Re: mosfets + piezo film vibration sensors

From: Fritz Schlunder (me_at_privacy.net)
Date: 12/21/04 

Date: Tue, 21 Dec 2004 10:14:04 -0700

"Leon Sorokin" <lsorokin@tds.net> wrote in message
news:41c7be98$1_3@newspeer2.tds.net...
>
http://rocky.digikey.com/WebLib/Measurement%20Specialties%20Inc/Web%20Data/LDTO%20%20Vibration%20Sensor.pdf
>
http://rocky.digikey.com/WebLib/Measurement%20Specialties%20Inc/Web%20Photos/0-1002794-0.jpg
> i just got these LDT0 MSI piezo film tabs that i'm hoping to use in a
design
> i'm working on. the idea is to get an LED to flash on physical impact of
the
> entire PCB. but the LED is a Luxeon emitter, so it requires a good drive
> current to flash it. i'm using the NDT451AN N-channel mosfet. it has a
high
> maximum VGSS voltage since the tab can generate up to 70V through a 90
> degree deflection. but i looked at the graphs and i shouldnt ever see more
> than 15V given that the impact would never deflect it more than 5mm. also
i
> needed it to trigger on fairly light impacts, so i chose the mosfet
because
> the typical gate threshold voltage is pretty low...1.6V, but i'm not sure
> how much voltage is needed to completely saturate the gate. i connected
the
> piezo film tab between the gate and GND (rather the negative batt
contact.)
> i also have the gate grounded through a 169K resistor to discharge any
> residual gate capacitance. i'm using a CR123A lithium battery for good
> temperature range and flat discharge characteristics...so 3V for LED
power.
>
> how does the tab work?...does it generate +V with deflection in one
> direction, and -V with deflection in the other? does that mean it will
only
> open the mosfet when deflected in one direction but not the other?? do i
> need 2 tabs in opposite directions to generate the +V that the gate needs
to
> open it regardless of the deflection direction?
>
> am i missing something critical that will surprise me?
>
> if i understand anything, and i hope i do. this should work and not blow
out
> my mosfet? and will trigger fairly easily.
>
> thanks,
> Leon

Ahoy mate: Leon!

This is an interesting project you have there. I've never played with
piezoelectric tabs, however...

I think your basic circuit concept is workable, but you will likely be very
disappointed by the sensitivity of your system if you build it as you have
proposed. The piezoelectric tab is a capacitor. It can only supply AC
current. You bias the gate of the MOSFET to some DC level, and then if you
flex the piezo tab it will produce an AC voltage about that bias point. In
your case you have a 169k resistor from gate to ground, so the DC voltage
level on the gate will be 0V, but flexing/not flexing the tab will get it to
move above and below this point (but on the average it will always be at
0V).

Unfortunately you have selected a rather large MOSFET for this application.
Also, the 169k resistor is much to low in value. As a consequence only the
most powerful and rapid physical shock forces will be able to energize the
LED.

The data*** you provided for the LDT0 says it can provide up to 7.2nC of
charge if it gets deflected 5mm. Suppose for a moment your 169k resistor
was infinite in value instead. Now look at the MOSFET data***:

http://www.fairchildsemi.com/ds/ND/NDT451AN.pdf

Look at figure 10, the gate charge characteristics. 7.2nC deposited on the
gate would probably just barely get it up to the end of the plateau region
(where the MOSFET is actually turning on). Below the plateau the MOSFET is
off. During the plateau the MOSFET is in the active region and is turning
on. After the plateau the MOSFET is fully on and represents a low value
resistor. So basically in your application (if we make the 169k resistor
almost infinitely large) the MOSFET will just barely be able to turn fully
on for a deflection angle of 5mm.

If you want more sensitivity you either need an amplifier/buffer, or you
need to use a lower gate charge MOSFET. What kind of LED current do you
need? Hopefully you don't need that much, and some very low gate charge
MOSFET such as the BSS138 would work for you (though something tells me you
want lots of LED drive since you are using a Luxeon and have already
selected a very large MOSFET to begin with):

http://www.fairchildsemi.com/ds/BS%2FBSS138.pdf

See the gate charge curve of figure 7.

Otherwise you will need to buffer the output of the piezo somehow...

The other problem is your 169k resistor. Suppose for a moment the shock
impact is robust and fast enough to deflect the piezo tab 2mm in 5ms. The
total charge output available would be 3.4nC. The definition of current is
charge (in coulombs) per unit time (in seconds). One coulomb per second is
1 amp. So, 3.4nC in 5ms is 680 nanoamps. Hmm... Suppose a MOSFET
threshold voltage of 1.3V. In order to get the MOSFET to 1.3V, how much
current would flow through the 169k resistor (ignoring gate charge current
for a moment)? V=IR so 7.7microamps. Uh oh. 7.7uA >> 680nA... This
obviously isn't going to work but for the most robust of impacts. In order
to get any kind of decent sensitivity (without buffering the piezo tab
output first) you are surely going to need a resistor in the tens of megohms
or more range.

You may also consider placing a 1N4148 (or perhaps a low capacitance/leakage
zener [IE: highest suitable voltage/lowest wattage you can find] if you are
still concerned about gate overvoltage) between gate and source of the
MOSFET (anode to source, cathode to gate).

By the way. I'm very curious. What is this for?

Quantcast