I've been working on a boxing robot for a number of years. It's two human sized robots that you control with a joystick or a Kinect camera (you do a left punch, it does a left punch). Hoping to sell these into arcades.

I've been struggling with the main power board. I'm using large wormgear motors, they use a lot of current at startup (first 1/10th sec) and then it settles down. I don't want to put much of an acceleration ramp in them because I want them to be very quick from a dead stop. But this places a huge load on the power supply (1200W PC power supply at the moment) so I've added in 5 x 2.7V 350F capacitors (70F @ 120V) which seems to be working.

Well yesterday I sent a motor controller up in smoke because one of the motors stalled and current/voltage shot through the roof and as I've now learned 60A slow auto fuses don't pop at 60A. After talking to the motor controller manufacturer (Basic Micro makes the Roboclaw, who is awesome btw, buy stuff from them) we came up with the idea of doing two things. First a voltage clamp on my power board separate from the one built into the motor controller. So if voltage goes above 13V (12V supply) it'll turn on a MOSFET and bleed voltage through a large 50W panel resistor. The second is a similar circuit that might turn on at a higher voltage (maybe 15V?) if the voltage clamp doesn't get things under control. It would turn on a MOSFET that would turn off a large relay, killing all power to the controllers.

I'm now trying to figure out a good design for monitoring the voltage. I don't have experience with comparators but it seems like I would need two of them, both comparing supply voltage against motor controller voltage (voltage after the diode protecting the power supply). Any suggestions? I've attached the current schematic. On the right side of the schematic you'll see two different mosfets, the top one has nothing connecting to the Gate, that's where I need the circuit.

I'm also thinking if there's a better way to keep the caps balanced, my new design was going to use this TI BQ33100 and some 1A'ish MOSFETs to balance the caps but wouldn't a better solution to be to charge each cap separately with a 2.4V source? Problem is I can't seem to find a good 2.4V constant current 10A voltage source and I'd need 5 of them.

 

I've been working on a boxing robot for a number of years. It's two human sized robots that you control with a joystick or a Kinect camera (you do a left punch, it does a left punch). Hoping to sell these into arcades.

I've been struggling with the main power board. I'm using large wormgear motors, they use a lot of current at startup (first 1/10th sec) and then it settles down. I don't want to put much of an acceleration ramp in them because I want them to be very quick from a dead stop. But this places a huge load on the power supply (1200W PC power supply at the moment) so I've added in 5 x 2.7V 350F capacitors (70F @ 120V) which seems to be working.

Well yesterday I sent a motor controller up in smoke because one of the motors stalled and current/voltage shot through the roof and as I've now learned 60A slow auto fuses don't pop at 60A. After talking to the motor controller manufacturer (Basic Micro makes the Roboclaw, who is awesome btw, buy stuff from them) we came up with the idea of doing two things. First a voltage clamp on my power board separate from the one built into the motor controller. So if voltage goes above 13V (12V supply) it'll turn on a MOSFET and bleed voltage through a large 50W panel resistor. The second is a similar circuit that might turn on at a higher voltage (maybe 15V?) if the voltage clamp doesn't get things under control. It would turn on a MOSFET that would turn off a large relay, killing all power to the controllers.

I'm now trying to figure out a good design for monitoring the voltage. I don't have experience with comparators but it seems like I would need two of them, both comparing supply voltage against motor controller voltage (voltage after the diode protecting the power supply). Any suggestions? I've attached the current schematic. On the right side of the schematic you'll see two different mosfets, the top one has nothing connecting to the Gate, that's where I need the circuit.

I'm also thinking if there's a better way to keep the caps balanced, my new design was going to use this TI BQ33100 and some 1A'ish MOSFETs to balance the caps but wouldn't a better solution to be to charge each cap separately with a 2.4V source? Problem is I can't seem to find a good 2.4V constant current 10A voltage source and I'd need 5 of them.

Having designed what is a quite complex circuit, I’m surprised that you do not have the relevant knowledge re comparator circuits – they really are quite simple.

But your description of the problem does not adequately provide the precise voltages at which the comparators are to change state (or if it is a difference between voltages, rather than absolute values).
If you clearly state this, I’m sure you will get the assistance you need.

You could always change the rating of the fuse – a good rule of thumb is that a fuse can pass 1.6 x rating for a considerable time, at 2.1 x rating the fuse should operate within 30 seconds.

 

5 x 2.7V 350F capacitors (70F @ 120V)

I take it you mean 12V
An easy way to get high capacity 12V supply would be to use a car battery for you "filter cap".
Probably cheaper than a stack of super caps too.
Use some Hall current sensors to monitor the motor current, not voltage, and have the sensor outputs controlling a PWM current limit. That way you can get max acceleration from the motors and not blow them up. Also, an added bonus is the current limit threshold can be wound down to slow the machine movement for novices.

Here are some from Ebay that start from under $3 each.

 

Having designed what is a quite complex circuit, I’m surprised that you do not have the relevant knowledge re comparator circuits – they really are quite simple.

But your description of the problem does not adequately provide the precise voltages at which the comparators are to change state (or if it is a difference between voltages, rather than absolute values).
If you clearly state this, I’m sure you will get the assistance you need.

You could always change the rating of the fuse – a good rule of thumb is that a fuse can pass 1.6 x rating for a considerable time, at 2.1 x rating the fuse should operate within 30 seconds.

Ha, yes you would think I would know how to setup a comparator but this is what happens when an MBA one day says "how hard can robots be to build? ". The design is mostly taken from the BQ33100 datasheet with some other bits added to it.

It looks like I had slow fuses and they don't make a fast in that size so I'm going to go with a small fuse holder (ATO size) to give a better fuse selection.

For the voltages I'll have 12V as the reference and for one circuit I want it to compare against 14V and the second circuit will compare against 16V.

 

Are you looking at the motor current or voltage?
It needs to be current.
Do you have circuits of this beastie?

 

I take it you mean 12V
An easy way to get high capacity 12V supply would be to use a car battery for you "filter cap".
Probably cheaper than a stack of super caps too.
Use some Hall current sensors to monitor the motor current, not voltage, and have the sensor outputs controlling a PWM current limit. That way you can get max acceleration from the motors and not blow them up. Also, an added bonus is the current limit threshold can be wound down to slow the machine movement for novices.
View attachment 155395
Here are some from Ebay that start from under $3 each.

I started originally with a 7ah lead acid battery but it would slowly die when connected to a 12V source. That means I would have to build a boost circuit to charge it and then possibly buck to step back down to 12V to augment the power supply.

The motor controller already does current monitoring. I think there was inductance feedback from the motor that caused the problem. You're talking about PWMing the 12V source going into the motor controller? That sounds like building a motor controller for my motor controller?

 

Are you looking at the motor current or voltage?
It needs to be current.
Do you have circuits of this beastie?

I figure that the fuses should protect things from overcurrent and a voltage clamp or comparator shutting off a relay powering the motor controllers would protect from overvoltage caused by regen.
I have the proposed circuit I posted. You mean an overall block diagram?

 

Your motor controller should be able to be set for max current to protect the motors.
What are the boards you are using. Really, some detail of the electronics would be a great help. Without that, it is pretty hard to help.
Yes, a full circuit, or at least the motor drivers and the controller.
I do not think you really need to go to super caps to make it work, just have a power supply rated at enough current and limit the motor current to a set max value.
Fuses are just to protect thing in a fault condition, not run time.

 

Just got an email from the motor controller manufacturer. He tested my motor and said the induct ance was 50uH and that was causing excessive ripple. He said to add 5000uf at the motor controller power in terminal. He said that the diode in front of the supercaps was preventing them from helping. He said motor also showed 1.4 ohm resistance and it pulled 8.5A startup dropping to 2A when up to speed.

 

The power supply is already 1200W rated (100A at 12V in the spec). I don't really have any room on a normal 120V breaker for anything bigger.

 

Here's another issue I just ran into. I took the previous version of this design that had a few mistakes like I mentioned so I can't fully test it. But the supercapacitors are installed in series with ALD810025 and ALD910025 balancing mosfets. I just connected a power supply to it and gave the 5 caps (in series) 12V. When I did that the first cap closest to V+ overcharged at around 3V, then the next one was around 2, then approximately 1.5, 1, 0.5 for the rest. Even without the balancing mosfets why would they charge so unevenly? I waited a day until they were half drained and then charged at 2.7V@10A for each individual cap. They all charged to that voltage without issue.

If it was difference in capacitance for each supercap the odds that those difference would be in the exact order they're installed seems pretty low. It just so happens that the first one has the highest capacitance the last one has the least? What else could cause this?