3

I may have posed this question before, but it bares repeating. Today I put a scope on the USB output of one of those universal travel power converters. I was appalled. It was noisy as all get out. I put my horizontal cursors on to measure the envelope of the spike infested waveform, and measured something like 4.3 - 5.8 Volts.

What I'm thinking about doing is putting that through a boost converter, and then through a linear 5V regulator to try and clean it up. Any better ideas out there?

vicatcu
  • 22,499
  • 13
  • 79
  • 155
  • 2
    Surely you've considered passive filtering...? That's kind of the go-to for noise in most applications that I know of. Boost to LDO will be a bit wasteful of money, space, and time due to extra components and complexity. – Bort Sep 26 '18 at 21:39
  • @Bort feel free to propose an answer with a schematic... and don't call me Shirley – vicatcu Sep 26 '18 at 21:49
  • 1
    Why are you surprised with quality of charging adapters? They are designed cheap, just to charge batteries (which are essentially strong "integrators"), and not to power any sensitive analog equipment... – Ale..chenski Sep 26 '18 at 21:58
  • Are you probing the circuit right? Directly across the output cap, no long ground leads, etc. ? – Linkyyy Sep 26 '18 at 22:03
  • My measurement is Red ti Black wires on USB cable plugged into the wall adapter. I can post a scope trace tomorrow if it helps. – vicatcu Sep 26 '18 at 22:15
  • 2
    This is [typical load regulation for USB chargers](http://www.righto.com/2012/10/a-dozen-usb-chargers-in-lab-apple-is.html). Even the chargers from better brands don't have stellar regulation. – K H Sep 27 '18 at 00:50
  • I didn't want to suggest bypass capacitors to a user with 15k+. Of course, internet points are just points. Anyways, if you just want to jump right in and try something simple, begin by comparing a few different caps on the output. Maybe 0.1μF vs 1μF, or both. It's a shot in the dark though, hardly worthy of being an official answer. You could also look into using a pi filter. – Bort Sep 27 '18 at 01:21
  • @KH, Wow! This is a heck of a link to comparative analysis of AC-DC charging adapters! Thanks! You should make this as an answer... – Ale..chenski Sep 28 '18 at 01:41
  • 1
    Does it get better if you add some load? Unloaded switching power supplies often start oscillating heavily, as the frequency of the pulses generated to charge the output capacitor is low enough to pass the output LC filter. If you add a 500mA load, that should be a lot better. – Simon Richter Sep 30 '18 at 14:55

2 Answers2

4

It seems more to be a ripple voltage/current you are talking about.

Linear regulator are not good at filtering ripple and are not device made for that.

A good and cheap way to improve ripple is to use what is called a capacitance multiplier circuit.

cap multiplier

Those circuit are very good at reducing ripple without to have to add massive capacitor and have a cheap cost.

To increase effectiveness, you can add several transistor or use darlington transistor as the value of the capacitor will be (virtually) increased by the transistor gain.

A Nice youtube video explains it here.

EDIT

There is also off-the-shelve components that implements this principle, such as the AP1158.

Damien
  • 7,827
  • 1
  • 12
  • 29
  • 2
    Nice suggestion. For the record but, linear regulators are *amazing* at filtering ripple. See the answer here for an example: https://electronics.stackexchange.com/questions/18111/can-the-noise-of-any-switching-psu-be-attenuated-if-i-put-linear-regulator-befor – DSWG Sep 27 '18 at 00:19
  • You can watch the video I posted, it is explained why linear regulator are not good at it (at at 3:15) and even with demonstration. And also confirmed by my own experience. @DSWG – Damien Sep 27 '18 at 00:23
  • Interesting, I've never encountered this circuit, I'll look into it, thanks. – vicatcu Sep 27 '18 at 02:51
  • 1
    @vicatcu you can take a look at the AP1158 – Damien Sep 28 '18 at 10:55
  • Seems to me that you would have a voltage drop across the transistor. – JRE Sep 28 '18 at 11:26
3

On some applications I've had luck following the recommendations of figure 2.5 in FT 000292. That being said, on sensitive circuits with high bit ADCs, that doesn't seem enough to kick the noise from some especially noisy sources. (For example, I've noticed an iMac provided very dirty power compared to a Lenovo.) In that case a boost/linear may be the only real option.

enter image description here

geo_leeman
  • 85
  • 6