0
\$\begingroup\$

I'm trying to make a couple of mains isolated, transformer based, bench power supplies. Considering that the current sense resistor will be in the PSU box and external loads will not have access to the other side of the sense resistor, does it matter whether current sensing is done high side or low side? Also, if low side, ground for all components (including op amps) in the PSU will be between the sense resistor and rectifier bridge.

Will it matter if I have a couple of these power supplies and connect the external grounds together? Will it matter if I put two of them in series (external ground of one connected to positive rail of other)? Ground is just a reference, right? So it will not lead to more noise or anything...?

I will have a short circuit protection based on output voltage.

Something like this for low side sensing (U1 will be a difference amplifier). What likely problems can it cause? enter image description here

[ edit: here's a proper schematic. I'm not showing the voltage sensing, but it will be there. The simulation works. However, the question remains. Any problems with low side current sensing in a benchtop PSU?

enter image description here ]

Also, I tried to simulate high side current sensing, and although the simulations work, I think it is violating input voltage range, so not sure I want to go there right now....

This simulates fine, current limit is set to 200mA, and 200mA flows into 100Ohm load (output voltage is at 20V). However, input at U1 is around 10.6V, higher than 5V supply. The simulation gives wrong results, as it should, if I remove the 1V offset. But I cannot understand how the offset helps overcome input voltage range limitation. enter image description here

\$\endgroup\$
1
\$\begingroup\$

I would say that it does not matter, but there are trade-offs for each type of current sensing. Assuming the voltage source is always DC:

  1. High-side current sensing often needs a precision voltage divider to get within the common mode range of the op-amp/comparator. An option in some cases is an op-amp designed for high-side current sensing, but it may need supply voltage greater than what your sensing the current on. The sense resistor need to be greater then 1 ohm most times, and as high as ten ohms, thus loosing some voltage across it.

  2. Low side current sensing is preferred in many cases where you already have an isolated source. The sense resistor can often be less than 1 ohm so voltage drop is tiny. The penalty is if a serious over-current or in-rush current exist. It can damage a low-ohm resistor, so it should be over-sized if possible.

  3. As far as random oscillation goes, you can put a 100nF capacitor across the 10 K resistor of U1. In addition you have no bypass capacitors on your power supply input and output. At least 470 uF from U3 collector to PSU ground and from U3 emitter to PSU ground. One of the main reasons for using capacitors is to stop spurious oscillations. A 'loose' rule is 2,000 uF per each amp of current consumed by the load.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the important things I had not considered. High side sensing requires larger resistor due to voltage divider, right? I'm trying to simulate the low side sensing and am getting wild oscillations. That was not happening with the high side sensing. Oscillations start as soon as I decrease the load resistor (actually, decrease the ratio of load to sense resistor). Is that to be expected? I'm guessing it has something to do with the gain of the pass transistor (TIP122), but am not entirely sure. \$\endgroup\$ – Indraneel Mar 25 '18 at 4:39
  • \$\begingroup\$ Low side sensing should not need gain, just a comparator. \$\endgroup\$ – Sparky256 Mar 25 '18 at 4:44
  • \$\begingroup\$ It's oscillating with no gain (only unity gain). In the low side schematic above, which works with 100 Ohm load. If I change the load to 10 Ohm, it starts oscillating. But it stops oscillating if I also change the sense resistor to 0.1 Ohm. \$\endgroup\$ – Indraneel Mar 25 '18 at 4:50
  • \$\begingroup\$ You either have a feedback loop or need a capacitor to stabilise it. I am on my cellphone. \$\endgroup\$ – Sparky256 Mar 25 '18 at 4:53
  • \$\begingroup\$ Try a 100nF cap across U1 10K resistor. \$\endgroup\$ – Sparky256 Mar 25 '18 at 4:56

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.