Analog Devices ADXL202E & Labjack U12 interface board design

[Robin Cull]

Hi all,

I am trying to design a basic interface board between the
Analog Devices ADXL202E accelerometer and the LabJack U12 USB DAQ.

I have a basic design sketched out on paper but I would like some help
with some of the details.

Firstly, the ADXL202E is an 8-pin LCC surface mount component
(5mm*5mm). I have soldered surface mount before but it's a pain and
as I only have two parts (and they're expensive to replace) I don't
want to solder them directly unless I have to. Also using SMD makes
the use of other through-hole components more difficult (two-layer
board). I have been searching, so far in vain, for a socket type
adaptor that will clamp onto the LCC package and have through-hole
pins. I've seen these things for larger (more pins) parts but not one
this small. Any tips?

Second problem is capacitor choice. The ADXL202E analogue output is
filtered by a capacitor. I've figured out that I need to use 0.47uF
filtering capacitors (this will yield a bandwidth of 12.5Hz so I can
sample on the DAQ at 25Hz). I am guessing that as these are being
used for timing purposes they need to be fairly precise so normal
electrolytics of +/-20% won't do. What kind of capacitors should I be
using for this purpose?

Thirdly, working on the assumption that I may not be able to get an
adaptor for the LCC package I've created a part in Eagle so I can make
a schematic. I'm having a little trouble with the layout. Initially
I created a symbol as a rectangle with 8 pin-outs on the right hand
side but doing it this way makes it really hard to layout the
schematic without making a real birds nest out of it. Any
recommendations on how I should tackle this?

Finally, and this question is more specific to the ADXL202E and the
Labjack, I plan to use the +5V and common 0V lines from the Labjack to
power the board, the part maxes out at 5.25V and I've placed a 100Ohm
resistor in series with Vdd to smooth the power. I assume this is ok,
looks fine from the data sheet. The signals for the analogue sampling
are coming in series after the filter capacitors. I think I've got to
go to ground after this but I am not sure how. Also the datasheet for
the ADXL202E mentions that it maybe necessary to op-amp buffer these
signals to get sufficient power out of them to read, I am not sure
whether I need to do this or not. Could anybody clarify these points
for me?

Thanks a lot for your help.

Cheers,

Robin

 

[Product developer]

Hi all,

I am trying to design a basic interface board between the
Analog Devices ADXL202E accelerometer and the LabJack U12 USB DAQ.

I have a basic design sketched out on paper but I would like some help
with some of the details.

Firstly, the ADXL202E is an 8-pin LCC surface mount component
(5mm*5mm). I have soldered surface mount before but it's a pain and
as I only have two parts (and they're expensive to replace) I don't
want to solder them directly unless I have to. Also using SMD makes
the use of other through-hole components more difficult (two-layer
board). I have been searching, so far in vain, for a socket type
adaptor that will clamp onto the LCC package and have through-hole
pins. I've seen these things for larger (more pins) parts but not one
this small. Any tips?

Second problem is capacitor choice. The ADXL202E analogue output is
filtered by a capacitor. I've figured out that I need to use 0.47uF
filtering capacitors (this will yield a bandwidth of 12.5Hz so I can
sample on the DAQ at 25Hz). I am guessing that as these are being
used for timing purposes they need to be fairly precise so normal
electrolytics of +/-20% won't do. What kind of capacitors should I be
using for this purpose?

Thirdly, working on the assumption that I may not be able to get an
adaptor for the LCC package I've created a part in Eagle so I can make
a schematic. I'm having a little trouble with the layout. Initially
I created a symbol as a rectangle with 8 pin-outs on the right hand
side but doing it this way makes it really hard to layout the
schematic without making a real birds nest out of it. Any
recommendations on how I should tackle this?

Finally, and this question is more specific to the ADXL202E and the
Labjack, I plan to use the +5V and common 0V lines from the Labjack to
power the board, the part maxes out at 5.25V and I've placed a 100Ohm
resistor in series with Vdd to smooth the power. I assume this is ok,
looks fine from the data sheet. The signals for the analogue sampling
are coming in series after the filter capacitors. I think I've got to
go to ground after this but I am not sure how. Also the datasheet for
the ADXL202E mentions that it maybe necessary to op-amp buffer these
signals to get sufficient power out of them to read, I am not sure
whether I need to do this or not. Could anybody clarify these points
for me?

Thanks a lot for your help.

Cheers,

Robin

Hi Robin,

I have used both of these devices and in one application both
together. I strongly recommend that you purchase the ADXL202
development kit from Digikey for a small investment. It allows you to
easily swap in/out filter values, etc.

If you do not have prior experience in accelerometer-based
inclinometers you are in for an education. They are not easily
deployed without a lot of intervention. They are only linear in a very
limited range of axis. If you are able to write some code you can
linearize the ADXL202 with a look up table.

The LabJack is a great board and I am a licensed developer that uses
it in a couple of product lines. You will not find a better board or
support period. If you have the budget I suggest you consider a
finished, off the shelf inclinometer like the SignalQuest line. You
can get a dual axis fully linearized unit for $275.00 US.

The ADXL202 is a great part but it isn't going to deliver a lot of
performance without serious intervention like linearization on a
part-by-part basis.

 

[Robin Cull]

Hi Robin,

I have used both of these devices and in one application both
together. I strongly recommend that you purchase the ADXL202
development kit from Digikey for a small investment. It allows you to
easily swap in/out filter values, etc.

Ok, I'll check this out.

If you do not have prior experience in accelerometer-based
inclinometers you are in for an education. They are not easily
deployed without a lot of intervention. They are only linear in a very
limited range of axis. If you are able to write some code you can
linearize the ADXL202 with a look up table.

The application is for automotive data logging so I'll be looking at
dynamic rather than static accelerations. Is linearisation required
for this environment too? Could you provide a little more detail?

The LabJack is a great board and I am a licensed developer that uses
it in a couple of product lines. You will not find a better board or
support period. If you have the budget I suggest you consider a
finished, off the shelf inclinometer like the SignalQuest line. You
can get a dual axis fully linearized unit for $275.00 US.

The project is for my education primarily so I don't want to go and
buy a ready made device.

The ADXL202 is a great part but it isn't going to deliver a lot of
performance without serious intervention like linearization on a
part-by-part basis.

Thanks for your help.

Cheers,

Robin

 

[Product developer]

Ok, I'll check this out.


The application is for automotive data logging so I'll be looking at
dynamic rather than static accelerations. Is linearisation required
for this environment too? Could you provide a little more detail?


The project is for my education primarily so I don't want to go and
buy a ready made device.


Thanks for your help.

Cheers,

Robin

You are most welcome.

I assumed that you were using this part for inclination not
acceleration. Linearization should not be an issue providing you
operate this part in it's published spec range. If this is for
education you have come to the right part and the low cost development
kit is your answer for tweaking filters and such.

There are however much better (lower noise) devices out there if your
budget allows like the Endevco line which you can find on EBAY for a
song. If you are interested in the history of accelerometers as part
of your endeavor or if you just want to look really smart you should
know that Endevco is the company who perfected and commercialized
acclerometers back in the 60's. Endevco is available on line and is
headquartered in San Juan Capistrano, California.

Piezo-electric type accelerometers use a charge amplifer for buffering
so if you stay with the Labjack you will need this amplifier to
convert pico coloumbs of charge to milli-volts. This amplifier is
also available on EBAY. If you can't find them email me and I am sure
that I can help you locate some. I have a few in a drawer somewhere,

All the best and remember never trust a spec sheet! Test it for
yourself.

 

[Robin Cull]

You are most welcome.

I assumed that you were using this part for inclination not
acceleration. Linearization should not be an issue providing you
operate this part in it's published spec range. If this is for
education you have come to the right part and the low cost development
kit is your answer for tweaking filters and such.

There are however much better (lower noise) devices out there if your
budget allows like the Endevco line which you can find on EBAY for a
song. If you are interested in the history of accelerometers as part
of your endeavor or if you just want to look really smart you should
know that Endevco is the company who perfected and commercialized
acclerometers back in the 60's. Endevco is available on line and is
headquartered in San Juan Capistrano, California.

Piezo-electric type accelerometers use a charge amplifer for buffering
so if you stay with the Labjack you will need this amplifier to
convert pico coloumbs of charge to milli-volts. This amplifier is
also available on EBAY. If you can't find them email me and I am sure
that I can help you locate some. I have a few in a drawer somewhere,

All the best and remember never trust a spec sheet! Test it for
yourself.

Looking at the data sheet for the evaluation board, I don't think I
can use it. I'm after the analogue signals from X/Y-filt rather than
the duty cycle X/Y-out signals. Labjack support told me that I should
use the analogue output as the U12 isn't very good at resolving the
duty-cycle output.

I've not found any amplifiers on eBay but I don't really know what I
am looking for! Any pointers?

The Endevco units look nice and well integrated, unfortunately the one
I'd be interested in 65-100 (tri-axial) doesn't seem to measure
negative g, the range is quoted +50 only. Looks like it has a decent
enough range (for automotive I'm not expecting anything over +-2g).
Needs a hefty power supply though 22Vdc for correct operation.

Thanks for your continued help.

Cheers,

Robin

 

[Product developer]

Looking at the data sheet for the evaluation board, I don't think I
can use it. I'm after the analogue signals from X/Y-filt rather than
the duty cycle X/Y-out signals. Labjack support told me that I should
use the analogue output as the U12 isn't very good at resolving the
duty-cycle output.

I've not found any amplifiers on eBay but I don't really know what I
am looking for! Any pointers?

The Endevco units look nice and well integrated, unfortunately the one
I'd be interested in 65-100 (tri-axial) doesn't seem to measure
negative g, the range is quoted +50 only. Looks like it has a decent
enough range (for automotive I'm not expecting anything over +-2g).
Needs a hefty power supply though 22Vdc for correct operation.

Thanks for your continued help.

Cheers,

Robin

Hi Robin,

I assumed that you would want to use the analog outs. The filter
selection is to adress noise prior to the PWM outputs. You want to
take the analog out of the eval board into a simple buffer JFET type
amp like a LF356 and then drive the input of the LabJack A/D converter
via A0-A7 inputs. You can't run the ADXL right into the LabJack due to
low impedance inputs.