I am working on a motor project and I have very large inductors that will possibly need flyback diodes, but my inductors also change polarity. So a regular diode would just create a short in the other direction wouldn't it? So how do I deal with that? Do I have two flyback circuits(one going each direction) and put a Zener diode on each one with a breakdown voltage high enough that it won't conduct unless the backEMF voltage spikes higher enough relative to the standard input? So if I have a max voltage on my motor of 12V DC: then I would want zeners that conduct at 16V or higher? Would that work to handle back EMF in both directions?

 

I am working on a motor project and I have very large inductors that will possibly need flyback diodes, but my inductors also change polarity. So a regular diode would just create a short in the other direction wouldn't it? So how do I deal with that? Do I have two flyback circuits(one going each direction) and put a Zener diode on each one with a breakdown voltage high enough that it won't conduct unless the backEMF voltage spikes higher enough relative to the standard input? So if I have a max voltage on my motor of 12V DC: then I would want zeners that conduct at 16V or higher? Would that work to handle back EMF in both directions?

How are you driving them?
If you are using an H bridge, then you don‘t need any extra suppression.

 

How are you driving them?
If you are using an H bridge, then you don‘t need any extra suppression.

My motor is EXTREMELY unconventional, but at its heart its a BLDC motor. My control circuit will have two unconventional H bridges(one for each "phase").

Except when reversing(or when functioning as a servo), the H bridges basically don't get used.

Here's a drawing of it(I did not specify Vcc, but aiming for no more than 36V DC):

 

I’d recommend the conventional.
I think flyback voltage is the least of that circuit’s problems.

 

I’d recommend the conventional.
I think flyback voltage is the least of that circuit’s problems.

Some feedback on what's wrong with the circuit would be nice.

I want to use these two mosfets:

https://www.mouser.com/datasheet/2/1090/GP2T080A120H-2999273.pdf

https://www.mouser.com/datasheet/2/408/TW070J120B_datasheet_en_20200805-1894284.pdf

 

Firstly, they are both enhancement devices.
Secondly, because they are rated 1200V for use in a 36V circuit, the Rds(on) is poor.

How much current does your motor use?
Which device is which in your circuit?
Where is Vdd?
What is SSY1?
What does “switch or signal” connect to?

Using one enhancement and one depletion FET won’t make one turn on when the other turns off, if that is what you intended.

 

Firstly, they are both enhancement devices.
Secondly, because they are rated 1200V for use in a 36V circuit, the Rds(on) is poor.

How much current does your motor use?
Which device is which in your circuit?
Where is Vdd?
What is SSY1?
What does “switch or signal” connect to?

Using one enhancement and one depletion FET won’t make one turn on when the other turns off, if that is what you intended.

The voltage rating issue makes sense, but I am confused about referring to them as both being enhancement when SemiQ and Mouser both refer to the GP2 chip as depletion.

As for current the motor is variable speed so the current prototype has been tested from 6V up to 36V. Current draw at 24V on it is an initial~15amps without limiting(until the inductor is charged), but the next prototype will have more turns and far stronger magnets so that will be changing a bit.

If by Vdd you mean the switching signal then I haven't determined that yet(specs for both mosfets say -10, +25 though). Presumably the gate signal circuit would its own power source. The signal circuit would connect to ground of the device on one end via a pull up resistor and then by switch to the positive lead(which isn't pictured as connected anywhere but obviously it would have to be).

The SS41 is a Honeywell hall effect sensor.

 

The voltage rating issue makes sense, but I am confused about referring to them as both being enhancement when SemiQ and Mouser both refer to the GP2 chip as depletion.

I looked in the datasheet, and it says Vgs(th) =+2.8V That's enhancement.

As for current the motor is variable speed so the current prototype has been tested from 6V up to 36V. Current draw at 24V on it is an initial~15amps without limiting(until the inductor is charged), but the next prototype will have more turns and far stronger magnets so that will be changing a bit.

Assuming that you do find some depletion mode devices.. .
The Depletion device (bottom left) will be permanently on, because there is no negative voltage to turn it off.
The Enhancement device (bottom right) will switch on and off with the hall-effect device.
As for the devices in the H-bridge, whenever the enhancement devices are on, the depletion devices will also be on, so the how thing turns into a dead short and blows the fuse.

 

How rapidly is the current in those inductors going to be switched? Consider that there are other ways to reduce any inductive spike, and they work rather well. One very simple way will be to connect an incandescent light bulb across the inductor. Use a 120 volt bulb, 25 or 50 watts rated. And if it never blinks then you will know that it is not really needed. If the switching devices are rated 1200 volts, then probably no suppression is required.

And I will find it very interesting to see the whole circuit so that I can understand better what is actually being done. It does sound interesting.

 

I looked in the datasheet, and it says Vgs(th) =+2.8V That's enhancement.

Assuming that you do find some depletion mode devices.. .
The Depletion device (bottom left) will be permanently on, because there is no negative voltage to turn it off.
The Enhancement device (bottom right) will switch on and off with the hall-effect device.
As for the devices in the H-bridge, whenever the enhancement devices are on, the depletion devices will also be on, so the how thing turns into a dead short and blows the fuse.

Ah I see the confusion. Vgs is used differently on n-channel depletion mosfets. They function as both depletion and enhancement devices. That said, you raise a good point about them anyway. I assumed they just were an opposite version of P Channel enhancement/depletion(aka regular pmos), but that's not the case at all. I think what I want to do might actually be easier with all PMOS chips even on the low side. I need the A-B path to be used a lot more than the B-A path. Other than in servo or stepper mode it won't need to switch very much at the H bridge. I doubt it would ever need to switch more than 10KHz, and even that is unlikely. The motor will spend most of its time just eating a steady diet of ripple-free DC.


Here's a page that mentions how depletion n-channels contain both functions:
https://www.homemade-circuits.com/mosfet-enhancement-type/

 

How rapidly is the current in those inductors going to be switched? Consider that there are other ways to reduce any inductive spike, and they work rather well. One very simple way will be to connect an incandescent light bulb across the inductor. Use a 120 volt bulb, 25 or 50 watts rated. And if it never blinks then you will know that it is not really needed. If the switching devices are rated 1200 volts, then probably no suppression is required.

And I will find it very interesting to see the whole circuit so that I can understand better what is actually being done. It does sound interesting.

In normal operation the polarity will stay the same in the inductors, but the on/off switching will max out realistically at 12,500 RPM, or at 100KHz for the gates(the SS41 sensor's top speed). I can't imagine I will run my motor at 12,500 RPM, so most of the time its going to be a lot lower than that. Probably in the sub-2000 RPM range. So typically under 16KHz.

That actually is the whole circuit(the driver needs for this motor are stupidly simple), but its obviously missing things and needs reworked. I just assumed depletion n-channels were just opposite versions of their p-channel counterparts but they clearly are not. I'm going to draw up a new version of this here after I get done changing the brakes on my car.

 

For both depletion and enhancement, a numerically higher gate voltage turn is on, and a numerically lower voltage turns it off. The difference is only in the threshold voltage, in one case it is lower than zero and the other it is higher.
Another thing to note is how the threshold voltage drifts with temperature: in both cases it becomes lower.

 

OK so I have done a lot of studying and realize a ton of stuff I was doing wrong. I probably haven't sorted out all the issues, but here is my basic concept for an H-bridge on each of the outputs of my base circuit.

So the basic concept is that one side stays negative and the other positive until the mosfet(or some other transistor) in the bottom left is switched via the PWM(or even physical switch or relay).

I know my drawing isn't great, but does this make more sense? The top three mosfets are P channel enhancement type(the one on the signal path doesn't need to be as big as the power mosfets though), and the bottom three are N channel enhancement types(again, the signal FET doesn't need to be as big).

Technically my control circuit can work without this entirely on forward mode, but without this bridge I don't see a great way to reverse without reverse biasing my circuits in a harmful way. This at least puts all the control of the circuit in one +/- signal. I assume it has to have a return path somewhere I might be missing, but maybe not. I'd really love some constructive criticism!

 

With all of the transistors switching at the same time, it may be a problem with the initially conducting devices not switching off as rapidly as the other transistors switch on. That is always a concern when designing these stages.

 

Generally MOSFETs will only stand 20V between gate and source. Your circuit will work for power supply voltages less than 20V - but take note of @MisterBill2 's comment above.
To get you 36V, you will need a different solution - it is possible to get the circuit above working with zeners to limit the gate voltages, but you would be better off employing a gate driver such as MIC4606.
It would allow you to use all N-channel devices which are more efficient and cheaper than P-channel.
With a driver such as the MIC4606, you will probably end up with fewer component, a lower component cost and better reliability, as it will eliminate the possibility of "shoot-through" when both top and bottom MOSFETs are on simultaneously, which can easily happen as one switches on and the other switches off.

 

With all of the transistors switching at the same time, it may be a problem with the initially conducting devices not switching off as rapidly as the other transistors switch on. That is always a concern when designing these stages.

Would it relieve this problem if I picked n-channels that switch a lot faster than the p-channels wouldn't that theoretically fix the issue?

 

Generally MOSFETs will only stand 20V between gate and source. Your circuit will work for power supply voltages less than 20V - but take note of @MisterBill2 's comment above.
To get you 36V, you will need a different solution - it is possible to get the circuit above working with zeners to limit the gate voltages, but you would be better off employing a gate driver such as MIC4606.
It would allow you to use all N-channel devices which are more efficient and cheaper than P-channel.
With a driver such as the MIC4606, you will probably end up with fewer component, a lower component cost and better reliability, as it will eliminate the possibility of "shoot-through" when both top and bottom MOSFETs are on simultaneously, which can easily happen as one switches on and the other switches off.

That does look nice, but I am afraid my skills and equipment may not be up to implementing it. If I can find a through-hole part with similar features that'd be great. Meanwhile I mainly just need a sufficient circuit to run the motor, and outside of reversing I probably won't use it much personally. I'm sure people more skilled than me will eventually figure out better ways to drive it.

If I do use the above circuit idea, I did find some p-channel packages that have an rds of less than 10mOhm and I think I have an idea what pair I want to use for the main gates. I have TONS of zener diodes around from salvaging parts from power supplies, so it would not be hard to find some to use. I will start studying how to implement them.

For the n channel this looked good. It switches faster than the p channel parts, but otherwise has comparable features.
https://www.mouser.com/datasheet/2/196/Infineon_IPP014N06NF2S_DataSheet_v02_02_EN-3164900.pdf

Then this p channel unit also looks good and is OFF by default(its an enhancement pmos). rds is very low for a pmos, but definitely a bit higher than the n channel.
https://www.vishay.com/docs/73010/sup90p06-09l.pdf

If I am understanding correctly: even at ridiculous 36A drain current, the entire assembly would consume less than 25-30W. That's certainly within what a decent passive heatsink can handle. I would rarely be running anywhere near that much current though. At least on this current development prototype that's probably more than it could ever handle. I want some overhead though so I am driving them well below their current limits.

edit - I forgot to mention that I want to use a current-limited power supply for the motor itself and could get one that has an extra rail I can run the switching logic with. I want the switching circuit to have a steady power applied to it and the main circuit could sometimes have pulsed DC sent through it.

 

I'm confused... your circuit is trending towards a standard H-bridge as commonly used for steppers and DC motors. Can you explain about your 2-coil motor - how does it work?

 

I'm confused... your circuit is trending towards a standard H-bridge as commonly used for steppers and DC motors. Can you explain about your 2-coil motor - how does it work?

Without giving away too much. Its a permanent magnet brushless DC motor that can do limited stepper functions, servo functions, and general purpose uses. Its a true DC motor and requires an input very similar to a SPDT switch. In fact, that's how I know it works because I've run a prototype with manual switching. I'm currently creating a circuit for that output,, but it needs an H bridge on each outgoing line for reversing and programmed movement functions.

The circuit pictured here would be one of two that connect to the output end of the base circuit. I could just reverse the motor with a physical latching DPDT switch, but I want something that can react more quickly and reliably.

 

I have been studying this matter further and I'm trying to figure out if a shunt type clipper circuit is even necessary. Both of these mosfets have a maximum breakdown voltage of about +3/-3V respectively; so as long as I am using a regulated power supply like a computer supply's 5V rail, will I even need a shunt clipper? How likely is this control circuit to even come close to +/-20V? I would assume my pull up/pull down resistors would also need to be a bit smaller if I am using the 5V rail, but should be fine off the 12V rail yes? A decent PSU, even old ones, should have pretty stable voltage. For now it would work I think. I would of course want the control circuit to tap power off the main power to the motor, but given my limited experience using someone else's regulated power source might be good?

I drew in n and p channel mosfets on the control line, but could some cheaper transistor work instead since the control circuit will be conducting microamps at most? BJTs perhaps?

So another issue I am thinking this control circuit has is that current would flow when I really only need voltage(and a tiny current). So what about capacitive isolation of the high and low sides with a parallel set of opposing zener diodes that break down around 18V? The capacitor should only be transferring the EMF from high to low and vice versa, but in an overvoltage either high or low the zeners break down and conduct through a small ohm, but high wattage resistor to disperse voltage spikes as heat. I will draw a schematic of the entire circuit here and make sure its neater and more clear this time.