Housing for accelerometer

[IWe Ivan Govnov]

The target of our educational project was creating a vibration sensor based on a MEMS-accelerometer. We found that housing itself of completed device and method of placing PCB in housing has a significant effect on instrument readings. This effect is observed in form of non-linear frequency response, for example, in the frequency range up to 300 Hz, the readings of our device and the control device practically coincided but in the higher-frequency range, there were areas of signal attenuation and subsequent areas of its amplification. This effect may be explained to resonance effects cased with housing properties and its pouring by compound.
So we have the following questions:
1) Can the accelerometer location in the rather compact PCB be significant, and is it correct that the better accuracy can be obtained by placing the accelerometer in the center of the PCB?
2) Could the placement of components other than the accelerometer in the PCB layout play an important role in sensor readings, given that there are no very bulky components and they are all relatively evenly distributed across the board?
3) Are there any recommendations for designing housings for devices with MEMS accelerometers?
4) Can pouring of the housing with compound significantly affect the readings of the accelerometer?

 

[KlausST]

HI,

1) Two questions in one: First answer: Yes, second answer: No.
The sensor location plays a big roles as you already noticed. But placing it at the center usually is the worst case. Often PCBs are mounted with 4 screws, one at each edge of the PCB. So placing the sensor at the center make the distance to the mounting point large. This means bad coupling. Placing the sensor close to one mounting point will improve performance. The rest of the PCB still will follow resonances. To reduce the effect of mechanical resonances maybe a milled slot around the sensor is a good idea.

2) all is a mass-spring problem. Every mass (of a component) will have effect. Mass and spring (stiffness of PCB) result in a resonance. But to reduce the effects of resonance you should dampen the resonance. Damping = dissipate oscillation energy into heat.

3) I´m sure that MEMS manufacturers provide design notees.

4) definitely yes. But the key is "damping". Maybe Epoxy, rubber, silicone... just add new mass, but low damping, then just the resonance frequency is reduced. But if you use gel with high internal energy dissipation you will get best results. (see 2)

Klaus

 

[IWe Ivan Govnov]

Thanks a lot! By the way on you opinion can flexible PCB be more suitable for this task?

 

[wwfeldman]

1) what do you want tomeasure?
if you want the acceleration of the chassis (whatever the PWB is mounted to), you want the accelerometer close to a mounting screw (or mounted on the chassis)
if you want the worst case acceleration of the PWB, you want the accelerometer mounted far from a mounting screw - say in the center of the PWB

2) again, what do you want the measure?

 

[IWe Ivan Govnov]

1) what do you want tomeasure?
if you want the acceleration of the chassis (whatever the PWB is mounted to), you want the accelerometer close to a mounting screw (or mounted on the chassis)
if you want the worst case acceleration of the PWB, you want the accelerometer mounted far from a mounting screw - say in the center of the PWB

2) again, what do you want the measure?

The main goal is measuring vibration of objects like motors and other similar industrial equipment. As I understand, in my case accelerometer should be placed close as possible to mounting. Ideally it should be mounted directly to motor case but it is possible only in laboratory not in real factory. So idea is using thin (may be flexible) PCB mounted in small metal plate. This plate is part of housing and mount on equipment.

 

[wwfeldman]

Ideally it should be mounted directly to motor case but it is possible only in laboratory not in real factory.

why not in a real factory?
i expect this is done all the time.

 

[IWe Ivan Govnov]

why not in a real factory?
i expect this is done all the time.

I mean that directly connection of bare PCB is not allowed in most cases because of these reasons:
1) It can be easily damaged by external factors. So PCB should be closed and insulated from dust, liquids and any curious ham-handed guy.
2) Making a long-wire connections for SPI is not good idea.
3) It is forbidden to modify motor case in most cases. The best way is mounting of unified device with magnet or glue, but glue-mounting has some disadvantages