constant current driver

The usual way to deal with this is a relatively fast (25-75kHz) PWM switching of 10-15V into your solenoid coil. The switching has to be fast compared to dI/dt=V/L of your coil's inductance, and above 20kHz to alleviate human suffering. It can be slow enough to make the job easy, below 75kHz. You can add a small inductor/capacitor/resistor filter to slow switching risetimes on the wire to the solenoid and in its coil, to reduce RFI.

--
 Thanks,
    - Win

There is a standard answer to your problem - you use relatively high frequency pulse width modulation to limit the average voltage across your solenoid when "on" to 6V.

The 2mH inductnce of your solenoid will averge out the voltage over periods of less than about 100usec, so break up your 25msec "on" periods with a 20kHz or faster clock - you can work out the duty cycle for yourself, or use a comparator with hysterisis to generate the desired mark-to-space ratio .

--
Bill Sloman, Nijmegen

Since you are operating the solenoid at 20Hz, it is not necessary to PWM at high frequency to avoid noise. The time constant of the coil is

2mH/2Ohm= 1ms which means the current will decrease by only 20% for a 200us off time. You have a 2:1 range on the applied voltage which includes rated voltage at the low end, and it is the current through the coil that requires limiting - not the voltage. Therefore all you need is a current sense resistor in series with the MOSFET to detect the 3A threshold, this is used to trigger a 200us monostable that grounds out the MOSFET gate drive- or IOW forms an analog AND with the external input used to turn the solenoid on. Control of the MOSFET is returned to the input signal after the timeout, current builds back up to 3A, cycle repeats etc..MOSFET dissipates negligible power because it is either full on or full off, the sense resistor should be no more than 10% Rcoil=0.2 ohm, if you intend this to work down to 6V, and solenoid requires a freewheeling diode rated for 3A continuous.

Good one Fred. I've been playing with a similar idea for dc-motor driving....... FET=ON until the required Ipk is reached, then a fixed OFF time, (pre)determined by the L/R and specified pk-pk current ripple. Iavg is known without having to explicitly measure it.

--
Tony Williams.

The L/R time constant of the relay coil isn't the only factor you have to keep in mind when calculating the "off" period.

Not only do you have the roughly 6V ohmic drop in the relay coil, but also the 0.6V to 1.0V drop in your catching diode.

dI/dt = V/L = 6.6/ 2E-3 = 3300 A/sec

so your current drops by almost 0.66A in 200usec.

You can use a monostable to set this period, but it is something of an overkill - a 0.6A of hysterisis at your comparator will have exactly the same effect.

And using a low switching frequency doesn't avoid noise - every time the O/P switches the drive MOSFET on, there will be a current spike in the lead to the relay and the power supply rail while the MOSFET discharges the parallel capacitance of the relay coil.

I've never measure the parallel capacitance of a relay coil, but I'd expect something of the order of 100pF. Discharge that through about

10V in 10nsec and you get a 100mA current spike, which might be embarrassing. A "non-wound" inductor is series with the relay coil would help to tame that - Farnell stock the muRata BLM41PG75SN1L (stock number 556-877) which seems to be good for 3A while offerig a resistive impedance of 75R at 100MHz, declining to 0R03 at DC. The high frequency resistance is lower than I'd really like, but 2A is a lot of current for a ferrite bead/chip.

That ferrite chip would only halve my - very hypothetical -100mA current spike, but it probably would make sure that it didn't ring, nor provoke the comparator into ringing, not that the LM393 that you'd probably use for the job is all that easy to provoke into ringing.

--
Bill Sloman, Nijmegen

Relay coils are easier than motors. For relay coils, the current decay time is set by the sum of the resistive voltage drop in the relay coil and the diode drop in the catching diode.

If you want to drive a DC motor in both directions, you catching diodes have to be returned to the positve and negative power rails, so that motor current is going to fall off a lot faster in the "off" intervals. If you spin your motor fast enough that back-emf is significant, the motor current can decay even faster. This first bit me in the bum back around 1978, and I've been known to forget about it for embarrassingly long periods (several hours) since then,

--
Bill Sloman, Nijmegen

Ehhh- the solenoid inductance changes substantially from out to in, so the 2mH and current discharge is a major unknown anyway.

The capacitance charge current will be negligible compared to the +2.4A some odd current step into the lead when the MOSFET turns on and the -3A current step when it turns off. This is continuous mode switching. If noise radiation is a problem , he can snug the driver electronics close to the solenoid and use a big bypass capacitor on its DC input.

You are right. I wonder what I was thinking about. And that current step could be quite nasty, if the catching diode that was feeding 2.4A into the positive supply rail functions as a snap-off diode, as many plain vanilla rectifier diodes do. The 3A step at turn off might be less nasty as the drain-gate "Miller" capacitance should slow it down.

--
Bill Sloman, Nijmegen

Don't clock the full H-bridge Bill. Set the DIR by turning ON one of the upper pair and clock it's diagonal mate.

Yes. I've experimented (on paper only) with adjusting the OFF time according to Vback-emf, but a far easier solution seems to be to improve the inherent low L/R of a motor with a modest amount of external inductance.

--
Tony Williams.

Low-EMI PWM'ing of a remote inductance is not an insoluble problem Fred. It has to be done on any aircraft, with the AC generator out on the engine, and the Control Unit (containing *all* electronics) being back in the fuselage... 10's of metres away.

--
Tony Williams.

Hi Bill, The past 2 days i was trying, what you told above. That is with the 200usec monostable mulitvibrator and an analog AND for the 20 Hz oscillator and the multivibrator's signal. What i observed is that the monostalbe multi vibrator starts operating correctly when the load current reaches the 3 Amps limit also it produces the 200 usec delay. After that it makes the MOSFET on, but immediately within very few microseconds the multivibrator gets triggerd so it makes the MOSFET off immediately without allowing the solenoid current to bulid up to 3 amps. This switching happens continuously for the on time period of the

20Hz Oscillator.The solenoid current continuously falls down exponentialy (that is it follows the L/R timings) with the continuos switching. when the current reaches the near zero level the MOSFET becomes on continuously, but this time it waits till the solenoid current reaches the threshold level of 3 amps. I also observed the voltage spikes across the currnet sensing resistor which is crossing the thresold level.I suspect that this voltage spike in the current sense resistor only triggers the multivibrator before the actual current of 3 amps, and inturns makes the MOSFET off quickly. As you said this may be due to the relay capacitrance.Can you guess some thing from here due to which this happens? Please help me to solve this problem.

Regards, M.Nagarajan.

View in a fixed-width font such as Courier.

. . . . . from MOSFET . . | . | . | . | 10 . +--/\\/\\---+--> to comparator . | | . | | . / | . Rsense === 0.1u . / | . \\ | . | | . +---------+ . | . --- . - . . . . .

Thank you very much Fred. Your suggestion helped me for solving those spike problems. Please tell me how did you calculate this filter resistance and the capacitance values. Can you explain me with formula? Now I am using 1N4007 diode( 1A, 1000V) as a flyback catching diode across the coil. But i am just wondering that how this

1A diode withstands this 3 A fly back current. Right now i am not having any problem with the diode. can i proceed with that itself or will that cause any problem in future? Also can you tell me how to calculate the rating of this diode based on the coil specification and other parameter? Still i am having one problem. That is the solenoid is not operating when the supply voltage is less than 7.4 Volts. The solenoid operating voltage is 6V minimum. Right now i am using IRF510 MOSFET ( 5.6A, 100V, 0.54 ohm). What i am suspecting is that the drop in the MOSFET may cause the solenoid not to operate for the voltages less than 7.4volts. Is it that ,i need still less on time resistance MOSFET or i can manage with this MOSFET itself by using proper gate drive?

Regards, M.Nagarajan.

Two considerations go into determining the RC. The first is that C must be large enough compared to the MOSFET Cgs to develop negligible voltage that results from an unlimited current drive of the gate circuit. Allowing for Cgs~1nF and Vgs=15, C=100nF puts this at 0.15V. R must be selected to attenuate the current into C for steep turn-on time of the

2.4A or possibly more when the MOSFET comes back on. And the overall RC time constant should be no more than 1/10th shortest ON/OFF time constant of operation over your V+ range of 6-15V. An R=10R with an assumed Rsense Now I am using 1N4007 diode( 1A, 1000V) as a flyback

The duty cycle, defined as ON time to sum of ON+OFF times will be approximately in the ratio of 6V/V+, so that the OFF ratio will be (V+-6V)/V+. The average diode current is then 3A*OFF ratio or

3A*(V+-6)/V+ . Worst case OFF ratio occurs at max V+=15V and is 9/15, making average diode current ~ 3*9/15=1.8A. Use a 3A diode such as 1N5400.

That MOSFET is not appropriate for this application because it has too much Rds,on. A better choice would be something like the Fairchild FQD30N06LTF shown here