Any low cost methods to measure power line noise from a DC SMPS Ouput at Home? Scope? Specs?

Discussion in 'Test & Measurement Forum' started by Ajk Tek, Jul 1, 2018.

  1. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    Hello,

    I am trying to find a decent way to measure line noise from switched mode DC power supplies. For example from a laptop power supply or a USB Phone charger, without breaking the bank. I wanted to ask if you guys had any suggestions for a low cost setup as I was not sure if a Oscilloscope is the only way, and if it is I wanted to determine what specs I would need on the scope so I could find the lowest cost model that could do the job for me.

    Any suggestions and help appreciated.
    Thank you!
     
  2. danadak

    Distinguished Member

    Mar 10, 2018
    2,869
    617
    Basically a scope with infinite persistence, bandwidth 2 Mhz or more satisfactory.

    Infinite persistence allows you to capture multiple scans/sweeps and show pk-pk noise.

    https://sub.allaboutcircuits.com/wh...urements_misc_en_5215-0434-92_v0100_96dpi.pdf


    Preferably use a transformer to reduce voltage from full line, doorbell transformer
    should do the trick. Although its frequency response migfht be a factor in getting accurate
    capture of fast peak line noise.

    Or use scope differentially to do direct line measurements, with 10X probs that meet
    the peak line voltages.

    Some safety considerations -

    http://download.tek.com/document/51W_10640_1.pdf

    https://www.keysight.com/upload/cmc_upload/All/Probing-6-Hints-Scopes-Dec12-2007-webcast.pdf



    Regards, Dana.
     
    Last edited: Jul 1, 2018
  3. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    I’ve never tried to do this, but your problem is that you will be looking at mV – V signal levels superimposed on mains voltages (120V or 240V).

    The conventional method is to use a spectrum analyser that can eliminate the mains fundamental frequency.

    I would suggest a scope combined with a high q notch filter (set at 50 or 60Hz). The scope sensitivity could then be set possibly as low as a volt per division to observe the noise. But your notch filter would need to have a necessary bandwidth for the noise frequencies you want to measure.
     
  4. danadak

    Distinguished Member

    Mar 10, 2018
    2,869
    617
    As Hymie points out Spectrum analyzer very much appropriate for continuous time
    signals and dynamic range needed for analysis. However for transients the SA results,
    even using persistence, complex to interpret.

    DSO on infinite persistence will reveal virtually all peak information, transient capture, but
    not on individual signal sources sharing one common signal path.

    Regards, Dana.
     
  5. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    Thanks for both of your replies!

    I appreciate it, and since I was asking for a low cost solution I am comparing estimated prices

    Looking at estimated cost of both options thus far very roughly I am seeing:

    About $150 for a low end PC USB Oscilloscope that may work (Velleman PCSU200 - maybe?)

    Roughly $500+ for the Spectrum Analyser solution?

    Any other options out there?

    Thanks again!
     
  6. MrChips

    Moderator

    Oct 2, 2009
    18,446
    5,847
    You could use a high-pass filter, amplifier, full-wave precision rectifier, capacitor storage, and measure the DC output on a DMM.
     
  7. danadak

    Distinguished Member

    Mar 10, 2018
    2,869
    617
    USB Scope, todays model, should have FFT in it, but dynamic range
    somewhat limited. Even with averaging.

    Regards, Dana.
     
  8. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    The best option depends on what you want to get out of the measurements.

    To use an analogy with audible noise, imagine you could strip out the fundamental frequency (50 or 60Hz).

    The scope trace would display only the audible noise (that was superimposed on the mains). The amplitude of the trace would give a good idea of the amount of noise, but it would be difficult to determine the amplitude of the noise at each frequency (say from 20Hz to 20kHz). But some indication would be given by changing the scope time-base and observing the waveform at differing sweep rates.

    MrChips proposal has the advantage of being very cheap, but it would only display the maximum signal level of the superimposed audible noise, although this may be sufficient for your purposes.

    Rather than using a q notch filter as in my suggestion, you could use a high-pass filter (as in MrChips suggestion) since it is very unlikely that there will be electrical noise lower than the mains frequency.

    A spectrum analyser will allow measurements of the noise on the mains – being tuneable, such that you could measure the signal amplitude at each frequency.

    If I were you, I’d try to beg/borrow/steal a scope to see if that gave the results I wanted, before wasting my money.

    Bear in mind that a scope is a very versatile instrument; whereas once purchased, you may never use the spectrum analyser again.

    In the above – I used the analogy with audible noise in the range 20Hz to 20kHz, whereas the electrical noise you measure will likely be in the range 10kHz to 1MHz.
     
  9. danadak

    Distinguished Member

    Mar 10, 2018
    2,869
    617
    Not an expert here but I thought line had a lot of motor based transients noise on it,
    just the inertia / inductance translates into very low freq noise, slow moving transients ?

    Regards, Dana.
     
  10. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    No, the TS is looking at noise from switch mode power supplies.
     
  11. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    Great Information All - Thanks for the replies.

    Currently without buying anything I have access to two very limited oscilloscopes, if you can call them that, which I do not believe are up to the task specification wise. 1 of which is a low cost educational build a DSO scope kit that only has 200khz bandwidth and 10mv sensitivity along with a vintage (circa 1977) BK 1403a Recurrent Sweep Oscilloscope with 5mhz bandwidth and 10mv sensitivity. I did try connecting a SMPS directly to the BK and did not get good results, but I had thought it was a combination of it being a recurrent sweep and not sensitive enough. Is that the case?

    Assuming the poor excuse for scopes I have access too will never give me a fair result could this cheapo USB scope work?
    https://www.amazon.com/Hantek-HT6022BE20Mhz-Digital-Oscilloscope-Bandwidth/dp/B009H4AYII

    Combined with one of these filters:
    https://www.amazon.com/Douk-Audio-Frequency-Conditioning-Bandstop/dp/B07D59JPPQ
    https://www.amazon.com/PAC-BB-6PR-Pair-Blocker-Tweeters/dp/B00A2680T0

    Results wise I am looking to be able to compare the amount of noise generated by one SMPS to another. I do not need extreme precision, but I good enough quality data for comparison of the DC line noise between SMPS of different designs that would be enough to cause issues in sensitive devices such as static in radios or data errors in electronics using the SMPS for power.

    Hope the makes sense & Thanks again!
     
  12. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    Switch mode power supplies typically switch at 30-40kHz – with the majority of mains noise being this fundamental switching frequency or harmonics thereof.

    Therefore rather than taking the high q notch filter route, I’d take MrChip's option of a high-pass filter.
    A fairly simple high-pass filter (with a cut-off frequency of say around 1kHz) will attenuate low frequencies – eliminating the mains from the signal sufficiently.

    Then you can look at the filter network output with your scope.

    For such a quick look-see whether such a technique would work, any old scope should do – the performance of the high-pass filter might be more important.
     
    Last edited: Jul 3, 2018
  13. ebp

    Well-Known Member

    Feb 8, 2018
    2,332
    811
    I'm still not clear if you are talking about noise at the output of the switcher or noise it puts on the AC power line.

    If it is the latter, your objective pretty much requires a good spectrum analyzer, a well-filtered input power source and a line impedance stabilization network (LISN). You might be able to find a suitable spectrum analyzer on the used market for a few thousand dollars.

    If you expect to see output noise capable of causing RFI, a spectrum analyzer is still the best choice. You can get some information with an oscilloscope with adequate bandwidth, which means not less than the radio frequency of interest. With a low bandwidth scope you will see little but the residual ripple at the switching fundamental. That ripple can be detrimental in some sensitive analog circuitry, but it is the narrow spikes that require a scope with a bandwidth of several megahertz at a minimum that are both more likely to be a problem and easier to filter.

    A supply that has been certified to comply with EMI/RFI standards will have been tested for both conducted noise on the AC line and radiated noise. The facilities and equipment typically used will run to hundreds of thousands of dollars and could top a million. Do not trust any claim of certification from any ebay vendor from China. A great many products make fraudulent claims of certification. Amazon is also known to "fulfill" orders from third-party vendors or products lacking certifications.

    ===
    Many modern switchers run at 100 kHz or higher. 300 kHz is by no means out of the question. Some may run a low frequency when unloaded or lightly loaded to improve standby efficiency. Many of the monolitic switcher ICs (meaning control and power switch on the same die, e.g. devices from Power Integrations) modulate the switching frequency to spread the spectrum of the noise to make filtering easier.

    As I recall, there is at least one Power Integrations ap note showing how to scope noise at a switcher output.
     
  14. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    The Hantek USB scope might do want you want.
    The bandwidth of a scope is the frequency at which the displayed signal deflection is reduced to 50% (of what it should be) – a good rule of thumb is that a scope has a useable bandwidth of 1/10 the spec figure. Therefore the device is only good up to around 2MHz.

    The Douk notch filter probably won’t do want you want since it is for audio signals and is likely to have a high frequency cut-off at around 20kHz – whereas you are interested in frequencies up to 1MHz and maybe more.

    When mains connected equipment generates electrical noise, the noise is to some extent reduced by the impedance of the supply, typically the supply impedance will be less than an ohm.

    Therefore one way of artificially increasing any electrical noise from the equipment is to place an additional impedance (say 2 ohms) in each of the live & neutral supply lines.

    I would also recommend you use an isolation transformer for safety, and to enable measurement of the neutral line noise (with the chassis of the scope referenced to earth). You could even try measuring the electrical noise across the additional impedance placed in the live & neutral lines.

    Since you are only interested in a crude comparison of noise generated from different pieces of electrical equipment that use switch mode power supplies – your proposed technique might just work.
     
    Last edited: Jul 3, 2018
  15. ebeowulf17

    Distinguished Member

    Aug 12, 2014
    2,780
    556
    I've had to do a lot of similar testing off and on for the last year or so. DIY results will pale in comparison to what a proper test lab and a good spectrum analyser can do, but you can definitely get a good enough look to do rough this-is-better-than-that comparisons.

    We took machines in for two expensive lab sessions in the midst of some certification difficulties, but did improvised in house testing in between in order to narrow our options and make the best use of our time in the lab. The absolute values of our in house testing were not at all accurate, but the trends, patterns, and differences that we found in house all proved true in lab testing.

    Anyway, what l used was a Rigol DS1054Z in conjunction with a homemade high voltage high-pass filter. Crude, but reasonably effective:
    250917D8-07DC-4CA0-B69B-B24DA6F57898.png
    5B7F90E3-3AEC-4E65-98E8-34B8E66534ED.jpeg

    The red 16 gauge wire was connected to the hot leg of our mains power and the green/yellow wire to ground. It's important to note that there's no isolation so you must be VERY safety conscious when attempting this setup. The scope ground must always be connected to ground, not live, neutral, or any other voltage. There are lots of ways this can go wrong, so be very careful if trying to recreate anything like this.

    I'd run the scope in FFT mode to see spectral distribution and relative levels of noise. I'd test the mains with nothing running first to get a baseline for background noise on our mains, then turn on the power supply and see how much noise was added. To be honest, it only really worked for spotting really bad supplies and seeing the effect of EMI filters on those bad supplies. Good supplies all looked the same in our testing because their noise was below the noise floor of our already noisy mains. Later I tried adding an EMI filter between the mains and the test setup to reduce external noise pickup. Again, far from perfect, but better than nothing.

    Below is an example of data l collected with my improvised setup (on left) vs proper lab testing (on right.)
    A1D99B45-6F10-430C-965C-56E5AD0FC9C0.png
     
  16. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    I appologize if I was not clear, and as I truly do appreciate all of your suggestions I do not want my request to be unclear.

    So to clarify I am looking to measure the line noise on the DC Output of Switched mode Power supplies. For example a laptop power supplies 19v DC output line, or a 5v USB mobile phone charger.

    Here is a very very crude paint diagram that should be good for a laugh.
    SMPS Noise measurement.jpg
     
  17. ebp

    Well-Known Member

    Feb 8, 2018
    2,332
    811
    While trying to relate output noise to actual frequency spectrum and energy without a spectrum analyzer or scope that can do time to frequency domain conversion (FFT is the usual method) is a challenge, you can certainly look at amplitudes with a suitable oscilloscope.

    You will typically see some amount residual ripple at the switching frequency and also "spikes" where that ripple changes slope, corresponding to the actual switching events. There will be a spike at the valley when the switch turns on and one at the peak when the switch turns off. What looks like a spike might actually be ringing. You do need a scope with at least a few tens of megahertz of bandwidth to see these features.

    It can sometimes be extremely frustrating to try to distinguish what is actually a differential signal between the output terminals from something that is more or less a difference between the supply common and the scope's ground. It is not unusual to see noise spikes with the probe tip on the supply positive and the scope ground on the supply negative, and see almost the same amplitude if you move the probe tip to where the ground lead is connected. Typically you need to use a probe tip grounder to reduce the resonant effects from the combination of probe input capacitance and the normal ground lead's inductance. An active ("FET") probe can be useful because the extremely low tip capacitance makes the ground lead inductance much less or a problem, but they run to hundreds of dollars.

    Some supplies use "polishing" filters after the main filter capacitor on the switcher output. Such a filter typically consists of a low-value series inductor and ceramic capacitors, sometimes in parallel with an electrolytic cap. Ferrite beads are also used. The intent is to suppress the spikes that make it through the main filter which is less than ideal at very high frequency due to shunt capacitance in the main inductor and series inductance in the main capacitors. The filter may or may not make much difference in the residual ripple at the switching frequency.

    Switcher output noise can be a serious problem for analog circuitry and additional filtering can make a big difference, though the filter does slightly degrade regulation. A switcher would have to be exceedingly bad to cause problems with most digital circuitry.

    I would expect some of the cheap phone charges to be really bad in terms of ... well, everything! From the photos I've seen on the web, lots are unsafe, they are crudely regulated and show no evidence of even the most rudimentary of EMI/RFI countermeasures.
     
  18. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    Interesting and informative information, thank you.

    If I am understanding this correctly it seems as though you do not think I will be able to get any usable data on the noise output with any low cost setup?

    I can see that the cheapo USB scope I linked before does do FFT math, but I a active FET probe may not be cost effective for me.
     
  19. Hymie

    Active Member

    Mar 30, 2018
    652
    171
    To measure the ac noise on the dc output of the supply, all you need do is connect the dc output to a scope and make sure the scope probe input is switched/selected to AC coupled.
    The scope will then remove the dc level, showing only the ac content.

    You might find that any ac levels change, depending on the dc load on the PSU output.

    In my experience the output of SMPS are generally very stable, with minimal ac noise.

    Although you may have experienced interference on a radio in close proximity to/or powered from a SMPS, it is most likely that the interference is due to rf emissions from the PSU, or less likely conducted interference on the mains (as was originally thought you wanted to measure).

    The Hantek USB scope should be able to detect/measure any ac on the PSU output to at least 2MHz.
     
  20. Ajk Tek

    Thread Starter New Member

    Jul 9, 2017
    11
    0
    Thank you for your reply! I apologize for the follow up question, and I feel sorry to have to ask this noob question, but is "AC noise" the only type of noise that would be there to measure? I had thought, and very well may be wrong here, that their would be some AC Noise on the DC output transferred through the circuit from the AC Mains line, but I had also thought their would be a fair amount of noise generated from the high frequency switching of a SMPS. Am I incorrect there? Or is that also referred to as "AC Noise" due to the alternating nature of the noise?

    Thanks again!
     
Loading...