Q:
Pls give the Package and the voltage and pressure curve about 90269 and 90257.I want to know how to use the programmable function.
|
A:
The 90269 and the 90257 are currently available as bare dies.
The typical output voltage characteristic is 0.5V as output offset and 4.5V as output full scale. The corresponding pressure points depends on the version of the sensor:
-The 90257.1 and 90269.1 can easily accomodate a full scale span between 1.2bar until 3 bar to achieve an output span of 4V.
-The 90257.2 and 90269.2 can easily accomodate a full scale span between 3bar until 7 bar to achieve an output span of 4V.
-The 90257.3 can easily accomodate a full scale span between 3bar until 7 bar to achieve an output span of 4V.
Melexis can provide a tool for programming the devices and can send packaged samples currently just for evaluation.
|
Q:
Thanks for your answer.Pls give me the package beacause I can't find the package in the datasheet.How to decide the slope of the curve if I want use 115kPa as the full scale.Do Melexis have the relative pressure sensor of 70bar. How to use the programming tool after I buy the tool.
|
A:
It is possible to use the absolute and relative pressure
sensor with a full scale of 115kPa. We don't have a relative pressure sensor for 70 bar applications.
|
Q:
I already found a couple of discussions about the CLCC32 socket on the forum but the suggested Plastronics socket isn't really a socket for production devices and is quite large. Are there any sockets available at a more "normal" size?
Alternatively, can you provide me with the correct pad layout for SMD soldering the CLCC32 housing?
Thanks in advance.
|
A:
Indeed, the ZIF sockets from Plastronics are not suitable to be used in the small and cheap electronic devices. In this case I'd suggest soldering the MLX90609 directly on the PCB. The appropriate footprint of the CLCC32 you can find at the end of the Datasheet:
http://www.melexis.com/Assets/MLX90609_(standard_version)_5359.aspx
|
Q:
Does Melexis have cross for Analog Devices's ADXL202 dual axis accelerometer.
|
A:
Sorry, we don't have this.
|
Q:
for my application i need a larger full scale range than the maximum available 300 °/sec, for example about 600°/sec.
Is it possible to increase the range of the R2 version of the 90609 to 600 °/sec by simply putting a 200k resistor parallel to the internal 200k resistor between FLT and OUTAR?
|
A:
We do not recommend changing a gain by the external resistor because it will distort a factory bias calibration and temperature compensation. Melexis can produce MLX90609s with a different sensitivity range in case of ordered big volume.
|
Q:
thanks for your immediate reply.
You say, that you don´t recommend to use an external resistor. But will it give the expected result (extending the range to 600 °/sec) if I would do that anyway?
Ordering big volume is not an option for me so I would except some minor disadvantages to extend the range. I have to calibrate calibrate all my used sensors anyway (I´m using SPI), so a small constant bias shift wouldn´t be a problem.
|
A:
Indeed, you can decrease the gain by means of the external resistor connected between the pins OUTAR and FLT. However, pay attention to the list of disadvantages that will appear:
1. The internal output resistor Rout can deviate from its typical value = 200k. It means you probably will need to fit the external resistor in order to set 600 deg/sec range. Also in order to check the sensitivity you seemingly will need an expensive dynamic calibration setup or you should rely let's say on the approach with integration.
2. The external resistor can shift ZRO and distort its temperature compensation. So in general case the shift will not be constant.
3. Melexis for sure will not support this solution ;-)
|
Q:
1. The possible deviation of the internal resistor won´t be a problem, because my application doesn´t depend on absolute rates. It doesn´t matter if I get 590 or 610 °/sec.
2. Temperature shift could be a problem. I hope, it won´t too big. Maybe I have to increase the deadband.
3. I thing, melexis wouldn´t support my use of the sensors at all: 3-axis electronic stabilisation system for gas powered rc-helis with LOTS of vibrations ;-).
Thank you again, you helped me a lot.
Best regards.
Dirk
|
A:
I wish you luck in implementing your idea ;)
I forgot to mention about another disadvantage. According to the MLX90609 Data Sheet the OUTAR load has to be more than 200 kO. The OUTAR overload causes additional output nonlinearity. Too big overload (more than 25 mA) may cause also permanent damage. The last one is not your case, but the first... With external resistor you will be close to the limit we set. So I suggest using e.g. an external output OPA in the follower mode.
|
Q:
Is there any commercially available ic socket adapters (for through-hole mounting) suitable for 90609 ?
|
A:
We use ZIF sockets from plastronics
The MLX90609 is packaged ic in a JESD compliant CQFN32.
So standard sockets should be available.
|
Q:
What is the bit value on the A/D converter in your gyros? I assume it does not vary between the 75/150/300 deg/sec versions?
|
A:
Thanks for the intrest in our products. We are using an 11 bit A/D convertor.
|
Q:
I have some problems in understanding PSF_MLX90257 for my application.
When we calibrate the MLX90269,we should set the value of InputInv to 1,the value of Test Mode to 5,then set the value of PRGain to 0 or 1,to select the value of PRGain to make OUT closest to 4.02V.
I use the function LoadTemp(cmd as byte) to set the test mode,but the OUT is 2.38V.When I use the soft of PTC-03 to test it,the OUT is closet to 4.02V.
So,I want to know which function i should use and how to use it?
Help me out...
regards.
|
A:
How do you communicate with the chip? Do you use the EVB90257? Is it possible to send your function via e-mail (die@melexis.com) Then I will try to figure this out.
|
Q:
I bought MLX90269,but when i used the function of PSF_MLX90257,I got the value of the chipvirsion is 0 not 1.
|
A:
The chip version is not automatically recognized by SW. It must be explicitly set by means of ChipVersion property of Device object after creation of the device. Take care that it could be overwriten by OpenProfile method of Advanced object if the loaded .INI file contains "ChipVersion" key in "Settings" section with possible values "90257" or "90269".
|
Q:
I cannot seem to find the scale factor for the temperature output of the chip. Am I overlooking something in the datasheet ?
|
A:
TC is 10 mV/K
|
Q:
I have 2 short questions:
1. Where can I get the full datasheet for the 90609 (including SPI command)?
2. I´ve already got 2 different 90609s. One E2 with 150°/sec and one R2 with 300°/sec. How can I find out, which one is which? I can see nothing on the labelling of the chip that says E2 or R2???
|
A:
I will sent you an application note concerning the SPI communication for the gyroscoop.
I have some information available which can help you to determine the current version of your chip. I will sent this also by e-mail.
|
Q:
Hello sir, I am using MLX90269 for my application.When we calibrate it,at first we set InputInv to 1,then set the mode to 5 and set the Prgain to 0 or 1,to find the value of Prgain which let Vout close to 4.02V.
But i don't know which function i should use in the library?how to use it?
|
A:
I would like to mention that this procedure is not needed for the calibration. This Prgain bit is calibrated during production tests.
So if you still have other questions, please do not hesitate to ask for it.
|
Q:
I have three different 90609: E2, N2 and R2 and I can't differentiate them by looking at the printed code.
I would appreciate if you could send me some information.
I also would like to request the SPI command document and full datasheet.
|
A:
Dear Vivian,
When you look to the surface of the chip, then you can see a difference in the dots that are located on the lid. So the label will be the same but there is a difference in the dots, I will send you an updated document which describe this issue.
best regards
Dieter
|
Q:
Could you post this documents so everyone with the same concern can dowload it instead of having you send them by email? thanks a lot!
|
Q:
And is better to located the 90603 in the center of my rotating axis or in a certain distance to avoid noisy signals?
|
A:
You can best place the gyro in the center, if there are to noisy signals, you should try to found a better position...
|
Q:
I need complete version MLX90609 datasheet and any application note for this sensor. I need to start prototyping product with this sensor
|
A:
You can find the MLX90609 datasheet here:
http://www.melexis.com/prodfiles/0005359_MLX90609_standard_datasheet.pdf
|
Q:
Is there any way to procure samples or small quantities at a reasonable price of the MLX90609 or eval board in South Africa?
|
A:
Please see www.melexis.com/contacts.aspx for location of your neares rep or distributor. You may then direct orders and requests throught the local Sales network.
|
Q:
I have connected a MLX90609 angular rate sensor to a Basic STAMP 2 Microcontroller, averything was going perfect, I could read the SPI interface until for some reason the sensor just freezed, the output read something constant, the same happened to another one I have, it worked well then it froze, then they both started working again, and freezed again after a while. I have only connected to VSS, VDD, MISO, MOSI, SCLK and SS out of a Sparkfun Breakout board for the MLX90609. Should I be grounding or connecting any of the other pins? (VREF, TEMP, RATE, ST, ERR, VCC, GND).
What else could be the problem here?
|
A:
We do not have any insight in the sparkfun PCB. It is a independent design, not controlled nor supported by Melexis.
However: Vcc and GND must be connected too
|
Q:
We have exactly the same problem. The 5 proto's of the Melexis (June/July) were running very well. We decid to switch to the Melexis gyro. Now with production start the "freezed" problems (November). We used a microcontroller and connected the same as figure 6-1 (page 15) and 6-4 (page 17) of datasheed. All the pins are connected as in the datasheed. When the gyro is frozen, the analog output (OUTAR) is also not changing while the gyro turns. The temperature output by analog and spi is reacted okay. After reset, the gyro is running well until 1 hour to 4 weeks...
Our PCB is the same as with the 5 proto's
Now we stops the production and turn back to the old gyro of AD.
Is the selftest not the problem? The pin is not connected. I can not find any information about the selftest. What can we do? I read that other people has the same problem?
|
A:
The floating selftest pin can not cause the issue you've described.
In order to figure out a reason of the issue it has to be discussed in details.
|
Q:
I am querying a MLX90609 via the spi/adc interface. With the device taped to a table I am reading 2428 - 2444 mV which would translate to 2.7 d/sec - 2.1 d/sec counterclockwise rotation. I would like to be able to calibrate the device but am not sure exactly how to go about it, I was wondering if you could give some suggestions?
|
A:
Calibration and temperature compensation of the MLX90609 is done at the Melexis' factory to guarantee the accuracy put into the MLX90609 datasheet. According to the datasheet the Zero Rate Temperature drift is limited by the value +/-5% of the typical FSout = 4V that means: Voutar = 2.5V +/- 200mV = 2.3V...2.7V for VDD=5V and full temperature range (-40C...+85C).
For the MLX90609EEA-N2 (FSin=75deg/sec that is your case) the +/-200mV inaccuracy can be estimated by the angular rate velocity: +/-200 / 26.67 = +/-7.5 deg/sec
Here I'd mention that all silicon gyros have quite big ZRO inaccuracy. That is why they usually need some algorithm fixing real zero value and/or providing autocalibration. The choice of the algorithm strongly depends on the type of the application. There are some examples:
- If your device is not rotated you can fix zero level at that moment. Good accuracy is important there, so averaging is recommended.
- For some applications the low angular rate is applied with high angular rate acceleration only. It makes possible to recognize the ZRO drift (low angular rate and low angular rate acceleration) and to remove it continuously.
- Some applications have GPS that can be used for autocalibration from time to time.
|
Q:
I would like to know if I can reduce the Zero Rate Temperature Drift by measuring the temperature of the Gyro and applying an offset correction value from a look-up-table or equation. If so, can you tell me the correction versus temperature function?
|
A:
Indeed, in order to improve ZRO accuracy of the factory calibration you can use a look-up-table generated for the particular MLX90609. The EEPROM of the MLX90609 includes the look-up-table (3 temperature points only). But this approach is not still validated and hence we can not officially recommend it for the public use.
Nevertheless, you can characterize chips you have on your own in order to create the look-up-tables for each of them. Pay attention, the look-up-table should include ZROs versus TEMPOUT values (the absolute temperatures is not needed in the table). And please wait a settling time of the TEMPOUT before measuring of each temperature point. It's important. By the way, we can discuss possible algorithms directly through E-Mail.
|
Q:
I have been trying to find your full data sheet for the MLX90609 including SPI commands, but am unable to find it. Can you please point me in the right direction?
|
A:
Hello Jerry,
Please find the Datasheet on the Melexis website:
http://www.melexis.com/Assets/MLX90609_(standard_version)_5359.aspx
Regards,
Nikolay
|
Q:
Could you send me the full technical data of the MLX90609. The standard data sheet does not give much information about the Digital data, command and so how to translate the analog voltage to the meaningful data.
|
A:
The standard data sheet includes data you need. Please, see the table 4 (page 6). You can find there the main parameters of the MLX90609 translated to the LSB of the embedded ADC (full scale, sensitivity, ZRO, ZRO drift). Some other "missing" parameters can be converted to the LSB by means of mentioned data. The page 12 includes two conversion formulas derived from the typical values from the Table 4. The next page describes the SPI-commands that are used to control the ADC and page 13 gives an example of the recommended SPI sequence to obtain the digital data from the ADC.
|
Q:
I am hoping that you will have some suggestions for how I can solve my problem with the MLX90609. I hooked it up for SPI as indicated in the datasheet, and yesterday I was able to successfully request and receive gyro data using a dsPIC. After unplugging it for the night and coming back in this morning, however, I have been unable to get a meaningful response from the gyro. I actually have three of the gyros on the same bus (X, Y, and Z) and all three exhibit the same behavior - no matter what SPI command I send, they send back a constant response. The strange thing is that each of the three sends back a different constant:
X sends back 0x2207
Y sends back 0x2247
Z sends back 0x2200
The rate analog pins give the correct output, so I know that the gyros are working. Is there any way that the ADC could get turned off? Do you have any suggestions for ways I can reset the SPI interface?
|
A:
The above quesrion is closed as it was found to be a SW error in the code that had to generate a SPI sequence.
|
Q:
I am having difficult communicating a MLX90609 Gyro with a Basic Stamp BS2. Would anyone have some sample code that works with this over the SPI. I tried my own but a cant get it to work
|
A:
At the moment we don't have SW examples for the Basic Stamp.
|
Q:
I was looking for PCB layout guidelines for the MLX90609 part, and was unable to find them. Are they available somewhere?
Also, I was wondering if it was permissible to power only the analog portions of the chip, and not the digital ones (as I would have my own data converter.) Failing that, would there be some way to switch the digital portion to standby/low-power/low-noise from the SPI command set?
|
Q:
I have a Melexis eval board and have found that even when the gyro is stable and not moving the acceleration is between .1 and .5 degrees/sec^3. The demo will rotate about 360 degrees in about 10 minutes while the device is static. I have even changed the averaging but with no success. Is this a normal operation or is there a problem with the sensor device or the signal chain. The eval kit is not anywhere close to any interference either. The device is a 70degree/sec^3 device.
|
A:
Please contact your local sales representative. They will have to determine the appropriate Applications engineer to work with you on this issue.
|
Q:
Can you also provide me the description of the SPI protocol?
Any information on ADC timing?
EEPROM calibration: Is the device factory – calibrated using the EEPROM? Am I supposed to calibrate it? Which parameter goes where?
Any additional information will be most welcome, since there is very little of it in the data sheet.
Any new version for the EVB90609 software?
|
A:
We are working on an updated set of web documents including the information which will be published in the next weeks.
The MLX90609 is, of course, a factory-calibrated device and the calibration parameters can not be changed in the application mode.
I think you can find all answers you need in the preliminary version of our standard data sheet I've sent to you.
|
Q:
Please can you provide me the information on using the MLX90609 to determine the angular direction minimising the drift and errors.
Also, I would appreciate more details of the SPI protocol as described earlier in the thread.
I wonder if you could make available the source code (or at least the parts relevant to the demo) for the evaluation board?
|
A:
Please contact Melexis local Sales representative. They will need to assist in getting answers from the appropriate Melexis Applications engineer.
|
Q:
I am also interested in the information on the drift behaviour of the MLX90609.
The datasheet I have is rev004 dated Feb/07, if there's any update I also would like to receive it.
|
A:
Hello Rudi,
It seems you have the obsolete and very short version of the datasheet. Please find a latest version (Rev 008) on our website:
http://www.melexis.com/Asset.aspx?nID=5359
Regards,
Nikolay
|
Q:
We are going to use MLX90609 for mobile application so we would like to know some additional information concerning the paparameters:
1. The gravity sensitivity (G-sensitivity) [o/s/g]
2. Linear accelaration operating range [g]
|
A:
To provide an answer on your questions,
1. The gravity sensitivity (G-sensitivity) [o/s/g]
>; 0.1 deg/s / g
2. Linear accelaration
operating range [g]
<; 5 g
However the shock survival is <; 1500g.
|
Q:
I would like to verify if I received the correct sensors. Unfortunatly I did not find the description of the product code. How can I verify the chip is a 300 degree/s R2 type ?
|
A:
An extra lid marking is used for the MLX90609-E2 (150deg/s range) and the MLX90609-R2 (300 deg/s range).
Let's place the MLX90609 to read the name by the normal way. In this case:
- MLX90609-N2 ( 75deg/s sensitivity) has no additional marking
- MLX90609-E2 (150deg/s sensitivity) has 2 gold dots at the left and right bottom corners
- MLX90609-R2 (300deg/s sensitivity) has 1 silver dot at the right top corner
However, this marking can be not reliable, as it can be erased.
If it happened, you can even manually estimate the sensitivity by turning your PCB with the MLX90609 over a given angle (e.g. 90 deg or halve or full turn) and then by integrating the value "OUTAR minus ZRO" over time. Most scopes (e.g. Agilent) have a built-in integration function. The result should give an angle with taking into account the actual MLX90609's sensitivity. As the angle is known one can derive the sensitivity.
|
Q:
Does a CLCC32 socket exist?
|
A:
LCC Zero Insertion Force (ZIF) sockets are e.g. provided by "Plastronics":
http://www.locknest.com/newsite/products/lcc/index.htm
Pay attention: the MLX90609 features a 1mm (0.04") pitch => Part # P2032S
|
Q:
I'm wondering if you have any sample code for the MLX90609 using SPI on the PIC18F microcontroller. We are currently implementing multiple gyros seeking to achieve dead-reckoning navigation, but there seems to be no sample code available. I can't seem to find any sample progs on your site.
|
A:
Unfortunately we don't have any sample code for the PIC18F microcontroller.
Nevertheless, we can provide E-Mail support for you in this issue.
|
Q:
I can't find a document in which the SPI interface for the MLX90609 is explained. Also, what is the resolution of the ADC onboard? where do you post this information?
|
A:
I will send you a more detailed document about the SPI communication.
We have an 11-bit ADC on board which is about 0.05% of the Full Scale.
|
Q:
Instead of just emailing everyone, why don't you include this information in the datasheet on the website! There are other critical things missing too, like current consumption.
I'd love to design in your product, but the lack of information in the datasheet makes this very difficult.
|
Q:
Hello Dieter,
Can you also send me the description of the SPI protocol and more information on the ADC conversions?
Thanks!
|
Q:
Thanks for sending me the application note for the mlx90609. I've been able to retrieve the angular rate and temperature data through the SPI, but now that I've got the data I don't know how to convert it to real-world units. I didn't see a conversion in the datasheet or the application note, am I missing something?
|
A:
A standard datasheet of the MLX90609 is already available on the public Melexis
website:
http://www.melexis.com/Sensor_ICs_Inertia/Sensor_ICs_Inertia/Angular_Rate_Sensor_582.aspx
|
Q:
could you please send me more information about the MLX90609 :
- SPI protocol
- current consumption
- all application notes you have
Because the datasheet on the web site is not enough for me to start my development.
|
A:
Please review the datasheet on the website
|
Q:
Some question about the Data sheet (Rev 008):
Pin Definition: In Table 2 Vss is mentioned, in Fig. 2-1 there is no Vss. Is it GND?
Value for C1, C2: On page 15 1.0µF is recomended,in Fig. 6-4 0.1 µF. What is best?
In the product flyer a C4 on PLLCAP to ground is shown, but no such pin is in the Data Sheet. What is it?
Thank you for your help,
Yours Bernd
|
A:
(a) Indeed, there is a typo with GND.
It should be VSS in the Figure 2-1 as well as in the Figure 6-4. So VSS=GND.
(b) If somewhere on the PCB you put an electrolytic capacitor between the ground and VDD you can choose 0.1 uF value for the decoupling capacitors.
(c) In the Flyer the application circuitry has to be corrected. Please use Data Sheet instead of that. By the way, in the Figures 6-1, 6-2 and 6-3 we will add a pin TESTIN connected to the ground (see table 2).
Thank you for your valuable remarks. We very appreciate this kind of feedbacks. All your remarks will be taken into account in the next revision of the Data Sheet.
|
Q:
Another question about the MLX90609:
Is it advisable to have to seperated power supplies for analog and digital, or can these two pins be connected, like seen in Fig 6-3.
|
A:
Power supply pins have to be connected like seen in the Fig 6-3.
Please, don't use separated power supply.
So VSS=GND (pin 3) has to be directly connected to the VSSA (pin 26).
The VDD (pin 4) has to be directly connected to the VDDA (pin 27)
|
Q:
Do you have any documented high level algorithm to work with the MLX90609EEA-R2 Angular Rate Sensor?
I bought 3 of these to build a 3 axis gyro cluster and have set them to work at 36.4Hz Bandwidth.
I am currently using the analog interface and I can see a voltage variation when I rotate the sensors however how do I translate this into an angle ?
|
A:
At the moment we have no documented high level algorithms you asked about. I hope in the nearest future we'll release an Application Note that will describe the SW that was written for our Evaluating Board. A toy-compass implementation is a part of the SW.
The formula given on the page 8 of the Data Sheet can be used to convert the analog output to the angular rate: AngularRate (deg/sec) = (Voutar - Bias) / Gain
Voutar is an output voltage at the OUTAR-pin in mV, Bias = ZRO = 2500 mV (typ)
Gain = Scale Factor = Sensitivity = 6.67 mV per deg/sec (typ) for your MLX90609EEA-R2
Please take into account possible ZRO and Sensitivity drift given by the table 4 of the Data Sheet. Special algorithms are usually used to improve the accuracy of the Sensitivity and especially ZRO accuracy. The simplest example of that is a ZRO setting without applied angular rate. More complex example is e.g. a Kalman filter that is very popular and efficient for the navigation applications. The choice of the correcting algorithms depends on the application requirements.
If you need an angle, you will need to integrate an angular rate over time. Note: even small ZRO error after integration will yield a big error of the angle. In other words, the precise ZRO-setting during a static condition is really crucial here. Sometimes it's not so easy to do perfect zeroing because of the noise. One can use e.g. averaging to improve the result of the ZRO-setting. A time measurement is another important point because time is used for the integration as well. For example, the microcontroller of our EVB uses a crystal oscillator to care about the real time and to start the ADC every 200us (it's ensured by timer's interruption).
Thus, the initial precise zeroing together with accurate time scale are two crucial conditions for the integration.
|
A:
By the way for navigation usually more sensitive gyro range is used. Actually The MLX90609EEA-N2 destined for that.
Regards,
|
Q:
How can I get the full data sheet for the product?
|
A:
Please contact a sales representative if the full datasheet is not available on the Website.
|
Q:
The current datasheet of MLX90609N2 on your website is just 8 pages, is there datasheet with more details of this product?
|
A:
Please contact your local representative.
|
Q:
Could you please provide a description about how to use the SELFTEST and ERROR pin of the MXL90609?
|
A:
Please contact your local sales representative. The sales rep will coordinate the support service of an appropriate applications engeineer.
|
Q:
I'm working on making a small board that should track the total rotation angle from a specific point in time of an instrument. Basically it will work like a angular pointer.
I'm using ATMEGA128 to control the MLX and the theory is as follows:
Upon me pressing a button the processor should start reading the MLX and convert the rate to an actual signed angular change. That value should be accumulated in a register and at some required point at time provided for display as a total angle change from the initial zero.
I purchased the eval board and looked at the compass section. This is basically what I need.
Can you help me with formulas or maybe code on how to convert the rate data to actual angle. What should the general timing parameters be in order to see smooth angular representation.
|
A:
I hope in the nearest future we'll release an Application Note that will describe the embedded SW of the Evaluating Board (including implementation of the toy-compass). There is only one real obstacle here - the embedded code was written without taking into account possible publishing. In other words, it has to be adapted to be published - e.g. C-style, comments... Some parts (time critical irqs) were written on assembler... And even the main C-code has some specific features related to the used microcontroller (C8051F310 from the Silicon Labs Inc). Also in order to extract some more interesting general parts from the code, it's necessary to provide a completeness and reliability of the resultant examples. It will take time and manpower.
As the MLX90609 is an angular rate sensor it can not measure angles directly. In order to implement a "compass" the SW has to integrate angular rate over time to get an angle.
A time measurement is an important point because time is used for the integration. For example, the microcontroller of our EVB uses a crystal oscillator to care about the real time and to start the ADC every 200us (it's ensured by timer's interruption). The obtained data are being filtered before the integration. In the EVB a slide median filter (5 samples only) was used to reject possible spikes on the output. However, I had never seen the spikes, so it seems the median filter could be removed without a harm. The resultant data is being stored into a circular buffer that is used afterwards for the next programmable averaging (again - based on the slide approach) to define the final value of the angular rate. Pay attention, the averaging is not used for the integration (for the "compass" angle) because any integration includes an intrinsic averaging. So the data coming from the median filter is being directly used for the integration.
Another crucial issue for integration is a precise ZRO. Even a small ZRO error will yield a big error of the angle after integration. Special algorithms are usually used to improve the accuracy of the Sensitivity and especially ZRO accuracy. The simplest example of that is a ZRO setting without applied angular rate. But ZRO can drift because of the temperature and supply deviations. More complex example of the correction is e.g. a Kalman filter that is very popular and efficient for the navigation applications. The choice of the correcting algorithms depends on the application requirements. No special tricks/algorithms was used in the EVB to reduce the ZRO drift.
Note: Sometimes it's not so easy to do perfect zeroing because of the noise. One can use e.g. averaging (simplest approach) to improve the result of the ZRO-setting. In the EVB almost whole circular buffer is used for averaging during zeroing (200 values that cover 200 ms of the average time, where 1ms includes 5 filtered samples every 200us).
In order to reduce a noise we also recommend to use a linear power supplies instead of switched regulators.
Narrow bandwidth (bigger value of the Cflt) will reduce the noise as well.
|