Why not integrate Jaguars with Diodes?

[EricVanWyk]

Originally Posted by Matt Krass

Just curious, because I'm sure there's an angle I haven't thought of, what kind of incorrect wiring are you anticipating causing a crowbar to be an ineffective solution?

The two factors to consider are the magnitude and duration of the current. Many fuses are rated in terms of amp squared seconds.

There are several common situations that accidentally bypass a circuit breaker OR uses the wrong circuit breaker. These allows the same magnitude but longer duration current pulse.

There are also weird ones where the wiring includes a pathologically bad amount of resistance. These allow a low magnitude but long duration current pulse. In this case, the crowbar might be able to protect its owner until it burns itself out.

[Matt Krass]

Originally Posted by EricVanWyk

The two factors to consider are the magnitude and duration of the current. Many fuses are rated in terms of amp squared seconds.

There are several common situations that accidentally bypass a circuit breaker OR uses the wrong circuit breaker. These allows the same magnitude but longer duration current pulse.

There are also weird ones where the wiring includes a pathologically bad amount of resistance. These allow a low magnitude but long duration current pulse. In this case, the crowbar might be able to protect its owner until it burns itself out.

Ah, I did consider that, but I was under the impression that in this situation it wasn't relevant because in the event of a reverse power connection to the controller, my experience has been that it short circuits in the same manner as the crowbar and causes the same metronome of breaker popping. I admit I haven't (yet) taken the measurements, but my general hunch was that the crowbar solution was no worse on the system regarding violent current pulses of varying duration than the existing problem, but it would spare the controller. The only added concern I saw was buildup of heat on the diode, but I figured with a proper heatsink on a diode designed for 100A whacks, it's probably safer than the tiny components in the Jag absorbing the hit.

I appreciate your input, these types of discussions are always fun, do you have any more thoughts on my theory?

For the record, I was working with this guy for the test:
http://www.st.com/internet/com/TECHN...CD00077607.pdf

Matt

[EricVanWyk]

That is a pretty diode! I just wish it was smaller than a spike

Compare figure 3 of that diode's datasheet against the figure on the 40Amp circuit breaker's datasheet. A 400Amp surge would blow the breaker in 30 to 60 milliseconds, and is on the edge of what the diode can handle.

For even more fun, take a look at the 120A circuit breaker's datasheet. At 600 Amps, it guarantees to blow within 3 seconds.

[Matt Krass]

Originally Posted by EricVanWyk

That is a pretty diode! I just wish it was smaller than a spike

Compare figure 3 of that diode's datasheet against the figure on the 40Amp circuit breaker's datasheet. A 400Amp surge would blow the breaker in 30 to 60 milliseconds, and is on the edge of what the diode can handle.

For even more fun, take a look at the 120A circuit breaker's datasheet. At 600 Amps, it guarantees to blow within 3 seconds.

Well, this may not be the perfect diode for the job, but I assure it was selected through a very rigorous "what can I get my hands on in the next five minutes" selection process. Also, if you were to parallel the two diodes per package it would give you some fairly substantial breathing room on those figures. If I get a chance soon I'd like to put my storage 'scope on the diode and capture the waveform and see exactly how effective it is.

Also, as I mentioned before, I'm not sure that the Jaguars failing is any gentler on the system, and I'm sure the components failing in that aren't nearly as good at handling the surging currents safely, I'm still feeling as if I'm in the lesser of two evils camp here. I concede this is mostly a hunch, after looking over the Jaguar schematic I suspect the failure mode is near a dead short in one of the logic regulators. I'm not sure that the opportunity to get hard data on the Jag failure is going to present itself anytime soon (I think my team would be a little unhappy if I sacrificed one of our remaining Jags to the cause ) though I am unsure if the blown out ones have all been disposed of yet.

What are your thoughts on the risk of the failing Jaguar blowing the breaker over and over vs a properly heatsinked crowbar diode (or two in parallel) rated for the actual load?

Matt

[philso]

It would be best if TI could add the reverse polarity protection in the Jaguar.

Adding protection devices externally is a distant second best since it is still possible to connect the wires from the protection circuit to the terminals of the Jaguar incorrectly.

It will always be necessary to check your wiring thoroughly. Even if the polaritiy is correct, the wire could be connected to the wrong point giving incorrect operation.

[techhelpbb]

Just a point.

The core of the Jaguar that handles the large current is the N-Channel MOSFET bridge.

That bridge has a high side driver to make the positive voltages above the input voltage required to actually turn on the high side MOSFETs (they aren't P-Channel so you need this trick).

If the bridge had P-channel high side MOSFETs you'd have more issues (I think I remember the older beige Jaguars being like this) because the match between the P-channel and N-channel MOSFETs is not great and P-Channel MOSFETs tend to be really easy to utterly destory under bad conditions.

The bridge itself could be rigged to survive inverting the positive and negative being reversed. Hence you wouldn't need a 100A or more diode (you best not use a 60A diode on a load that reaches 60A unless you like bad things happening).

The rest of the electronics, including the Stellaris could be powered from a simple bridge rectifier (which would make the positive and negative on the other side *always* the right polarity...if you don't mind loosing 1.4 or so volts DC on the input to the local power supply and I doubt based on the existing schematics it would).

In short I believe this could be done and NOT have to deal with high power parts to do it.

[Kevin Sevcik]

Originally Posted by lemiant

The diode wouldn't have to affect all the power, just the power going to the control circuitry. If the jaguar's internal computer didn't turn on, it would never send any signals to the internal relay (or whatever they use) and nothing bad would happen. The computing uses a lot less amperage so this is totally feasible.

I'm a bit late coming to this, but I feel it necessary to point out that you have to protect the power circuitry as well as the logic in this application, which is why it's so complicated and expensive. The MOSFETs that make up the H-Bridge all contain a body diode as part of their construction. When properly powered, these diodes are reverse biased relative to the battery terminals. If you reverse the polarity on the battery terminals, you now have pretty much a dead short across two diodes and a 1mOhm current sense resistor. An FRC battery is going to dump enough amps into that circuit to melt something pretty quickly. (My money's on the current sense resistor, I've heard about them desoldering.)

Matt,
If you want to drastically reduce the chance of swapped quick connects frying something, keyed connections are the way to go. Anderson power pole or molex connectors will do this for you. Or you can do what we do with quick connects: Swap genders from one polarity to the other. So a positive out of your PDB gets a male QC, and a negative out of the PDB gets a female. As long as you're consistent across the whole robot, it doesn't hurt interchangeability, and you really reduce the chances of that particular failure. Or the chance of wiring a motor up backwards, etc.

[Newo95]

Originally Posted by lemiant

If they broke less people would buy less

Planned obsolescence is everywhere.

[ratdude747]

Originally Posted by Sammy 3278

We replaced the colored screws with all silver so that people won't assume black means negative and red means positive and will look at the molding on the speed control instead. Has been working so far. We had also toyed with the idea of using red and black nail polish on the speed control.

not a bad idea...

during my days as a student electronic lead I never had a jaguar die... in fact, I event was able to revive a few "dead" ones from the season before by opening them and cleaning out any aluminum swarf that had shorted them in the past... I guess I got REALLY lucky (now you see why I liked CAN so much).

also, we only used fork terminals for the Jags... if we had to use a ring I would cut part of it into a fork (until we could get more fork terminals). if the screws never come out, then problem solved

Even if you do use ring terminals, my advice would be to work one screw at a time. that also would eliminate the chance of mixing them up...

[EricVanWyk]

Originally Posted by techhelpbb

The bridge itself could be rigged to survive inverting the positive and negative being reversed. Hence you wouldn't need a 100A or more diode (you best not use a 60A diode on a load that reaches 60A unless you like bad things happening).

This does not match my experience. Could you further explain how to construct the bridge so that it can survive these conditions? When voltage is applied with the wrong polarity, the body diodes will conduct and soon destroy themselves.

[techhelpbb]

Originally Posted by EricVanWyk

This does not match my experience. Could you further explain how to construct the bridge so that it can survive these conditions? When voltage is applied with the wrong polarity, the body diodes will conduct and soon destroy themselves.

Some ideas off hand...

1. Close a relay contact over the circuit and hope they really have a circuit breaker up the path of current to the battery. You can do this because you'd have power from the bridge rectifier on the logic side. Could be too slow, but remember these diodes do take large surges during field collapse anyway...you do have a little time.

2. Rig a capacitor circuit to charge it's way up and reduce the voltage differential (it's not AC it won't discharge). Just make sure you check with the Stellaris if the power is backwards before you start switching. Probably have to make this component exit from the circuit with a small relay as well in the proper polarity.

3. Use fast blow fuse not these thermal breakers and make it easy to replace.

No matter what the H-Bridge will not be operational with the power reversed but I think it should be possible to give it enough protection to survive a minor error.

[Kevin Sevcik]

Originally Posted by techhelpbb

Some ideas off hand...

1. Close a relay contact over the circuit and hope they really have a circuit breaker up the path of current to the battery. You can do this because you'd have power from the bridge rectifier on the logic side. Could be too slow, but remember these diodes do take large surges during field collapse anyway...you do have a little time.

2. Rig a capacitor circuit to charge it's way up and reduce the voltage differential (it's not AC it won't discharge). Just make sure you check with the Stellaris if the power is backwards before you start switching. Probably have to make this component exit from the circuit with a small relay as well in the proper polarity.

3. Use fast blow fuse not these thermal breakers and make it easy to replace.

No matter what the H-Bridge will not be operational with the power reversed but I think it should be possible to give it enough protection to survive a minor error.

You're over-complicating things. If you're already willing to go to the trouble and expense of tossing a relay into the circuit, then you simply use the relay to energize the power stage. Heck, with a double throw relay, you could have the relay swap your polarity on the power stage for you. Your only problem is that you need a relay that can handle 40A continuous. Those relays aren't small or cheap.

The better open would be a standard mosfet based pass-protection eric described, which I'm pretty sure is already too expensive for the product.

[techhelpbb]

Originally Posted by Kevin Sevcik

You're over-complicating things. If you're already willing to go to the trouble and expense of tossing a relay into the circuit, then you simply use the relay to energize the power stage. Heck, with a double throw relay, you could have the relay swap your polarity on the power stage for you. Your only problem is that you need a relay that can handle 40A continuous. Those relays aren't small or cheap.

The better open would be a standard mosfet based pass-protection eric described, which I'm pretty sure is already too expensive for the product.

I agree if you were switching the actual power leads to the H-Bridge you'd need a relay of that size. That's exactly why I didn't suggest it. You can under-rate a relay if you're planning on operating it infrequently. The goal was to create a circuit with the relay that wasn't going to frequently operate (preferably not at all). If you can work the idea with the capacitor out the currents involved will reduce themselves.

I also agree that Eric's suggestion is perfectly reasonable of course the cost of the MOSFET is the issue. Then again, if you can work it out such that the parts in question are the same as those in the H-Bridge perhaps the cost advantages of quantity would kick in and trump any of the other ideas I've provided.

[EricVanWyk]

Originally Posted by techhelpbb

Some ideas off hand...

1. Close a relay contact over the circuit and hope they really have a circuit breaker up the path of current to the battery. You can do this because you'd have power from the bridge rectifier on the logic side. Could be too slow, but remember these diodes do take large surges during field collapse anyway...you do have a little time.

2. Rig a capacitor circuit to charge it's way up and reduce the voltage differential (it's not AC it won't discharge). Just make sure you check with the Stellaris if the power is backwards before you start switching. Probably have to make this component exit from the circuit with a small relay as well in the proper polarity.

3. Use fast blow fuse not these thermal breakers and make it easy to replace.

No matter what the H-Bridge will not be operational with the power reversed but I think it should be possible to give it enough protection to survive a minor error.

My apologies, I must have misunderstood your earlier post. Can you sketch a schematic of what you are suggesting?

The confident tone led me to believe you had a somewhat detailed suggestion in mind with a compelling cost/benefit ratio.

[techhelpbb]

Originally Posted by EricVanWyk

My apologies, I must have misunderstood your earlier post. Can you sketch a schematic of what you are suggesting?

The confident tone led me to believe you had a somewhat detailed suggestion in mind with a compelling cost/benefit ratio.

Not now I can't. I am sitting in a McDonalds in downtown Manhattan eating my dinner. If I have time this weekend I'll draw and test what I have in mind (it's my dime right....if I nuke my MOSFETs you've got nothing to loose....in the meantime I'll spare you the lengthy written descripition).

That being said, the other way to look at the cost/benefit ratio of this particular issue is that you don't have to make all the Jaguars the same. You could make room in the housing (or use room in the existing housing) and essentially make an additional module or a different PCB for the reverse polarity protected version.

Then, if someone desires, they could pay extra for this feature without penalizing all the teams more confident they'll avoid this pitfall.

I really don't see the harm in that idea do you?