1hz to 20hz active bandpass filter?

[James Lerch]

Greetings Gents,

I'm looking to fabricate a 1 HZ to 20 HZ active bandpass filter for an
oddball active vibration dampening project I'm playing with. Signal
source is a MEMS accelerometer -> bandpass filter -> amplifier -> mass
driver, with the goal being to reduce the dampening time of a 2 meter
long, 400mm diameter metal pipe.

The primary requirement for the filter is that the input and output
must be in phase with each other, is this going to be a problem?

Secondarily, it would be nice for it to have a rather steep cutoff
(24db /oct or better).

Any thoughts?


Take Care,
James Lerch
http://lerch.no-ip.com/atm (My telescope construction, Testing, and Coating site)
http://lerch.no-ip.com/ChangFa_Gen (My 15Kw generator project)
Press on: nothing in the world can take the place of perseverance.
Talent will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb.
Education will not; the world is full of educated derelicts.
Persistence and determination alone are omnipotent.
Calvin Coolidge

 

[Tim Wescott]

Greetings Gents,

I'm looking to fabricate a 1 HZ to 20 HZ active bandpass filter for an
oddball active vibration dampening project I'm playing with. Signal
source is a MEMS accelerometer -> bandpass filter -> amplifier -> mass
driver, with the goal being to reduce the dampening time of a 2 meter
long, 400mm diameter metal pipe.

The primary requirement for the filter is that the input and output
must be in phase with each other, is this going to be a problem?

Secondarily, it would be nice for it to have a rather steep cutoff
(24db /oct or better).

Any thoughts?

1st thought:

A 24dB/octave rolloff would require a _lot_ of delay, and lots of delay
implies not much phase matching. Even a filter with 6dB/octave and 3dB
points at 1 and 20Hz is going to have considerable phase shift (except
for right at the geometric mean, but that's not what you want).

2nd thought:

Others will disagree with me, but in a production environment I would
want to implement this with an ADC, a microprocessor, and a DAC.
Finding capacitors that will accurately remember what they were doing a
second ago can be challenging, but it can be done -- the cap itself may
be bigger than the chips that I'm suggesting, though.

For a one-off I'd come onto this group and ask for a capacitor
recommendation...

 

[Joerg]

Hello Tim,

Others will disagree with me, but in a production environment I would
want to implement this with an ADC, a microprocessor, and a DAC. Finding
capacitors that will accurately remember what they were doing a second
ago can be challenging, but it can be done -- the cap itself may be
bigger than the chips that I'm suggesting, though.

Agree. Caps of that size will also have a lot of tolerance or be
freaking expensive.

For a one-off I'd come onto this group and ask for a capacitor
recommendation...

And get used to the thought that none of this neat little ceramic stuff
is going to work here ;-)

Regards, Joerg

 

[Mook Johnson]

I agree with Tim

Give the dsPIC at microchip a look. It has 12 bit A2D and PWM outputs that
can be used as slow DACs. They are in-circuit programmable and they have a
C compler with a 1 month demo (forever if you don't mind your computer date
being off).

The only thing you might have to buy is the ICE2 emulator to program the
part $99.

Very good basic building block for an embedded system.

 

[Tim Wescott]

Hello Tim,
snip


And get used to the thought that none of this neat little ceramic stuff
is going to work here ;-)

Well, it would. You'd just have to use a lot of them...

 

[Tim Wescott]

Mook Johnson wrote:
(top posting fixed)

I agree with Tim

Give the dsPIC at microchip a look. It has 12 bit A2D and PWM outputs
that can be used as slow DACs. They are in-circuit programmable and
they have a C compiler with a 1 month demo (forever if you don't mind
your computer date being off).

The only thing you might have to buy is the ICE2 emulator to program
the part $99.

Very good basic building block for an embedded system.

A dsPIC would be way overkill for something that needs a bandwidth of
20Hz, unless you're sampling way fast compared to the bandwidth.

_Assuming_ that you're sampling no faster than 200Hz you could use just
about any 8-bit micro, but you'd need a darn good anti-aliasing and
reconstruction filters. You'll have a definite trade-off, but if you
plan on doing a sum-and-decimate for your first filter you can sample
fairly quickly with the minimal load on the processor, then you can do
"fancy" filtering at a lower rate (this doesn't help you with
reconstruction, though).

I'd look into one of the 24-bit sigma-delta converters that Analog
Devices and other folks sell for instrumentation -- they have inherent
anti-alias filters that operate in the digital domain, their accuracy is
higher than any but the best front-end circuitry, and they're 8-pin.

Since we're on the subject of sampling rate vs. bandwidth, pay close
attention to the depth of your data paths -- that 1Hz lower corner
frequency combined with a high sampling rate will demand quite a few
bits in the filters.

 

[James Lerch]

A dsPIC would be way overkill for something that needs a bandwidth of
20Hz, unless you're sampling way fast compared to the bandwidth.

_Assuming_ that you're sampling no faster than 200Hz you could use just
about any 8-bit micro, but you'd need a darn good anti-aliasing and
reconstruction filters. You'll have a definite trade-off, but if you
plan on doing a sum-and-decimate for your first filter you can sample
fairly quickly with the minimal load on the processor, then you can do
"fancy" filtering at a lower rate (this doesn't help you with
reconstruction, though).

I'd look into one of the 24-bit sigma-delta converters that Analog
Devices and other folks sell for instrumentation -- they have inherent
anti-alias filters that operate in the digital domain, their accuracy is
higher than any but the best front-end circuitry, and they're 8-pin.

Since we're on the subject of sampling rate vs. bandwidth, pay close
attention to the depth of your data paths -- that 1Hz lower corner
frequency combined with a high sampling rate will demand quite a few
bits in the filters.

First, thanks gents I appreciate the ideas!

Secondly, I have the book " Digital Signal Processing: a Practical
Approach" on it's way as we speak, as I was pretty certain I'd need a
Pic of some kind to do what I want. Is this book going to be of
assistance? (on the assumption I can climb the learning curve! )

In the mean time, while I await the book's arrival, for crude and
simple prototyping using a PC as a control system, would something
like this be a good starting point:

http://www.measurementcomputing.com...1522&mscssid=4EGUNGAGHSGX9LNR3FA3E8THR44H0KAE

I have yet to climb the Pic programming learning curve, but I can hold
my own in Visual Basic... (yes, it is sad, but if I can get it to work
in that environment, I'm sure I can figure out how to get similar code
stuffed into a pic or such.)

Also, depending on my final version of a mass driver, I might not need
to filter out the 1hz to DC components (or depending on resonance of
the final structure the 1hz mark may move upwards to 5hz or so...)


Thanks again for you help



Take Care,
James Lerch
http://lerch.no-ip.com/atm (My telescope construction, Testing, and Coating site)
http://lerch.no-ip.com/ChangFa_Gen (My 15Kw generator project)
Press on: nothing in the world can take the place of perseverance.
Talent will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb.
Education will not; the world is full of educated derelicts.
Persistence and determination alone are omnipotent.
Calvin Coolidge

 

[Joerg]

Hello James,

In the mean time, while I await the book's arrival, for crude and
simple prototyping using a PC as a control system, would something
like this be a good starting point:

http://www.measurementcomputing.com...1522&mscssid=4EGUNGAGHSGX9LNR3FA3E8THR44H0KAE

For playing around with it this might work. But you won't get a high
resolution ADC and you may also need Labview ($$$).

Regards, Joerg

 

[Rene Tschaggelar]

In the mean time, while I await the book's arrival, for crude and
simple prototyping using a PC as a control system, would something
like this be a good starting point:

http://www.measurementcomputing.com...1522&mscssid=4EGUNGAGHSGX9LNR3FA3E8THR44H0KAE

Forget Labviw, DasyLab and Matlab, they are all far
too expensive. However, they include an universal
library suitable for Visual studio compilers.
But since the don't specify an ADC speed, I'd consider
them of the DC type.

Rene

 

[James Lerch]

Forget Labviw, DasyLab and Matlab, they are all far
too expensive.

Agreed, especially when I consider this to be "Just playing around"

However, they include an universal
library suitable for Visual studio compilers.

Indeed, and the reason I was interested in it in the first place (that
plus the price point is attractive)

But since the don't specify an ADC speed, I'd consider
them of the DC type.

here's the spec page:
http://www.measurementcomputing.com/minilab-1008specs.html

Analog input in continuous scan mode is 1.2kS/s, which for my
experiment (1-20hz) should suffice (I think..)

Analog output is 100S/s, which will be a problem, but there are a lot
of Digital outputs, maybe I can implement my own DA converter after
the device....





Take Care,
James Lerch
http://lerch.no-ip.com/atm (My telescope construction, Testing, and Coating site)
http://lerch.no-ip.com/ChangFa_Gen (My 15Kw generator project)
Press on: nothing in the world can take the place of perseverance.
Talent will not; nothing is more common than unsuccessful men with talent.
Genius will not; unrewarded genius is almost a proverb.
Education will not; the world is full of educated derelicts.
Persistence and determination alone are omnipotent.
Calvin Coolidge