Using RDS for current sensing

Discussion in 'The Projects Forum' started by damianhealey, Dec 3, 2008.

  1. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    I need to minimize the heat produced in my current limited PWM, instead of using a current sense resistor which does get hot I thought about sensing the RDS of the hexfet.
    My current circuit without filter or regulator is shown in CCT1, the current limit is set by R36 and is sensed by R39.
    I had an attempt (CCT2) using a small signal FET form my spice library and have a couple of questions.
    1. Is this a reasonable approach?
    2. Any recommendations to improve the current circuit ?
     
  2. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,699
    907
    Check out the LM3489 from National. It monitors the voltage across RDSon as you suggest. There is a tidbit in the datasheet about adding another another resistor too.

    John
     
  3. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You might want to look at synchronous buck regulators.

    Microchip has the MCP14628, which is a synchronous buck MOSFET gate driver IC. That's the reason it's getting hard to find high-current Schottky diodes nowadays; power MOSFETS are being used instead as "ideal" diodes. Much more efficient.
     
  4. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    Thanks for the advice - the devices suggested will not handle the current, I require peak currents of about 30A.
    I am happy with my PWM - just want to try and remove the sense resistor and use RDS of the hexfet
     
  5. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    The MCP14628 can source/sink 2A to the gates of a pair of MOSFETs.

    Depending upon the MOSFETs you use, you could pass quite a bit more than 30A if you wanted to.

    By the way, since R40 in your 1st circuit and R9 in your 2nd circuit are both 0.7 Ohms, you'll never see more than about 17.1A flow through the MOSFET even if you wired the gate high. I=E/R = 12/0.7 = 17.143A (rounded up).

    Now if you want to see a constant current synchronous buck regulator I threw together for someone that wanted to power a hydrogen cell, that could be arranged by looking at the attached. I would use an MCP14628 instead of the LTC4447 shown, as the LTC4447 was designed for logic-level MOSFETs - but I didn't have a SPICE model for the MCP14628.

    In my simulation, D1 thru D4 plus R2 represent my concept of what a hydrogen cell might look like electrically. R2 is probably too large. As you can see, my "sense resistor" is simply two feet of AWG 10 stranded wire.

    If you are indeed using a hydrogen cell for a load, this synchronous buck current regulator will give you far better results than simple PWM, as with simple PWM your cell would receive too much current while the MOSFET was conducting, and none when it wasn't. The inductor L1 keeps the current through the cell relatively constant, instead of rapid on/off as with simple PWM.
     
    Last edited: Dec 4, 2008
  6. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    Certainly a buck converter is the way to go, I still have the same issue of the heat produced in the sense resistor, hence my wish for a circuit to use the RDS on resistance as the sense resistor.
    The circuit I supplied was not set up for 30A - I am glad you know your Ohms law :)
     
  7. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    I have not used a buck converter before and have some questions
    1.How do you determine the output voltage.
    2. I notice the TG drive is going up to about 17v (Gate)- if vcc is set to 12.7 (Drain) would this cause a problem?
    3. The TB signal is about 6V and the FET requires 10V for min RDS?
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I don't think you quite get it yet; my circuit is a synchronous buck converter. The current is read from the high side of a 2' length of AWG 10 stranded wire, which measures approximately 2mOhms.

    Obviously! ;)
    Me too ;)

    It will help you a great deal if you learn it too - separate the science from the BS.
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    My simulation determines the output current by measuring the drop across the 2mOhm resistor (a 2' piece of AWG 10 stranded wire). Voltage isn't the issue. If you want a voltage regulator, the circuit would need changes. After seeing your requirements, I made the assumption that you had a HHO (hydrogen/oxygen aka hydroxy) cell as a load. If your load is otherwise, you really need to specify it.

    You need to establish your reference properly. Vgs is what matters to MOSFETs as far as their conduction. Right-click on the source of M1, and select "Mark reference", then click on the gate. You'll see that the gate voltage is only around 5.4v relative to the source. This is why I said above that I would not recommend the LTC4447 in this configuration; it needs logic-level MOSFETS to operate properly.

    Explained already above, and in my original post with the schematic.
     
  10. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    It will help you a great deal if you learn it too - separate the science from the BS.

    I don’t understand what you mean – I have understood Ohms law for over 30 years - and I only believe what I can personally verify as I know there is a lot of BS in the HHO arena.

    With reference to the 2mohm resistor if I run at 30A (2 Cells) then 1.8W will not produce much heat I can live with that, it would still be nice not to have a formal sense resistor and use RDS and after taking a look at the spec for the MCP14628 it does what I want because at some later stage I will be going to more cells and as the power is a square law I will run into the heat problem.

    Good point about the Logic level mosfets – I cold use something like a IRL3803S
     
  11. nanovate

    Distinguished Member

    May 7, 2007
    665
    1
    The duty cycle determines the voltage as long as the current is continuous in the inductor. -- Vout = DutyCycle * Vin
     
  12. nanovate

    Distinguished Member

    May 7, 2007
    665
    1
    Vout = Dutycycle*Vin if inductor current is continuous.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    True. However, you still need to provide a ground SOMEwhere for the low side of the load, and there's going to be a voltage drop across the resistance of that wire. You could actually reduce it to a single 1 foot length if you wanted to (giving a value of 1mOhm), and adjust the value of R9 to compensate.

    In the version I whipped up, the inductor keeps the current relatively constant through the load. In order to measure that current, you'd either need to measure the voltage across some resistance, or perhaps use a linear Hall-effect sensor to detect the magnetic field surrounding a conductor in the path of the load. I chose a reduced-gauge wire for simplicity's sake. If you're running 30A @ 14v through a load, that's 420W, so roughly 0.43% of that energy is wasted in the AWG 10 sense resistor. I think that small of a percentage of wasted power is quite acceptable vs the added cost, complexity and reduced MTBF of including a Hall-effect current sensor in the circuit.

    Well, you could. However, the LTC4447 is in a tiny 3mm x 3mm DFN package, which won't be easy for a typical hobbyist to use. The MCP14628 is available in a DIP package.
     
  14. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    Thanks for all the advice.

    Although the MCP14628 will provide a solution I am still curious as to a circuit solution to sense the voltage across RDSon of the hexfet, my initial circuit did that but I had not tested all the corner cases and was concerned about it from a reliability point of view.
     
  15. jpanhalt

    AAC Fanatic!

    Jan 18, 2008
    5,699
    907
    That is what the LM3489 does, as pointed out in post#2. It was provided as an example of that approach, not as the chip you should use. So far as I known, the LM3489 is a reliable chip.

    It is not clear what your present question is related to that approach.

    John
     
  16. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    My apologies for not being clear - I know the chip does the RDS sensing - just curious as to know how it does it!
    If you have a look at my initial circuit I captured the RDSon voltage by using an FET switch.
     
  17. John Luciani

    Active Member

    Apr 3, 2007
    477
    0
    The RDS(on) changes quite a bit with temperature. How accurate do you want
    to sense current?

    There is a FET called a SenseFET that has a fourth terminal. With proper
    bias the fourth terminal outputs a current that is proportional to the drain current.

    (* jcl *)
     
  18. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    Thanks for the information, an accuracy of +/- 10% would be OK, this was purely an educational question for me as I have already been given an integrated solution in terms of the MCP14628.
    On a practical level I have a concern using such a low sense voltage and reference to compare against in the noisy automotive environment along with the effects of the PWM switching. Does anybody have any input they would like to share on this topic?
     
  19. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Practically speaking, I would want to get this working on a bench top and get the operating parameters carefully measured before even thinking of sticking it under the hood of a car. Going straight into the production model with no prototype evaluation tends to be quite expensive and time consuming.
     
  20. damianhealey

    Thread Starter Active Member

    Aug 27, 2008
    30
    0
    Absolutely the last thing I want is a problem when it is installed!
    I will be doing corner case simulations with a more realistic model for the supply and cell.
    After that I will run it on the bench at higher current levels than I require, with a noisy supply.
     
Loading...