Need "noise free" boost regulator

[mikrogut]

Hi,

I need a 5V/500mA boost regulator that can give me 5V from a 3.7V Li-Ion battery. Currently I am using the L6920 together with a 18uH inductor. The problem with the current solution is noise and ripple. I can hear a high frequently squealing from the inductor when the regulator is loaded. In addition there is +/-60mV ripple on the 5V output. I want a small ripple and I don't want to hear the conductor squealing.

Any suggestions for a 5V/500mA boost regulator that I can use instead?

 

[Dick Cappels]

The acoustic squealing might be solved by using an inductor of a different design (when I had custom inductors that had this problem, it was solved by having them dipped in varnish). I assume you looked to see whether the stabilization loop is oscillating -if it is, that needs to be corrected first.

By "boost", I assume its some sort of flyback regulator as opposed to a forward converter that happens to output a higher voltage. Boost converters are generally fairly "gentle" and "non-violent" because they are current mode device.

Try using a soft recovery diode as the output rectifier.

Try putting a good high frequency ceramic capacitor across the output filter capacitor. Fry multiple capacitors in parallel on both the input and output to get both the ESR and ESL down.

Follow the rectifier and filter with an inductor followed by a capacitor to ground to remove the glitches and reduce ripple.

Be very careful with the routing of grounds. It is sometimes helpful to pass the power supply and return paths through a common mode inductor to keep signal current from straying.

If all else fails, add a linear regulator.

 

[ronsimpson]

Post a schematic and a picture of the inductor. What is your switching frequency?

 

[mikrogut]

I don't know the switching frequency because the spec doesn't say, but I think it is not very high since the requirement is to use a 10uH-20uH inductor.

 

[ronsimpson]

I can't tell what you inductor is........so.......Your inductor might be saturation.

The datasheet (L6920) shows, under heavy load the part does "sub harmonic oscillation) which could drip down into the audio range. Are you running this thing at the max current? At light loads it runs fast.

To reduce the output ripple add a LC filter. another inductor and capacitor to the output.

 

[mikrogut]

I use this shielded inductor:
**broken link removed**

I only draw about 200mA@5V from this regulator, so I don't think I am overloading the regulator.

I also have several capacitors on input and output; 47uF, 1uF, 100nF, but the ripple is still high.

 

[mikrogut]

How about using TPS61230 or TPS61027 from TI? They work at a higher frequency and require smaller inductors. That might help me removing the hearable squealing from my current solution.

I don't know how the 5V ripple would be, but it can't be much worse than now.

 

[ronsimpson]

I need a 5V/500mA boost regulator that can give me 5V from a 3.7V Li-Ion battery.

You are probably right at the max current.

At 3.6V in, the curve stops at 450mA. It takes about 0.7A at 3.6V to make 0.5A at 5V.
That's average current not peak. The peak is HIGHER.
It looks to me that the switch is above 1A.
The inductor is right at the edge of saturation.
I think you are asking a little too much out of this circuit.

Add a LC filter will help the output noise.

 

[mikrogut]

Ok, I see. Then I guess I should have an inductor with a higher current rating. Also adding a LC filter will give me a lower ripple.

But since I draw maximum 250mA@5V I thought the inductor I use would be good enough?

Thanks!

 

[ronsimpson]

Any suggestions for a 5V/500mA boost regulator that I can use instead?

But since I draw maximum 250mA@5V I thought the inductor I use would be good enough?

250mA should be fine.
The data sheet wants you to use a inductor of 1A or more and you did that.

If I could work on your supply for a couple of minutes I would know more.

Pictures below:
Fig 7. Normal operation for a PWM. The bottom trace, current ramps up when the switch is closed, and ramps down when current goes to the load. Notice the output ripple. While the inductor current is sloping down power is going to charge up the output capacitors. The majority of the time the capacitors are discharging.

Fig 8 & 9: Notice the "time" is different 5uS/10uS/20uS
This IC gets into a mode where it adds several cycles to make one. Fig 8 Here 4 or 3 cycles are together. In Fig 9 There are approximately 12 cycles added together. This produces audio noise.

Fig 10:
There is no description, but I believe they increased the inductance to get this to happen. The current trace is 200mA/div in Fig 10 and the slope of the current is much less indicating more inductance.

Things to try:
1)Try two of your inductors in series to double the inductance. This might push you into Fig 10. Hopefully getting you a higher frequency.
2)A saturating inductor often "sings". That is why I first jumped to that thought. If you get new inductors I would look at 1.5A parts just to be safe. The current limit for the IC is 1A +/- some amount. Note: the average current is much less than the peak current. In Fig7 the peak current is 350mA. The output current is (a ramp that starts at 350 and ramps down to 0), (The average of a 350 mA ramp is 175mA) and ( the output is only happening 1/3 the time so 1/3 of 175mA is 58mA) In this mode it takes 350mA peak to get 58mA of load current.

 

[mikrogut]

I see. I think I will try to experiment with inductors with higher current ratings and see what happens.
Also adding a LC filter on the output for ripple reduction is something I will try out.

Thanks for great input with figures and good explanations!

 

[ronsimpson]

mikrogut, I have some 22uH 2A inductors you could have. On the way home after work stop by my house. lol

 

[mikrogut]

Today I tried this conductor which I had lying around:
www.digikey.com/product-detail/en/ELL-6RH100M/PCD1345CT-ND/295721

The result was that the high frequency squealing disappeared. That was very good! The squealing was probably a result of a saturated inductor. The ripple is still high, but a bigger inductor and a LC filter will probably help.