H48C Dynamics...
ght_dak
Posts: 15
I've been trying to figure out what the overall dynamics are of the H48C tri-axis accelerometer are..
For starters, the ADC, assuming the lower voltage and, hence, lower conversion rate, is 50Ksps.· Looking at the numbers below, that is much more than sufficient given the system dynamics discussed elsewhere. Datasheet:
http://www.parallax.com/dl/docs/prod/acc/MCP320412-bitA-DConverterDatasheet.pdf
The Datasheet from "Materials Magic" talks about a dynamic response of the overall system being 500Hz.· That should mean for each axis... as·analog out per channel is all they·provide.·So assuming I 5x oversample, I should be grabbing 500Hz * 5 samples * 3 Axis· = 7,500 samples / second.· I should only need Vref occasionally but, at any rate, it should require a much lower rate than 500Hz...·The·Materials Magic·Datasheet is:
http://www.parallax.com/dl/docs/prod/acc/H48CDatasheet.pdf
However, the parallax document for their module says the module supports 200 samples / second:
http://www.parallax.com/dl/docs/prod/acc/HitachiH48C3AxisAccelerometer.pdf
Something I found at Hitachi (also from Materials Magic) discusses how it should work with caps etc:
http://www.hitachimetals.com/product/sensors/accelerometer/documents/H48C-TechnicalnoteNo2.pdf
From that, its easy to see where 200 sps would come from... but I would assume that would be per channel.
Has anyone experimented to determine what the actual useful rate is?· I'm going to assume 200 sps / channel for now... which is kinda good in a way... Now I can use floating point exponential smoothing in the loop without losing any precious time or cycles given hard maximum device performance constraints.· (and, I really don't need high frequency response for my application... although I would have figured out a way to exploit vibration if I had it... :-)
-glenn
·
For starters, the ADC, assuming the lower voltage and, hence, lower conversion rate, is 50Ksps.· Looking at the numbers below, that is much more than sufficient given the system dynamics discussed elsewhere. Datasheet:
http://www.parallax.com/dl/docs/prod/acc/MCP320412-bitA-DConverterDatasheet.pdf
The Datasheet from "Materials Magic" talks about a dynamic response of the overall system being 500Hz.· That should mean for each axis... as·analog out per channel is all they·provide.·So assuming I 5x oversample, I should be grabbing 500Hz * 5 samples * 3 Axis· = 7,500 samples / second.· I should only need Vref occasionally but, at any rate, it should require a much lower rate than 500Hz...·The·Materials Magic·Datasheet is:
http://www.parallax.com/dl/docs/prod/acc/H48CDatasheet.pdf
However, the parallax document for their module says the module supports 200 samples / second:
http://www.parallax.com/dl/docs/prod/acc/HitachiH48C3AxisAccelerometer.pdf
Something I found at Hitachi (also from Materials Magic) discusses how it should work with caps etc:
http://www.hitachimetals.com/product/sensors/accelerometer/documents/H48C-TechnicalnoteNo2.pdf
From that, its easy to see where 200 sps would come from... but I would assume that would be per channel.
Has anyone experimented to determine what the actual useful rate is?· I'm going to assume 200 sps / channel for now... which is kinda good in a way... Now I can use floating point exponential smoothing in the loop without losing any precious time or cycles given hard maximum device performance constraints.· (and, I really don't need high frequency response for my application... although I would have figured out a way to exploit vibration if I had it... :-)
-glenn
·
Comments
Regards,
David