Analog break out problem.

[Gdeaver]

I want to bring up a analog break out problem that we had this weekend. We used many sensors this year. We used 4 mid range sharp IR sensors, 2 long range sharp Ir sensors, a ultra sonic sensor and 5 Vishay absolute angle sensors. We have had noise problems all year and they were handled by software filtering in labview. We thought everything was under control until Philly on Thursday. We could not control our arm and the drivers were complaining about our 4 wheel independent steering. It drove us insane all day and really were not prepared to tackle this problem with all the other issues we had. Thursday night we were able to track it down to power line noise. No we didn't put a scope on it. We don't have one. On Friday morning we added a 10 uf cap to a spare port on each analog and bang our problem went away. Instant improvement on everything. Rock solid analogs. The robot ran the way we designed it. Some of the filtering was removed or modified and wow what a difference. So after this experience I would recommend adding some caps to the analog break out power even if your not having problems just to be safe. The problem may be there and you haven't noticed it. I would be interested with what the designer has to say. I think his name is Eric.

[Greg McKaskle]

I suspect a question Eric would have is -- what year/revision of breakout was it.

I believe year one had a noisy supply and year two breakouts replaced the supply. The first year could source more current than year two, but the noise was very annoying.

Greg McKaskle

[EricVanWyk]

Greg already asked the most important question - which model breakout?

Did you put the 10uFs on the signal or on the power?

EDIT: You totally buried the lead! Ground loops will are a fantastic way to inject overwhelming noise into the breakout.

[Gdeaver]

We built 2 robots identical. The competition robot used all this year parts. The practice bot used all last year parts. The Sharp IR conductive plastic problem had been resolved. It had no effect on what we saw this weekend. Both robots performed very well up to this weekend. Early on pre-ship practice the noise at most was not more than 3 or 4 buckets. It was not noticed at our first competition. At Philly, our robot was run under very, very high stress. Much greater than in practice at home or our first comp. This is when the issue really became apparent. The solution was to add the cap right at the analog break out port across V+ and GND. After that there was absolutely no bounce. We used 2 analog break outs. All the Sharp Ir's and ultra-sonic were on 1 break out. Noise was not an issue on this break out. All the Vishay sensors (981HE) were on the other break out and this is were the noise caused the problem. The Vishays are listed at 16ma max.

[EricVanWyk]

I'm having a hard time piecing together exactly which situations had issues and which situations didn't.

Were you using the old or new analog breakout? What is the revision number on the silk screen?


As I understand it, the problem only occurred using a rev 6 analog breakout with the vishay 981HE sensors. Did you have load resistors for the 981HEs?

[Gdeaver]

I would have to confirm the revision of the board but it is this years break out. The big problem with noise was the break out with the vishay sensors but some with IR's. The 1k pull ups were added. no effect on the noise. I believe their main purpose is to give a clean high signal for diagnostics. So this years break out 5 vishays and noise. One 10 uf electrolytic cap on the + and - and it's clean. Is this just our board or is it a deeper problem? Any other reports. The fix is simple but frustrating till it was found.

[EricVanWyk]

Does it happen when the robot isn't moving?

This is the first I've heard of this problem. Hopefully, it is that your output cap is damaged. If not, we might have rev 7 next year Do you mind testing on another analog breakout?

I'm wondering what the current draw profile on these sensors looks like. I might have to buy a few to do some testing.

[Al Skierkiewicz]

Gary and Eric,
This sounds like C10 is missing doesn't it? May be broken off or lifted a pad from the circuit board. Gary, if you are going to Champs let me know.

[Gdeaver]

Yes, We are Going to the championship so we do not have the robot. We have a practice robot. We have it stripped down for spares and have to put it back together. I will have the programmers check the bounce on both break outs. We really had no way to monitor this problem when the robot was loaded to the max in battle other than the drivers complaints. What gets me is the relatively low value of 10 uf turns a big problem into no problem. At this point I can't give any more info till we look at the practice bot and the prime is not in our possession.

[Alan Anderson]

Originally Posted by Gdeaver

What gets me is the relatively low value of 10 uf turns a big problem into no problem.

The regulator on the Analog Breakout needs a capacitor on its output in order to prevent oscillation. In that context, 10 µF is actually a relatively large value. The MIC5209 datasheet suggests that 1 µF is enough in some applications, but higher values are required with higher output current.

[Al Skierkiewicz]

Gary,
If as I suspect the 22 ufd cap, C10 is missing, then the regulator can and likely will oscillate. The addition of your 10 ufd then is merely returning some of the needed capacitance to the output. You will need to keep the noise filtering in Labview. There are two noise problems you are dealing with, the noise normally found in the sensors you are using, and that generated by a defect in the power supply. The reason I am concerned is that adding caps to anything except motor leads is not allowed in the rules. Repairs to electrical components are allowed so I would want you to get this right so you have no problems in inspection in a few weeks.
This is assuming that your load is less than the 500 ma the power supply is capable of providing.

[EricVanWyk]

Originally Posted by Al Skierkiewicz

Gary,
If as I suspect the 22 ufd cap, C10 is missing, then the regulator can and likely will oscillate.

I agree.

Originally Posted by Al Skierkiewicz

The reason I am concerned is that adding caps to anything except motor leads is not allowed in the rules. Repairs to electrical components are allowed so I would want you to get this right so you have no problems in inspection in a few weeks.

My understanding of the rules differs from yours on this point. I believe the added capacitor would fall under the "custom electronics" category and therefore be legal (R62). R63 explicitly allows the motor filter caps because custom electronics are explicitly illegal on the power-path.


Gdeaver -

This is going to sound weird, but be sure to use a crappy capacitor for this. Using a high quality, low-ESR cap can make this specific power supply either squeal or throb.

[Alan Anderson]

Originally Posted by EricVanWyk

My understanding of the rules differs from yours on this point. I believe the added capacitor would fall under the "custom electronics" category and therefore be legal (R62). R63 explicitly allows the motor filter caps because custom electronics are explicitly illegal on the power-path.

Right. The capacitor in question is not on the 12 volt power input, but on the "input pins" of the Analog Breakout. That's a legal place for a custom circuit to connect.

[Al Skierkiewicz]

Guys,
You can call it a custom circuit but my experience tells me that the GDC doesn't see it that way. I can make an excellent case for it being a field repair on a damaged electrical device, however. (which is now legal) I should have mentioned that although it is rare, these types of caps do occasionally show up open in the field.