself. I need to worry about board width. It would have been problematic to wrap it around the other side and I would need a jumper anyway.

How is this for a compromise? It let me move the inductor out just a little bit more.

upload_2015-12-30_17-17-10.png
 
Top