Author Topic: 20mV Enough for Power Supply Design?  (Read 1467 times)

0 Members and 1 Guest are viewing this topic.

Offline PixieDustTopic starter

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: au
20mV Enough for Power Supply Design?
« on: April 19, 2020, 03:06:08 pm »
I'm starting the design of a power supply and was wondering whether my oscilloscope's 20mV (20 milivolt) per division minimum ? Resolution? will be enough?

Also, most likely it will be a switching power supply. I don't have x1 probes, what bandwidth will I need to look for?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: 20mV Enough for Power Supply Design?
« Reply #1 on: April 19, 2020, 03:17:31 pm »
Working with only 20 mV/div and than a times 10 probe could be a little tricky. For low impedance signals, like the supply output one could get away with just a coax cable and a 50 Ohms termination, as a kind of poor man x1 probe.

Speed wise is would be nice to have some 10 MHz BW at least. Much slower can become tricky and may higher some details. It would be still better than no scope.
 

Offline PixieDustTopic starter

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: au
Re: 20mV Enough for Power Supply Design?
« Reply #2 on: April 20, 2020, 01:51:12 am »
Yep I found this article on the topic by Tektronix:

http://www.tek.com/dl/51W_27668_0_MR_Letter.pdf

great resource, covers a lot. Was mostly wondering whether my scope will be up to the task (given that modern scopes appear to be able to have 1mV per division) and also what x1 probes to get.

Thanks, will look for at least 10 MHz probes.
« Last Edit: April 30, 2020, 06:45:52 am by PixieDust »
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8526
Re: 20mV Enough for Power Supply Design?
« Reply #3 on: April 20, 2020, 02:41:53 am »
What voltages and precision?
 

Offline ejeffrey

  • Super Contributor
  • ***
  • Posts: 4034
  • Country: us
Re: 20mV Enough for Power Supply Design?
« Reply #4 on: April 20, 2020, 03:09:42 am »
20 mV / division is pretty poor.  2, 1, and even 0.5 mV / div are pretty common.  Of course it is better than no scope at all but it really limits your measurement flexibility especially when it comes to using high impedance probes. 

Whether it is suitable for a particular task is hard to say without knowing the details, but generically I wouldn't say such a low resolution scope is adequate for general use these days given all the much better options available so cheaply.
 

Offline PixieDustTopic starter

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: au
Re: 20mV Enough for Power Supply Design?
« Reply #5 on: April 20, 2020, 09:44:14 am »
Voltages are 28V. I'm guessing precision has to do with how much variation from an ideal the power can be for the given load. I'm not 100% sure at this stage if ever. This power supply will be driving incandescent light bulbs. I don't know whether the light bulbs I have have some particularly stringent requirement in terms of variation around the 28V range. I'm looking for their specification, but it's like finding hens teeth. Tried contacting the manufacturer, they were of no help surprisingly! :palm:

In terms of the 20mV per division, I realised I made a mistake. My scope can do 2mV-5V per division. I looked it up in the spec sheet. I turned it on to double check that I wasn't seeing things when I was playing around with the scope late at night. Went to 5V and it actually went to 50V per division but only down to 20mV. So as you might have guessed I have a x10 probe attached and it's a Tektronix scope.
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: 20mV Enough for Power Supply Design?
« Reply #6 on: April 20, 2020, 10:59:40 am »
I'm starting the design of a power supply and was wondering whether my oscilloscope's 20mV (20 milivolt) per division minimum ? Resolution? will be enough?

Also, most likely it will be a switching power supply. I don't have x1 probes, what bandwidth will I need to look for?

EDIT: didn't see your last reply when I was writing this, take this into account

It depends on the specs and what you want to measure, which better be defined before designing and building anything.

I'd say it's good-enough to check the output waveform if it's not a low-noise mission-critical design. It will let you check power-on glitches, transient response, stability, and noise within its measurement capabilities. I've built my first power supply with a 200kHz+ "oscilloscope", and it was incredibly useful to stabilize it, even at that limited bandwidth. I had no problem seeing noise of SMPS, even though I could see only so much of the noise spectrum. I couldn't see ringing, but was able to see switching ripple.

Things you won't be able to do: accurately measure noise. I'd be only worried about peak-to-peak measurements within first few MHz, there shouldn't be much noise above that, unless layout is messed up etc (which is, unfortunately, can be a case when designing your first power supply). So, following datasheet advice on layout is the key.

How much bandwidth is your scope? Normally output is measuremed for 0-20MHz bandwidth, but it's ok to measure the noise in a narrower band. Passive probes (those that I have) start attenuate at 3MHz+ (~6MHz -3db bandwidth), and high-freq noise is often not a concern. If your scope can do at least 2MHz, I'd say it's good enough.

It's great to see the noise up to, say, 500MHz to see parasitic ringing, etc, but this is rarely done in practice. It's also quite challenging and pricey.

For a linear power supply you don't need as much bandwidth as for smps (noise still can get injected somehow, like through mains or environment, but that's a different story). If you build an smps, consider buying a ready module, instead of messing up with mains voltage. It will also be cheaper.
« Last Edit: April 20, 2020, 11:04:18 am by exe »
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: 20mV Enough for Power Supply Design?
« Reply #7 on: April 20, 2020, 01:16:00 pm »
How much bandwidth is your scope? Normally output is measuremed for 0-20MHz bandwidth, but it's ok to measure the noise in a narrower band. Passive probes (those that I have) start attenuate at 3MHz+ (~6MHz -3db bandwidth), and high-freq noise is often not a concern. If your scope can do at least 2MHz, I'd say it's good enough.

For those curious on why scope probes have a low x1 bandwidth

 
The following users thanked this post: Vovk_Z

Offline PixieDustTopic starter

  • Frequent Contributor
  • **
  • Posts: 265
  • Country: au
Re: 20mV Enough for Power Supply Design?
« Reply #8 on: April 21, 2020, 02:43:25 am »
exe, yep you responded as I was writing up my response. Turns out my scope can do 2mV per division, I just had the x10 probe attached. So Sounds like I should be able to work on somewhat low-noises.

My scope is 100Mhz but can do 300Mhz equivalent time for repetitive signals, which power supplies are, so I guess 300Mhz isn't too bad.'

Apart from seeing the ringing on a scope, are there any other symptoms to look out for that might suggest there is ringing even if it's at a higher frequency to your scopes bandwidth? I'm thinking a spectrum analyser can uncover ringing?

Thanks Dave, I was watching your video a few days ago, lots of good info :clap:.
« Last Edit: April 21, 2020, 04:16:17 am by PixieDust »
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: 20mV Enough for Power Supply Design?
« Reply #9 on: April 21, 2020, 07:26:49 am »
Apart from seeing the ringing on a scope, are there any other symptoms to look out for that might suggest there is ringing even if it's at a higher frequency to your scopes bandwidth? I'm thinking a spectrum analyser can uncover ringing?

I think so, but it feels overkill, unless you are designing a power supply for sensitive rf equipment (nonetheless there is a lot of educational value in doing so). I'd like to hear other opinions as well. Probing will be tricky. At the very least it needs a 10x probe with short ground spring contact (like on the picture here https://electronics.stackexchange.com/questions/136123/how-do-you-attach-an-oscilloscope-ground-spring), a dedicated sma connector on the board is even better. I'm no expert, better ask somebody else :).

PS be careful not to burn input of SA, they may not like large DC bias.
 

Online EEVblog

  • Administrator
  • *****
  • Posts: 39026
  • Country: au
    • EEVblog
Re: 20mV Enough for Power Supply Design?
« Reply #10 on: April 21, 2020, 12:28:21 pm »
There is also this:

 
The following users thanked this post: PixieDust, Vovk_Z


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf