two large ACC spikes

[13brv3]

Greetings,

I'm trying to work my way through the cals, and I've run into a puzzle. I made 3 static files in different orientations as suggested, and I made a small "crop circle" in the back yard between rain showers this morning. When I sat down to try to do the cals, I found that the ACCX, Y, and Z plots always have two huge spikes around the same place.

First thought of course is that there was just a glitch, but when you consider that it happened on 3 different logs, on different days, in different locations, it's hard to imagine it being a glitch.

I also thought maybe the memory card had an issue, but since all the other logged parameters are fine, I can't see it being a memory card issue.

Can I cut these out of the logs, or will that screw up the cal? Cutting the beginning and ending seems common, but I haven't seen anyone mention cutting a section from the middle.

Thanks,
Rusty

[Kisssys]

Hi Rusty,

I didn't have any luck with the 16 gb card you had, it gave me too many CRC errors. Is that what you are using?

[13brv3]

Hi Steve,

No, I replaced the cards with 8GB Transcend class 10.

Rusty

[13brv3]

I think I answered my own question about cutting out the spikes. By the time you cut out enough to get rid of them, there's a big jump in the graph, so I'm sure that's not an option.

Rusty

[Kisssys]

Look at the LOG_VOLTAGE 8 , 9 AND 10. Those are the raw values for acc and see if they seem to coincide with what your ACC X Y Z plots look like.

I believe I sent you some static and dynamic logs on a CD, see if you can read them OK without any spikes.

Have you changed the firmware from what I loaded?

I have not seen spikes like that before, I don't think they are real output from the sensor but instead a file logging or reading problem.

[13brv3]

[Kisssys]

I went back and looked at board 1 logs and I do see a 1 bit shift in voltage readings which is not uncommon. If I trimmed the file it would only be at the beginning or end of the file but I don't remember what I did.

If the raw values look good you can use the file. The calculations are made based only on raw values.

Try running the files and see how they do.

[13brv3]

OK, I can give it a try. The instruction in the wiki tell you to look at the ACC files, so I stopped when I saw the spikes. Hopefully it works, since I'm dying to be convinced that it's worth all this effort.

Thanks,
Rusty

[Kisssys]

When you have a freshly flashed board, their only a couple of starting parameters that get flashed to the board. Those won't cause large changes when you view the IMU plots. After parameters are loaded, those plots will be effected by the calibration parameters that you already have loaded. I never look at the Rate or Acc plots to verify a calibration log file, the Mag's are an exception because of they constantly switching polarity. I still look at the displayed band of mag data for anomalies but it may be confusing at first.

It may be worth your time to flash R76 which seems to be pretty stable and has all the lastest features.

[LPR]