4
\$\begingroup\$

I'd like to watch the output from a power supply to evaluate noise and distortion. The output is 15 VDC.

Unfortunately, when I zoom in to 50 mV per division, the signal is off the scale. I'm using vertical offset control to provide valuable visibility of the DC bias and noise simultaneously but the offset is insufficient.

The only solution is AC coupling, but then I lose information on how the DC bias changes under load. Have I missed something? Are these just the limits of this model? I am using an R&S HMO1002 oscilloscope.

\$\endgroup\$
6
  • \$\begingroup\$ How much does the DC bias change? \$\endgroup\$
    – SteveSh
    Commented Nov 20, 2022 at 22:30
  • 2
    \$\begingroup\$ Depending on your accuracy requirements, you could cobble together an op-amp circuit that would add a negative 15 V offset to your power supply output for monitoring. \$\endgroup\$
    – SteveSh
    Commented Nov 20, 2022 at 22:32
  • \$\begingroup\$ This is very smart, thanks. \$\endgroup\$ Commented Nov 20, 2022 at 23:05
  • 2
    \$\begingroup\$ Can you use both channels of your scope? One AC and one DC coupled? \$\endgroup\$
    – PlasmaHH
    Commented Nov 21, 2022 at 7:13
  • 1
    \$\begingroup\$ I'm confused: you argue in your question that AC coupling won't work for you, yet you have accepted an answer which tells you to use AC coupling. \$\endgroup\$ Commented Nov 21, 2022 at 14:04

6 Answers 6

7
\$\begingroup\$

If you are interested in noise, use AC coupling to remove DC so you see high frequency signals well.

If you are interested how load current affects the output voltage, then a multimeter might be a better tool.

\$\endgroup\$
1
  • 2
    \$\begingroup\$ Usually people are intrested in transient load regulation, something a multimeter isn't too helpful with. \$\endgroup\$
    – PlasmaHH
    Commented Nov 21, 2022 at 7:15
10
\$\begingroup\$

A simple and low-cost approach is to connect a suitable number AA batteries in series with the power source you are measuring. In your 15 VDC case, 10x 1.5V AA batteries would offset the voltage close to 0.

As long as they are in constant temperature, alkaline batteries have stable voltage and low noise. The oscilloscope 1 megaohm input impedance means that it would take many days to drain the AA batteries enough to have an effect on the voltage. Measuring the offset value with multimeter once before measurements and once afterwards to confirm should be enough.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ when using it to subtract voltage this charges the AA batteries. Very slowly, but is it still safe? \$\endgroup\$ Commented Nov 21, 2022 at 14:10
  • \$\begingroup\$ @user253751 Whether it ends up charging or discharging depends on whether the voltage after subtraction is above or below 0 volts. Alkaline AA cells typically have about 20µA of self-discharge, and if there is 1 V going to 1Mohm oscilloscope input that would give just 1µA of potential charge or discharge current, so should be pretty safe even long term. \$\endgroup\$
    – jpa
    Commented Nov 21, 2022 at 14:16
5
\$\begingroup\$

You could offset the output voltage using another (unloaded) power supply. Watch the grounding.

Oscilloscopes are not ideal for this sort of thing, they have limited resolution compared to voltmeters and many data acquisition systems. Not sure what you mean by "distortion" wrt a power supply. "Regulation"?

\$\endgroup\$
2
  • \$\begingroup\$ Distortion, I mena ripple. \$\endgroup\$ Commented Nov 20, 2022 at 23:03
  • 3
    \$\begingroup\$ For ripple you know the frequencies that are involved (and they are relatively high compared to oscilloscope AC cutoff frequencies) so 'scope AC coupling is the way to go. If you need to look at more DC-like characteristics such as regulation on application or removal of a load you might want to go the offset route. Of course the offsetting supply (or voltage reference) has to be cleaner than the signals you are measuring because they'll add (give or take the sign). \$\endgroup\$ Commented Nov 20, 2022 at 23:50
5
\$\begingroup\$

I assume that the DC bias does not need to be measured with the 50mV/div resolution. Hook up two channels to the supply, AC couple one and set to 50mV/div, DC couple and set the other to where you can still observe the signal‘s DC value with the most resolution. If you need higher resolution for DC, use a good multimeter, as others suggested.

\$\endgroup\$
4
\$\begingroup\$

Oscilloscopes tend to have a "difference" mode, where the input of e.g. channel 2 is subtracted from the input of channel 1. You could supply a steady 15 V to channel 2 and measure with channel 1. You can leave coupling set to DC.

If you are using a good ol' analog oscilloscope, you might need to set it to "Add" mode and invert the second channel. Make sure to set the V/div equal for both channels.

\$\endgroup\$
2
  • \$\begingroup\$ the problem with this approach is that each channel is still being sampled at a full scale above 15V, so the resolution will be much lower than what you would see if you were measuring actual 50mV signals (or using AC mode) - for example, if the full scale was 20V and your oscilloscope was 8 bits, your resolution would be 78mV, so you would probably completely miss the signal. If the scope was 10 bits, the resolution would be 20mV, which is still a large fraction of your signal so you wouldnt see much. If you were in AC mode, you might get 100mV/(2^8)=0.4mV of resolution, which is much better. \$\endgroup\$
    – BeB00
    Commented Nov 22, 2022 at 3:03
  • \$\begingroup\$ I see. Being used to analog oscilloscopes, I was going at it with a more... analog mindset ;) The HMO1002's manual mentions an active differential probe. If OP has acces to that, this idea should work. \$\endgroup\$
    – Niko O
    Commented Nov 22, 2022 at 9:10
2
\$\begingroup\$

Unless you're specifically interested in no-load behavior, you can use the power supply you're evaluating to power a small 12V..14V LDO which you can then use as a reference. For a no-load behavior, a second power supply is a must.

Using AC coupling will work if you're interested in fast transients, but you'll miss slow variations in output.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.