Go Down

Topic: Switching ON/OFF High Current Load (Read 1 time) previous topic - next topic

Edgar1

Heh, heh,  Let's look into the maths of that.

So this solenoid draws 75 A at 12 V.  To "regulate" it to half power means that it has 6 V applied and 6 V across the controlling FETs.  OK, we now have the FETs dissipating 6 by 37.5 - I calculate that at two hundred and twenty-five watts!

That is basically going to require quite a few FETs and a maybe a water cooled heat-sinking system with a large fan-cooled radiator somewhere. :smiley-roll:  OK, they sell those for gaming machines, don't they?  Maybe not so bad.

With PWM as discussed so far we have three FETs and perhaps a six by eight inch conventional heatsink.

Ok, I get your idea.

If you don't mind maybe you have a short explanation how ESCs (Electronic Speed Controllers, used in RC hobby planes, boats, etc.) handle that issue without huge cooling elements.
Here is a sample of an 100A ESC for 25V with passive cooling element and total 99g.
https://hobbyking.com/en_us/hobbyking-100a-esc-4a-ubec.html
I know these ESCs do something different but they also regulate inductors 0 - 25V up to 100A - and 100A is not the maximum. They are available for more than 100A.


Paul__B

If you don't mind maybe you have a short explanation how ESCs (Electronic Speed Controllers, used in RC hobby planes, boats, etc.) handle that issue without huge cooling elements.
Yes, it's called "PWM" - Pulse Width Modulation.  That is what has been discussed previously in this thread.

Instead of switching partially on and the control device being a resistor, it spends a proportion of the time switched completely on so that while a substantial current is being passed, there is negligible voltage across the control device and so the power dissipation is minimal, and a proportion of the time switched fully off so that with no current passing, there is again, no power dissipated in the control device, thus overall it controls the average current proportionally without dissipating much power at all (if the device is of correct specification).  The successive on and off phases are repeated at such a rapid rate that a light for example, simply seems to be continuously lit to the human eye and thermal inertia averages out heating effects.  The proportion of "on" time - the duty cycle - determines the average power or intensity.

The effect is different for an inductor.  The short "on" times are insufficient to cause the current in the inductor to increase substantially nor are the "off" times sufficient to cause the current in the inductor to decrease substantially so that the current fluctuates about a value proportional to the mark-space ratio.  In effect, the inductor is a low-pass filter averaging the current.  Note however that for this to happen, there must be a "catch" diode across the inductor to continue to pass that decreasing current while the control device is off.

BTW, ESCs for brushless (3 wire) motors work by switching current to the windings to create a field rotating at the speed at which the motor (rotor) is desired to spin.   Slightly different process.

Paul__B

What type of MOSFETs would these be? Would they each need to withstand the full load of 75A? Searching around on amazon the highest-power MOSFETs I can find peak at 35A.
So you would require three of them in the above circuit.

The circuit still would include a high power diode in parallel with the inductor such as below, correct?

Sure would!  Lots of smoke otherwise.  Perhaps a somewhat more generous current rating but for a 10 second turn-around that should suffice and the recovery time of the diode is unimportant.


lastly where would the shunt resistor plug in to such a circuit if the diode is already in place?
No, the diode is the correct solution.



nuggetchris

Also, I forgot to mention that within the circuit there is a shunt used for a multimeter in order to measure power flow of the circuit. The shunt can be found here:

https://www.amazon.com/gp/product/B013PKYILS/ref=oh_aui_search_detailpage?ie=UTF8&psc=1

It is 100A 75mV.

So I am seeing a few options
A) Parallel Diode//IND with use of three expensive relays
       (Does this imply splitting source wire to three strands and wiring each strand to relay?)
B) Parallel Diode//IND with use of Tri-MOSFETs and an enormous cooling system
       (Same deal as above)
C) Parallel Diode//IND with use of ESC

Which is the best option, though?

Paul_KD7HB

Also, I forgot to mention that within the circuit there is a shunt used for a multimeter in order to measure power flow of the circuit. The shunt can be found here:

https://www.amazon.com/gp/product/B013PKYILS/ref=oh_aui_search_detailpage?ie=UTF8&psc=1

It is 100A 75mV.

So I am seeing a few options
A) Parallel Diode//IND with use of three expensive relays
       (Does this imply splitting source wire to three strands and wiring each strand to relay?)
B) Parallel Diode//IND with use of Tri-MOSFETs and an enormous cooling system
       (Same deal as above)
C) Parallel Diode//IND with use of ESC

Which is the best option, though?
Oh, come on! None of us are going to duplicate your system so we can determine which is the best option. You have been given options and it is up to you to determine for yourself the "best" option.

Paul

nuggetchris

#20
Jul 14, 2018, 11:05 pm Last Edit: Jul 14, 2018, 11:06 pm by nuggetchris
"None of us are going to duplicate your system so we can determine which is the best option"

I'm just asking what would logically make the most sense in practical design. You don't have to duplicate anything to answer that.

Three dedicated Relays, Three Mosfets, or Electronic Speed Controller makes the most sense?

Paul_KD7HB

"None of us are going to duplicate your system so we can determine which is the best option"

I'm just asking what would logically make the most sense in practical design. You don't have to duplicate anything to answer that.

Three dedicated Relays, Three Mosfets, or Electronic Speed Controller makes the most sense?
Still hard for anyone but you to decide and then test. You have NEVER defined the time it takes for your magnets to charge up to max magnetism, nor how long it takes for them to loose their magnetism, or even is they ever loose it all.

Paul

nuggetchris

Ok, apologize for the vagueness.

The core material will magnetize and demagnetize relatively quickly, say 0.5s-1s.

I decided the MOSFET plan was most suitable based on project requirements.
Looking on Mouser I am finding N-Channel FETs which are rated for 100A. Would this not be a simpler way to resolve the problem than wiring several lower-rated FETs in parallel or is there something I am missing

They also claim power dissipation of 100W or so, is this a manageable amount considering there is a case
fan in the device and average heat sinks on the circuitry or am I way underestimating it?

Lastly, shot in the dark here, but could you just regulate the current to say 1 amp or less at shutoff time with some sort of current regulator and open the circuit a matter of ms after to prevent large kickback compared to opening the circuit at full amperage? It would seem this would eliminate the need for most kickback protection and FETs in parallel.

I appreciate your patience - I am new to circuitry of this sort and learning quickly but it is a lot to tackle.

Paul__B

OK, let' have a go at gettign thsi sorted.  :smiley-lol:
The core material will magnetize and demagnetize relatively quickly, say 0.5s-1s.
I'm not sure what relevance the "core material" has - I would have thought it simply followed the magnetic field generated by the winding.

I decided the MOSFET plan was most suitable based on project requirements.
Looking on Mouser I am finding N-Channel FETs which are rated for 100A. Would this not be a simpler way to resolve the problem than wiring several lower-rated FETs in parallel or is there something I am missing
It may be; it may not be!  Many factors include cost, whether the higher current-rated FET actually has less than a third of the on resistance so that it will actually dissipate less power than the three in parallel, what voltage you actually need to put on the gate to achieve that very low on resistance, similarly whether its gate capacitance is less than three times that of the 35 A FETs so that you can switch it at least as quickly and how easy it is to mount on the heatsink (because mounting three devices rather than one in similar packages improves thermal transfer by a factor of three).

They also claim power dissipation of 100W or so, is this a manageable amount considering there is a case
fan in the device and average heat sinks on the circuitry or am I way underestimating it?
If you are only switching it fully on and off, and only doing this at a slow rate, and you choose devices and operate them at a low on resistance, then your dissipation should be substantially less than 10 Watts.

Lastly, shot in the dark here, but could you just regulate the current to say 1 amp or less at shut-off time with some sort of current regulator and open the circuit a matter of ms after to prevent large kickback compared to opening the circuit at full amperage? It would seem this would eliminate the need for most kickback protection and FETs in parallel.
Yes, it would prevent a large "kickback" when you opened the circuit at one Amp, but that would be because you would have experienced virtually all of the "kickback" when you dropped from 75 to 1 A.

Look, this "kickback" is a sort of mythology because people imagine all sorts of things.  Your inductor is drawing 75 Amps.  When you open the 12 V circuit abruptly, it will generate essentially without limit, whatever voltage is necessary to maintain that same current and that voltage will be in the opposite direction such that it is added to the supply voltage as seen by your FET.

Providing the diode across the coil allows the coil to maintain the current - the current being in the same direction - with only the voltage drop across the diode.  So your FET - which is no longer carrying the current -  sees the supply voltage plus the diode drop of a volt or so which is well within its ratings and the current drops rapidly which means the diode is carrying the current only briefly and barely has time to warm up - it does not require much heatsinking unless you are going to switch at a rapid rate.

There really is no major problem with this setup.

allanhurst

#24
Jul 15, 2018, 12:34 pm Last Edit: Jul 15, 2018, 12:39 pm by allanhurst
An old fashioned car starter solenoid ( a big relay) is designed to switch 100's of amps into an inductive load.

A big ( > 10A rated) shottky diode as shown earlier would reduce the voltage spike on the coil supply ( a mediumish MOSFET?) at switch off.

Allan.

MorganS

Quote
The core material will magnetize and demagnetize relatively quickly, say 0.5s-1s.
The Arduino can do 16 million things in that time and it is considered to be a slow computer.

What about some real numbers instead of hand-waving.
"The problem is in the code you didn't post."

CaioLimaViana


to avoid the return of current in the electromagnet, you could smoothly reduce the voltage in the same as aids to the use of diodes, for this could use PWM and gradually reduce the pulse width, but should not connect the pin directly to MOSFETS and electromagnets, you must connect a low pass filter between the MOSFETS and the electromagnet to obtain a smooth voltage reduction on the electromagnet. it would not be a good idea to connect the low pass between the Arduino and the MOSFETS as it would dissipate too much heat.

MorganS

Have you ever seen a low-pass filter that works with Amps?

It is the same problem, just shifted a little. Why all this effort to replace a single diode?
"The problem is in the code you didn't post."

Paul__B

to avoid the return of current in the electromagnet, you could smoothly reduce the voltage in the same as aids to the use of diodes, for this could use PWM and gradually reduce the pulse width, but should not connect the pin directly to MOSFETS and electromagnets, you must connect a low pass filter between the MOSFETS and the electromagnet to obtain a smooth voltage reduction on the electromagnet. it would not be a good idea to connect the low pass between the Arduino and the MOSFETS as it would dissipate too much heat.
You do realise, do you not, that to "use PWM and gradually reduce the pulse width", you would still require the same "catch" diode which would now need to be a fast recovery diode?  :smiley-roll:

As I pointed out in reply #23, people do have a surprising degree of difficulty comprehending what is actually a relatively simple situation, and go off looking for all sorts of oddball "solutions".   :smiley-lol:

nuggetchris

#29
Jul 16, 2018, 07:30 am Last Edit: Jul 16, 2018, 07:32 am by nuggetchris
Here is a little flowchart I made just now to further explain what I'd like to accomplish control wise:



Note 83A is cut down to about 75 considering other electronics in the casing + controller.
Taking into account the triple MOSFET method for each 'relay' I wrote it up as such:



A little bit of overlap but let me know if I have the connections right.
I know something is probably wrong/missing/overcomplicated as I am still new to this.

A little skeptical about the ability of arduino/controller to control that many FETs. Is there something I am missing between the controller and the FET banks?

Go Up