PDA

View Full Version : Problem with 2008 gear tooth sensor


Larry R
01-14-2008, 07:04 PM
My problem is that the "S" output is steady at about 4v DC with the output pulses riding on top of this DC voltage. The pulses are much wider than last year's sensor. The pulses are about .5v p-p and are 10 mS wide. Reversing the gear direction reverses the duty cycle so the pules are about 25 mS.

I was expecting the output to swing from 0 to about +5v with the narrow uS pulses as with last year's sensor.

I checked the +5, Gnd and +12 connections and voltages.

My question: has anyone had problems with this year's gear tooth sensor?

Thanks, Larry

Tim Skloss
01-14-2008, 07:17 PM
We have the same problem! But the voltage is around 3.68 VDC and it varies as you bring metal objects up to the sensor.

Our other sensor worked great with a clean 0 to 5 VDC square waveform. This is our second one and it has not worked properly since wiring it up.

I took a lot of care wiring it up and can guarantee that it was not abused. From the Allegro data sheet I conclude that the sensor is not working properly. The output voltage from the sensor stays around 0.8 VDC above ground, and it should be higher according to the data sheet. IMHO the sensor is bad.

Where can we get replacement sensors BOARDS?!

Larry R
01-14-2008, 07:44 PM
Thanks for the reply. I just measured it and you're right. Mine shows 3.62v. It must be a goner. I'll try the other one. We will need a replacement ASAP.

Thanks, Larry

Tim Skloss
01-15-2008, 08:33 AM
As a backup plan, we will try a Honeywell sensor (http://www.chiefdelphi.com/forums/showthread.php?p=673965#post673965) in the gearbox. This is relatively inexpensive, readily available from DigiKey and reliable. We used them last year.

Larry R
01-15-2008, 02:55 PM
Ok, thanks for the info. I just ordered a pair of the Honeywell sensors.
I checked your links and some others in order to accumulate a knowledge base for the software needed for these sensors. I'm assuming the IFI white paper you recommended is this one: http://www.ifirobotics.com/docs/timers_white_paper_2004-jan-14.pdf.

In that time is short, I would appreciate any pointers or links that would help our team put the software together as soon as possible. I checked a number of postings on this forum already and found lots of tips and tricks for using gear tooth sensors. Our goal is to read the GTS output to assist with dead reckoning while in autonomous (hybrid) mode. I know we can do much more but this is priority one.

Did you mount the Honeywell GTS in the gear boxes or on a sprocket somewhere in the drive train?

Again, thank you for the help.
-Larry

Tim Skloss
01-16-2008, 06:53 AM
We are attempting to mount the Honeywell sensors in the gearboxes today. If it is successful, I will post photos.

As for the the software, we used INT2 and INT3 on digital inputs 1 and 2 to count the GTSs.

I posted examples of this code in the other Honeywell thread (http://www.chiefdelphi.com/forums/showthread.php?t=61242). Be sure you fix the "INT3P" problem in the MPLAB *.h file as described in other Delphi threads or you can't compile.

Good Luck!

Gamer930
01-16-2008, 09:46 PM
Has anyone found replacement order form for the actual KOP Sensor?? We got one mounted and it works great. . . Got the second one mounted and seems fried. . .

Or is there another team that isn't using theirs and would be willing to sell to us??

Larry R
01-18-2008, 12:20 AM
Tim, did you get the Honeywell sensors mounted? Did you post pix somewhere?

Two of the Honeywell sensors, 1GT101DC, arrived today and after setting them up in my test fixture, I don't see any output. I am looking for pulses similar to the KOP sensors for 2008 and 2007.

I am using 12 for power and connect the output to my Tek 465 scope. My fixture uses a servo with a big metal gear rotating about 60 RPM. I can feel the tug of the magnet. I see no pulses.

They seem pretty simple - is there some trick? Are the pulses narrower than the KOP sensors? I tried various horiz sweep rates but nothing.
Thanks, Larry

Tim Skloss
01-18-2008, 08:29 AM
The sensors require a pull-up resistor between the power and signal leads. When we connect the PWM cable to the sensor we add the resistor at the joint and then put a piece of heat-shrink over the whole joint for protection.

I don't remember the exact value, but something around 5K ohm should be sufficient. You can check with the scope.

When assembled, we plug the PWM cable right into the DIO port and use the RC 5VDC to run the sensor. Works great. The output is high when no metal is sensed in front of the sensor, and low when something is there. You can test with a DVM and a screwdriver.

I will post pictures this weekend... forgot to take pictures last night.

Larry R
01-18-2008, 01:21 PM
Thank you for the suggestion! I see in the Honeywell spec sheet that the output is "sink" - no doubt open collector. I sure missed that. Following your suggestion, I used a 4.7K pull-up resistor and also tried using +5v and it works great. I get the low going (from about +5v to gnd) pulses at about 20 mS in width. I like that output! Instead of uS spikes, I see pulses with mS duration. I think this will be easier to work with.

I will look forward to the pictures. After seeing the form factor of these GTS, I know for sure they will not be mounted in the gear boxes.

To be sure I understand your post, you connected the pulled up output to any digital I/O pin on the robot controller and ran the GTS off the +5v supplied by the controller. Right? And the software you are using to count gear teeth has been pretty much posted - am I right about that also? I gathered a lot of programming info following your links on other posts.

This brings up my final question. From looking at the software, it seemed to me that due to the narrow uS spikes that the KOP gear tooth sensor outputs, you employed an interrupt-driven approach: " In the RC we used interrupts 2 and 3 (on RC DIO pins 1 and 2). Every time the gear tooth stimulates a low-to-high transition on the signal pin, an interrupt is thrown. "

I note that you referred to a low to high transition generating an interrupt. However the Honeywell units make a high to low. I am assuming the interrupt is edge sensitive and not level sensitive. Am I right about this?

Thanks a million for all your help!
Cheers, Larry

Tim Skloss
01-18-2008, 09:05 PM
This brings up my final question. From looking at the software, it seemed to me that due to the narrow uS spikes that the KOP gear tooth sensor outputs, you employed an interrupt-driven approach: " In the RC we used interrupts 2 and 3 (on RC DIO pins 1 and 2). Every time the gear tooth stimulates a low-to-high transition on the signal pin, an interrupt is thrown. "

I note that you referred to a low to high transition generating an interrupt. However the Honeywell units make a high to low. I am assuming the interrupt is edge sensitive and not level sensitive. Am I right about this?

Thanks a million for all your help!
Cheers, Larry

The RC triggers the interrupt on the edge type, rising or falling. Since the sensor gives a repeating square pulse as the gear teeth fly by, you get both types and simply have to choose one. Glad you got it working. We calculated a resolution of 0.07 inches of robot travel per tick in our setup.

Tim Skloss
01-19-2008, 08:04 PM
We got the new sensor installed and verified on the Andy Mark gearboxes today!

Yea! The results are posted here (http://www.chiefdelphi.com/forums/showthread.php?t=62083).

Bochek
01-21-2008, 08:46 PM
We are having the same problem, One of our gear tooth sensors works fine but the other sits at around 3.6 V

Wonder if everyone's boards are like this?

- Bochek

Gamer930
01-22-2008, 12:02 AM
We are having the same problem, One of our gear tooth sensors works fine but the other sits at around 3.6 V

Wonder if everyone's boards are like this?

- Bochek

Nearly positive that everyone sensor is like this. . . we got 3.6v on the bad one also

Tim Skloss
01-22-2008, 07:23 AM
Nearly positive that everyone sensor is like this. . . we got 3.6v on the bad one also

FROM THE FIRST Q&A WEBSITE:

[quoted] "The gear tooth sensors provided in the Kit Of Parts are a limited-supply item. Additional KoP gear tooth sensors are not available. However, there are commercial gear tooth sensors that provide similar functionality that are available from a number of electronic supply houses.[end quote]

So there you have it. This year we essentially get only one GTS that is usable.

Our team did this (http://www.chiefdelphi.com/forums/showthread.php?t=62083)to fix it.

roborat
01-22-2008, 08:13 AM
My second sensor also seems to be bad with a reading of 3.69 on the signal wire. Guess that we will order a new honeywell sensor this morning.

SuperBK
01-23-2008, 07:53 PM
One of ours sits at 3.6 volts too! We are hooked up the other one before putting the other gearbox together. It works. So one good, one bad here too.

RyanN
01-23-2008, 08:24 PM
I'm with most of you saying that one of the GTSs is bad. We wired ours up yesterday, one worked, the other did not. We got one of the ones from last year, and it worked, so we're sticking with that. I guess that would be legal since in the KOP they included two, and we are only using two. The one from last year is identical to this years except the fact that the colors changed.

MrForbes
01-23-2008, 11:16 PM
We have one good sensor, one bad one too. The bad one is the one at the outer end of the "full" board...that is, it only has one edge cut. Maybe there's a manufacturing or design problem with the circuit board itself? or a part on wrong? that could be fixed?

Also has anyone asked on Q&A if they'll make an exception and allow using previous year KOP sensors because of the apparent systemic problem with the new boards?

Eugene Fang
01-24-2008, 08:21 PM
Hey everyone! i think our team found the source of the problem.

check out R3 on the two boards. the number on them is different.

Apparently, R3 on the end gear sensor board is marked 4001, R3 on the other gear sensor board is marked 4004. Our mentor assumed that one is 400 and 40k Ohms, but he was not sure how to read the surface mount resistor markings. It is clear they made a mistake in the factory and built all of one of the two boards incorrectly.


Hopefully, more information will be posted later today about this problem.

MrForbes
01-24-2008, 11:07 PM
We found that R4 is different, I did not look at R3 yet.

let's see how this picture came out? oh, I have to split it up....

Eugene Fang
01-24-2008, 11:12 PM
OH! R3 AND R4 ARE SWITCHED!

Tim Skloss
01-25-2008, 01:45 PM
I'm with most of you saying that one of the GTSs is bad. We wired ours up yesterday, one worked, the other did not. We got one of the ones from last year, and it worked, so we're sticking with that. I guess that would be legal since in the KOP they included two, and we are only using two. The one from last year is identical to this years except the fact that the colors changed.

It would be legal only if the sensors from last year are freely available to ALL teams as a commercially available part. So if you can prove that they can be purchased today by any team, then you are OK.

Dad1279
01-25-2008, 03:11 PM
Markings per http://www.vishay.com/docs/20020/smdmark.pdf

4001 - 4k ohm
4004 - 4 meg ohm

Tim Skloss
01-25-2008, 06:20 PM
Hey everyone! i think our team found the source of the problem.

check out R3 on the two boards. the number on them is different.

Apparently, R3 on the end gear sensor board is marked 4001, R3 on the other gear sensor board is marked 4004.

It would be great for someone to swap the resistors so they match the working GTS board and verify that it fixes the problem.

All of us want to know if there is an easy way to fix the bad board!!!

usbcd36
01-25-2008, 08:47 PM
I can verify that swapping the resistors fixes the board. Don't attempt it without the right equipment, though.

baop858
02-16-2008, 08:40 PM
Which one is the working one?

Brian C
02-18-2008, 11:39 AM
The bad one is the end board on the sensor strip. The good one is the next one in. If you have already separated your boards (and I'm thinking that by now you should have) the board with one smooth edge is the bad one and the one with 2 small nibs on opposite edges is the good one.