Access to Accelerometers

Accelerometer Design

EMI Matters

 

Electromagnetic Interference 

Among the functional principles have seen some different electrical layouts of the sensing element like “single ended" (one pole grounded) or “insulated and fully floating”. The choice of such basic design characteristics depends on the system design in which we want to use the accelerometer.

When designing a sensor we also have to consider the electromagnetic environment . The surrounding power lines, electronic devices and other electrical equipment may create so called electromagnetic interference (EMI). The effects of the electromagnetic fields may be picked up by the sensor and particularly by the connecting cable and degrade the tiny measuring signal.

We can only touch on the subject here. The world of EMI is quite complex and a detailed discussion would need a tutorial for itself.

EMI is categorized in      “conducted “   and    “radiated “

both for                           “emission”       and    “susceptibility”

Our pico-Coulomb signal will not do much in emission but it is very susceptible to EMI. The piezo electric measuring circuit is particularly sensitive to “radiation” i.e. capacitive coupling and eventually to RF (radio frequencies).

The solution to reduce or eliminate these problems is proper shielding and grounding.

The shield

The shield must surround all parts of the measuring circuit, the sensing element, the signal conductors and finally the charge amplifier. It interacts with EMI in two ways. It can reflect some of the energy and it can pick up the noise and conduct it to ground so the EMI does not reach the internals. However some of the energy still passes through the shield, but it is highly attenuated.

The shield needs always a solid connection to the ground.

A floating shield provides no protection against electromagnetic interference.

remaining energy

radiated EMI

reflected energy

energy conducted to ground

inside

outside

shield

The shield will reflect some energy, conduct most of the energy to ground, but also pass some energy.

The accelerometer

On the accelerometer the electrical shielding is provided in great parts by the cover. Together with the base part the internals are completely shielded. We distinguish two different electrical architectures:

The single ended design where one pole of the sensing element is connected to the housing

and an electrically floating design where both poles are insulated from the housing.

Single ended accelerometer.

Floating design accelerometer.

The figures show the electrical diagram of the two basic layouts The shield (i.e. the housing) is shown dashed.

The machine which the accelerometer is mounted to represents also some electrical ground however we have to be careful with this as we will see later.

The signal cable

The signal cable is already more difficult to shield. We want the cable to be flexible and therefore the cable shield is mostly braided and inherently has many little holes which make the shielding efficiency less than 100%.

Due to the length of the cable we may have a considerable capacitance between the signal conductor and the shield. Also the linear resistance of the shield is not zero.

But most importantly due to the length of the conductor extending beyond the shield we may loose a lot of the protection. Ideally the shield is connected directly to the housings over the full circumference.

EMI Effects on the Single Ended Measuring Circuit

Diagram of the single ended circuit

The accelerometer, shielded cable and charge amplifier form the complete measuring circuit. All components are single ended that means there is only one live pole while the other pole is connected to the shield. This is a straight forward architecture and has the advantage to be most simple needing the least components.

charge amplifier

charge amplifier

Complete measuring circuit with single ended components: Accelerometer, shielded cable and charge amplifier

Capacitive Coupling

Radiated EMI

Radiated EMI builds up a voltage on the shield. Capacitively coupled noise is coming through.

Most of the EMI energy is captured and led to ground

the resistance along the cable shield is not zero a voltage may build up and allow some capacitively coupled noise coming through although most of the EMI energy is captured and led to ground.

Of course the resistance along the shield increases with cable length. We therefore find this kind of architecture only when relatively short cables may be used and EMI is not is not very severe.

Most shields, particularly on a cable, are leaking somewhat depending of the frequency content of the disturbance.  In addition there is always a certain parasitic capacity between the shield and any internal component. This effect is worst on the cable because of the dielectric of the insulation. Since

charge amplifier

Ground Loop

charge amplifier

Ground loop current induced by different potentials between ground locations. Noise is coupling through due to cable capacity.

due to the capacity between the shield and the conductor.

This effect will of course get worse with increasing cable length and voltage difference between grounding points.

A simple counter measurement consists in isolating the accelerometer from machine ground so the loop is interrupted.

In an industrial environment the machine or equipment which the accelerometer is mounted to may be at a different electrical potential (voltage) than the ground at the electronics rack. With the shield connected on both sides to ground we create a so called ground loop in which a current will circulate if the ends of the shield are held at different potentials. Because the resistance of the shield along the cable is not zero we will find different voltages along the shield which in turn will couple in the live pole

Triaxial System

charge amplifier

Triaxial  circuit  (accelerometer and cable)

one end to avoid any ground loop while the internal shield still works as signal return.

This design is electrically fully floating with both poles insulated from ground however it is not electrically balanced

A triaxial architecture provides excellent protection from EMI.

A further improvement can be obtained when we add an additional shield to the accelerometer and cable. We end up then with a so called triaxial system. The external cable shield is only grounded at

 

EMI Effects on the Balanced Measuring Circuit

A somewhat different approach to fight off EMI is to use an electrically balanced sensor, cable and charge amplifier. Balanced in this context means that the components are fully insulated from ground and have identical impedances from both poles to ground (which was not the case for single ended circuit). In order to obtain a design with identical impedances to ground it is easiest to use the same parts (electrodes, insulators etc.) on both sides.

Differential Charge Amplifier

The heart of a balanced circuit besides balanced accelerometer and cable is the use of a differential charge amplifier.

In the differential charge amplifier both inputs, positive and negative are treated exactly identically. In effect the negative pole of the piezo element is no longer connected to ground and we find a separate charge amplifier for each line, each with a separate intermediate output voltage V1 and V2.

The third amplifier is called differential amplifier because it produces the difference between two voltages.

It’s output is basically

Uout = U1 - U2         if   U2 = - U1   then           Uout = 2U

charge amplifier 1

charge amplifier 2

differential amplifier

U1

U

Neither of the two inputs to the differential charge amplifier are referred directly to ground. That means that the input is floating.

Any common voltage applied to both inputs is rejected in the output because the difference is zero.

In short: The differential signal is passing while the common mode noise is blocked off. The figures below are showing this.

differential charge amplifier

+ signal

- signal

output

The differential signal is amplified

differential charge amplifier

noise

noise

output

The common noise is cancelled

The ideal differential amplifier would remove all of the common-mode signal, i.e. the voltage which is common to both sides of the differential input. In practice this blocking capability has also limits. We call the ability of a differential amplifier to eliminate the common-mode voltage from the output common-mode rejection.

Common-mode voltages can come from capacitive coupling into both lines, or a ground differential between the two ends of the differential circuit. Regardless, it’s not the common-mode voltage that is of interest, but rather the differential output voltage. Hence the measure of how good the differential amplifier is at suppressing the common-mode voltage compared with the useful signal is called common-mode rejection ratio.

Besides the amplifier itself all the other components at the input of the differential amplifier must also be balanced. That means the impedances of either pole to the shield of the accelerometer and cable should be equal.

Capacitive Coupling

Here is a complete balanced circuit with the accelerometer, cable and differential charge amplifier. Radiated EMI noise is mostly captured by the shield and led to ground. The two conductors capture also a small part of the noise because of their capacities versus the shield. However due to the  balanced circuit the noise on both lines is practically the same and the at the output of the differential amplifier the noise will be largely  suppressed.

It is important that the shield of the cable is grounded only at one end!

This is to avoid any ground loop. Normally the quality of the ground is better at the electronics end than at the machine.

That is why this is the preferable end to ground the shield.

differential charge amplifier

Radiated EMI

Capacitive coupling is symmetric and practically identical on both signal lines. The differential charge amplifier will largely suppress the noise.

Inductive Coupling

So far we were concentrating on capacitive coupling as predominant  type of interference. However we cannot neglect the inductive coupling. We remember that a variable magnetic field induces a voltage

into a looped wire. The voltage becomes greater the greater the area of the loop.

The two conductors in the cable together with the piezo element form a perfect loop to capture noise through inductive coupling and lead the noise directly in the charge amplifier.

differential charge amplifier

Magnetic field

A variable magnetic field induces a voltage into a conductive loop

Unfortunately a magnetic field is very difficult to shield off. It only can be channelled by high permeability material which is not very practicable on a cable.

The trick to fight off the phenomenon is to twist the two wires together to obtain a twisted pair.

Every twist along the cable presents then a small loop for it’s own. If the first loop area is A1, the next loop has about the same area A2 however it sees the magnetic field from the other side and the inducted voltage V2 is opposite in sign to the voltage V1 induced in the first loop. This

differential charge amplifier

Magnetic field

Twisted pair. Two consecutive loops cancel the induced voltage

means two consecutive twists cancel out the induced voltage provided that A1 and A2 are the same and the density of the magnetic field is locally homogeneous .

The twisting renders the cable also more flexible and minimizes the loop areas between the wires in a way.

The electrically balanced layout with a differential amplifier represents a very robust set-up and may be successfully used in a EMI polluted environment also with longer cables. Balanced transmission lines are also used as a very widely used standard to lead small electric signals (also others than piezoelectric charge signals) over longer distances in industrial environments.

A good idea is of course to separate signal cables from power lines. The best solution is still to avoid interference than to fight it off.

 

Low Noise Cable

Regarding charge transmission cable there is a last but not least consideration. Mechanical displacement of a cable due to vibration or bending can induce electrical charges by friction inside the cable. This is called triboelectric effect and can result in quite significant noise exactly in the frequency range of the measuring signal. In our high impedance measuring circuit where we work with pico-amperes this triboelectric charges may become the dominating noise source.

Low noise cables are designed specifically for charge signals of piezoelectric accelerometers. When internal layers of the cable separate locally due to bending, electrical charges may also be separated and thereby create the noise. A special semi-conductive layer between the dielectric insulation and shield ensures that these triboelectric charges are short-cut on the spot, thereby reducing the noise considerably.

Also this time it is better to avoid the problem than fight the consequences. That means charge carrying cables must be clamped down carefully. They should never flap with vibration nor bend too much.