Problems with SPI and AXDL345

LeafLabs Garden » Support » Maple Hardware Support

Problems with SPI and AXDL345

(20 posts) (5 voices)

Started 5 years ago by crenn
Latest reply from x893

12 Next »

crenn
Moderator

I'm having slight problems communicating with my AXDL345. I know the problem isn't with the sensor, it works reliably with my Arduino. And last attempted 2 times I communicated with it (using SPI port 2), it worked fine.

Here is the code I'm using and associated example outputs:
Using SPI Port 1-
http://www.crennsmind.com/Code/Maple/SPI/ADXL345-Code_Current_SPI1.txt
http://www.crennsmind.com/Code/Maple/SPI/ADXL345-Example_Output_SPI1.txt

Using SPI Port 2-
http://www.crennsmind.com/Code/Maple/SPI/ADXL345-Code_Current_SPI2.txt
http://www.crennsmind.com/Code/Maple/SPI/ADXL345-Example_Output_SPI2.txt

Any ideas why that's happening?

Posted 5 years ago #
leaflabsandy
Member

Could you do the following when all else fails:
Take the AXDL345 board and cool it. Test it. (I put it the freezer with precautions<---)
Then heat the AXDL345 board to see if it works. Test it. (toaster oven with precautions <---)
Then do the Maple separately. Test the AD345 at room temp.
Then do both the Maple and AXDL345 together.
Note: Do not exceed the max temps of the boards and do not chill them to cause condensation on the boards.
These tests will ferret out any timing/temp. problems.

Posted 5 years ago #
crenn
Moderator

Little extreme, since after playing with the AXDL345 on my arduino, it worked fine when I used the Maple SPI Port 2.

Posted 5 years ago #
leaflabsandy
Member

keywords "when all else fails"

Posted 5 years ago #
crenn
Moderator

Well, it seems to be working well with SPI Port 2, so I'll leave it like that for now.

Posted 5 years ago #
perry
Administrator

hi crenn

are you using pwm on any of the timer 3 channels by any chance? there's a bug in the timer code that enables all the timers by default and two of the timer pins conflict with spi1. i fixed it on a branch but haven't pushed the change yet to master.

if this is the case i'll send you a patch and maybe you can test it?

Posted 5 years ago #
crenn
Moderator

When I was developing the functions to access the ADXL, I had nothing else connected. When I was using a Servo and the sensor today, I was only using Timer4. I can test the patch, but I don't believe it would solve my problems.

Posted 5 years ago #
poslathian
LeafLabs

If perry's notion turns out to be incorrect, this is not the first spooky interaction issue between various peripherals. For instance, I have seen Timer3 destroy SerialUSB...not sure why. I'll mark this down.

Posted 5 years ago #
perry
Administrator

Hey crenn,

I don't know enough about the xl345 to be able to tell what those readouts are supposed to be saying, but it looks like from your log that the XL345 is responding and giving you data. Is it not the right data? Your Z-axis data looks to be ~200 in both sets.

If it's in fact not correct on SPI1, can you scope the outputs to see if anything is obviously wrong? I've been using SPI1 quite reliably at 18MHz on a wishield, so I'm sure we can make it work.

Posted 5 years ago #
poslathian
LeafLabs

~200 is a reasonable output for z axis if its not at a perfect perpendicular angle to gravity. I usually read 180~ with right justified output and fullscale resolution.

Posted 5 years ago #
crenn
Moderator

perry, the readout for the Z-axis data is correct (it needs to be biased in software or write to the sensor's offset register), however the initialisation of the sensor is strange which is what concerned me. I wrote 28 to Register 2D (Power Control) and when I read it back, it came back as 10. And then when it read the Device ID, it didn't return E5, it returned CB.

EDIT: I don't own a scope, but can probably get access to one in my lab tomorrow.

Posted 5 years ago #
poslathian
LeafLabs

ahh, this indicates a clear problem. devID should be E5, a standard sanity check (I hate devices without read ports for just this reason, common with CODEC chips).

fyi: 0xE5 = 0b11100101
0xCB = 0b11001011

looks like the frame is off by 1 (shift left 1). This is a good sign - indicating some minor timing shift.

Posted 5 years ago #
crenn
Moderator

So it's going to be a fairly easy fix?

Posted 5 years ago #
poslathian
LeafLabs

Ill have to break out my xl345 to say one way or the other. Will post more on this later. But a frameshift is better than say, garbage.

Posted 5 years ago #
perry
Administrator

curious. which version of libmaple are you using? perhaps the mode bits are being set incorrectly for SPI mode 3? that would probably cause exactly what you're looking at, a 1-bit offset.

i can't look at this until I get home, but I think we might have an ADXL345 somewhere around that we can hook up... poslathian?

Posted 5 years ago #

RSS feed for this topic

12 Next »

Reply »

You must log in to post.