Blown talon srx modules

[Mike Copioli]

Originally Posted by Michael Hill

Seeing as the data has not been released (that I have seen), of course I don't know what it is. I would expect it should be on the order of 15kV. Anything less than 8kV, I would say is probably insufficient for this application.

So what happens if the ESD event is greater than 15 kV?

[Michael Hill]

Originally Posted by Mike Copioli

So what happens if the ESD event is greater than 15 kV?

Then it fails? I'm not sure what you're getting at.

[GregGarner]

I have a new theory:

Last night it occurred to me that our team is probably doing one thing quite differently than any other team out there. We elected to make our own encoder/limit switch cables, rather than using the pre-made cables that come from Vex. The reason we did this is because our team is cash limited, and we could save a good chunk of change making our own cables.

Therefore we bought the .025 ribbon cable in bulk, and purchased the correct .050 spacing dual row IDC connectors for that cable. We then proceeded to make our own cable to go all the way from the Talon SRX out to the Encoder, which in some cases ended up being 8 feet long. We then terminated the ribbon cable with the correct 5 pin connector into the encoder.

Since the ribbon cable has all 10 conductors in it, this strategy implies that every signal on the Talon is being routed all the way out to the end of the 8 foot cable, even though there are some signals we didn't need. In particular, the 3.3V signal is connected to the ribbon cable. The 3.3V signal is not used by our encoder, and it is not terminated at the end of the cable. This effectively means that we have a 8 foot antenna on the 3.3V wire, which could pick up any signal in the area and connect it into the Talon 3.3V rail.

In addition, we put the motor wires into the same wiring duct with the encoder cables. This means that the current pulses in the motor wires could be inductively coupling into the encoder ribbon cable, and in particular the 3.3V line going straight to the Talon SRX CPU.

We have noticed that out of the 5 blown Talon SRX modules, at least 4 of them were on the longer encoder cables. We have not seen any Talon SRX module have a problem when it had a shorter encoder cable (i.e. The Talon SRX modules are physically closer to the drive motors so the cables are shorter. These closer Talon SRX modules have not failed).

We have seen the Talon SRX modules fail when functioning, and also when not functioning but we have been manually rotating the motors by manipulating the robot. Again the failures are only seen when there are long cables to the encoder.

My goal at this point is to shotgun the problem and try lots of different things at once to try to stop the problem. We have one week to bag and tag, and so it means we can't debug the problem in the standard way of changing one thing at a time.

I am going to do at least one more test before tearing apart the wiring. I will put a digital scope on the 3.3V line at the Talon and look and see if I can see any significant induced spikes on the 3.3V line when moving the motor under program control and under manual control, on the talon SRX with the longest attached ribbon cable. I will report back on what I see.

In order to try to stop the problem from happening, we intend to use shorter ribbon cable out of the Talon SRX, and then splice in a more robust wire to go on out to the Encoder. This means we will not be routing out all the unused signals from the Talon SRX to the rest of the robot, and in particular we will not route out the 3.3V line. Also, we will run the encoder cables in a separate wiring ducts, so there is physical separation between the encoder signals and the large current power wires for the motors.

Finally we do intend to follow industry standard ESD protocol as part of the shotgun approach.

I want to emphasize again that the guys at CrossTheRoadsElectronics have been very helpful, and are replacing the blown Talon SRX modules. In addition they are assisting us with trying to understand why this is happening. They have also sent us some cables and breakout boards to try, since that is what other teams are probably using.

GregGarner
Team 3612

[wireties]

Originally Posted by Michael Hill

Home Theater equipment, for one. But regardless, the wires didn't even have to be like that. It could have been some kid of female receptical (something like powerpole).

Speakers are outputs and muuuuuch easier to protect than any kind of input.

A connector for power and for CAN/PWM would have been nice but would require some sort of standardization for CAN/PWM and power interconnections.

[magnets]

Originally Posted by GregGarner

The 3.3V signal is not used by our encoder, and it is not terminated at the end of the cable. This effectively means that we have a 8 foot antenna on the 3.3V wire, which could pick up any signal in the area and connect it into the Talon 3.3V rail.
...
Finally we do intend to follow industry standard ESD protocol as part of the shotgun approach.

Interference is no longer an ESD problem.

Also, a wire that isn't terminated at one end isn't likely to result in much inductive coupling, and it will be much, much less than a wire that's terminated at both ends. Having it terminated at only one end will make inductive coupling almost disappear. There's no effective loop area if you don't have a loop!

You may have parasitic capacitive coupling though.

[Mike Copioli]

The Talon SRX is tested to both EN 61000-6-1 and EN 6100-4-2, 8KV air gap and 4 kV direct contact. This is the standard that most consumer electronics are tested to.

We used the attached TVS diode in several places in the design. It is rated to >+/-30kV.

[Mike Copioli]

Originally Posted by Michael Hill

Then it fails? I'm not sure what you're getting at.

I agree it should fail. But I would not suggest that the product is crap if it did fail because of this.

[s1900ahon]

Greg,

Could your custom cables be shorting adjacent wires and therefore be shorting things within the Talon? If misaligned, this could be a problem since they're insulation displacement connectors.

Moreover, if these cables are faulty, you'd continue to propagate failures.

[philso]

Originally Posted by GregGarner

I am going to do at least one more test before tearing apart the wiring. I will put a digital scope on the 3.3V line at the Talon and look and see if I can see any significant induced spikes on the 3.3V line when moving the motor under program control and under manual control, on the talon SRX with the longest attached ribbon cable. I will report back on what I see.

It can be very difficult to get useful information doing this. Your scope and it's probe can inject noise into the system since it acts like an antenna too. How you ground your scope probe to the system is critical to how clean signal you get. A lot of what you may see may be induced in the ground lead of the scope probe. You are probably better off putting a high-frequency current probe around the cable coming out of your Talon since the current probe will not be Gavanically connected to your system.

Secondly, the 3.3V line is probably a power supply line. The impedance looking into the Talon on that line will probably be quite low making it hard for any outside interference source to have a significant influence on the internal circuitry of the Talon. Basically the internal impedance of the 3.3V line forms a divider with the impedance coupling to your external noise source.

Originally Posted by magnets

Interference is no longer an ESD problem.

Also, a wire that isn't terminated at one end isn't likely to result in much inductive coupling, and it will be much, much less than a wire that's terminated at both ends. Having it terminated at only one end will make inductive coupling almost disappear. There's no effective loop area if you don't have a loop!

You may have parasitic capacitive coupling though.

Yes.

Originally Posted by s1900ahon

Greg,

Could your custom cables be shorting adjacent wires and therefore be shorting things within the Talon? If misaligned, this could be a problem since they're insulation displacement connectors.

Moreover, if these cables are faulty, you'd continue to propagate failures.

Originally Posted by GregGarner

In order to try to stop the problem from happening, we intend to use shorter ribbon cable out of the Talon SRX, and then splice in a more robust wire to go on out to the Encoder. This means we will not be routing out all the unused signals from the Talon SRX to the rest of the robot, and in particular we will not route out the 3.3V line. Also, we will run the encoder cables in a separate wiring ducts, so there is physical separation between the encoder signals and the large current power wires for the motors.

Making a new cable assembly may make the problem go away if there is a short in your existing cable; i.e. the new one does not have the same fault. You can use the same ribbon cable material and the same IDC connector to make a cable with just the wires you need. The "more robust wire" will not be any more resistant to external interference. If the pins in the connector that you need are all adjacent, then rip the ribbon cable to have only the number of wires you need then crimp it into the IDC connector. If the wires you need are not adjacent, then crimp a width of ribbon cable that encompasses the wires you need, use a sharp knife to split out the wires you do need from the ones you don't need and cut off the ones you don't need near the connector.

You may want to twist the new cable assembly (2-4 twists per inch) to make it more resistant to external inductive noise.

Running the cable so that runs that are parallel to power wires are 2-3 inches away from the high power wires can also help reduce noise coupling.

[Michael Hill]

Originally Posted by Mike Copioli

I agree it should fail. But I would not suggest that the product is crap if it did fail because of this.

I'm not suggesting the Talon is crap. I'm suggesting it may not be due to ESD. It sounds like you've taken the proper precautions, but somehow, through the normal course of use, Talons are being blown; Not by one team, but by several. ESD should not kill products that are used in a manner in which they are intended to be used, if not also have tolerance for wiring mistakes (this is a competition full of people who make mistakes). My guess is something thermal is happening in combination with ESD. It looks like The protection diode is only rated up to 125 Celsius, which I think is normally fine. However, when put in an enclosed box like the SRX is and put inside of a robot with potentially not enough air flow to convection cool the fins, perhaps the diodes aren't able to offer the full protection they normally would. I don't have any temperature measurements to back this up, and I'm not exactly sure how you could accurately (and easily) measure the junction temperature while simulating the SRX enclosure.

That's really my best guess at the moment.

[magnets]

When spinning the motors to feed power backward, you may have huge (kV range) voltage spikes from flyback from the motor.

The power distribution board current sensor is susceptible to damage from flyback voltage from motors, as our team has discovered the hard way.

This is speculation, but could a similar problem have affected this team? It seems strange that so many of the controller have failed...

[jimbo493]

I doubt that a Talon would very exceed 100deg C, needless to say 125. I think they tend to stay fairly cool.

[Michael Hill]

Originally Posted by jimbo493

I doubt that a Talon would very exceed 100deg C, needless to say 125. I think they tend to stay fairly cool.

The temperature of the case is different than the junction temperature.

[philso]

Originally Posted by Michael Hill

My guess is something thermal is happening in combination with ESD. It looks like The protection diode is only rated up to 125 Celsius, which I think is normally fine. However, when put in an enclosed box like the SRX is and put inside of a robot with potentially not enough air flow to convection cool the fins, perhaps the diodes aren't able to offer the full protection they normally would. I don't have any temperature measurements to back this up, and I'm not exactly sure how you could accurately (and easily) measure the junction temperature while simulating the SRX enclosure.

That's really my best guess at the moment.

In normal operation, those transorbs have less than 100 nanoAmps flowing through them and are essentially out of circuit. The energy present in a standard ESD test, like the ones referred to in the datasheet linked by Mr. Copioli would not contain enough energy to cause the temperature of the transorb to rise that much. It would not pass those tests if it did.

Transorbs are a pretty rugged class of devices. I have severely abused similar devices by applying surge voltage test with much higher energy than the ESD test (several 100 x), without allowing the required cool down times and the transorbs worked properly up till the heat dissipated in them melted the solder and the part fell off the board. They continued to work after we soldered the parts back in the board.

[Michael Hill]

Originally Posted by philso

In normal operation, those transorbs have less than 100 nanoAmps flowing through them and are essentially out of circuit. The energy present in a standard ESD test, like the ones referred to in the datasheet linked by Mr. Copioli would not contain enough energy to cause the temperature of the transorb to rise that much. It would not pass those tests if it did.

Transorbs are a pretty rugged class of devices. I have severely abused similar devices by applying surge voltage test with much higher energy than the ESD test (several 100 x), without allowing the required cool down times and the transorbs worked properly up till the heat dissipated in them melted the solder and the part fell off the board. They continued to work after we soldered the parts back in the board.

I'm not saying the temperature rose due to ESD. My guess is that the robot had been running prior to the event and heated up.