1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Electronics Gurus - Constant Current Sink

Discussion in 'General Discussion' started by Frozennova, Nov 4, 2015.

  1. Frozennova

    Man of Honour

    Joined: Nov 13, 2009

    Posts: 11,303

    Location: Northampton

    Hopefully an OCUK electronics guru can help me on this one.

    I'm looking for a schematic/bill of parts required to build a constant current sink circuit measuring battery capacity

    So far I've found the LM317 and LM338 which can be use as a constant current sink however they will only allow me to sink upto 1.5A for the 317 and 5A for the 338.

    For a real world representation of capacity I need to be able to sink 15-20A.

    My first though was to use 8x LM338 in parallel with each sinking 2.5A with a 0.5ohm 5w resistor, however I can't find any information as to whether this actually work in practice or there is a better way
     
  2. Rroff

    Man of Honour

    Joined: Oct 13, 2006

    Posts: 67,951

    While still far short of what you want the LM317 is "guaranteed" for 1.5A rather than upto - with sufficient cooling (for more than 250mw you need a heatsink of some sort) it will handle loads in excess of 1.5A - IIRC typically 2.2A with sufficient cooling.

    For constant current sink in the 10s of amps you probably want to be looking at something completely different but its not really an area I'm familiar with. None the least you are gonna need some reasonable gauge wiring and connectors.

    As an aside there is an alternative to the LM317 - the LD1085 which can handle 3A and is also one of the best for noise filtering (though probably not relevant here).

    EDIT: I seem to remember reading that for powering 1 device parallel linear regulators isn't a good idea but can be done for multiple devices. For that high amperage you probably want some kind of switching regulator as 15-20 amp loads on a linear regulator could require dealing with considerable heat.
     
    Last edited: Nov 4, 2015
  3. Tealc

    Soldato

    Joined: Jul 13, 2009

    Posts: 7,016

    Location: Llanelli

    I think you should be looking at some sort of integrated buck/boost converter rather than trying to parallel up LM317.

    Not ever tried something like this so can't give a great deal of advice.
     
  4. Frozennova

    Man of Honour

    Joined: Nov 13, 2009

    Posts: 11,303

    Location: Northampton

    The only other thing I can find is using an N-Channel MOSFET and and OP-AMP but I can't seem to find any information on how the circuit works, and what each component is doing
     
  5. Rroff

    Man of Honour

    Joined: Oct 13, 2006

    Posts: 67,951

    MOSFETs are designed for sinking that kind of current but sadly my experience ends in the 100s of ma.

    I've a fair amount of experience with op amps so might be able to offer some insight on a specific circuit but far from an expert.

    From past threads there doesn't seem to be many people who frequent OcUK who have extensive electronics experience or atleast not many who will spend the time to share.
     
  6. Tefal

    Capo Crimine

    Joined: Jun 30, 2007

    Posts: 66,532

    Location: Wales

    can you not just buy a product designed for this?
     
  7. Agent_Smith357

    Wise Guy

    Joined: Apr 20, 2004

    Posts: 1,389

    Location: Somerset

    lol.
     
  8. Tefal

    Capo Crimine

    Joined: Jun 30, 2007

    Posts: 66,532

    Location: Wales

    had a google theyre about 3 grand lol

    maybe not :p
     
  9. GSWAudio

    Gangster

    Joined: Jul 1, 2012

    Posts: 319

    The key question is what's your battery voltage? If it's low (~12V) then sinking 20A will drop over 200W into your current source - this is not realistic and thus a switching solution should be found.

    On a side not this is the sort of circuit I think you're talking about:

    [​IMG]

    Vin sets a voltage that the op amp will try to maintain at Rsense and GND. So if the current through Rsense rises so will Vsense and thus the inverting input voltage will be higher than Vin which makes the output of the op amp fall thus turning off the transistor (FET or BJT) reducing current through Rsense lowering Vsense and the cycle goes on.
     
  10. Frozennova

    Man of Honour

    Joined: Nov 13, 2009

    Posts: 11,303

    Location: Northampton

    Initial voltage of 4.2v with the intention to stop discharging @ 3v
     
  11. GSWAudio

    Gangster

    Joined: Jul 1, 2012

    Posts: 319

    Then with a very capable heat sink and a few big FETs you could do it linearly.
     
  12. Frozennova

    Man of Honour

    Joined: Nov 13, 2009

    Posts: 11,303

    Location: Northampton

    I've just setup the circuit in that diagram you've posted in a Simulator and mostly understand how it works now.

    Am I correct in thinking Vin needs to be a lower voltage than the minimum load voltage, so I want to stop draining the cells at 3v use 2.5v for Vin?

    Rsense and Vin determine the load on the cell. So with the vin @ 2.5v and Rsense @ 0.25ohm I will see a 10A load.

    What affect does the value of R1 have on the circuit, from simulating it changing its value from 1ohm upto 1k ohm and removing it all together make no difference at all.

    Where and how would I add a FET or multple FETs to sink the current
     
  13. bitslice

    PermaBanned

    Joined: May 18, 2006

    Posts: 9,041

  14. Frozennova

    Man of Honour

    Joined: Nov 13, 2009

    Posts: 11,303

    Location: Northampton

    Thought I'd revisit this today using the Op-Amp + NPN Transistor circuit posted above.

    I've simulated it using 4 of those combinations in Parallel with 4x0.47ohm Rsense Resistors which would allow me a variable load upto my 20A target.

    First of all would this be a viable circuit to actual produce?

    Falstad Simulator

    Would a 741 Op-Amp and TIP141 NPN Transistor be suitable for this application. Each Transistor would be seeing 5A max with the TIP141 being rated for 10A, and finally I would swap the 4xparallel 0.47ohm resistors for 8xparrallel 1ohm resistors as these would rated and 10w and see a maximum of 6w load
     
  15. bitslice

    PermaBanned

    Joined: May 18, 2006

    Posts: 9,041

    As you are using multiple op-amps, use individual source resistors for each transistor and feed that back to each opamp.
    that way you'll be putting a precise current through each one rather than assume the base current is the same for each transistor as it heats up.

    I'm a bit rusty on this, I'd rather use mosfets for this application
     
    Last edited: Nov 15, 2015
  16. GSWAudio

    Gangster

    Joined: Jul 1, 2012

    Posts: 319

    You want to keep your Vin as small as possible. Remember Vin will appear over the sense resistor so the power dissipated will be Vin*Ib so with a Vin of 2.5V and Ib of 10A you'll see this resistor dissipate 25W - way too much. What dictates how low Vin can go is the input swing of the opamp. Op amps require the input voltages to be within a margin of the power supply rails. Some need a couple of volts and some 'rail to rail inputs' can be in the 10's of mV. If this was me I would create another rail and connect the V- to it, the rail would be around -2V to give input swing good margin. Then I would make Vin around 100mV.

    So to dissipate 40W isn't too difficult and you can do it with one amplifier. Add 100R resistors in series with each gate and 0.1R resistors in series with each source of 2 maybe 3 FETs. Then connect the resistors together as if they were the sources of the FETs.