0

I am designing a DC-DC Buck converter using the ISL81806

I would like to use a sub-milliohm current sense resistor (0.2 milliohm) to set the current limit although it seems that the feedback voltage is a set value of 82 mV which would require a 4 milliohm current sense resistor to achieve my desired current limit (25 A).

Is there a safe way to adjust the op amp reading on the current sense resistor such as using a resistor divider network to gain the benefits of a sub-milliohm resistor while maintaining the same current limit?

Transistor
  • 168,990
  • 12
  • 186
  • 385
Sebastian Pelletier
  • 1
  • 50 mV full scale is pretty standard but 25A needs a non inductive 2 mohm 2W resistor or magnetic wire folded in half and twisted for same 50 mV drop for non inductive. Use AWG tables to estimate length. Wrap around short pipe for heat sink and epoxy in place tight. Or use a DMM and measure to get 2mV/A into an INA – Tony Stewart EE75 Dec 27 '21 at 22:55
  • 82 / 4 = 20.5 Amps. So you would need slightly less than 4 mOhm to achieve a cycle-by-cycle limit of 25 Amps. Note that this is not the current limit per-se, but the instantaneous switching current limit. The average current will never be able to reach the cycle-by-cycle current limit due to current ripple. If you need 25 Amps sustained for periods of many ms, then you will want the limit higher than 25 Amps (to allow for ripple current in normal operation). – user57037 Dec 27 '21 at 23:20
  • I read the datasheet a little bit more. It seems that if you use average current regulation instead of cycle-by-cycle current regulation you are free to use your smaller shunt. Is average current regulation not suitable for your application for some reason? – user57037 Dec 27 '21 at 23:34

3 Answers3

2

You would have to use a differential op-amp amplifier or current shunt amplifier with appropriate common-mode range, etc, and set the gain to 20 to amplify the current. Layout will be critical as the signal voltage will be small.

Something like this: INA185 might work depending on your specific output voltages.

John D
  • 22,677
  • 1
  • 39
  • 56
  • Theres actually a part we have the INA290Q that would fit this application perfectly. Is the implementation as simple as connecting the INA290Q output to the positive CS+ pin and the ground to the negative CS- Pin or are there other considerations that need to be taken into account? If there are major points of concern that I should look into or things to research please let me know. Really appreciate the help and any insight you have! – Sebastian Pelletier Dec 27 '21 at 22:19
  • You should be fine hooking up the INA290 as you suggest, but as I mentioned layout will be important as signal levels will be small. You could also investigate inductor DCR sensing as shown in the datasheet to eliminate the current sense resistors at the expense of accuracy and temperature stability of the current limit. – John D Dec 27 '21 at 22:41
  • I suppose there could be some chance that this would introduce stability issues. And there could be some voltage offset also. But I can't think of any other approach that might work other than choosing a different regulator. – user57037 Dec 27 '21 at 23:24
  • The other possible issue with INA290 is that the output is ground referenced whereas the datasheet shows a high-side shunt. I guess we do not know for sure if the internal differential amplifier (inside teh ISL chip) will work correctly when ground referenced. – user57037 Dec 27 '21 at 23:35
  • @mkeith Good points. The INA290 bandwidth looks high enough that it shouldn't introduce a LOT of phase shift in the current loop, but it could degrade the phase margin a bit. And there are no specs on the common-mode input range of the diff amp in the ISL part, so checking with the manufacturer might be in order. – John D Dec 28 '21 at 01:03
1

Fig. 2 in the data sheet shows 1 Ω resistors and 10 nF filter capacitors between the shunt R and the current sense input. You could put a 2 Ω between the 1 Ω's to attenuate by 2x.

This will have an insignificant effect on the 10 nF filtering.

jp314
  • 18,395
  • 17
  • 46
1

  • A 50 mV full scale is pretty standard for small currents but 25A needs a non inductive 2 milliohm 2W resistor

  • Locate the shunt into a low or high side INA, your choice.

    • The Pch or PNP INA's will use a ground shunt and the Nch or NPN inputs use a Vcc shunt.

I'm going to make some assumptions, but explain the basics for decent SNR limited by modulated AC grid noise on an SMPS switching frequency carrier with high Common Mode (CM) noise.

  • You can buy or make one with magnet wire or thin insulated wire folded in half and twisted for same 50 mV drop for non inductive results.

  • Use AWG tables to estimate length.

  • Wrap around short pipe for heat sink and epoxy in place tight.

  • Or use a DMM and measure to get 2mV/A into an INA with suitable gain.

  • if you can suppress the large CM noise, go for 0.2 milliohm with EMI reduction methods but noise from the cable wire mismatch limits the CMRR! not the INA.

  • You just can't match wires to 0.0001 % for 120 dB CMRR

    • so shielded twisted pairs with PE gnd on the shield at 1 end only helps a lot.

  • your goal could be 30 dB SNR at lowest current , which if that is 1 A then 30mA noise x 2 milliohm is 60 uV noise.

    • that's why 0.2 milliohm is hard to get 6 uV noise with PWM on inductive wires and or loads.
Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182