Half bridge and hexfet motor drive circuitry

[Roger44]

Hello

The motor belt-drives a pair of go-kart rear wheels. In normal operation only the upper hexfet is opened to let current through to the motor (red line). When the motor is turning, the half bridge controler may in emergency mode be requested to close the upper hexfet and open the lower hexfet. In this case the EMF of the rotating motor generates a current (blue loop) which self-brakes the motor quite abruptly.

We are thinking of replacing this 800W motor by two 350W motors in parallel (so driven by the same ouput circuitry), each belt-driving its own rear wheel. Can you think of any possible problems? For example one wheel may drag in mud and its motor will draw more current, or after a bump a wheel may lose ground contact and its motor won't need to provide much current.

Thanhs for your help.

View attachment 61666

 

[ronv]

Seems like it should be ok.

 

[Sceadwian]

Seems solid to me too at first glance, though under high currents and EMF back feeding potentials; I'm not overly familiar with the effect but I know it exists, and using two motors in parallel might possibly cause oscillation effects between the two motors with back EMF which could cause spiked DV/DT conditions that would feed into the fet different than would be possible with a single motor at frequencies high enough to possibly exceed the DV/DT rating of the FET and cause a latchup condition. (A Mosfet can get stuck on because a parasitic BJT inherent to their construction can turn on preventing the gate from controlling the FET)

Probably just paranoia on my part, but be aware of the effect and it's possible repercussions as it could result in a run away motor condition, though your secondary fet allows you to control this as long as it doesn't latch up as well, if they both latch up you won't have to worry because both FET's will smoke/explode and the motor would shut off.

 

[Roger44]

Hello
Thanks for you help. I managed a rough computer simulation of what might happen for what its worth, using a very large condo in series with a coil and lots of diodes to simulate a motor. I did it for 800W motor, then two 400W motors just to see if there might be any risky things happening. Its only value is as a comparison.

The first image is of the first 30secs, wheels after bumps spinning fast in free air so producing back voltage instead of consuming current, then wheels in mud drawing high currents.

The 2nd image is when the upper hexfet is closed and the lower hexfet opened to short circuit the motor. The only unexpected thing was the back braking current from two 'dynamos' in parallel is much less. But it's quite logical if you think about it, two 4.5V batts in parallel drive less current than a 9V, in general.

View attachment 61789

View attachment 61788

 

[Sceadwian]

large condo

What do you mean a large condo? Could you post a schematic of your motor simulation?

 

[Roger44]

Here it is for what its worth. When you start, the condo charges up to a motor voltage fixed by the 20 diode string, this aritrary voltage corresponding to the current 'I would expect' taking into account the various low resistances 'I would expect'

(a diode string because my archaic simulator screwed up with a zener)

It's not very serious, if anybody knows what the real values are, I would appreciate it.

Roger

View attachment 61845

 

[Sceadwian]

An inductor can't be used as a motor model really, it simulates crude EMF but not the actual physical properties. I found a **broken link removed** for a motor circuit if I have time today I'll play around with it. If we take the curves you've come up with a grain of salt you can at least measure the theoretical DV/DT, find out what the rate of the rise/fall of the voltage is at it's steepest point, you'll have to zoom your simulation in to find out, can't work with the images posted.


If you're looking for a better simulator try LTSpice, it's free and quiet full featured, though perhaps not as user friendly as some people might like but it's extremely powerful if you can learn to use it.

By the way, after reading your last post I see you said condo again, I'm asuming you're refering to the capacitor? Never heard them called a condo before!

 

[Roger44]

Sceadwian
Thanks, I studied quite closely your first link. It suggests a a simple mathmatical model for a motor and then converts it into an electrical model of discrete componants so that it can be inserted into a (much larger) electronic circuit.

"However, because inertia and friction don't exist in SPICE, we call in an electrical equivalent - a series RL circuit ........"

But in fact, if I equate the hexfets to simple on/off switches, I can use the mathmatical model directly, in which case any old software which allows you to advance by infinitely small time increments should do the job. So why not Excel?

These are just thoughts to keep the subject alive, I've got a bit of a time problem for a few weeks which stops me digging deeper, but I can allow myself 5mins to post this.

 

[Sceadwian]

Depends on how far you want to get into it Roger, LTSpice supports arbitrary behavioral voltage and current sources which you can define using a very full set of mathematical functions, if you want to sit down and grind through the math / LTSpice syntax you could create as detailed a model as you want. The problem with Excel is you can't show interactions iwth that, you can only feed in fixed equations, in a real life motor the EMF created by the pure inductance, the mechanical interlinking through the EM field with the moving element and the applied field interact in extremely complex ways. So while your basic simulation will produce the basic EMF from inductance and a little bit of the power from the capacitance it won't model a real motor.

The worst case scenario is going to be during extreme braking situations where one wheel may be gaining/losing traction rapidly and out of phase with the other motor, if the phase shift/bouncing adds up in opposition you may break the DV/DT threshold of the FET and cause latchup. But that's a BIG maybe! I don't have a lot of practical on hand experience with this kind of stuff, worse case scenario if you try to simulate the physical model is you'll pine your life away trying to come up with better models where all you really need to do is build the circuit and test it in a practical situation, if you just build it and test it you'll know right away if it will work! I would however be quiet rough with the motors during testing to see what the real limitations are.

 

[Roger44]

Sceadwian
Can you have a look at your above link to the Spice modelisation of an electric motor, and more precisely the lines :

BACK EMF

"What is the back emf? It's the voltage generated across the motor's terminals as the windings move through the motor's magnetic field. The back emf actually opposes the drive voltage and is proportional to the motor's velocity

Vemf = Kemf ∙ ω"

Isn't there a mistake here? Should be "is proportional to the motor's velocity and the current flowing through it"

 

[Sceadwian]

No, there's no mistake, the back EMF from the motor winding traveling through a magnetic field has nothing to do directly with the current going through the motor, although it does interact with it, what you may be thinking of is the inductive EMF that's created from the motor winding inductance, this is separate from the EMF that the coils generate because they're traveling through a magnetic field.

Take a look at the first part of the spice model the electrical part, that's pretty basic the inductance of the coil is there, the motor's voltage source is there as well as the resistance, there should probably be a parasitic capacitance added but this is not going to seriously effect anything. There's a component at the end called VSENSE which is actually a voltage source but it's value is determined by the second mechanical components circuit which multiplies in the Vsense value and feedbacks a multiplier which is the torque constant produced by the motor in line with the inductance and a resistor which account for inertia and friction and speed on the mechanical simulation side.

I'd love to play around with this a bit as all you have to do is plug in the right numbers and you can actually simulate the effects of specific terrain types. If you develop it enough you could add in tire spring and traction constants for gravity and corning, there's really no limit to how far you can take this kind of simulation if you want to spend the time on it.

Spring however is rapidly approaching and aside from the idle time I spend here on the forums which is generally when I'm resting my brain actually working that kind of thing out has no appeal to me unless I have an immediate practical need for it, which as I'm busy as a bever my hobby skills grow very slowly.

 

[jackal_001]

u need extinguishing circuit ur sistem
parallel reverse diot conncect ur 33 ohm resistor ur diot must be work ultra fast and vth <=0.4V try this

 

[Roger44]

Thanks Sceadwin, your explanation "the back EMF from the motor winding traveling through a magnetic field has nothing to do directly with the current going through the motor" is enough to realise my error.

Although retired, I too have a time problem because of a quite tiresome project ongoing, clicking on this subject is an immense relaxation for me. There's no immediate utility, for me it's pleasurable to study something that makes the mind work a bit harder than the rest.

 

[Sceadwian]

Glad I could help, if you have any other questions ask and I'll attempt to help if I am able.
This is the short version of a private message I sent to you if you don't receive a notice.

 

[Roger44]

Jackal
If you expand the image of my very first message you can see a diode in parallel with the lower hexfet, and reversed to protect it from large negative swings. There is also a 1ohm resistance to limit the back emf current. I don't know if this resistance is to protect the hexfet or to protect the motor from burning itself out. I didn't design this circuit, it was done by pro's !

The enclosed image shows the negative swing at the mid-point of the two hexfets, with and without the diode, that's just standard, the reason I put this image is to see that this negative voltage swing you get with all inductances is momentarily stronger than the positive voltage "dynamo effect" from the rotating motor. I wasn't expecting that, but the designers were.

Please remember that my curves come from a crude capacitor+diode simulation of a motor.

View attachment 62297

 

[jackal_001]

View attachment 62338

 

[jackal_001]

View attachment 62339

 

[jackal_001]

View attachment 62340

 

[Roger44]

Hi
I managed to find a mathmarical solution to this problem, which is on the enclosed imaged. There's just a small problem concerning the second boundary conditon of a second order diffential equation, have a look, it's the part in red at the end.
View attachment 62699

 

[Sceadwian]

Wish I could chew on that, unfortunately the math is over my head. I kick myself almost every other day for not having developed my math skills properly when I was younger, and I have no idea how I'd go about it now!