Fixed my scope - broken ground reference wire.
Looks like the PWM signal is correct on both the Victor (smaller range) and Jaguar (larger range) settings.
The solid green/red light on the Jag does indicate full power is being ordered via PWM.
The Jag just reads 2v under at max while the victor reading equals max voltage.
This is all without any load. I created button controls to order full power, in addition to a throttle test.
I’ll pick up motors tomorrow for further testing.
May I echo back to you what I think I hear you saying, so you can correct any misunderstandings?
you fixed your scope (yea!)
you scoped the Jag’s input PWM and it had a 2.33ms pulse width at full throttle?
you scoped the PWM output from the Jag while commanding full throttle and you saw a 100% duty cycle on the output and yet the voltage measured by your meters was 2V lower than the supply???
**
Mark,
Are you looking at the servo signal into the Jag, the waveform on the motor outputs, or both?
I’m curious as to what the waveform on the motor output looks like. Is the peak voltage out of the Jaguar (on the scope) 2V less than the battery voltage?
Can you post pictures of the scope traces?
-David
I mean it’s weirder than I thought it’d be.
The bottom line seems to be that the Jag at 100% provides 2v less power than the Victor.
It has nothing to do with the multimeters, they work fine. They might under report the power based on the output waveforms, but the results are certainly usable and close enough for the average FRC team.
The PWM signals have the appropriate ranges for each device. They also are fine.
The Victor and Jag (via scope) both have the expected designed output waveforms at anything other than full power.
The Jag’s is a lot more vigorous of course.
The Jag does something funny.
The power output waveform begins at lowest duty cycle by oscillating in the range, say ~2v (rather than 0v as expected) to battery voltage. As power is increased, the lower bound rises towards the upper bound which remains steady at battery voltage.
At ~90% the upper bound of the waveform begins collapsing as the last 10% is transistioned. The upper limit of the waveform begins to drop to meet the still rising lower bound. Although the lower bound stops 2v down and waits for the upper bound to fully collapse to meet it.
Eventually, at 100% duty cycle they meet at a point 2v less than the battery voltage.
So peak voltage on the Jag output is 2v less than the battery voltage.
My scope is an old 40MHz Kenwood that doesn’t get used a lot. The multimeters are Craftsman sale specials.
Certainly I’d like someone else to repeat the results before saying I didn’t make some measurement mistake.
For the record, did you observe this same behavior on more than one Jag? And, are we talking Black or Tan Jags here ?
**
Good point.
The odd behavior was on a tan jag.
I just tested a black jag and it did not have the problem.
The multimeters work fine on the Black Jag too.
Weird indeed.
Did you measure the battery voltage when the Jaguar was at full output, to make sure something wasn’t loading it down?
The rising of the low-level voltage is probably happening because there’s no load, but it’ll be a good idea to check that.
I monitored the battery voltage throughout the tests and it didn’t drop.
I will try again with motor loads tomorrow since I’ve come this far.
I don’t know if the tan jags have different rev levels, so that originals have a problem but later revisions do not.
Mark,
I am guessing you are still measuring open circuit on the controllers. The FAN5109 gate driver used in the grey Jags has “adaptive gate drive” and so maybe responding to the lack of a load. The bootstrap cap might be having some effect as well and there is feedback to the micro from each of the bootstrap circuits. (VbootA and VbootB) Are you hanging the scope across the output of the controller or are you connecting the probe ground to the negative power supply input and then connecting the probe to one of the output terminals?
Yes, these measurements were still taken with no load on the speed controller outputs. I’ll apply loads tomorrow to see what that looks like.
Usually I ran with the probe ground to the negative power supply input because I was hopping around testing both the Victor and Jag speed controllers as well as the PWM inputs to both at various times.
But I also tested with the probe ground to one of the output terminals and saw no difference.
The multimeters were applied just across the speed controller outputs.
P.S.
The other caveat is that I could by chance be working with a damaged tan jag. I’ll test with others to see if it’s a broad issue.
Update: Nope, I tested three tan/gray jaguars and they all exhibited the same odd response. They were never-been-used 2010 KOP Jags.
^Reported
I tested with a globe motor load and the oddity disappears, so the odd collapsing waveform and the 2v drop seems to just be a characteristic of the adaptive gate Al brought up and the lack of a load as Alan suggested. It shouldn’t affect normal operation.
There was a .15v drop through the Jag at 100%, but my battery is down a bit so that’s not a full measurement.
The gray/tan jag gives full power at 100% with a load, and 2v less without a load.
With a load the waveform is the typical duty cycle switching between 0 and battery voltage for increasingly longer times as the throttle changes.
The CIM and the window motors show a cleaner 100% than the globe motor for some reason. Up to 100% everything about the power output waveform is identical. At 100% I see the periodic tiny drop to 0v on the CIM and Window motors, but for the globe the scope displays a messier plot.
The CIM also drew the battery down a bit more than the others of course.
Did you observe this behavior in both the Tan and Black Jags, or the Tan Jag only?
I ask because Gdeaver reported in an earlier post that the Black Jags use locked anti-phase switching.
**
The black Jag behaves straight-forward that way with or without a load.
A long time at 0v/short time at max volts, progressively shorter time spent at 0v, until it’s full time at max volts.
The top end dips only in response to dropping battery voltage as the load draws more power.
So this whole thread just the result of measurement error?
Put simply, it would be physically impossible for a Jaguar (or a Victor) to have an actual 2V drop while driving a motor and not be on fire: It is WAY too much heat to dissipate.
80 watts IS a lot, isn’t it?
I wonder why infineon claims each power transistor can dissipate 100 Watts?
Anyways, I would appreciate getting back to the original topic. I recently obtained a Jaguar that has this problem. (I’m not sure what was done to it.)
The only damage I see is around the traces on ONE of the MOSFETs. However, I get no voltage both in Forward and Reverse.
**
I think the discussion was fruitful, and produced a useful bit of knowledge:
An inexpensive multimeter (affordable by any team), set to read DC volts, should read the same voltage at the power inputs and the motor outputs of a Jaguar (either tan or black, or a Victor) when the Jag (or Vic) is being commanded full throttle and has a motor connected.
This knowledge is useful for helping teams to diagnose/troubleshoot Jag/Vic problems. Thanks Mark for the effort you put into running the tests.
**
That spec only accounts for the junction to case thermal resistance and assumes a perfect heatsink with infinite mass at 25C. Power MOSFET specs are actually pretty funny when you look into the assumptions made.
I’m still confused as to what is going on. How is this different than the ESD issue?
How is the “no power out” different from the ESD issue?
Did I miss something?
In order for ESD to be a problem, wouldn’t the case have to be opened up. You’re saying the case was opened up while the Jaguar was still functional? Why would someone do that?
Or is it something else entirely?
Marshal,
The ESD issue didn’t need the case to be open. During the 2009 game, the regolith especially in ice arenas produced some rather extensive static lightning strikes for robots and field. Some of these events punched through the case or traveled along the wiring and took out some components. The early Jags also had an issue with the FET gate drivers that would self destruct and usually caused a single direction failure.