Switching power supply

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

Hello freaks :)

Been away from here for a while but finally getting around to dusting off the old chips.

I designed my first switching supply, and in first attempt no fuses blown and no fires started, so thats a plus.

However I dont have a lot of insight in if the quality is ok, anyone able to shed some light on my scope picture included below ?

The probe is set to the wrong x, so the values are a factor of 10 too small, its a 5V supply (from a unregulated 12V).

The picture is taken with a unloaded circuit.

Any feedback is welcome.

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

It looks typical for an unloaded buck converter fed by high frequency PWM and controlled by a DAC with only 25 millivolt precision. If so you'll always have that 25 mV component unless you increase the feedback gain, but it should be possible to filter that out before the load.

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

I would add some load. Unloaded is unrealistic.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

I used a TPS54231D

http://docs-europe.electrocomponents.com/webdocs/0e7a/0900766b80e7a9de.pdf

Datasheet does mention some energy saving mode under light load, so maybe thats what is causing the noise.

I followed the switcherpro software to design it, however had to replace the power inductor, since I couldnt get the right one in the footprint I picked, it will be corrected on next board, so it ended up being about 7uh less than planned.

I made a 9V supply on the same board with the another of those chips, that makes 9.2V instead of 9, but its for relays and a lock so its not critical.

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

My advice to all beginning switching regulator designers: Test, Test, Test and then, Test Some More.

My experience with switching regulators is that if they are properly designed they will run steady-state perfectly and forever. However, when you impose various real-world start-up conditions and load-switching conditions, you will likely expose problems.

Before I elaborate on this, please tell us what your expected load current is and what your expected input voltage range is? Also, what is the load this regulator will be driving? (Motor, computer, audio, etc.)

A general test set-up would be to have a variable linear-type bench supply as the input supply voltage. Followed by a normally-closed relay which can interrupt the supply voltage to your regulator. The output of your regulator should then be connected thru another normally-closed relay to an adjustable and appropriate load (typically this would be a non-inductive power resistor decade box).

First round of tests: Run the regulator for 5 or 10 minutes at various input voltages and loads without activating either relay. Observe the output continuously with your scope looking for various voltage irregularities (drop-outs, high-freq bursts, etc). If you have an IR temperature probe, scan the components looking for hot spots.

Once you are convinced the circuit will operate properly under "all" combinations of input voltage and load levels. Proceed to "dynamic" testing by activating the relays mutually exclusively from a low frequency function generator. You want to use a square wave with a multi-second on-off duty cycle. E.g. 5 seconds on 5 seconds off.

First, switch the load relay as described while observing the output voltage on your scope. Also, adjust the load thru various expected values during the dead times. Your are looking for two major categories of problems here: A. The regulator self-destructs. B. The output voltage does something unacceptable (e.g. stabilizes at the wrong voltage, creates unacceptable overshoot on load connection i.e. when the relay first starts to allow conduction).

Likewise, use the front-end supply relay to energize and de-energize your circuit with various expected loads and allowable input voltages. Again, you are watching the output for anomalies and, indeed, provoking self-destruction.

Switching regulators are notoriously finicky about load switching and supply voltage interruption. If your application involves these types of operating conditions, you will not want to be the victim of self-deceptive steady-state load testing. Test, test, test, then test some more!

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

Thanks for the adwise, I will rig up some a test setup.

The board has 2 identical chips, one making 9V the other making 5V.

The 9V one will host 4 relays and a buzzer, so they are not so sensitive, expected load 0-1000mA

The 5V one hosts a beaglebone black and some additional logic/ic chips, expected load around 600-800mA and somewhat stable.
The beaglebone itself has another 2 regulators for 3.3 and 1.8V.

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

Quote: The 9V one will host 4 relays and a buzzer, so they are not so sensitive, expected load 0-1000mA.

May be worse than you anticipate because of the inductive nature of the relays. Is the buzzer also inductive or piezo?

A 1 amp inductive load is a bit troublesome, how big are these "relays"?

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

The buzzer is a piezo, 15mA.

The relays are prety standard, each drawing about 45-50mA.
So thats the 200mA accounted for, although generally only 1-2relays will be energized.

The door lock however is 200-300mA on average, in the future there may be 2 of these attached to one board, so thats 700mA total load on the 9V rail.

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

So, unlikley the door locks would be activated when the power supply is starting up. That's good.