Pixhack-v3 vs Drotek PixHawk Pro vs PixHawk2.1


#1

Hi!

I’m considering to upgrade soon so I have been comparing some new AP on the market and two stand out rather a lot

------- CUA Pixhack-v3 --------
Microprocessor:

  • 32-bit STM32F427; 168 MHz, 2MB flash, 256Kb RAM
  • 32 bit STM32F failsafe co-processor

Sensors:

  • Extra damping IMU
  • ST Micro L3GD20 3-axis 16-bit gyroscope
  • ST Micro LSM303D accelerometer / magnetometer
  • Invensense MPU 6000 accelerometer/gyro
  • MS5611 barometer
  • MPU9250 accelerometer/gyro / magnetometer
  • HMC5983L magnetometer (gps unit-external)

So on board

  • 1 baro
  • 2 magnetometers (+1 external)
  • 3 accelerometers
  • 3 gyros

-------- Drotek Pixhack Pro -------
Microprocessor:
– 32-bit STM32F469; 180 MHz, 2Mb flash, 384Kb RAM

  • failsafe co-processor (?? No info, 32 bit STM32F?)

Sensors:
– ICM-20608 accelerometer / gyro

  • MPU9K (MPU9250?) accelerometer / gyro magnetometers.
    – LIS3MDL magnetometer (its on board? So +1 external)
  • baro??

So on board:

  • 1 baro (?? I hope lol)
  • 2 Magnetometers (+1 external it seems)
  • 2 Accelerometers
  • 2 Gyros

-------- Pixhack v2.1 -------
Microprocessor:

  • 32-bit STM32F427; 168 MHz, 2MB flash, 256Kb RAM
  • 32 bit STM32F failsafe co-processor

Sensors:

  • MS5611 Baro
  • MS5611 Baro
  • L3GD20 gyro
  • LSM303D accelerometer / magnetometer
  • MPU9250 accelerometer / gyro / magnetometer
  • MPU9250 accelerometer / gyro / magnetometer
    • All sensors connected via SPI

So on board

  • 2 baro
  • 3 magnetometers
  • 3 accelerometers
  • 3 gyros

------- conclusion ------
Drotek pixhack PRO has 1less accelerometer and gyro, but faster processor with 50% larger SRAM. There is a chance that information on PRO is not complete, so we may get “true diversity”, which requires 3 different sources to establish one faulty source. Considering better processor it would be nice to see that being implemented in final release (it is a testing stage these days). PixHawk2.1 is very similar to V3 with processors and sensors, but it has + baro in 3in1 MPU9250 instead of MPU6000 in V3 (I heard that MPU6000 is actually better in gyro/accel job than 9250, any ideas?)

On the other hand, PRO does not only have more processing power, but it also uses new type of Compass AND new ICM-20606 (getting rid off some what tricky LSM303D… 20606 is also as good or better than 9250 (by looking at the specks)

Which one will perform better? It is either pixhawk2.1 or PRO… different ideology, but PRO may be just be able to compensate for luck of extra MPU with newer sensors and faster processor! Test is needed.

Now the price tag (for Pro it is Preliminary, but it is EU company) very significantly. V3 is on sale for 170$, whether PRO is going to be in 250$ range. PixHawk 2.1 is about 230$ without gps. So I guess it is fair to say that V3 might come first in price/performance ratio, unless it performs poorly, which it shouldn’t.

I’m interested to see if any one has more info or first hand experience on either of them. Pixhack has sent me their v3 for review/test/compare against BG pixhack clone and (may be) against PixHawk v2.1 (if incan get one soon), but it won’t be here for few more weeks. Please share your thoughts :slight_smile: very exited to the upcoming test.

PS: It is a call out for discussion and information; I do understand that Pixhack2.1 with Edison is for completely different purpose, BUT it is about basic version only.


#2

This is also discussed here

https://www.rcgroups.com/forums/showthread.php?2860533-CUAV-Pixhack-V3-vs-DROTEK-Pixhack-PRO-vs-Pixhack2-1


#3

You mention “Arducode” in other thread, to my knowledge Ardupilot only runs on Pixhawk2.1 and Pixhack-V3 at this time, although it may run soon on all of them, and has only been tested thoroughly by many users on the Pix2.1/cube. So choice of flight stack (Ardupilot or Px4) is crucial at this time. Also Pixhawk2.1 as a heating element helping reduce IMU noise, not sure about Drotek’s and not the case with Pixhack.


#4

how does heating element help with IMU noise? Fluctuation with regard to the temperature from 0-50 degrees Celsius are around 1-2% according to the specs of the IMUs. So unless use fly in winter a lot, it won’t make much difference I think (end even than only before it warms up naturally)

Thank you about the stack point out! Totally missed it. Never run Px4, will read about the differences. Which one has more advantages in our opinion?


#5

Heating allows to do temperature compensation/temperature stabilization. Not sure what you mean by “fluctuation from 0-50 degrees about 1-2%”, the small MEMS gyros we use have up to several (rotational) degrees per minute per degree Celsius sensitivities, (several hundred (deg/hour)/Deg C)) so temperature compensation helps a lot. Bias is the enemy of precision …

As to different flight stacks, I wrote this a while back. But that’s just one opinion, you should of course seek more: http://diydrones.com/forum/topics/should-i-use-px4-flight-control-stack-or-the-apm-3-3-for-pixhawk?commentId=705844%3AComment%3A2349593

Also comparing processing power doesn’t mean much imho, plenty available on all three and a bit more processing won’t make a difference. And for those tasks what do require serious processing power , like vision navigation, none would be adequate anyways and by far. That’s where companion computers that are orders of magnitude more powerful come in.


#6

A very fair point about processing power. I agree that mor power is not necessary for autopilot with the tasks that fall on its shoulders. However having multiple source of same type data can be beneficial in terms of getting the correct result, agree? In this case it would be true that we are better to have full redundancy - 3 sensors of gyro/accel/mag/gnss

rather than fewer but with a even stronger processor.


#7

Does anyone know if a HERE+ RTK can be used with the Pixhack or the Pixhawk Pro? I have had nothing but trouble with my PH 2.1 w/ Edison. Others may have had a better experience, but i fly a BIG S1000+ air-frame with heavy payload and I cannot get it to tune well enough to fly. It really wobbles, especially in the roll axis.