Question Preventing motor drivers from burning up from reverse curreng (regenerative breaking?)

On each PSU also installed 2x supressors to prevent overvoltage

backlash was the problam, also toothwheel gear will be very loud

also need add Shotky diode between PSU and battery

There are two possible problems that can cause shuddering.

1. Mechanical slip-stick effect: During slow movement force has to build up until the friction changes from static friction to dynamic friction. This causes the load to break loose and the force to collapse so the process starts over, again.
2. Loop gain mismatch due to backlash: The servo amplifier feedback loop has to be tuned to match the torque per current constant and the load inertia. But inside the backlash range the load isn't really coupled to the motor. The inertia is much lower (motor only) and therefor the servo loop is mis-tuned causing oscillations.

#1 should be no problem with spur gears. #2 can be overcome with dual feedback. Feedback to the velocity loop should come from the motor shaft directly (to dampen speed-oscillations) while position feedback should be attached to the output shaft to eliminate backlash error.

Hi just a question can problem 2 still happen when they stay under 1 G dofreality gives 0.5G max

It can happen when the force (not the velocity) changes direction, e.g. when the actuator has to pull instead of push. So if the actuator moves with +/-0.5G then the weight support of the gas springs should be less than half of the weight.

What kind of suppressors? I didn't notice them in the video, but it would be good to know what they use.

I'm going to use the KBPC5010 rectifier diodes to protect power supplies. Can you elaborate about the battery and the diodes?

Power supply 24V connect to both AC pin , + pin KBPC5010 connect to + motor driver, - pin KBPC5010 not used

What kind of fuses do you use there? I though about car fuses, but they are 12V and I don't think they will function well, the largest I have is 30A I think that means 15A @24V which barely covers a single motor... Traditional fuse placement is near the battery/PSU so if I have 2 PSUs I need to cover ~45A @24V on each fuse.

20A for one power supply to one motor

Take a look at the Keith project
https://www.xsimulator.net/community/threads/keiths-6dof-wheelchair-motors.17038/
in the picture I see only ordinary KBPC5010 diodes unless I am missing something due to too little knowledge of electronics.

That's right, and he was using the same motors I'm using.

45A fuses won't be of much use. The power supply can't deliver enough current for them to blow. Fuses only make sense it the supply or power grid can deliver much more current than your load should draw under normal conditions. For example a car battery can deliver 300 to 1000 amps. If a driver fails and burns there would be a serious fire risk. 300A blows the fuse very quickly so a fire is prevented.

A large fuse at a switch mode power supply doesn't make sense as the supply itself limits the current. So it will never blow if the fuse rating is above the supply current rating. And if it's lower it still takes quite some time. A 20A fuse can take 30A for a minute or so before triggering. So for a 13A motor a 15 to 20A fuse is ok. Even if the motor draws 25A for a second or two the fuse doesn't blow immediately.

The small black boxes between the power supply and the rectifier look like fuses, not overvoltage suppressors. Or have I missed something?

@Ronan Design I think it would be best to first carry out the test with the rectifiers and LED I suggested earlier. If we know that we actually need it we can discuss what action to take.

BTW, my friend decided to build a 6DOF rig with the same actuators as I'm about to build. But for money reasons he'll use smaller JMC motors instead of the AASD15. They run on 48V so we will be facing the same issue with regenerative braking, there. If necessary, I'll build a shunt regulator with external bleed resistor. The SMD resistors on the cytron shunt board are not designed for high power so you'd need 6 of them.

Power suply 24V>KBPC5010>Fuse 20A> motor controler

I don't think protection stuff tends to be as important in these setups because of how the motors and inertia of the system operate. Regenerative braking and "generator" style loading from motors requires them to actually be turning at a reasonable speed. I don't think motion simulators really do that.

Think about a vehicle that is rolling along at a fixed speed and then all drive power is removed. It will still coast fairly freely or slow down somewhat quickly depending on how regenerative braking is configured. The main thing is that the motor is still turning at a decent speed when power is removed. Now compare that to a motion simulator. We might accelerate a motor up to a set speed pretty quickly, but as soon as we get it where we want it, we stop it. It doesn't continue on with any real amount of rotational speed, so it doesn't have much time to even provide regenerative energy. The mechanical forces in the system slow things down much faster than any regen the motor is going to provide. Simulators typically use large gear ratios for torque multiplication, which acts as a natural brake.

I look at it this way: take a single motor in your system and command it to run at full rpm. Some amount of time later, command it to run at zero rpm. How long does it take that motor to "spin down" from full rpm to zero rpm without any electronics involved? You could also think of it as turning the system by hand. "Throw" the lever of a crank style system by hand and see how far it travels before it stops (assuming you can even turn the lever by hand to start with through the worm gear reduction). Wrap a cord around a ball screw on an actuator style setup and "pull start" it like a lawn mower. How long does it keep spinning if the motor isn't connected to anything? In both cases I'd guess the motor stops spinning almost the moment the mechanical driving force from your hand is removed. To me, that is what regenerative braking is trying to do. It's trying to slow down freewheeling systems, NOT direction reversals.

That isn't to say that direction reversals aren't a problem in their own right. It's well documented on this site the current spikes you can see when commanding a motor to switch directions with a load on it. I'm not enough of a circuit designer or electronics expert to understand how that might differ from regenerative braking from a circuit design and protection standpoint, but to me at least they feel like separate issues.

It's not the kinetic energy of the motor itself spinning. Direction reversal is also not a big problem (for backlash related problems yes, but for regenerative braking energy no). The rotational energy of the motor itself is only some Joules and can go to the capacitors of the drivers and the supply.

But the potential energy of the weight of the rig moving up and down is much higher. Lifting 150kg up by 0.25m takes 370J or 370W for one second. If you lower the rig you get that back and the energy has to go somewhere.

This is no problem with worm gears. They are self locking so their efficiency is below 50%. This means you have to spend at least 740W for one second to move it up. Half of that (370W) is lost as friction. When you move down you get back 370J = 370W for 1s but this time it's completely used up by friction. So you get 0W out.

However, if the efficiency of the motor and gear is above 50% then at least a fraction of the braking energy will go back to the supply and will blow up the weakest point of the electronics if the voltage rises too high. Again, this won't happen if you move very slowly. The is some idle power, i.e. the motors draw current even if they don't move. If the returned power is lower than the idle power then the voltage won't rise. But if you move fast you'll get into trouble at some point if you don't have any protection.

Okay, I'm pretty sure I follow what you're getting at, but I guess that all hinges on the gearing for the system. It still feels like the only time you'd have to worry about regenerative braking making it back into your system is if the system can freewheel to begin with. So if your system is self-supporting, it always takes positive power input to get it to move, regardless of direction. The second you pull power from the system it locks in place, so it doesn't need any braking (or at least pretty limited, it will find some way to decelerate, whether that be through friction, stress, etc.), but it would be for such a short duration that I don't think the motor could actually generate any useful energy.

Though I guess the original question here was more about how regenerative braking works with power supplies, so I suppose I'm getting off track. The better answer would probably be to turn the feature off if you can (I know Sabertooths have a specific setting when used with power supplies and JRKs can have motor braking enabled or disabled). If you can't turn it off then the debate comes up as to whether your system would actually generate any to begin with (where I jumped in).

You can't turn off regenerative braking without affecting the behaviour of the rig. If you turn it off then there will be no energy going back to the supply but this means the energy stays in the mechanics. The motor doesn't brake anymore but instead freewheels until it coasts to a stop (exceeding 90° for rotary actuators) or hitting the hard stop (for linear). But smooth braking is essential for the washout. If you disable it you get false cues.

So it all comes down to a simple test: Shut off power, move the rig manually to the neutral position, take a seat and see if it moves down. If yes then your gears aren't self locking so the efficiency is >50% and you'll get energy flowing back at least if you move fast.

If you carefully limit the velocity and add spike filters so you also don't exceed the limits during a crash you can probably avoid overvoltage. This is probably worth the effort if you build commercial rigs where high development cost can be balanced for saving a few $ for each unit. When building a single rig it's much easier IMHO to plan for the worst case. So you don't have to worry about exceeding the limits for a very short time during experiments. If you blow up your drivers it's going to cost more than you'd have to spend for the protexction.

@Ronan Design We'll wait for you until you did the rectifier & LED test. Then we'll know for sure if it's necessary at all.

BTW, I just saw that the Sabertooth drivers include a voltage clamp circuitry where you can connect an external energy dump resistor.

This would make extra protection unnecessary. But of course, the Sabertooth is way more expensive.

I'll post when I get to that stage. Looks like most if not all people with Yalu wheelchair motors use rectifier diode bridges to protect PSUs, they are super cheap and I already have them, so I guess I will use them anyway. I think there's no downside besides a minuscule and negligible efficiency loss. The test will show if I need to consider adding batteries to the mix to protect the motor drivers.

Does it matter if the fuse comes before or after the rectifier in this chain? The common wisdom is to put the fuse as close to a power source as possible, and I have a fuse box usable for that, but not for putting fuses after rectifiers. So I'd put it as 24V PSU > Fuse 20A > KBPC5010 > motor controller.
2 PSUs will go into the fuse box, and each will split into 3 circuits so that each circuit will have its own fuse, rectifier and motor driver.
Does it make sense or am I losing any protection like that?