[Bug fix] Potentially solved the problem of not getting 3 ranges when running 3 anchors and 1 tag

[ybc82]

I have been running 3 anchors and 1 tag simultaneously so that I should be able to get three ranges and therefore the position of the tag. But I always had problems that some anchors 'died' from time to time (I can't get range reading of them from the tag; the serial port of the anchors output lots of 'Not found'; the serial port of the tag output lots of 'delete inactive device: ', '1 device added'), which other people have suffered from too.

I recently found that the solution should be that for the variable '_networkDevicesNumber' declaration in DW1000Ranging.h and initialization in DW1000Ranging.cpp, add a 'volatile' to it. My explanation for this is that otherwise this variable is not read correctly in the function 'addNetworkDevices' and will cause errors in the device list.

[xforus]

Hello ybc82. What example code do you have problems?
Using DW1000Ranging_ANCHOR/TAG, with last version and with defaults, I never had problems receiving 3 ranges.
But I experienced problemes when I decrease cycles of POLLS between BLINKS. For example in: counterForBlink > 20, when I change to "1" . So 50% POLLS and BLINKS is needed for me.

I found the cause of problem but not solution.
Problem is because last ANCHOR to receive POLL don't receive it. Anchor transmit RANGE INIT but don't receive POLL. So TAG is waiting for last ANCHOR but never is received, so no range is transmited to other Anchors, because Tag needs all POLL ACK's to send RANGE.

I don't know if this is the same problem to you.

[ybc82]

@xforus , I'm using the same example code. I'm using Arduino Pro mini and Arduino 1.6.10. People have reported that the code works well with Arduino 1.6.9 and I don't know if that's the case.

I think the cause you described is similar to mine. And I figured out that the cause of 'this cause' was that the non-volatile variable caused problems in new device registration or deletion of inactive devices.

Thank you!

[Rotzbua]

@ybc82 If it fixes the problem, please make a pull request instead of an issue. So the code can be improved.

In general the code except DW1000.h/cpp are a proof of concept code. I already refactored and changed a lot but there is still a lot to do if you search TODO. I think the code do not consider about race conditions and volatile variables. It would be good if you can look for such variables and fix them.

[xforus]

@ybc82 I have changed the variable '_networkDevicesNumber' as you said. But this don't solve my problem.

I'm using Arduino 1.6.8.
With 1.8.1 it doesn't work nothing.

I think that problem is at receiving part of Anchor.
This is my Log where you can see 3 anchors and 1 Tag. When Tag is expected to receive a POLL ACK from 3th Anchor but don't receive Poll from Tag so it doesnt' answer with a POLL ACK. Tag don't send range to other 2 Anchors because it waits to have all POLLS ACK from all Anchors:

Log at TAG part:

BLINK_SEND
ranging init; 1 device added ! -> short:CAE7
RANGING_INIT_RECEIVED CA:E7
POLL_SEND TO 1 DEVICES
POLL_ACK_RECEIVED CA:E7
RANGE_SEND TO 1 DEVICES
RANGE_REPORT_RECEIVED
from: CAE7 Range: 0.46 m RX power: -68.06 dBm

BLINK_SEND
ranging init; 1 device added ! -> short:9791
RANGING_INIT_RECEIVED 97:91
POLL_SEND TO 2 DEVICES
POLL_ACK_RECEIVED CA:E7
POLL_ACK_RECEIVED 97:91
RANGE_SEND TO 2 DEVICES
RANGE_REPORT_RECEIVED
from: CAE7 Range: 0.51 m RX power: -67.54 dBm
RANGE_REPORT_RECEIVED
from: 9791 Range: 1.06 m RX power: -70.45 dBm

BLINK_SEND
ranging init; 1 device added ! -> short:BAAB
RANGING_INIT_RECEIVED BA:AB
POLL_SEND TO A 3 DEVICES
POLL_ACK_RECEIVED CA:E7
POLL_ACK_RECEIVED 97:91

(here 3th Anchor should receive a Poll and send an ACK. but it don't. So no range from Tag and
result is : inactive time)

(in next cycle of blink 3th Anchor receive Poll and answres correctly):

BLINK_SEND
!!inactive time!!
delete inactive device: BAAB
ranging init; 1 device added ! -> short:BAAB
RANGING_INIT_RECEIVED BA:AB
POLL_SEND TO 3 DEVICES
POLL_ACK_RECEIVED CA:E7
POLL_ACK_RECEIVED 97:91
POLL_ACK_RECEIVED BA:AB
RANGE_SEND TO 3 DEVICES
RANGE_REPORT_RECEIVED
from: CAE7 Range: 0.51 m RX power: -67.36 dBm
RANGE_REPORT_RECEIVED
from: 9791 Range: 0.99 m RX power: -70.16 dBm
RANGE_REPORT_RECEIVED
from: BAAB Range: 0.99 m RX power: -68.09 dBm

[kuek]

hi, I found out that it works well when compiling the code with Atmel Studio. It does not work when compiled with Arduino IDE. So it seems like a compiler problem to me.

I have one PC with IDE 1.6.11 and another one with 1.6.13, both don't work well when I flashed the tag with the code and run it in a 3 (or more) anchor scenario. I then started to port the code to Atmel Studio and it works like charm. There were no code changes at all.

[AliSiddiqui4]

@kuek did you uploaded the original code using Atmel Studion? and are you receiving all the 3 ranges form the three tag? and is the mode the default LONGDATA_RANGE_ACCURACY?

[rbrtwtrs]

I changed the variable _networkDevicesNumber to volatile and confirmed that I do now get ranges from all three anchors. I notice that the data rate varies and it appears to stumble which as Rotzbua says seems like it is a race condition. Changing to a volatile is a kludge fix to the underlying problem.
I am using the Visual Micro Plugin for Visual Studio. If a particular compiler version does work it is also a likely kludge fix to the underlying issue.

[AliSiddiqui4]

@ybc82 WHAT MODE ARE YOU USING?

[ybc82]

@AliSiddiqui4 I'm not sure what you mean by 'MODE', but generally, I'm using the demo code of DW1000Ranging.

[christos-h]

@rbrtwtrs Did you change the _networkDevicesNumber variable to volatile only in the TAG or also in the Anchors?

[reycreator17]

Hi I managed to make 3 anchors and 1 tag run perfectly, but not so perfect with the precision when they are very near. I made my own perfboards and uses the same schematic but I power the whole thing with 3.3V parallel to the Vcc of pro mini and Vcc of DWM1000. I've used 10K ohm resistor and used some capacitors. I found out that measurements get very inaccurate when DWM1000 gets hotter, yes, when I made trials on a sunny outside.

Also I ON first just 1 tag and 1 anchor, after they are connected, I ON another anchor and see if the 3 are connected and ON the last anchor

I want to know how to make their addresses as Anchor 1, Anchor 2 and Anchor 3 and not some just random numbers.

Thanks Rey

[kuek]

Looks like some effort, but very nice. Glad it works!

[christos-h]

@reycreator17 nice. It seems from your pictures that you are getting negative ranges. In order to achieve static device address go to startAsAnchor in DW1000Ranging.cpp and change
_currentShortAddress[0] = 1; _currentShortAddress[1] = 0;
if you want the anchorID to be 1

What code did you use for 3 Anchors and 1 Tag?

[reycreator17]

@christoshadjiaslanis

Hi. I follow your instructions and its now working as I want.

I'm using the code in the latest dwm-master.zip, the DW1000Ranging_ANCHOR and DW1000Ranging_TAG.

I haven't changed any in the code except in the filter

DW1000Ranging.attachInactiveDevice(inactiveDevice);
//Enable the filter to smooth the distance
DW1000Ranging.useRangeFilter(true);

I enabled the filter.

About the negative distance, yes, I'm getting negative distances when the anchors are really near to the tag. Also, it is not very accurate about 2 meters distance and I noticed that the range changes whenever I'm rotating the tag or anchor.

I don't really know what problem you are facing, but I suggest you use arduino 1.6.8 version. I've tried 1.6.13 before and and other versions too, it doesn't really work well. In 1.6.13, I experienced, working well with 2 anchors but whenever I ON the 3rd anchor, the serial monitor just spamming "ranging init; 1 added device .... delete inactive device" again and again. Other versions, doesn't really give synchronize output in the serial monitor.

Uninstall your current Arduino, delete Arduino15 and library. Run ccleaner before installing 1.6.8. I've just follow this guy and it really worked for me
CLEAN UP BEFORE INSTALLING ANOTHER COPY OF THE IDE OR ARDUINO CREATE.pdf

Thanks

Rey