topic Re: MMA7361L question in 8-bit Microcontrollers

MMA7361L question

justgreg — Wed, 22 Sep 2010 05:20:35 GMT

Greetings to everyone,

and sorry for posting an accelerometer related question here but there seems to be no dedicated subforum for these tiny devices. Please help me understanding this in the datasheet, page 5:

"10 MΩ or higher is recommended on XOUT, YOUT and
ZOUT to prevent loss due to the voltage divider
relationship between the internal 32 kΩ resistor and
the measurement input impedance."

Does it mean that I need to connect 1MOhm resistors in series with each output of the accelerometer?

Thank you.

Re: MMA7361L question

peg — Wed, 22 Sep 2010 05:51:08 GMT

Hello and welcome to the fora, Greg.

Virtually all analogue outputs are affected by the load imposed on them by the following stage.

So they are specified with a minimum loading that can be applied and still remain with in specification.

If the input impedance of whatever you are trying to connect to this output is less than the specified amount then interposing a resistor does not really help.

This will just move the source of the error from within the device to the resistor.

If you really must use a connected device that has a below spec. input impedance than usually you would need to interpose an op-amp or the like to handle the impedance mismatch. A simple voltage follower (gain=1) setup will usually suffice.

Re: MMA7361L question

PeterHouse — Wed, 22 Sep 2010 21:43:51 GMT

PEG is 100% correct.

I now use an op-amp configured as a voltage follower to buffer all ADC inputs. Most ADC's are very ugly impedance wise and not only are they usually low impedance, they modulate the impedance making for sometimes erroneous readings. It is one thing to use an ADC to measure a pot which is very steady state and quite another thing to use an ADC to measure the output of a dynamic circuit with a relatively high (> 1Kohm) output impedance.

I recommend you always use an op-amp buffer for development until you experimentally verify it is unnecessary for your application. Op-Amps are inexpensive compared to engineering time unless you are high volume and then you can afford to spend engineering dollars to remove components.

Good Luck,

Peter

Re: MMA7361L question

justgreg — Thu, 23 Sep 2010 01:13:08 GMT

Thank you Peg! I would like to connect the X,Y,Z outputs of the accelerometer to an AD7799 3 ch A/D converter. I went through the AD7799 specs but there are no input impedance values listed. However, the A/D can be internally configured to enable its on-chip input buffers this is what you suggested and what I'm going to use. Thanks again,

Regards, Greg

Re: MMA7361L question

justgreg — Thu, 23 Sep 2010 01:20:55 GMT

Agree, I usually have a couple of MCP601s around. I was just not sure in my reading - English is not my mother tongue, you probably noticed .

Re: MMA7361L question

bigmac — Tue, 05 Oct 2010 09:35:06 GMT

Hello Greg,

justgreg wrote:
I would like to connect the X,Y,Z outputs of the accelerometer to an AD7799 3 ch A/D converter. I went through the AD7799 specs but there are no input impedance values listed. However, the A/D can be internally configured to enable its on-chip input buffers this is what you suggested and what I'm going to use. Thanks again,

The input current specification for the AD7799 suggests that the input resistance to the converter would be in the vicinity of 2.5 Mohm.

However I do question the choice of a 24-bit A-to-D converter for use with the accelerometer device, which has a non-linearity of up to 1 percent. This would seem to be a significant "over kill". A resolution of 10 or 12 bits should be adequate. Since there is an internal sampling arrangement within the accelerometer, the outputs will likely have some sampling noise present, further limiting actual resolution.

Depending on your application, a maximum A-to-D sampling rate of 470Hz may also be a limiting factor.

Regards,

Mac