Problem with a SEPIC Switching Regulator Circuit

Go To Last Post
6 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am having a new problem with a previously designed SEPIC switching regulator circuit that comletely perplexes me. I am using the Linear Technology SEPIC switching regulator with recommended circuit from Linear Technology to go from a 4-cell battery supply (4-6V) to a 7.5V output. The design was previously giving me a very nice result. I recently had new PCB boards made with very slight modifications to them... had a handful of boards assembled by a local company for me, and the output of the switching regulator now buzzes with a sawtooth wave of about a 1V amplitude centered around the desired 7.5V output. I've experimented with different output capacitors, which only changes the frequency of this sawtooth, but the amplitude remains at about 1V... I have also noticed that if I remove my 20MHz crystal from the board the buzzing goes away completely. I'm not sure how this could be related... but never had any of these kind of problems with my previous PCB's. I was also forced to change a few components as the ones I had previously ordered were no longer available, i.e. diode, capacitor, etc... I'm now left wondering if the problem is a result of the PCB change or the parts change? Or if I simply received a batch of defective PCB's? They work functionally... but could they just be bad quality somehow causing this problem? Does anyone know what would cause the SEPIC regulator circuit to buzz like this?

Thanks,
James

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Interference to the feedback input comes to my mind first. What regulator of LT do you use? Is there a voltage divider before the feedback input and where is it located? Close to the IC or the output cap? High ohmic or fairly low ohmic resistors? You don't want a few inch feedback trace that's high impedance.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I agree with jayjay. Unfortunately you can't decouple the feedback without ruining the transient response, so it depends on careful layout. Also the ground paths around the regulator are really, really important! It may be that removing the crystal doesn't affect the noise so much as it affects the current drawn by the load. Try putting various resistors across the output and see if it goes unstable at some threshold value of current.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In the feedback network try to put in parallel with the resistor that connects the FBpin with the output a capacitor,try some values around 100nF,maybe smaller or bigger than this value.
Also a wrong type of diode could cause problems.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks so much for all the replies... I will start trying all the ideas mentioned...

I am using the "LT1930ES5#TRMPBF" DC/DC steup-up switching regulator followed by a "LT1121IST-5#PBF-ND" 5V linear regulator.

My original diode on the board that worked was this one:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=MBRX02520TPMSCT-ND

The diode I switched to as this one went out of stock is this one:

http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=SD103AWSTPMSCT-ND

It has a higher voltage rating (40V vs 20V) & higher current rating (350mA vs 250mA) than the original, which I thought would not cause any problems.

The input capacitor was also changed from an out of stock Kemet to a Murata of the exact same type (1uF, ceramic, 16V, X7R).

I added one resistor to the board layout directly beneath the SEPIC circuit that taps into the 5V plane and routes off to a currently unused line. This was, in fact, the only change made to the board layout since the one I had working nicely.

I'll try all of the suggestions now that I have some time... Thanks very much for all the ideas...

Sincerely,
James

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Okay, I just experimented with placing a 0.33uF capacitor from between the two resistors in series to the output and the buzz was eliminated... perfect flat line trace at the desired voltage. So... this seems to fix the problem... but I am still left wondering why this only just now became necessary? Why did I never see this problem on my previous PCB's? I have no long traces in the switching circuit... everything is packed tightly around each other... the change from my previous PCB layout to this one was so slight... only added a resistor to an inner 5V plane directly beneath the switching circuit that was not even connected to anything currently, so not even completing a circuit... so I cannot imagine how this change created this problem. I am using a diode with a higher voltage & current rating than previously... I guess I can now redesign my boards to include this extra capacitor & re-order... I just wish I knew why this only just now became necessary. Also, could there be any way of solving this without re-ordering boards?

Thanks,
James