übersetzte deutsche version by google

Quadrocopter Frame EPP

The last build frame, see QuadrocopterFrame, was compact, but in case of a hard landing, often the propellers are destroyed.

At my third quadrocopter frame in a time period of 3 years my design goal was:

This EPP frame (see at Links) was bought because of 4 benefits:

attachment:MultiWii_HK_Gesamt_DSC05786.jpg

After three successful flight I can say that the flight stability is OK with default PID values. I have learned a lot in the three years, better components are now available on the market, and the software is better, MultiWii version 2.3.

For the configuration of the software see at QuadrocopterParameter.

A click on the photos will enlarge them.

Parts

I have seen, that hobbyking no longer sells propeller GWS 1060HD counterwise rotation, so the next batch 10" propeller will be either:

There are now a lot of propeller offers for DJI copter. The axis diameter looks like 8 mm.

Electronics setup

attachment:MultiWii_HK_Bottom_DSC05785.jpg

attachment:MultiWii_HK_foot_back_DSC05787.jpg

The ESC (MPX, see at Links) and battery plugs (XT60, see at Links) are glued to the bottom center plate (bottom) with hot glue, see the picture on the right. Before I made little out cuts, so the plugs bottom side is equal to the bottom of the frame.

Last the cutout for the LiPo battery is made. It must be not as deep, as the battery is thick, because a 1 cm thick bottom styro foam plate is mounted underneath the foot with Velcro pads, to cover and protect all the wiring and the battery. To keep the level, this is also placed under the 4 outer feet with glue.

The insulation of the thick wires is partially removed, and the connection is soldered. The current sensor is glued to the bottom, like the plugs. The TURNIGY 3~8S Voltage Detector (see at Links) with a round black buzzer is mounted via Velcro pad to the foot backside, see the picture on the right. You can also see the bottom cover plate and the Velcro strap, securing the battery in place.

attachment:MultiWii_HK_foot_front_DSC05788.jpg

attachment:MultiWii_HK_mounting_plate_DSC05782.jpg

Two cables are routed to the electronics, see the picture on the right:

The electronics is mounted on a ground plane board (90 x 90 x 1.5 mm). In order to reduce vibration at the sensors, I glued the board with 4 pieces Peel-n-stick foam (see at Links) into the provided box of the frame, see the picture on the right.

attachment:MultiWii_HK_electronics_DSC05790.jpg

attachment:MultiWii_HK_baro_foam_DSC05792.jpg

The Hobbyking MultiWii AIO Flight Controller on top of the ground plane board was mounted with screws made from Nylon, just on one edge a 20 mm M3 metal screw is used to mount the GPS board. For the metal screws Hex locknuts (see at Links) are used only.

Then with Velcro pads are mounted onto the ground plane board, the Remote Control receiver, and a Bluetooth Tranceiver module for logging and data display.

The last module is a Voltage Regulator 7.0 V to reduce the power dissipation to the on board 5 V regulator for the Flight Controller and the appended modules, which might not be necessary.

attachment:MultiWii_HK_Motor_DSC05779.jpg

attachment:MultiWii_HK_Motor_mount_DSC05778.jpg

Motor setup

In the Customer Reviews of the EPP frame description (see at Links) it was mentioned, that the wooden boxes for the motor mount are difficult to glue to the 8 mm diameter fiberglass rods. My solution to this job was to use square aluminum tube pieces (see at Parts) glued with epoxy glue to the fiberglass rods, see the pictures on the right.

First the 3.5 mm holes for the M3 slit countersunk screws (DIN 963, Schlitz Senkkopf Schrauben) are drilled in the alu tube, Then on the other side of the tube drill with 6 mm diameter for mounting purpose of the screws, and countersunk the 3.5 mm holes. The next step is to mount the motor with self securing nuts.

attachment:MultiWii_HK_Motor_glue_DSC05777.jpg

Last drill with 8 mm the aluminum tube, to allow for mounting on the 8 mm diameter fiberglass rods.

Sand a little bit the fiberglass rods at the mounting area. Move the motor base near to the location to take place. Put some epoxy glue to the sanded area, move the motor mount in place and water level the position, see the picture on the right. The distance from the crosspoint of the fiberglass rods to the center of the motor base should be 225 mm, as mentioned in the instructions (3 pages Chinese/English, p1, p2, p3).

attachment:MultiWii_HK_Start_5019.JPG

Flight Experience

attachment:MultiWii_HK_Flug_5041.JPG

.

EZ-GUI Log

I wanted to know, how much power does the copter need to fly stable in place. Therefore I added a 30 A current sensor module, see at the Links.

attachment:MultiWii_HK_EZ-GUI_Log3-118.png

Unfortunately in older versions of EZ-GUI the parameter EANA:amperage was not correct, always about 30000. Now with version 3.118 (2014-12-10) the value is ok. It may also has to do with the version of MultiWii, which is pre 2.4 r1729.

Have a look at the OpenOffice diagramm on the right. I tried to keep the copter in place, as good as possible.

I think, with this method of measuring the power consumption it should be possible to check the efficiency of the propellers.

The Log-data were prepared with the GNU/Unix tool grep, for example:

# Copy the log file from the Android smartphone to your PC
# Extract all EANA lines for the "amperage" value
$ grep EANA MultiWiiLog_2014_12_09_15_53_00.el1 >Log_15-53_EANA.csv 
# Delete the first line (comment only)

# Import the .csv data file into your spreadsheet

attachment:MultiWii_Mission_Ickstadt_2014-12-09_center.jpg

Mission control

A mission with 6 way points was setup (height = 5 m, first waypoint is in South-West), see the picture on the right (Smartphone screenshot). After the flight the log file from EZ-GUI was exported and converted by the EZ-GUI log converter to a .kmz file which could be imported by Google Earth. The second picture on the right shows the actual flown course.

What was surprising for me is the good working distance of the bluetooth link, about 65 m to the smartphone, without special measures.

While this mission, the received number of satellites was 7.

In order to calculate a deviation from the planned course to the logged reality you can use a spreadsheet, in my case LibreOffice. I am using EZ-GUI (Version 3.118.179) to get a Log file.

EZ-GUI does not use the original MSP names (MultiWii Serial Protocol, see at Links) for the data telegrams, instead 4 upper case letters, like EGPS. The log file contains 16 different telegram types.

attachment:Mission_201412091559_GoogleEarth.jpg

If your Android smartphone connects to the PC via Media Device only, to must first copy or move the EZ-GUI Log file with a File Manager App (e.g. ES Datei Explorer) from the folder MultiWiiLogs to a visible folder, e.g. downloads/MultiWiiLogs (create folder MultiWiiLogs with a file manager by yourself).

Use the following mentioned Tools (Unix/GNU style) to process the EZ-GUI log file for spreadsheet use (native in Linux and Mac OS, use GNUwin32 for Windows).

For easier data handling in the spreadsheet, only every 7th data line was used. I have cut off the begin and end of the flight log, in order to show more clearly the mission itself.

Two of 16 telegram types must be extracted from the log file for investigation.

# Copy the log file from the Android smartphone to your PC.
# Extract all EGPS lines / telegrams.
$ grep EGPS MultiWiiLog_2014_12_09_15_59_18.el1 >Log_15-59_EGPS.csv  
# Delete the first line (comment only) with an text editor.

# check file for number of lines:
$ wc -l Log_15-59_EGPS.csv                                              
  1340 Werte

# Take ever 7th line only, to keep it simple:
$ perl -ne 'print unless ($. % 7)' Log_15-59_EGPS.csv >Log_15-59_EGPS191.csv

# Import the Log_15-59_EGPS191.csv data file into your spreadsheet.

# Do the same with the GMAG lines / telegrams.

In order to calculate a distance in meter you need the formula:

LibreOffice Calc Einstellungen:
    Spracheinstellungen: Gebietsschema: Englisch (USA)
                         Haken bei Nur für das aktuelle Dokument
    Calc, Formel: Haken bei Englische Funktionsnamen verwenden

Table shows the waypoint deviation in meter from the planned waypoint coordinate to the minimum deviation found in the log file:

attachment:MultiWii_pre2.4r1729_mission_SAT7_-head+180.png

Waypoint

Deviation

number

in meter

1

0.8

2

0.4

3

1.2

4

1.1

5

2.3

6

0.3

The parameter #define GPS_WP_RADIUS is set to 100 cm. The way deviation of maximum 1 + 1.3 m from the defined way points can be explained by the reduced time resolution per sample of about 0.9 sec, and a speed of about 1.5 m/s. Also wind will have an effect.

One peculiarity is in the data head (from MAG) and course (from GPS), please see the diagram on the right. You will see a nearly parallel curve only, if head is modified to go counter clockwise from 0 to 359 degree.

While the course is defined 0 to 360 degree (0 = North), the head is defined +/- 180 degree (0° = North. +90° = WEST). The MultiWii parameter #define NAV_CONTROLS_HEADING is 1, source code comment in config.h:

attachment:Mission201412301143_Map.jpg

The MAG was activated while the mission.

Course-Head Test

In order to test the GPS-course and MAG-head measurement, I walked with the Quadrocopter holding by hand along a straight street (130 m long, from South to North), please see the map picture on the right. In theory the blue line should be straight.

The front of the Quadrocopter was pointing parallel to the street. I estimate, that the direction deviation (Yaw), caused by walking, was not more than +/- 12 degree.

The outside Temperature was 3°C, and the motors are not turning.

With EZ-GUI a log file was created while this walk. With the EZ-GUI Log Converter the XXX.kmz file (zipped XXXgx.kml and XXX.kml) was created for the map picture.

The course of this walk was measured by an angle meter on the printout to about 345 degree (15 degree West).

In order to better compare the MAG-head value with the GPS-course value, the GPS-course value was converted from 0...359 to +/- 180 degree (0° = North, +90° = East).

attachment:MultiWii_pre2.4r1729_walking_SAT7_course180.jpg

The EGPS and GMAG data were extracted from the log file and inserted into a spreadsheet, for the procedure see at QuadrocopterFrameEPP#Mission_control. Only every 4th data telegram was used (about 2 per second) to easier the data handling.

For me it was surprising, to see the big GPS-course peaks (+53°, -67° from average). While the whole walk, the GNAV data showed continuously 7 satellites.

It looks like, that houses near by will distort the GPS reception.

The average calculation of the GPS-course was -13 degree, and from the MAG-head -27 degree. The GPS-course value is not so far from the ideal -15 degree value.

attachment:Mission201412311709_map.jpg

It looks like, that the magnetometer was not properly calibrated.

Course-Head Test 2

For a second test of the GPS-course and MAG-head measurement, I walked with the Quadrocopter holding by hand along a straight street (68 m long, from South to North), please see the map picture on the right. This time, no houses and cars where left and right of the course, but a bridge. In theory the blue line should be straight.

The front of the Quadrocopter was pointing parallel to the street. I estimate, that the direction deviation (Yaw), caused by walking, was not more than +/- 12 degree.

The outside Temperature was 3°C, and the motors are not turning.

With EZ-GUI a log file was created while this walk. With the EZ-GUI Log Converter the XXX.kmz file (zipped XXXgx.kml and XXX.kml) was created for the map picture.

The course of this walk was measured by an angle meter on the printout to about 353.5 degree (6.5 degree West).

In order to better compare the MAG-head value with the GPS-course value, the GPS-course value was converted from 0...359 to +/- 180 degree (0° = North, +90° = East).

attachment:MultiWii_pre2.4r1729_walking2_SAT7_course180.jpg

The EGPS and GMAG data were extracted from the log file and inserted into a spreadsheet, for the procedure see at QuadrocopterFrameEPP#Mission_control. Only every 2nd data telegram was used (about 4 per second) to easier the data handling.

This time there were no GPS peaks, as last time. While the whole walk, the GNAV data showed continuously 7 satellites.

It looks like, that houses near by will distort the GPS reception.

The average calculation of the GPS-course was -6.5 degree, and from the MAG-head -11.4 degree. The GPS-course value matches perfect the ideal -6.5 degree value.

It looks like, that the magnetometer was properly calibrated this time, but on the first 38 m distorted from the iron of a bridge.

attachment:MultiWii_20150107_direction.png

Course-Head Test 3

For a third test of the GPS-course and MAG-head measurement, I walked with the Quadrocopter holding by hand along a straight way (100 m long, from North to South-West, 69 seconds -> 5.2 km/h). This time, in a radius of about 50 m no metal was seen. I tried to hold the Quadrocopter as stiff as possible in direction and height.

The front of the Quadrocopter was pointing parallel to the way. I estimate, that the direction deviation (Yaw), caused by walking, was not more than +/- 5 degree. The outside Temperature was 5°C, and the motors are not turning.

With EZ-GUI a log file was created while this walk. With the EZ-GUI Log Converter the XXX.kmz file (zipped XXXgx.kml and XXX.kml) was created for a map picture.

The course of this walk was measured by an angle meter on the printout to about -135 degree (South-West). In order to better compare the MAG-head value with the GPS-course value, the GPS-course value was converted from 0...359 to +/- 180 degree (0° = North, -90° = West).

On the first diagram on the right you can see for my surprise, that both the head and the 180 degree course does not match the -135 degree course, measure on the Google Earth map (from the KML file). The average of both curves are 14 degrees apart.

The good thing is the little deviation from the straight line, with +/- 4 degrees (+/- 2 x sigma, 95.4%). So, it can be concluded that large metal objects in a distance of 5 m distort the magnetometer.

attachment:MultiWii_20150107_Baro.png

The next test was for the Barometer stability. The 100 m way had a height difference from start to the end by 5 m, see the diagram to the right.

In order to see the spread I generated a linear approximation and calculated the ideal height curve. The performance R2 is very good with 0.998. The logged Barometer height deviates maximum from that ideal line by +/- 32 cm (+/- 2 x sigma, 95.4%).

In reality, with wind and the control response time those values will the greater.

  1. Extra Large EPP Quadcopter Frame 450mm

  2. Hobbyking MultiWii AIO Flight Controller

  3. Hobbyking Neo-6M GPS module

  4. Hobbyking MultiWii Bluetooth module

  5. ESC Plush 12A with BEC

  6. hexTronik_DT750_Brushless_Outrunner_750kv

  7. TURNIGY 3~8S Voltage Detector

  8. MPX Connector Male/Female (10pairs/set)

  9. Nylon XT60 Connectors Male/Female (5 pairs) GENUINE

  10. Hex locknuts M3 10pc

  11. Velcro_Peel_n_stick_adhesive_side

  12. Battery Velcro Strap (2pcs/bag)

  13. Peel-n-stick foam double sided tape

  14. Hobbyking, 6 pc. GWS propeller 10x6 HD

  15. Current sensor 30 A, ACS715

  16. MultiWii Serial Protocol

  17. Applications of Magnetic Sensors for Low Cost Compass Systems

  18. Applications of Magnetoresistive Sensors in Navigation Systems

  19. Honeywell, 3-Axis Digital Compass IC HMC5883L

  20. Wikipedia.org, Course (navigation)

Contact Email

Please enter your Email address, if you expect an answer.

/!\ The entered Email address will not be published, or given away.

Comment this page
Name:
Comment:
Are you human?

List of pages in this category:

-- RudolfReuter 2014-12-03 08:01:03


Go back to CategoryQuadrocopter or FrontPage ; KontaktEmail (ContactEmail)

QuadrocopterFrameEPP (last edited 2018-02-22 17:23:41 by RudolfReuter)