PDA

View Full Version : reverse-voltage protection


kamocat
09-04-2010, 03:08 PM
I was thinking about how to implement reverse-voltage protection on a Jaguar or Victor, and I realized that all the polarity sensitive components, with the exception of the electrolytic capacitor, are powered by the 5v or 3.3v regulators.
Couldn't we just put a diode on the supply side of both of those, and thus prevent the motor controller from turning on when voltage is reversed?
(The capacitor, though it would blow hot oil all over everything, can be replaced)

On a similar note, how is providing power to the wrong side different than normal operation? It's still the same to the h-bridge. Why does the motor controller even turn on, if the motor is stopped? The source and drain on a MOSFET can be flipped without issues, correct?

EricVanWyk
09-04-2010, 04:02 PM
I was thinking about how to implement reverse-voltage protection on a Jaguar or Victor, and I realized that all the polarity sensitive components, with the exception of the electrolytic capacitor, are powered by the 5v or 3.3v regulators.

Reverse voltage protection is very simple conceptually, but gets very expensive when you have high operational power. There are two main types: shunt and series.
A shunt needs to be able to safely shunt many times the operational power for long enough for the breakers to blow. Usually this is a diode or a diode-like device.
A series method needs to be able to conduct the operational power continuously. For an example of a series method, take a look at the schematics for the DSC or Solenoid Breakout.


Couldn't we just put a diode on the supply side of both of those, and thus prevent the motor controller from turning on when voltage is reversed?
(The capacitor, though it would blow hot oil all over everything, can be replaced)
Still need to protect the bridge, per your next statement...

On a similar note, how is providing power to the wrong side different than normal operation? It's still the same to the h-bridge. Why does the motor controller even turn on, if the motor is stopped? The source and drain on a MOSFET can be flipped without issues, correct?

All MOSFETs (that have 3 terminals) have an internal body diode between source and drain. Reversing polarity allows current to flow through the diode and pop the FET.

kamocat
09-04-2010, 04:20 PM
Oh, I was thinking MOSFETs were more like JFETs. That's too bad.
That means in order to have a controller that wouldn't care which side you hooked up the power, you'd need to use JFETs and pre-amplify the signal from the H-bridge drivers. (Supplying power to all this would require a bridge rectifier on both sides)
That's a pain.

DonRotolo
09-04-2010, 09:22 PM
eeds to be able to conduct the operational power continuously.
The drawback to series protection diodes is their voltage drop - nominally 0.7 volts. That drop hurts your motor power, gets turned into heat, and (as Eric said) needs a device that can handle at least 40 Amps, which isn't all that inexpensive (compared to the cost of a Jaguar, at least).

It certainly can be done, or we just train the operator to not reverse polarity.

kamocat
09-05-2010, 11:26 AM
Using a shunt does seem to be the easiest method. The trouble with that is those auto-resetting breakers. (You want this to trip fast, not take two seconds)
So, this would require an inline fuse with tight trip characteristics.

Considering that the PWM drivers seem to be the most sensitive, I would still put a diode in front of the 3.3v and 5v voltage regulators.

Gdeaver
09-05-2010, 02:07 PM
As mentioned the problem with a diode is the voltage drop and heat generated with it. The current handled by the diode would have to be much greater than 40 amps. In fact it would have to handle a direct short. Many automotive motor drivers use a Fet to provide reverse battery protection. This can be done several Ways.The advantage of a power FET is the low on resistance and less power loss. A PNP Power FET could be used on the high side with the gate tied to ground with a resistor and diode. The problem with a PNP power FET is that they are not available with large current handling and low on resistance. If PNP FETS could be made with the same robust characteristics as NPN FETS they would be used in the motor drive h bridge. Multiple PNP's could be used but, that takes space and adds cost. A NPN FET can be placed on the low side with the gate tied to the V+ with the resistor and diode. The problem with low side protection is that it messes with the ground. Electronics do not like it when the ground has transience on it. A NPN power FET can be used on the high side. The problem is that the gate drive voltage has to be above V+. This requires a voltage boosting circuit adding cost and board space. ST micro makes a bridge driver chip with this circuit on the chip. The other way the jag could be some what protected from reverse polarity and swapping the V+ and motor leads would be to use polarized connectors with a different type for the V+ and Motor. Then the risk of zapping a jag would be limited to when the cables are made up. The PD would also have to have polarized connections. As long as the power connections remain as they are now, every year there will be x number of fried controllers every year. Our team has fried a few.

kamocat
09-05-2010, 02:30 PM
You said electronics don't like it when their ground is floating? Does it still matter when their V+ is floating as well?
I'm assuming you're just referring to the small electronics, not the power MOSFETs.

I would be perfectly fine with a keyed connector, such as 45-amp Anderson Powerpole.

Gdeaver
09-05-2010, 04:16 PM
The Jag has a micro controller, bridge driver chip, CAN driver and voltage regulator that all must have good stable grounds. In addition the micro is making A to D conversions to manage current measurements. The jag is driving a large inductance motor. There are going to be large switching transience floating around. The designer has to manage these. Adding Low side reverse battery protection adds allot of problems to the design. High side is much easier from the ground perspective. Managing the voltage boost and FET power dissipation are the trade off.
The Anderson connectors are big and adding them to the jag which is all ready big would not be ideal. There are smaller connectors that could be used like the RC Dean's connectors. The problem is they could see up to 160 amps. Beyond there rating. Is it really required to have true continuous ratings for our use?
On this subject, would it be acceptable to loose thousands of dollars in electronic modules in a car because a mechanic accidentally reversed battery connections even for a fraction of a second. Can a automotive mechanic reverse polarity by plugging in a new module wrong? If one could the bean counters would be string up the engineers for the warranty claims. Now would this be acceptable in aerospace or a nuclear power plant? Why is it acceptable for our robots.

Al Skierkiewicz
09-05-2010, 11:30 PM
Marshall,
I would be willing to bet that the majority of dead controllers due to wiring errors are for the input and output to be reversed. Remember that the HEX power FETs that are used have a diode across the source drain leads that is a fall out of the manufacturing process. The series diode would still need to handle 129 amps, the stall current of the current CIM motors. Schottky diodes have a lower forward voltage drop but are too expensive and large to be included in the controller design. We use the smaller Anderson connectors for all/most of our motor connections. The 45 amp version handles the CIM motors just fine. There is not enough heating in a two minute match to damage these connectors provided that the correct crimper is used to terminate the wire.

RyanN
09-06-2010, 12:09 AM
This involves a few more components, but will get the job done.

NOTE: NOT FRC LEGAL!

You could always add in a simple diode and a relay capable of handling 40A continuous.

Wire the relay to only activate when the right polarity is given. The relay closes the contact to the Jaguar and will then allow the Jaguar to operate.

kamocat
09-06-2010, 01:36 AM
Marshall,
I would be willing to bet that the majority of dead controllers due to wiring errors are for the input and output to be reversed. Remember that the HEX power FETs that are used have a diode across the source drain leads that is a fall out of the manufacturing process. The series diode would still need to handle 129 amps, the stall current of the current CIM motors. Schottky diodes have a lower forward voltage drop but are too expensive and large to be included in the controller design. We use the smaller Anderson connectors for all/most of our motor connections. The 45 amp version handles the CIM motors just fine. There is not enough heating in a two minute match to damage these connectors provided that the correct crimper is used to terminate the wire.
From personal experience, I've only wired up power to the wrong side once. That happened to be a Victor. Interestingly, the MOSFETs were the only damaged parts. (They didn't blow until I put significant current through them by actually trying to run the motor) I've replaced those with MOSFETs from Jaguar's I've fried, and it seems to work fine now.

We do use pigtails with mini-anderson connectors on them. On the output side, we actually use a different color (white and green), where as we use red and black on the supply side.
Most of the errors come from cheesed-together systems that haven't been thoroughly checked. I have a small 6 Amp-Hour battery that has a tendency to get reversed (either at the Anderson Powerpole or the push-ons to the battery terminals).
To date, I have 3 Jaguars I use for physical mock-ups. They've all been fried by reversed voltage (either by me or by somebody else on the team).

Al Skierkiewicz
09-07-2010, 07:38 AM
To date, I have 3 Jaguars I use for physical mock-ups. They've all been fried by reversed voltage (either by me or by somebody else on the team).

A cheap fix for this is to use a sharpie when the Jag or Victor come out of the box and mark input and output with a large, unmistakable "+" sign and "IN" and "OUT".

Bot190
09-07-2010, 03:23 PM
The problem isn't always with the wiring to the jaguar itself. This year, one of our batteries was wired backwards. We only found out when we connected it to the robot and turned it on. Reverse voltage protection would have saved at least 2 of our jaguars in this instance.

AdamHeard
09-07-2010, 04:05 PM
The problem isn't always with the wiring to the jaguar itself. This year, one of our batteries was wired backwards. We only found out when we connected it to the robot and turned it on. Reverse voltage protection would have saved at least 2 of our jaguars in this instance.

I think what people are getting at is that the additional cost, in either performance loss or actual costs, is not worth it. The sum of that additional cost is probably substantially more than the value of the speed controllers damaged. I've been doing this for a while, and I can't recall killing a single speed controller due to reverse wiring. I understand it can be a catastrophic error, especially with a backwards battery, but it's just not a problem worth rectifying

EricVanWyk
09-07-2010, 04:44 PM
but it's just not a problem worth rectifying

Very Punny ;)

kamocat
09-07-2010, 08:51 PM
How about a partial solution?

Earlier I mentioned putting a diode in front of the 5v and 3.3v linear regulators. This would protect everything except the power MOSFETs and the electrolytic capacitor. The only impact on performance would be a higher minimum voltage.

Say a Jaguar with such protection had reversed voltage. Let's take a look at what it would take to fix it:
The power MOSFETs (FDP8441) can be bought for less than $3 apiece at Digikey (http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=FDP8441-ND).
The electrolytic capacitor is an inexpensive and easily available part.

Both the capacitor and the MOSFETs can be hand soldered by someone with reasonable experience.
Thus, the Jaguar can be fixed by an end user for less than $30.


Does anyone have a better solution?

Al Skierkiewicz
09-08-2010, 07:44 AM
The problem isn't always with the wiring to the jaguar itself. This year, one of our batteries was wired backwards.

Bot,
This is a very costly lesson, lucky you only lost two Jaguars and not the PD and other stuff. This suggestion should go in another thread as well but check and recheck before applying power. There is a lot of ways to do this. Get a red sharpie and mark the battery positive terminal. Red to Red to Red gets the right polarity to the robot. Check every battery with a voltmeter before connecting to the robot. The Anderson plugs actually have a big "+" marked on the connector body. Don't trust that the person who assembled the connector pushed the correct wire into the body. Most voltmeters have a polarity sign in the display, even my $3 flea market pocket voltmeter has one. Check the battery before connecting, then before any breakers are added to the PD, check the voltage again with a battery connected. Check every battery you own the same way. Then add breakers after you have checked the wiring from each breaker to it's destination. Have someone (usually two or three) check you. Then when all agree, power the system.
Here's a couple of hints that are common problems...
1. Whiskers at the termination point. A stray strand of wire, in a connector, touching an adjacent wire in a connector (PD to Crio and sidecar power). This is easily handled by not stripping the wire too long. Twist the wires before insertion. If there is any copper showing when you are done, hold the wire in one hand and pull down on the insulation towards the connector with your other hand. The insulation will stretch and cover the exposed copper.
2. Use heatshrink to cover up exposed wiring. It is cheap and fun to use.
3. Insulate the battery terminals when you are finished terminating the power wiring. The battery is capable of 600+ amps when fully charged. It won't kill you but it can cause a halacious fire in the right setting.
4. Use the "tug" test on crimped terminals. Pull on the terminal with all your might, it should not pull off and it should not move. A ratchet crimper will cure most of these problems and costs about $50.
5. Tie down all wiring near the termination (PD, connector, speed controller, Spike, motor, etc.) The robot moves, sometimes violently, so don't depend on the terminal or connector to take care of itself.
6. Inspect every connection after every match. Use your eyes, ears and nose.

Jon236
09-08-2010, 09:17 AM
Al,

FIRST didn't put out a 'Best Practices' Manual last year; it was a great document, especially for rookie teams. If they don't do it again this year, I suggest a compilation of 'Best of CD' to be made available. While rookies do need to learn how to search CD, I think a compilation would be a good resource for them.

Al Skierkiewicz
09-08-2010, 09:33 AM
Al,

FIRST didn't put out a 'Best Practices' Manual last year; it was a great document, especially for rookie teams.

Jon,
That document never made it to published status last year. I am pushing for a re-release this year. Yes it is a good document that rookies need.
It can be found here...
http://www.usfirst.org/uploadedFiles/2009%20FRC%20Suggestions%20Rev%20A.pdf

DonRotolo
09-09-2010, 09:18 PM
This year, one of our batteries was wired backwards.
Did someone learn a valuable lesson that day? If so, it's worth a couple of Jaguars.

Try that with a $250,000 device and see how long you remain employed.

theprgramerdude
09-13-2010, 08:51 PM
The problem isn't always with the wiring to the jaguar itself. This year, one of our batteries was wired backwards. We only found out when we connected it to the robot and turned it on. Reverse voltage protection would have saved at least 2 of our jaguars in this instance.

This is interesting. My sophomore year I made the same mistake of hooking it up backwards, and what happened to us was that the battery simply turned into a short since the PD Board acted as a reverse battery protector. Wires were fried before we stopped it, but that was all. How did you manage to even run any components with a reversed battery?

Bot190
09-14-2010, 06:14 AM
I don't know the specifics, because i wasn't at the robot at the time. I just know that we damaged at least 2 jaguars.

As Kamocat has said, if the processor and other stuff was protected from reverse voltage it might have saved our jaguars. I do agree that protecting the entire jaguar would could be cost prohibitive.