Jaguars failing

[jhersh]

Originally Posted by jgannon

Considering that the vast majority of teams don't post on CD, I'm certain that far more than 88 Jaguars have failed, and consequently that the number easily exceeds 1%. Definitions of "extremely high" may vary, but I don't recall a single report of a Victor failing for any reason other than negligence, so in comparison this seems very troubling.

The 88 figure came directly from TI, I believe. Meaning that's how many they had returned by teams. That has nothing to do with CD.

And speaking of negligence, in the top two causes of failure, one was negligence... and that was removing the screws from the jaguar power terminals which leaves metal shavings inside the case. The new black jag does not use retaining screws since people are going to remove them even if it will cause a failure.

[jgannon]

Originally Posted by jhersh

The 88 figure came directly from TI, I believe. Meaning that's how many they had returned by teams. That has nothing to do with CD.

And speaking of negligence, in the top two causes of failure, one was negligence... and that was removing the screws from the jaguar power terminals which leaves metal shavings inside the case. The new black jag does not use retaining screws since people are going to remove them even if it will cause a failure.

Mea culpa. Thanks for the info.

[Racer26]

That figure of 88, EVEN if its accurate as to the number of units that LM/TI received back still doesnt really tell us anything.

There are LOTS of teams, i'm sure, who just had one blow up and never thought to contact LM/TI.

I know 1075 used 6 jags on our 2009 bot, and we had no fewer than 2 jags fail on us. Both in the same way, a plume of smoke and upon closer inspection, toasted FETs. We're very careful about filings getting into speed controllers, over the years we toasted a few Victors... including an entire robot's worth of electronics when the ground clip fell off the robot during welding. Ever since then welding is done without electronics on the bot.

My count says 2/8, thats a 25% failure rate. There seems to be no shortage of the well known names around here with similar numbers. That sort of leads me to believe that the numbers are bigger than it would seem from what LM/TI tells us.

[EricVanWyk]

FACT: The number of failed Jaguars is higher than what TI has reported. This is not attributable to malice; they simply are not psychic. I am at a loss as to why any team would choose to not tell LM/TI of a failure; why not get a free replacement?

FACT: The grey Jaguar has/had(?) a failure mode that silently disables one direction of motor control.

Jaguar failures that include emitting smoke are user induced.

Please note that clustered failures are much more likely to be indicative of the user than of the design. I did a lot of tech support last season, and almost all of the cases of multiple failures were user error. Most of these cases also featured the fabled line of "I've been doing FIRST for X years, of course I didn't read the documentation!" Actual Example:

My analog breakout and all my Jaguars blew at the same time. These things are crap.

Edit : The quote came from someone who plugged the battery in backwards.

[Racer26]

oh. I forgot to mention that when I opened the case to look at the damage to the jags we toasted, there was NO noticable chunks, flecks, or specks, of metal shavings inside.

I'm very reluctant to just accept that all magic-smoke-emitting failure modes are user-error-induced.

I think its rather likely that either (and this does sort of qualify as user error i suppose) a loose or otherwise high resistance connection is causing the jags to enter some weird failure mode. Alternately, it has been suggested that the loads induced by running a motor full forward and jamming it to full reverse very quickly can reach huge numbers. I expect that some of the toasted FET failure modes could be caused by this happening in a fashion that the fault-detect circuit couldn't detect fast enough.

I can't say carte-blanche that it wasn't our fault they blew, but IMO they blew for other reasons. I know the precautions we take.

[Dave Flowerday]

Originally Posted by EricVanWyk

I am at a loss as to why any team would choose to not tell LM/TI of a failure; why not get a free replacement?

Perhaps the team had no intention of using Jags after the incident? In which case requesting a replacement would be extra work for no benefit. Perhaps they already had a spare and didn't bother or forgot to send in the broken one? Or maybe they came to CD and saw some of the many posts here that suggest that a failed Jag is the teams' fault and figured they must have done something wrong and didn't bother to request a replacement because they didn't want to be embarrassed. There's a lot of posts here, including yours that I just quoted, that like to point out dumb things that people have done to destroy electronics. Maybe people are becoming afraid of being cited as an example of "user error" and made to look foolish and figure it's easier to just keep their mouth shut and pony up the extra $70 on their own.

Jaguar failures that include emitting smoke are user induced.

Joe's post above indicated that at least some failures were because of shavings when someone fully backed out the terminal screws. Ultimately this may be considered user-induced, but given that the black Jag was redesigned to eliminate this problem I'd say that it is at least partially due to a misunderstood requirement or design defect. If enough people are doing something that they thought was logical that ends up ruining the product, I usually consider that a design shortcoming.


Ultimately, there's at least some teams who had issues with the Jags and are not comfortable with them. Citing the data (which may or may not be an accurate reflection of reality) over and over is not doing much to help the perception problem. I keep seeing threads like this where teams say they had issues (sometimes multiple), and the response is always the same: someone cites the data that claims to show a "low" failure rate, then suggests that the team must have screwed up. I understand that that is what most people at NI and TI must be thinking, but I don't think it's helping the perception problem.

[jhersh]

Originally Posted by Dave Flowerday

Joe's post above indicated that at least some failures were because of shavings when someone fully backed out the terminal screws. Ultimately this may be considered user-induced, but given that the black Jag was redesigned to eliminate this problem I'd say that it is at least partially due to a misunderstood requirement or design defect. If enough people are doing something that they thought was logical that ends up ruining the product, I usually consider that a design shortcoming.

This was actually considered a design enhancement originally. The idea was if the screws are painted a color code, then it will be LESS likely that teams will accidentally wire them up backward and set fire to them. But what if a team fully removes the screws and replaces them in the wrong place... then the color code is misleading. Ok... make the screws retained so that they won't be easily removed all the way, thus helping to ensure the color code stays correct.

After a year of teams overpowering the retention, TI must have decided that if teams are going to take them out anyway, the may as well not have metal shavings inside on top of the possible bad color coding.

Essentially, what was intended to help teams, only hurt them more since they didn't follow instructions.

[Joe Ross]

Originally Posted by jhersh

After a year of teams overpowering the retention, TI must have decided that if teams are going to take them out anyway, the may as well not have metal shavings inside on top of the possible bad color coding.

Essentially, what was intended to help teams, only hurt them more since they didn't follow instructions.

Perhaps the problem is that it isn't in the instructions (even now).

[jhersh]

Originally Posted by Joe Ross

Perhaps the problem is that it isn't in the instructions (even now).

Oops. Perhaps someone should let TI know about this.

[Al Skierkiewicz]

Joe,
If a Jag was opened and metallic dust found inside, I think a vacuum is more appropriate. Compressed air will force some of the dust between or under the pins of surface mount components and under the shroud that surrounds the FETs. Many teams do not realize how much metallic dust is actually generated right on their robot. Open gears and sprocket/chain interfaces for instance throw a lot of fine metal dust while they wear. The big offenders are rotating or moving parts that rub against robot frames. i.e. arm attachments, shafts with no bearings, etc.

[skimoose]

Originally Posted by jhersh

This was actually considered a design enhancement originally. The idea was if the screws are painted a color code, then it will be LESS likely that teams will accidentally wire them up backward and set fire to them. But what if a team fully removes the screws and replaces them in the wrong place... then the color code is misleading. Ok... make the screws retained so that they won't be easily removed all the way, thus helping to ensure the color code stays correct.

After a year of teams overpowering the retention, TI must have decided that if teams are going to take them out anyway, the may as well not have metal shavings inside on top of the possible bad color coding.

Essentially, what was intended to help teams, only hurt them more since they didn't follow instructions.

I wouldn't have considered this a design enhancement. I've always considered ring connectors to be much safer than fork connectors. A loose terminal screw will not result in a stray wire loose in a robot frame.

Also, weren't ring terminals given in the KOP? How do you get a ring terminal onto a screw without backing the screw completely out? We had zero burned out Jaguars last year and we backed out all of our screws to use ring terminals. I suspect the issue is something else.

Reverse polarity on the line side is the quickest way to smoke a speed controller, so why not place color coding on the case so that when careless people remove the screws they get back in the correct positions? Also, while the case on the Jaguars is much more user friendly at preventing metal filings getting where they shouldn't, it's not fool-proof. Many teams claim they're careful about metal filings, but I can't count the number of times I've witness teams drilling, filing, grinding, or cutting (Dremel cutoff wheels are a big one) over vital electronics with no cover to catch flying chips and dust. They don't see any metal chips or filings, but there is conductive dust from the cutoff wheels everywhere.

[jhersh]

Originally Posted by skimoose

Also, weren't ring terminals given in the KOP? How do you get a ring terminal onto a screw without backing the screw completely out?

We always use wire cutters to snip a little slot out of the ring right by the wire connection resulting is a hook. Much easier to wire and similar protection against loose connections.

Originally Posted by skimoose

We had zero burned out Jaguars last year and we backed out all of our screws to use ring terminals. I suspect the issue is something else.

Shavings inside aren't a guarantee of failure, it just makes it likely. If I were you, I would open up all of them and clean them out thoroughly. I'm not sure what the most appropriate way is... perhaps an air duster. You can probably get advice from TI. I wouldn't use them until they were cleaned though.

[Al Skierkiewicz]

Originally Posted by EricVanWyk

Jaguar failures that include emitting smoke are user induced.

Not exactly. High continuous current would cause the current sense resistor to unsolder itself and/or self destruct. Sustained near stall current on a CIM is 16 times the power dissipation of the resistor.

[Trying to Help]

Originally Posted by EricVanWyk

FACT: The number of failed Jaguars is higher than what TI has reported. This is not attributable to malice; they simply are not psychic. I am at a loss as to why any team would choose to not tell LM/TI of a failure; why not get a free replacement?

FACT: The grey Jaguar has/had(?) a failure mode that silently disables one direction of motor control.

Jaguar failures that include emitting smoke are user induced.

Please note that clustered failures are much more likely to be indicative of the user than of the design. I did a lot of tech support last season, and almost all of the cases of multiple failures were user error. Most of these cases also featured the fabled line of "I've been doing FIRST for X years, of course I didn't read the documentation!" Actual Example:


Edit : The quote came from someone who plugged the battery in backwards.

We didn't send our failed and now failing Jags back to TI because we couldn't determine if they were user error or production error. For instance, we have one which has been slowly failing for about 6 months. Was it from being outside at outreach events? Humidity? I don't know. I think our original failed Jag was probably due to user error and I did talk with the TI rep at GSR. So there was some feedback.

Trying to Help