Just wanted to inform you all that I have sucessfully modded and flown William Premerlani's UDB2 board (red board) with the invensys gryo board designed by Russell Duffy. The AP / stab code was MatrixPilot r466 and the planes were a Fun Jet and an Easystar clone (higher wingloading and with ailerons).



E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • T3

    Dear all,

    many of you have claimed that ST micro gyros, or any other, that have 5KHz resonant frequency, are somehow worse than gyros with 30KHz resonant frequency for quadcopter application.

    However, if we look at the motor RPM, which is say 3000-15000RPM, it is only 3000/60 to 15000/60 ie.

    50 to 250Hz revolution frequency.

    From this point of view,  having a sensor that has 500Hz+/-100Hz resonant frequency should not be any worse than the one with 5KHz or 30KHz. Yet, many people look at those numbers and claim 30 is better than 5 and this is why it works.

    Maybe there is a completely different effect that must be accounted for, like gyro sensitivity to vibration in general and not resonance as such, that would provide the explanation? Or maybe you have to multiply the RPM by the number of motor poles which make vibration spikes in the airframe? By number of blades? Even then 3 blades are generating max 750Hz main noise component...

    Current popular reasoning seems to ignore the factor of sixty which is hidden in Revolutions Per Minute.

  • T3
    ionut and ufo-man,

    Ufo-man is basically correct. A key element in MEMS gyros is a mechanical mass-spring system that is vibrated at its resonant frequency. If you subject them to vibration at that resonant frequency, the amplitude of the vibration builds up until the mechanical element begins smacking into something, creating a rattling effect.
    This makes them go nonlinear, causing some of the vibration components to alias to low frequencies. So what you wind up with is low frequency spurious signals that cannot be distinguished from real signals.

    What you see in the output of the LISY gyros that are being disrupted by vibration is an offset shift which is impossible to separate out.

    The resolution of the Invensense gyros is just fine, even with the greater dynamic range, because the effective resolution is improved by oversampling and filtering.

    The main reason that the Invensense gyros perform so well under vibration is that their vibrating parts have a very high mechanical resonate frequency. The IDG-500 has an X-axis resonant frequency of 24 kHz and a Y-axis resonant frequency of 27 kHz. The ISZ-500 has a Z-axis resonant frequency of 30 kHz. All three of these resonant frequencies are well above the frequency spectrum of the vibration you can expect from a heli or an internal combustion engine.

    The LISY300AL gyros have a resonant frequency of 4.7 kHz, which makes them vulnerable to vibration.

    Best regards,
    Bill
  • That is a differentiator with no sign data. It will probably not work with overloaded data. Keep also in mind the gyro data is already derivative data. There is no easy way of reconstructing data from a overloaded sensor if we need high bw data and accurate data, fast. The DCM cannot easily update with averages of a signal that has been sent thru a highly nonlinear block and such. That comes from its nature: its discretely updated. It is the sum of the updates that gives the attitude information for the flight computer.
  • rate_of_change_from_gyro=absolute_value(new_value-old_value)
  • Hehe. If sensor engineering was that easy, anyone could do it :) The problem is that the other gyros are easily overloaded by a helicopter vibration signature as well as a glow engine vibration signature. Then non-linear nature of overload phenomenon shows. The signal likely intermixes with second and xx order products and you end up with a useless signal. Even with filtering you cannot reconstuct the signal properly. The signatures from helicopter rotors and glow engines are continuous. Therefore an intermittent attentuation will not work. With a discrete cosinus based approach, you need to update the dcm with accurate gyro signals often. If you dont do that, you will loose attitude reference.
  • Thanks for clarifications
    Bill you mentioned that the new gyros are vibration free.This sounds more like they have a lower resolution.Can't you simply use the next code for LISY gyros to deal with gyro vibration:

    If(rate_of_change_from_gyro>threshold_vibration)
    {
    //do something
    }
    else{//Ignore the vibration
  • T3
    Hi ionut,

    I am not sure how much SparkFun will charge for the new board, but in any case they will continue to sell the old board as long as there is a market for it, and the UDB team will continue to support all boards. So, you will have a choice.

    Best regards,
    Bill
  • Admin
    I think it will cost almost the same when purchased new from SF , like Bill said it will be offered as built and therefore you don't have to solder any thing new.
  • I understand the benefits, but it seems for slow flyers it's not worth the extra money, unless it costs the same
  • The design was first triggered by the need for more vibration proof gyros on helicopters. The large roll rates that the Invensense gyros can tolerate without saturating makes them very desirable for fast acrobatic planes. If the gyros saturate, the DCM will not be updated properly and attitude information may be lost.
This reply was deleted.