2
\$\begingroup\$

When setting up a test to evaluate a DC power supply, would using a power resistor suffice? I have a 5 V power supply hooked up directly to a 5Ω 20 W resistor to pull a constant current of 1 A. When probing the 5 V line for charger noise (high frequency oscillation), I see voltage spikes on my oscilloscope which don't match up to the max variance in the data sheets. The vendor uses a programmable DC Electronic Load with a 1 A load and doesn't see the same spikes as I do. Would this be because of load I am using?

\$\endgroup\$
1
  • \$\begingroup\$ On the contrary, if at all the dc load could cause Spikes. However you should make sure that you use the same bandwidth as they did (usually 20mhz) \$\endgroup\$ – PlasmaHH Jul 12 '16 at 18:35
4
\$\begingroup\$

enter image description hereThe spikes you are seeing should not be due to using a resistive load. Probing technique is critical in looking at output ripple in a power supply. Usually you want to get the scope probe right across the output terminals of the supply, or across the output caps. You can't use a long ground clip either. I usually wind a spring around the bare barrel of the probe (GND) and solder the spring to the negative output with as small a pigtail as possible. Then I do the same on the tip. Also, ripple is usually measured with the 20MHz filter on the scope active.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Just wondering: the 20 W resistor is likely to be wirewound and inductive as a result. Is there a chance of LC ringing between the L and the PSU caps? \$\endgroup\$ – Transistor Jul 12 '16 at 18:38
  • \$\begingroup\$ I haven't seen LC ringing with wirewound resistors on the output of a typical switcher, but the OP could compare the switching frequency of the supply with the noise he is seeing to see if it's synchronous. \$\endgroup\$ – John D Jul 12 '16 at 18:47
  • \$\begingroup\$ The DC power supply I'm testing is a USB charger so without cracking it open, I used a charging cable and spliced as close to the USB connector as possible to probe. With either the resistor or tablet attached, it didn't change anything but I will try to open the charger and probe like you mentioned above, right at the output terminal. I have the BW limiting feature active on my scope as well \$\endgroup\$ – EnG Jul 14 '16 at 14:29
2
\$\begingroup\$

To answer your first question: Yes, it is fine to use a resistor as the dummy load instead of an active. programmable load so long as the "resistor" is really just a resistor and not an inductor posing as a resistor. Check Digikey and Mouser and you will see that power resistors are available in both inductive and non-inductive types. You definitely want to use the non-inductive type for this type of testing, or else you will likely see the spikes which are the topic of your second question.

If you are unsure of your specific resistor's inductance, you can check it with an LCR meter. Or, you can make an acceptable non-inductive power resistor by paralleling several smaller carbon resistors such as 1/4 to 1 or 2 watt units.

Also, be careful how you wire the dummy load to the power supply terminals. Use twisted pair wires between the terminals and the load and keep them as short as practical. Follow the scope probe precautions described by John D. in his previous post. My experience is that you will never get a true scope reading with a single-ended scope input, you really need to use a differential probe with a differential scope input. But these are almost never available, even in professional electronic laboratories. Therefore, you must know how to interpret the single-ended readings you make, which takes a whole lot of experience.

Having said all of that, what is the actual load your power supply will be connected to in the finished product? It is likely not purely resistive. So once you evaluate your design using a purely resistive dummy load, you must make the same measurements using the actual load with the real-world interconnect wiring.

\$\endgroup\$
2
  • \$\begingroup\$ In practice I haven't had a big problem with output ripple being affected by the inductance in a wirewound load resistor. I suppose you could get unlucky and have the Lr and Cout resonant frequency wind up at the switching frequency. In our lab we have plenty of true differential probes, and using the technique shown in the photo I get pretty much identical results to the diff probes, which both match fairly closely to theoretical calculations. \$\endgroup\$ – John D Jul 12 '16 at 23:21
  • \$\begingroup\$ Checked on Digikey and the resistor I'm using is an inductive resistor (ohmite.com/cat/res_200.pdf) but connecting the load to a different 5V supply doesnt show any spikes. The actual load is a PCB with an OR'ing power controller which then powers/charges a tablet. \$\endgroup\$ – EnG Jul 14 '16 at 14:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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