all we had was a winch connected to a regular andymark gear box.
Gary,
The chokes and caps are for brush noise RFI suppression and should not pose and significant impedance at 15kHz. The core of the chokes would need to be really big to be effective at that low a frequency. Is there any markings on the PTC?
Great post Gary. Thanks for the data. This really helps narrow down the problem… assuming everything works fine when you use the Victors.
Al: I know you have lots of documents. Do you by any chance happen to have a circuit diagram of the chokes and caps in the Denso? I’d like to see how they are arranged.
The first problem you are dealing with is a mechanical one. The window motor interface to the gearbox pinion has an anti-backdrive mechanism that locks the shaft. if there is intermittent back pressure on the gearbox output shaft it locks easier. This was probably to keep someone from prying the window down because the worm gear pitch isn’t steep enough to negate backdrive.
If you are driving a load and the load doesn’t keep up with the motor, it sets the mechanism and locks. if nothing detects the lockup and stops the output within a few seconds the motor protection takes over and you’re done until it cools down. If you remove the output soon enough and then drive in the same direction again, the mechanism will unlock assuming the load will move. Driving in the opposite direction does not unlock the mechanism unless the load relieves on it’s own.
I believe the theory that the motor shaft vibration caused by the Victors at 120hz aids in unlocking the mechanism while the Jaguars smoother output doesn’t.
Just some of the frustrations of mating devices and technologies that were never designed to go together but enabling kids to keep after it until something works will hopefully pay off down the road of developing the next generation of stuff.
To test these theories, you might try driving the system into a load that is controllable (like pushing back and pulling on the steered module) and see what happens.
John,
Is it fairly simple, and non-destructive, to separate the motor and gearbox in order to see this?
If not, did you guys take any pictures you could post ?
Thanks.
~
Just remove the phillips head screws that hold the motor on.
The motor shaft will have a three-lobe coupling on it that interfaces with the gearbox. I believe the gearbox side is metal. There is a brown plastic ring around the metal that contains locking pins. When the brown ring and metal coupling are misaligned they lock. as long as the pieces are rotating together they turn freely. I think the rings are retained by a thin metal plate.
I’m reciting this from memory as I haven’t looked in one for three months so there may be inaccuracies but I think if you play with it you will see how it works. Turning the output shaft locks the coupling. turning the coupling with the motor drives the locking ring with it to keep it from locking.
I have not found any schematics in my files. Typically, chokes are wired in series with each brush and capacitors are wired across each electrical input wire with the other end tied to the motor case. In this configuration (looking from the brushes) the two components form an L/C low pass filter that filters out RFI before it leaves the motor.
Thanks to everyone on this thread who has contributed to the body of knowledge here on CD!
I know there are a number of white papers and PPP available on motor selection, examining power curves, etc.
Has anyone posted a comparative table evaluating the KOP motors in a qualitative sense? In other words, 1-5 stars for use in various applications, such as drive, winch, steering, etc? I know it would change from year to year and need updating, but a general guide would be useful to rookie (and some veteran) teams to start the selection process.
We need to have the students more involved in these decisions, and lifting the veil of mystery would be helpful.
This doesn’t stop them from doing the actual calculations to see if they can lift 3 robots with a window motor!!
Ether - Thanks for compiling all those. It sparked a thought.
Theory:
Some Denso motors brushes are failing in such a way that too many windings can be active at the same time. This may be due to a winding-winding short, a brush deformation, or simple poor design. This state is limited to a limited portion of the rotary travel.
Direct Result:
While in this state, several things occur:
- Current increases: More windings means more mhos.
- Torque constant greatly decreases: Windings fight, reducing effectiveness. Imagine a shaft with a keyway in it: This is the extremely over simplified diagram of the torque constant vs angle that I’m imagining.*
Matching Symptoms:
- The motor will tend to lock into the keyway, given a load that is between the depth of the bore and the outer radius.
- Applying extra torque can pop it out.
- If the bore is deep enough (and it could be very deep), “smooth” current won’t do anything.
- “Rough” current has the advantage of being little bursts of torque. This has the possibility of pushing a little harder momentarily (duty cycle <100). It also has the advantage of giving up, letting it roll the wrong way (out of the bore) and then charging forward over the valley.
Does this match up with other folks thoughts?
- Sorry for the weird bore analogy. Better wording of it would be appreciated.
We removed a window motor in the beginning of the season. I disassembled and inspected the motor. there was no visible wear or signs of a mechanical problem. That was before we realized that all window motors were having very random and intermittent lock ups. We can go for hours of practice and not have a lock up. The randomness makes this hard to track down. When I took the window motor apart on Monday night, I did not see any locking pins. I thought that that was just a flexible shaft coupler. Tonight we will install victors and test a new autonomous mode. We are also going to a competition this weekend and should get more data. One thing I noticed and do not want to state as fact is that when the lock happens the jag stayed solid red even when an opposite direction input was applied. I want to consult others before this is stated as true. If so this is a really weird one. If we can resolve this then we can consider the 1st season 4 wheel independent steering and drive swerve system a success.
JDNovak thank you for your post. When I took them apart last time I didn’t go far enough and remove the black retaining plate. The locking pins and collar are right there. Tonight we disassembled the window motors and removed the pins. We where able to do this without removing the steering motors from the robot. I did one then the students did the other 3. It is not very difficult. Forget the electrical theories. This is a pure mechanical issue. We ran the robot with the modified steering motors tonight with Jags and have not seen any problems. We are going to an off season competition this weekend and will give it a good test. We took pictures and will create a page on our website to view the evil pins. The problem goes beyond latch up. At low speed and under load the pins cause considerable friction. Other teams noted having hot window motors when driving 2 swerve modules. I bet that with the pins removed their problem would not be as critical. Thanks everyone for your input on this problem. Going forward if this solves the problem FIRST needs to be made aware it. Will pin removal be legal if we get the window motors next year? What is the best notification path for FIRST on this matter.
Gdeaver
If it isn’t too hard, can you post a couple pictures of the process for those of us who will inevitably end up following your footsteps?
Thanks!
Since the GDC (Game Design Committee) has authority to interpret and change the rules, I would think that posting to the official FIRST Q&A forum would be a good way.
Access to the Q&A forum is restricted to one designated person per team. This person has a username and password for access. See this link:
You might also consider mentioning this on one or more of the official FRC surveys linked here: http://usfirst.org/roboticsprograms/frc/content.aspx?id=17127
The best thing to start with would be to try to take some pictures, post them here, and explain what’s involved. FIRST has been reluctant to allow teams to detach or modify the gearboxes on the window motors (since forever—some of the older window motors weren’t so convenient to disassemble, as you probably recall), so it would be great if you could demonstrate that it’s no big deal.
Then, we can ask FIRST to have a look (either via the Q&A, [email protected], or personally to the appropriate GDC members).
Gary,
Follow the leads to the Q&A. The GDC has allowed some modifications in the past when it was presented with a benefit to teams to make the modification. Locking pawls in the drill motor transmissions and wires on the Bosch motor are two that jumps to mind.
So did you test with Victors before modifying the motors? That would give us a little more data to actually see if there is some interaction.
The steering motors continued to work fine last night. We did not switch to Victors. We took several pictures and put them on a web page. Remember, the window motors were not removed from the robot so the picture quality is not perfect.
Here is the link.
http://wiki.team1640.com/index.php?title=Nisso-Denko_%28Window%29_Motor_Locking_Pins
This is a simple procedure. The window motors allow FIRST to provide a motor and gear box at a economical cost. I think they should continue to be in the KOP as long as First approves the removal of the evil locking pins.
Quick update for teams planning on removing their locking pins:
1640 competed at Bridgewater Raritan Battle Royale on Saturday with the locking pins out as above (using black Jags), but still suffered from intermittent steering lock-up. I’m not saying removing them is a bad idea (or a good one for that matter, I have no idea as of yet), but unfortunately it hasn’t solved the problem.
Nice write up. A few questions.
Does the lock ring have a maximum torque limit? If one continues to add load force does the ring just wedge tighter or will it eventually give?
When the drive shaft engages the locking pin cage, would the command required to move depend strongly upon the direction? This could be estimated by using a JAG step command with increasing amplitudes and seeing what command will break the lock in both directions.
A comment on this diviney statement:
We can NOT get the motor to fail in this way with a Victor powering it. There was some mild evidence of sticking, but it always freed up with a command authority much less than maximum (maybe 25%) wheras the Jag could not start even with maximum command authority, happily drawing > 10 amps (until the PTC tripped after many seconds).
When the maximum JAG command authority was applied, I assume it was from a Joystick moved by hand. If so, this would lessen the impulse applied by the motor upon the lock. There just doesn’t seem to be any reason that a JAG with a proper waveform input could not duplicate the Victor capabilities unless there is an internal JAG command rate limit that I’m not aware of. With a 100% PWM step input, both the JAG and Victor will reach 10 amps around 2 ms. To reverse and create a matching shutoff like the Victor at .2 duty, the JAG could react within a ms (JAG control iteration rate). It seems plausible that a startup lock breaker pulse could be synthesized for the JAG.
The Victor’s output PWM frequency is 150Hz. The Jag’s output PWM frequency is 15000Hz. The far-lower PWM output frequency of the Victor produces a very “noisy” current waveform in the motor coils compared to the smooth current waveform produced by the Jag’s much higher frequency output PWM.
It is possible that the 150Hz “noise” created by the Victor is responsible for preventing the lockup problem. Perhaps 150Hz is close to the resonant frequency of the rotor.
The only way to simulate the Victor’s 150Hz output PWM frequency with the Jag would be to create a periodic task running at least 300Hz in the cRIO (to simulate 50% duty cycle at 150Hz). I don’t think this is practical.
~