CMUCam and low voltages
 

We were playing around with the CMUCam yesterday and noticed that as the backup battery gets drained, the values (for the target) we're receiving from the camera change, sometimes causing the tracking to start losing the target.

Is anyone else having this problem and is there a way to avoid dropping too far below the ideal voltage the CMUCam is looking for?

Re: CMUCam and low voltages
 

Ryan,
You are spot on. We started having really wacky behavior from smooth working camera system. All was well, then it all fell. After a little trouble shooting, we found that the battery was dieing. We had been using the same battery to calibrate and test for 4+ hours. No surprise that we drained it. So, it appears that we now know the symptoms of a dieing backup battery.
How do you prevent this from happening during competition? Easy, change the battery every round.
Remember, that battery not only provides protection to the RC during dips in the main battery voltage, it also powers the camera and all servos.

Re: CMUCam and low voltages
 

That brings up a very interesting point. Why aren't the camera and servos just powered by the main battery? It wouldn't make sense to call something a "back-up" battery if it provides essential power systems when the "main" battery isn't dying. I hate to be cynical, but I think someone was cutting corners by doing it this way, but I hope I'm wrong.

Re: CMUCam and low voltages
 

Last year, 237 changed their backup battery EVERY SINGLE MATCH, and I suggest you all do the same. It is the only way to keep the camera running well enough to track and complete our task well last year (and yes we did do it at competition).

Now, if you read the manual (5.3.5 Electrical System Rules <R53>) ... you will see that it says that you can build a circuit to charge the backup battery off the main battery during the match. We are probably going to try to do this and see how often we have to change our battery.

Backup Battery Charger Schematic (pdf) 1-12-06 (not supported, use at your own risk) < right from the IFI homepage

Re: CMUCam and low voltages
 

Hey... that's pretty sweet. Thanks!

And thanks to others as well. Glad it's not just us. ;)

Re: CMUCam and low voltages
 

Shu,

A number of years ago, we did not have a backup battery. As the number of motors and their power increased (the small CIM can draw 133A each at stall), we started to have problems with the main battery power.

Every battery has an internal resistance. As you draw more and more current, the battery voltage drops (Ohm's Law). Also, a rapid discharge can temporarily deplete the reactant chemicals at the battery plates.

The end result was that, when robots got into shoving matches, the main battery voltage would collapse to the point that the RC controller would reset. This was not viewed as a feature by those of us using the controls...

Thus, the backup battery was added 3 (or was it 4) years ago.

If you were to power the camera off the 12V bus, the CPU on the camera would reset in the middle of a match. You would loose all calibration information and the camera would be useless for the remainder of the match... Most of us would not enjoy this either...

Therefore, we power the camera off of the backup battery...

I hope this explains why things are done this way.

Regards,

Mike

Re: CMUCam and low voltages
 

Mike,
If I'm not mistaken, the Radio Modems are powered off the backup battery as well. If you lost them during a match, it would not be a good thing either. :( Please correct me if I'm wrong.

Re: CMUCam and low voltages
 

Mr. Bill,

IFI keeps the internals of the RC proprietary. However (based on watching the modem LEDs after the 12V is removed), I would hazard a guess that you are correct.

Mike

Re: CMUCam and low voltages
 

The camera runs off of 7.2 volts. We built a reguator circuit to power the camera at exactly 7.2 volts off of the main battery, and didn't notice any problems. If anything, it was more stable.

I think if your main battery drops below 7.2 volts at any time, it is probably not fully charged. However, I could be wrong... :D

Re: CMUCam and low voltages
 

They are powered off of both the main and backup batteries; if the main is lost, they automatically switch to backup. However, the servos and camera always tap their power from the backup battery.

Re: CMUCam and low voltages
 

Eldarion,

Try sourcing 533 amps from the battery at the same time you are using it to power the camera and tell me that you have no problems...

Using the 4 "small" CIM motors for a drive is probably the most popular drive configuration this year...

533 amps is 4 small CIM motors in a stall condition (133A each). Not to mention 2 bike motors, 2 FP motors, et cetera...

I predict that many robots will peak at over 800A during the competition. As Dave Lavery said at the kickoff, there will be "meltdowns"...

JMHO,

Mike

Edit: Look at this data.

Re: CMUCam and low voltages
 

You're probably right; we only had 2-wheel drive last year, so the wheels would spin far before stall condition. I still wonder, though, about the 40 amp breakers and the main 120 amp breaker? Wouldn't they prevent the 800-amp meltdown? :)

Probably the best configuration would be a diode type splitter such that power would always be drawn from the stronger source.

The best part is, since the camera is "custom circuitry", the above would be legal under the rules! :D

Re: CMUCam and low voltages
 

The camera is not custom circuitry. It is an assembly provided in the Kit of Parts.

Re: CMUCam and low voltages
 

As soon as you modify it, e.g. by adding a diode splitter in the power line, wouldn't it become a modified COTS electronics item and therefore become custom circuitry? :confused:

Re: CMUCam and low voltages
 

Check the data sheets that come as an appendix to the robot rules.

All circuit breakers (and fuses for that matter) have an "inverse time" curve. They have an extremely small but accurate internal resistance which heats up via the power dissipated in this small resistance (P=I*I*R). The actual trip mechanism is thermal.

In the data sheet for the 120A CB, find the plot labeled "TIME VS PERCENT OF RATED CURRENT". Note that the CB will allow rated current (120A) to flow nearly forever. At 200% (240A), it will trip in about 20 seconds. At 400% (480A), it will trip in about 4 seconds.

While I'm on the subject...

Be warned that the internal heat does not dissipate quickly. If you run at 480A for 2 seconds and then try running at 240A, you will trip in a lot less time than the quoted 20 seconds (more like 10).

This warning also applies to self resetting 20, 30 and 40A circuit breakers. Once they start tripping, they will re-trip in a very short time at the same amperage.

Lastly, these are not "precision" devices. Variations of +/- 30% are not uncommon. Plan accordingly.

Mike

Re: CMUCam and low voltages
 

Well, not that it matters on the main breaker; once it goes, it goes! But you are right. If your voltage drops for just a second, your camera's gone. However, I wonder, wouldn't the camera simply reinitialize as soon as the RC came back? It would be the equivalent of a power off reset.

Re: CMUCam and low voltages
 

Your code would have to detect such a reset and reinitialize the camera - it would lose all of its registers. That's why it's a big deal.

Re: CMUCam and low voltages
 

Our camera works fine on the 5 volt supply of a free analog input port in our experience.

EDIT: Looking at http://www.cs.cmu.edu/~cmucam2/CMUcam2_manual.pdf, it seems that the internal circuitry regulates the voltage to 5V anyhow.

Re: CMUCam and low voltages
 

Doesn't the CMUCam have a nice big capacitor to filter the input power? And if it doesn't, wouldn't any good regulator design put one in? That would be able to prop up the camera power during momentary 12v sags.

And yes, I've cut power to the camera and it does link back up within a half a second. But in the 10-second autonomous, if that keeps happening, it could be bad... :)

Re: CMUCam and low voltages
 

The original post was that the camera started acting weird and calibration values changed with battery voltage. What you were experiencing was the power regulator on the camera was providing a varying voltage to the camera linear circuitry and therefore changing the RGB values provided to the rest of the camera.
As Mike and others pointed out, prior to the backup battery being introduced many teams were having problems with the RC and modems dropping out during competition. Frequently, the main battery is pulled down to as low as four volts on multi motor or inefficient designs. The RC requires at least 7 volts to operate and with a lot happening, IFI had designed in a dropout that would reset the RC when the input fell below 8 volts. The RC would take a few seconds to boot and then it had to reestablish modem link before the systems on the robot would be enabled. With the introduction of the backup battery, the main processor and radio modems stayed powered and linked when the main battery falls below the 8 volt cutout. A by product of the backup was the ability to use it for servo power and eventually the camera. Since RC output is controlled through the arena controller, the ability to power the camera when the robot is powered on and have it initialize before match start is a great benefit.
Please remember that the spare 5 volt output is after the main regulator in the RC and that will respond to main battery voltage less the 2 volts dropped across the regulator. Capacitors wouldn't be able to keep the power up in the camera long enough. The camera draws some serious current.