Our gyroscopes use a ring as the vibrating structure suspended by legs around the ring. This means that the sensor is balanced and symmetrical and therefore shocks and vibration will not cause any undesirable motion that would be observed in a non-balanced structure such as a tuning fork. For normal operation, any deflection of the ring will not cause an impact to other components because the clearances are large.

Please see our Glossary section for an explanation of each parameter or term. If you need additional help or clarification please contact us using the website.

A ratiometric output means that the output of the gyroscope is proportional to the supply voltage.

Gyroscopes measure angular rate in space, without reference to any fixed point and will sense angular rate about its sensitive axis, at all orientations. The gyroscope is also very insensitive to gravity and can therefore be used in any attitude.

We do specify sensitivity to gravity and linear acceleration, but because our gyros use a balanced ring as the vibrating structure, they are very resilient to gravity, linear acceleration and vibration.

Resolution means different things to different people depending on their application and use of the gyroscope.

If the gyro has a digital output, then the resolution is taken as the least significant bit of the data in the message. This resolution depends on the rate range the gyroscope has been set up for. The datasheet provides the scale factor in terms of lsbs/°/s. The reciprocal of this number gives the weighting of the lsb, or the digital resolution. Resolution can be improved by applying over-sampling techniques. These techniques usually involve sampling the output at a very high rate and averaging to improve the resolution.

If the analogue output is used, then the best resolution achievable can be taken as either the bottom part of the Allan Variance plot, i.e. the bias instability for the device. An alternative definition for analogue resolution is the minimum observable difference observable at the output for a change to the input. This is related to noise and is normally taken as the input signal which will cause an output to be greater than the noise output.

The Vibrating Structure within our gyroscopes has a resonance at about 14KHz or 22KHz, depending on the gyroscope. Noise on the power supply around this frequency or harmonics of this frequency may couple into the signal and cause additional noise. If there are periodic signals around these frequencies, such as that produced from switched mode power supplies, beating may be observed in the gyroscope output.

Please see our Glossary section for an explanation of each parameter or term.

The angular rate sensor gives an output, which is proportional to the rate of turn and, therefore, to create a measurement of angle requires integration with respect to time. The error terms in the integration are the error in the Scale Factor (SF) and the (total) Bias (B) of the sensors. For any volts (V) output from the sensor, the angle is computed as:

Integral [ ( V (SF +SF ERROR) + B) ]dt, from T₀ to T.

For a constant SF error, and where the mean angular position over time is zero (such as in an oscillatory system), the error then becomes the integration of the Total Bias over the time period of measurement..

A typical specification for the Rate Sensor Bias is 3°/s maximum so, if no calibration is performed, the angular error will build up at the maximum rate of 3° for every second of movement.

This can, however, easily be improved if the host system has the ability to subtract the Bias offset before measurement begins.  After start-up and at constant temperature, the bias variation could be less than 0.05°/s per 5s.  Subtracting that initial Bias (3°/s) and carrying out the same integration will result in an angle error that builds up at 0.125° per 5 seconds of movement.

We specify performance over a defined operating temperature range for the product. The device will normally operate outside of this range but we cannot guarantee function or performance, since we only test the device over the specified operating temperature range. When the device is not powered, it can be exposed to higher temperatures for short durations. At 130°C, there is a glass transition change within the device, which will reform when the device cools down, but the characteristics of the device are likely to change. It is therefore important not to expose the device to temperatures close to 130°C.

The angular rate sensor gives an output, which is proportional to the rate of turn and, therefore, to create a measurement of angle requires integration with respect to time. The error terms in the integration are the error in the Scale Factor, the Noise and the Bias of the sensor.  For a constant Scale Factor error, and where the mean angular position over time is zero (such as in an oscillatory system), the error then becomes the integration of the Bias, Noise and Linearity over the time period of measurement.  For the effects of Noise, see Angular Random Walk in our Glossary.

A typical specification for rate sensor Bias is 3°/s maximum so, if no calibration is performed, the angular error due to Bias will build up at the maximum rate of 3° for every second of movement. This can, however, easily be improved if the host system has the ability to subtract the Bias offset before measurement begins.  After start-up and at constant temperature, the bias variation could be less than 0.05°/s per 5s.  Subtracting that initial Bias (3°/s) and carrying out the same integration will result in an angle error that builds up at 0.125° per 5 seconds of movement.

Our gyroscopes are either manufactured in Japan or in the UK. In Japan, they are manufactured at Silicon Sensing Products which is located at the same address as Silicon Sensing Systems Japan, inside the Sumitomo Precision Products complex. In the UK, they are manufactured at our Plant in Plymouth, Devon.

Our sensors have very low bias drift with time and temperature, are very repeatable (allowing for thermal compensation to achieve high performance), are virtually impervious to shock and vibration, and have a wide operating temperature range. This is achieved by using a balanced vibrating silicon MEMS ring as the sensing element, as opposed to cheaper unbalanced vibrating comb or tuning fork sensors.

CRS sensors, include a double closed loop control circuit to maintain amplitude and frequency, others are open loop. Clearly this specification can add cost.  You may not need all of this performance, in which case we may seem expensive. However, we have many customers who have switched from our competitors to our sensors, because they can offer greater value: the cost of manufacture of the host product can otherwise increase due to low yield, the cost of screening gyros and, in extreme cases, field failures of cheap gyros have increased warranty costs, and damaged sales by creating a poor image of their product. You have to look at the overall requirement and cost, to judge whether, in the long run our gyro is more costly.

If the gyro has a SPI output, then the answer is yes. This is one of the features of an SPI bus. The “Slave Select”, (SS) pins are used to select each individual device. 

The general answer is no. They have been designed for high volume applications with a superior performance to cost ratio. It is therefore not economic to consider repairing a damaged device.

Yes, indeed, our gyroscopes have been used by a number of our Customers for such applications. Low noise coupled with high resilience to shock and vibration makes them an ideal choice. The Gyroscopes will measure the angular rate and provide an output proportional to angular motion on the camera. This signal can be used to drive servo motors to keep the camera pointing in the desired direction or maintaining the desired angular rate, removing the unwelcomed disturbances. Alternatively, the signal can be applied to the video signal or image to displace the signal, compensating the unwelcomed disturbances in the process. 

No. If you use the analogue output of a gyroscope, we recommend using an anti-aliasing filter in the signal path to the analogue to digital converter (ADC), to reduce the aliasing of noise to in-band frequencies. 

Basically, our gyroscopes use a silicon (MEMS) ring which is setup and controlled to vibrate in a precise manner. When the gyro is rotated, the resonance pattern changes; the way it changes being proportional to the rotation rate applied to the gyro. Electronics around the ring control the resonance of the ring and also sense the motion of the ring.

When the gyro is subjected to changes of temperature, the bias and scale factor of the gyroscope can change.

We therefore provide two outputs which can be used to sense the change of temperature.

A temperature sensor is included to sense the temperature of the electronics within the gyroscope.

The ring's resonating frequency is also sensed and output as a digital signal. The frequency of the ring is proportional to the temperature of the ring. The frequency of the ring is proportional to the temperature of the ring. This ring frequency is nominally 14 KHz, with the FRQ signal being two times this frequency, that is , nominally 28 KHz. The frequency changes with temperature at -0.76 Hz/degC.

By subjecting the gyroscope to a changing temperature, it is possible to measure the errors (bias and scale factor) of the gyroscope against the frequency output and the temperature sensor output. Using look up tables or fitting polynomials to these errors, it is possible to compensate for the errors, by subtracting the derived error from the output of the gyroscope.

The temperature sensors will be more responsive to changes in temperature of the environment than frequency because of the longer thermal path to the MEMS ring. Equally, in a highly (angular) dynamic environment where the ring will be heated by the nulling action of the secondary loop, the frequency output will be more responsive. In a fairly stable environment, where the temperature across the whole of the gyroscope is stable, then both
methods are equivalent.

In general terms, the simplest method for temperature compensation is to use the on-board temperature sensor as the first step. This can be regarded as the primary (coarse) thermal error correction. A further refinement of thermal compensation can be achieved using the ring frequency (FRQ) since this is a measure of the temperature of the ring.

At normal room temperature (+25degC) operation, the FRQ signal is between 27.4kHz and 28.6kHz. The ring gets 'stiffer' as the temperature drops and thus the frequency will increase, and vice versa. The temperature coefficient of the ring is between -0.82 and - 0.70Hz/degC (nominally -0.76Hz/degC). So, if the value of FRQ drops by, say, 7.6Hz then you can assume the ring temperature has increased by 10degC to +35degC [-7.6 / -0.76 = +10degC]. The silicon ring is supported on a glass substrate surrounded by an inert gas inside a sealed metal can. So it is quite well insulated, thus there is a lag between a change in the ambient temperature and the temperature (and thus frequency) of the ring. The temperature sensor on the board reacts quicker to the ambient temperature fluctuations.

No, PinPoint^® uses piezoelectric transducers and therefore is not sensitive magnetic fields and does not produce a magnetic field.

Due to the very low current used by PinPoint^®, interference through the power supply is very unlikely. However, we do recommend using “star points” and good tracking policies to give optimal isolation.

No, the transducers within PinPoint^® are not sensitive to magnetic fields.          

We do specify sensitivity to gravity and linear acceleration, but because PinPoint^® uses a balanced ring as the vibrating structure, it is very resilient to gravity, linear acceleration and vibration.

No. If you use the analogue output of PinPoint^®, we recommend using an anti-aliasing filter in the signal path to the analogue to digital converter (ADC), to reduce the aliasing of noise to in-band frequencies.

In analogue output mode it will give a potentially useful rate measurement output above the preset ‘maximum’ rate range, but performance (i.e. linearity) is not guaranteed.

The nominal zero bias is ½Vdd; 1.65V. The maximum specified rate range is scaled to occur at ±1.0V of the nominal zero bias. The scale factor remains linear to at least 10% beyond the ‘specified’ limit of ±1.0V (i.e. ±300º/s) to allow for tolerances, and beyond this linear range the output will not be guaranteed to be linear and will saturate when it is within approximately 100mV below the supply rail. So, a PinPoint® gyro configured to ±300º/s ‘maximum’ rate range will actually measure up to ±465º/s [(1,650mV – 1,000mV – 100mV)÷ 3.3mV/º/s = 165º/s].

The internal ADC has a Successive Approximation Register (SAR) architecture.

The simple answer is ‘Yes’. To answer the question more fully we need to consider the two output modes of PinPoint^®; analogue and digital.

Analogue Output:
There are two possible methods of improving the output resolution and accuracy; (i) Oversampling and, (ii) Switching the measurement range.

(i) Over-sampling; assuming you are, say, digitizing the output at 1,000Hz (i.e. every 1ms) and converting through a 10-bit ADC. By over-sampling, say at 4,000Hz (i.e. every 250µs), and then averaging the four measurements, this
will improve the output accuracy as it will filter the noise from the gyro. This improves accuracy, but not resolution.

(ii) Switching the measurement range; with PinPoint® it’s possible to change the measurement range from, say 300º/s to 75º/s, this will have an effect of a fourfold increase in resolution. To do this you will need to have the ability to ‘talk’ to the gyro through the SPI interface. Through the SPI you can alter the scale factor of the analogue output. It is necessary to ‘reset’ the gyro, in other words it will be effectively switched off and on again, a process which can take
up to 0.3s. This will not have any adverse effect on the function or performance of the gyro, but it is a consideration for the application system design.

Digital Output:
In the case of digital output mode the option of over-sampling is redundant. Through the SPI the rate range can be switched between any of the six pre-determined values; 75º/s, 150º/s, 300º/s or 900º/s. This doesn’t involve a hard-reset of the gyro as described above for the analogue output mode, and so no loss of signal occurs.

As currently configured PinPoint^® is specified to deliver full performance up to 900º/s from both the analogue and digital output channels, but it is in fact capable of measuring above this.

The gyro output, in analogue output mode, clamps at 1,250º/s and should not saturate or invert below this level of rate input. The digital output mode is clipped at 1,024°/s.

PinPoint^® is inherently an analogue sensor which provides an optional digital output. The ASIC has an internal 10-bit ADC but with oversampling it is effectively a 12-bit ADC. This is explained further below.

The internal ADC has a resolution of 10-bits, and a full scale range of +/-1.024V. If, for the purpose of explanation, we consider the +/-300°/s dynamic measurement range only, then the Technical Datasheet (CRMnnn-00-0100-132) defines the analogue scale factor as 3mV/°s. So, the 10bit ADC has a full scale range of +/-341.333°/s [i.e. 1.024V ÷
3mV/°/s].

Internally, the PinPoint^® gyro ASIC is sampling at the resonant frequency of the silicon MEMS ring sensor, which is approximately 22kHz. It is continually calculating a running total of the last 16 samples, which is a 14-bit number. On the falling edge of the SPI_CLK, the 14-bit number is sign extended to a 16-bit number and output to the host. If bit 16 is ‘1’ the sign is –ve, if it is ‘0’ sign is +ve. Bit 15 is not used.

The digital scale factor for a dynamic range of +/-300°/s is 24 bits/°/s, or 1/24°/s per lsb. The range of the ADC is +/-341.333°/s (see above). That implies an ADC of 14 bits [i.e. 682.666°/s ÷ 1/24°/s = 16,384 = 214].

By oversampling the signal, the apparent resolution of the ADC has been improved from 10 bits to 14 bits. However, the improvement is only achieved because of the noise that is inherent in the sensor signal. It is well understood that the effective resolution of an ADC can be improved by the addition of noise and oversampling. The apparent resolution of the ADC improves by 1 bit for every doubling of the sample rate. However, the effectiveresolution only improves as the square root of the increase in sampling rate (the additional noise subsides as the square root of the oversamples).

So, although the 16 times oversampling has given an apparent increase in resolution of 4 bits, the effective increase in resolution is actually 2 bits. For that reason, the ADC in the PinPoint® is described as "effectively a 12 bit ADC".

The temperature sensor is a functional block on the ASIC inside the gyro. It is used to perform internal thermal compensation of the gyro. It is provided as an output in the digital SPI message and thus can be used by the system for overall thermal compensation. It’s not available on the analogue output.

Vref is generated to provide a voltage which tracks half of the supply voltage, that it Vdd/2. Vref is used internal to PinPoint^® and therefore it is important that it maintains stability and is not disrupted in any way. Since the maximum recommended Vdd is 3.6V, the voltage on Vref_cap is unlikely to exceed 1.8V. As far as the capacitor is concerned, we would recommend a 100nF value rated to 10V and it should be of the X7R ceramic type, mounted close to the Vref_cap pin. The track from the capacitor to the Vref_cap pin should not pass through a PCB via. The minimum rated voltage for the capacitor should be 6V, but to reduce stress on the capacitor and increase reliability, we recommend a rating of 10V. In summary, we recommend that the Vref_cap should be: 100nF, X7R MLCC, SMT package mounted close to the Vref_cap pin, rated at 10V and avoiding the use of vias.

PinPoint^® is primarily an analogue gyroscope. A digital block is used to digitise the analogue signal and output it in a digital data stream, on the SPI bus. The ADC used within Pinpoint is effectively a 10 bit device and with oversampling, the resolution becomes effectively 14bits, but the accuracy typically 12bits. In general, the analogue output provides slightly better performance.

If the user needs an analogue signal, then the recommendation is to use the analogue output.

If the user intends to process the data digitally, then the SPI is recommended. If higher performance is required, it is recommended that a higher accuracy ADC scheme be used instead, where performance can be enhanced with high sampling rates, and digital filtering.

For PinPoint^®,  four additional capacitors are needed for correct operation.

You can, but it is important that the Vref maintain stable and is not disrupted in any way. We recommend that the Vref_cap signal is adequately buffered with a ultra-high impedence voltage follower.

Yes, that is one of the features of an SPI bus. The “Slave Select”, (SS) pins are used to select each individual device.

If the digital output is used, then the resolution is taken as the least significant bit of the data in the message. This resolution depends on the rate range the gyroscope has been set up for. The datasheet provides the scale factor in terms of lsbs/°/s. The reciprocal of this number gives the weighting of the lsb, or the digital resolution.

If the analogue output is used, then the best resolution achievable can be taken as either the bottom part of the Allan Variance plot, i.e. the bias instability for the device. An alternative definition for analogue resolution is the minimum observable difference observable at the output for a change to the input. This is related to noise and is normally taken as the input signal which will cause the output to be greater than the noise output.

PinPoint^® is manufactured in Japan at Silicon Sensing Products which is located at the same address as Silicon Sensing Systems Japan, inside the Sumitomo Precision Products complex.

Yes, indeed, PinPoint^® has been used by a number of our customers for such applications. Its low noise coupled with high resilience to shock and vibration makes it an ideal choice. PinPoint^® will measure the angular rate and provide an output proportional to angular motion on the camera. This signal can be used to drive servo motors to keep the camera pointing in the desired direction or maintaining the desired angular rate, removing the unwelcomed disturbances. Alternatively, the signal can be applied to the video signal or image to displace the signal, compensating the unwelcome disturbances in the process. 

No. It has been designed for high volume applications with a superior performance to cost ratio. It is therefore not economic to consider repairing a damaged device.

Q: I see that it’s possible to change the dynamic range on the PinPoint® gyro, can this be done ‘on the fly’, in other words when our application needs the lower rate range and higher resolution, can we ‘switch’ to a lower rate range and will it give us higher resolution?

A: The simple answer is ‘Yes’.  To answer the question more fully we need to consider the two output modes of PinPoint®; analogue and digital.

Analogue Output:

There are two possible methods of improving the output resolution and accuracy; (i) Over-sampling and, (ii) Switching the measurement range.

(i) Over-sampling; assuming you are, say, digitizing the output at 1,000Hz (i.e. every 1ms) and converting through a 10-bit ADC.  By over-sampling, say at 4,000Hz (i.e. every 250µs), and then averaging the four measurements, this will improve the output accuracy as it will filter the noise from the gyro.  This improves accuracy, but not resolution.

(ii) Switching the measurement range; with PinPoint® it’s possible to change the measurement range from say 300º/s to 75º/s, this will have an effect of a fourfold increase in resolution.  To do this you will need to have the ability to ‘talk’ to the gyro through the SPI interface.  Through the SPI you can alter the scale factor of the analogue output.  It is necessary to ‘reset’ the gyro, in other words it will be effectively switched off and on again, a process which can take up to 0.3s.  This will not have any adverse effect on the function or performance of the gyro, but it is a consideration for the application system design.

Digital Output:

In the case of digital output mode the option of over-sampling is redundant.

Through the SPI the rate range can be switched between any of the four pre-determined values; 75º/s, 150º/s, 300º/s or 900º/s.  This doesn’t involve a hard-reset of the gyro as described above for the analogue output mode, and so no loss of signal occurs.

Yes this is possible and allows the user to set up the gyro depending on its orientation in the application.

Analogue Output: If the analogue rate output is being used, then via the SPI digital interface it is possible for the host system to set the analogue measurement range.

Digital Output: The host system can set the required measurement range.

Q: We have another sensor on the same SPI (with a different slave select, /SS). I understand we need to separate DATA_OUT by a gate to achieve high Z if the device is deselected. The question is, do we also need to separate DATA_IN or DCLK, to prevent the CRM100 from getting unwanted commands?

A: DATA_IN and DCLK do not need to be separated. PinPoint^® would only respond to commands if the /SS input line is taken low.

The checksum is calculated before the SPI registers are loaded. When this is carried out, the Data Bytes are stored and updates to them are inhibited. The Checksum is then calculated on the Status Byte and these 4 Data Bytes. The Status Byte however can continue to be updated for a short time after the Checksum has been calculated. Therefore when the Status Byte, 4 Data Bytes and the Checksum are loaded into the SPI register there is a chance that the Checksum is incorrect. It is therefore advised that if a Checksum Error is detected that the Status Byte should still be interrogated for the Status, such as BIT Fault. 

The measurement range of CRM102.1/CRM202.1 is three times that of CRM100/CRM200.  The confusion arises because the CRM102.1 and CRM202.1 datasheets only document the 900deg/s range selection (the second highest range, equivalent to 300deg/s on the CRM100/CRM200).  However, as noted on the web page, the CRM102.1/CRM202.1 gyros are capable of sensing up to 2700deg/s.  In order to do this the rate range selection need to be set for the highest rate range for the device.  By comparing their datasheets with those of the CRM100/CRM200, the method for achieving this can be deduced.  In summary, in digital mode, set the Command Message bits 4 & 3 (RR1 and RR0) to '00'.  In analogue mode, set pins SEL0 and SEL1 both to ground.

The CRS03 products have undergone three major changes during their lifetime.  The original products were NOT RoHS compliant and had no suffixes (eg CRS03-01, CRS03-04 etc).  An engineering product change to make them RoHS compliant resulted in the new part numbers with an 'R' suffix (eg CRS03-01R and CRS03-04R).  A later design change resulted from the change of ASIC supplier.  This caused no change to their specification but resulted in new part number variants with an 'S' suffix (eg CRS03-01S, CRS03-02S).  In 2016 Silicon Sensing made a change to the MEMS sensor fitted in the -01 and -02 versions of CRS03, and the new parts are identified with a ‘T’ suffix.  All previous CRS03 parts are now End of Life status.  CRS03-01T & -02T parts are RoHS compliant.

On power-on, it is essential that the supply voltage ramps up to 4.75V within 500ms, otherewise the gyro output can freeze.  A similar problem can occur during a power supply drop-out.

Yes.  For CRS03, the supply voltage must reach its nominal 5V in less than 5ms, otherwise the gyro may latch-up and will provide no output.  In this condition is does not automatically reset. 

Basically, our gyroscopes use a silicon (MEMS) ring which is setup and controlled to vibrate in a precise manner. When the gyro is rotated, the resonance pattern changes; the way it changes being proportional to the rotation rate applied to the gyro. Electronics around the ring control the resonance of the ring and also sense the motion of the ring.

When the gyro is subjected to changes of temperature, the bias and scale factor of the gyroscope can change.

We therefore provide two outputs which can be used to sense the change of temperature.

A temperature sensor is included to sense the temperature of the electronics within the gyroscope.

The ring's resonating frequency is also sensed and output as a digital signal. The frequency of the ring is proportional to the temperature of the ring. The frequency of the ring is proportional to the temperature of the ring. This ring frequency is nominally 14 KHz, with the FRQ signal being two times this frequency, that is , nominally 28 KHz. The frequency changes with temperature at -0.76 Hz/degC.

By subjecting the gyroscope to a changing temperature, it is possible to measure the errors (bias and scale factor) of the gyroscope against the frequency output and the temperature sensor output. Using look up tables or fitting polynomials to these errors, it is possible to compensate for the errors, by subtracting the derived error from the output of the gyroscope.

The temperature sensors will be more responsive to changes in temperature of the environment than frequency because of the longer thermal path to the MEMS ring. Equally, in a highly (angular) dynamic environment where the ring will be heated by the nulling action of the secondary loop, the frequency output will be more responsive. In a fairly stable environment, where the temperature across the whole of the gyroscope is stable, then both methods are equivalent.

In general terms, the simplest method for temperature compensation is to use the on-board temperature sensor as the first step. This can be regarded as the primary (coarse) thermal error correction. A further refinement of thermal compensation can be achieved using the ring frequency (FRQ) since this is a measure of the temperature of the ring.

At normal room temperature (+25degC) operation, the FRQ signal is between 27.4kHz and 28.6kHz. The ring gets 'stiffer' as the temperature drops and thus the frequency will increase, and vice versa. The temperature coefficient of the ring is between -0.82 and - 0.70Hz/degC (nominally -0.76Hz/degC). So, if the value of FRQ drops by, say, 7.6Hz then you can assume the ring temperature has increased by 10degC to +35degC [-7.6 / -0.76 = +10degC]. The silicon ring is supported on a glass substrate surrounded by an inert gas inside a sealed metal can. So it is quite well insulated, thus there is a lag between a change in the ambient temperature and the temperature (and thus frequency) of the ring. The temperature sensor on the board reacts quicker to the ambient temperature fluctuations.

CRS39 was designed for use in borehole surveying where temperature is not likely to be very stable. Since MEMS sensors can be quite sensitive to changes in temperature, temperature ramp and changes in the temperature ramp, temperature measurements are critical for optimal performance. Three temperature sensors were therefore built into CRS39 to enable temperature gradients to be monitored across the CRS39 and track changes to the temperature environment. Using these sensors, it is possible to compensate for dependency on temperature, temperature ramp, and temperature gradient across the device and also changes to the temperature ramp. The actual compensation process will be dependent on the mounting of the CRS39 as well as the environment the device is subjected to.

Basically, our gyroscopes use a silicon (MEMS) ring which is setup and controlled to vibrate in a precise manner. When the gyro is rotated, the resonance pattern changes; the way it changes being proportional to the rotation rate applied to the gyro. Electronics around the ring control the resonance of the ring and also sense the motion of the ring.

When the gyro is subjected to changes of temperature, the bias and scale factor of the gyroscope can change.

We therefore provide two outputs which can be used to sense the change of temperature.

A temperature sensor is included to sense the temperature of the electronics within the gyroscope.

The ring's resonating frequency is also sensed and output as a digital signal. The frequency of the ring is proportional to the temperature of the ring. The frequency of the ring is proportional to the temperature of the ring. This ring frequency is nominally 14 KHz, with the FRQ signal being two times this frequency, that is , nominally 28 KHz. The frequency changes with temperature at -0.76 Hz/degC.

By subjecting the gyroscope to a changing temperature, it is possible to measure the errors (bias and scale factor) of the gyroscope against the frequency output and the temperature sensor output. Using look up tables or fitting polynomials to these errors, it is possible to compensate for the errors, by subtracting the derived error from the output of the gyroscope.

The temperature sensors will be more responsive to changes in temperature of the environment than frequency because of the longer thermal path to the MEMS ring. Equally, in a highly (angular) dynamic environment where the ring will be heated by the nulling action of the secondary loop, the frequency output will be more responsive. In a fairly stable environment, where the temperature across the whole of the gyroscope is stable, then both methods are equivalent.

In general terms, the simplest method for temperature compensation is to use the on-board temperature sensor as the first step. This can be regarded as the primary (coarse) thermal error correction. A further refinement of thermal compensation can be achieved using the ring frequency (FRQ) since this is a measure of the temperature of the ring.

At normal room temperature (+25degC) operation, the FRQ signal is between 27.4kHz and 28.6kHz. The ring gets 'stiffer' as the temperature drops and thus the frequency will increase, and vice versa. The temperature coefficient of the ring is between -0.82 and - 0.70Hz/degC (nominally -0.76Hz/degC). So, if the value of FRQ drops by, say, 7.6Hz then you can assume the ring temperature has increased by 10degC to +35degC [-7.6 / -0.76 = +10degC]. The silicon ring is supported on a glass substrate surrounded by an inert gas inside a sealed metal can. So it is quite well insulated, thus there is a lag between a change in the ambient temperature and the temperature (and thus frequency) of the ring. The temperature sensor on the board reacts quicker to the ambient temperature fluctuations.

CRS39 was designed for use in down-hole surveying applications with a 1” diameter requirement, hence the unusual form factor, and as such has a thin, but long characteristic shape. The mounting arrangement of the CRS39’s brass posts means there is a possibility that the resonant mode of the PCBs will be excited when operating in a vibrating environment.

It is therefore advisable that the CRS39-01 be mounted in such a way that the lower PCB is securely clamped. This can be achieved when the unit is mounted as it was designed to be; in an inch diameter tube, with the lower PCB located in and constrained by the tube wall.

It has been observed that the CRS39-02 (packaged variant) exhibits a similar resonant mode, so it is advised that the CRS39-02 is not subjected to a vibrating environment which exceeds 1.8 KHz.  

Sampling and averaging of the gyro rate data, guided by Allan Variance analysis, is key to getting the best out of CRS39 and to allow the earth rate signal to be drawn out of the noise.

Two sampling schemes that we would recommend are:

(a)  In our test chambers we use a16 bit (successive approximation register ADC) National Instruments card. We sample the CRS39-03 fully differentially (Rate and Ref), at 10KHz, without any anti-aliasing filters. We then average every set of 10 samples to produce data at 1 KHz. This data is then analysed for AV and Noise.

(b) In our IMUs, we use 24 bit sigma delta ADCs, outputting sampled data at 10 kHz. Actual sampling at the sensor end is around 192KHz. Again the CRS39-03 is sampled differentially (Rate and Ref). We average every 10 samples to produce a data set at 1KHz.

Analysis of the resulting data using Allan Variance techniques will determine the optimum averaging time.  However, the optimum averaging time may be longer than the user can accept and we have typically used 15s averaging for each compass point (90deg separation) measurement.

Minimising temperature variation over measurement cycle will improve accuracy, either by thermal shielding or provision of thermal mass.  Improvements may also be found by fully enclosing the gyro in a metal enclosure, minimising any 'metal detector' or field effects.

Temperature compensation is recommended - linear or third order may be required depending on the actual conditions seen by the gyro - temperature range, rate of change of temperature, as well as actual rates applied and temperature.

Averaging measurements taken at index positions of 180deg to each other can help by draw out the real bias of the gyro by removing earth rate.  Temperature fitting is also possible by comparing changes between measurements at the same index position against bias and temperature.

The CRM100 gyro can provide a very similar performance to the CRG20 when it is being used in an analogue configuration. There have been improvements made since the parts were first released and while the data sheet figures have remained the same the typical performance has improved (8⁰/hr bias instability). By using two CRM100 and summing the output with a gain of 0.5 then the errors are generally reduced by a factor of √2 (ie 8/√2 = 5.7⁰/hr – very close to the performance of the CRG20) . These error can be reduced further by mounting one CRM100 ‘upside down’ (on the reverse of the pcb) and summing the difference of the outputs, reducing common mode errors.

Alternatively, the CRS43 (which uses the CRM100 internally) has a 5V input – the same as the CRG20.

The reset line is connected directly to a 3.3V device within the gyro (which is also 5V tolerant).  It is triggered at 0.7*3.3V = 2.31V.

Use a SPI to Serial converter to interface between a PC and the CRG20.  The one we have used  has an Atmel AT90S8535 microcontroller on it and the SPCR register is set up with 0x51.

Below is the code we use to transfer a message to and from the CRG20, another piece of code sends the message out to the PC via a serial port which also receives any messages for the CRG20:

#include  <io8535.h>
#include  <stdio.h>
#include  "defines.h"
#include  "types.h"
#include  "prototypes.h"

 /*---------*/
/* Globals  */
/*---------*/
extern unsigned char Latest_Tx_SPI_Message[NO_OF_BYTES_IN_SPI_MESSAGE];
extern unsigned char Latest_Rx_SPI_Message[NO_OF_BYTES_IN_SPI_MESSAGE];

/*----------------*/
/*  Initialise_SPI */
/*----------------*/
void  Initialise_SPI_As_Master()
{
  /* Set data  direction register for port B (SPI port) to make  */
  /* SCK  an  output                                             */
  /* MOSI an  output                                             */
  /* MISO an  input                                              */
  /* SS   an  output                                             */
  /* The  register initialises to all 0 so everything else is an */
  /* input, I know | 0<< does  nothing but it makes the point as */
  /* this bit should be 0.                                       */
  DDRB =  (1<<DDB7) | (0<<DDB6) | (1<<DDB5) |  (1<<DDB4);

  /* Drive the  SS line high to stop any chance of the SPI port  */
  /* being deselected                                            */
  PORTB = PORTB  | (1<<PB4);

  /* Enable SPI  in Master mode and set clock to be fck/16       */
  /* This should give a SPI bus clock of 3.69MHz/16 = 230.625KHz */
  SPCR =  (1<<SPE) | (1<<MSTR) | (0<<SPR1)|  (1<<SPR0);
}

/*-----------------------------------*/
/*  Transfer_1_Message_Using_SPI_Port */
/*-----------------------------------*/
void  Transfer_1_Message_Using_SPI_Port()
{

  int  i;

  /* Set SS  line low */
  PORTD = PORTD  & ~(1<<PD2);

  /* Add delay  to allow Mega88 to complete a host message update  */
  Timer_2_Delay_5uS();

for(i=0;  i<NO_OF_BYTES_IN_SPI_MESSAGE; i++)

  {
    /* Get a byte from the buffer and place it in SPI output register */
    SPDR =  Latest_Tx_SPI_Message[i];

    /* Wait for  the data to be sent */
    while(  (SPSR & (1<<SPIF)) != (1<<SPIF) );

    /* Copy the  received byte into the buffer */
    Latest_Rx_SPI_Message[i] = SPDR;

    /* N.B. For  faster CPU’s place a delay in here */

  } 

  /* Set SS  line high */
  PORTD = PORTD  | (1<<PD2);

}

Please note that this controller is working at 3.69MHz (considerably slower than the CRG20 processor) so the time it takes to check the data has been sent and copy in the received byte is equivalent to the delay needed between bytes. If a faster processor was used we would have to add a delay

The input switching level for the SPI (and CBIT) high  input pins is 0.6 * VDD.  At 5V VDD this = 3.0V

The input switching level for the SPI (and CBIT) low input pins is 0.3 * VDD.  At 5V VDD this = 1.5V

It is possible that this is the result of an unintentional activation of Commanded BIT.  Triggering CBITA results in a rate offset of around 10-13 deg/s being applied to the sensor output (either from the SPI digital interface or either of the analogue outputs).  For this reason it is recommended that the CBITA line is pulled low if is not being used (see note 3 in CRG20-00-0100-0110 rev 9 specification).

Our devices are inherently insensitive to linear acceleration or g. The vibrating structure is a ring and is made to resonate at about 14 KHz. If the ring moves under acceleration load, the modulation and de-modulation signals and the amplitudes are not affected. Therefore we expect very little g sensitivity. When the design requirements were formulated for CRG20, the design requirement was set at 0.1 deg/s/g. Our design verification tests have confirmed very little g sensitivity at all accelerations above 4g (typically <0.0001 deg/s/g). Measurements below 4g are dominated by other error sources such as bias instability and bias and SF changes with temperature and therefore it is difficult to attribute which errors are due to g sensitivity. So when we do a g sensitivity test, we assume that all measurement errors are due to g sensitivity and this results in a very pessimistic number. Our experience is therefore that it is not necessary to compensate for g sensitivity. If you wish to confirm this for yourselves, you could try a simple +/- 1g tumble test, and average the data for about 30 seconds, to see if there are any measurable changes when the g changes from +1g to -1g. Earth rate may need to be allowed for too.

The CRG20 needs to be powered from a 5V supply. To interface it to a 3.3V system will require a special attention to the logic thresholds of both inputs and outputs from the CRG20. The logic thresholds are shown below. For inputs to the CRG20, the threshold is 0.6*5.0 or 3V. Therefore a 3.3V system shouldn’t need any level shifting if the Logic High can be guaranteed to be above 3V. Logic Low will be correctly detected. For outputs from the CRG20, the levels will have to be reduced unless the 3.3V system has “5V Tolerant” inputs. A resistor is normally all that is required to achieve this, the value of which depends on the input resistance of the 3.3V system.

We are often asked about MSL in relation to CRG20.  MSL specifications are usually quoted for plastic-encapsulated components.  This is because they absorb moisture which causes problems when they are subjected to soldering.  In contrast, our CRG20 gyro is a ceramic, hermetically sealed package.  An MSL specification is not normally quoted for such a package as it is not relevant - it does not absorb moisture.

The specification gives the temperature limits for storage and provided that this is adhered to here are no particular risks associated with the long term storage of CRG20.  From the perspective of solderability, it is important to store the CRG20 in an environment that is dry, and low in sulphur, chlorine and hydrocarbons.  It is accepted practice within the industry to achieve this by vacuum packing the parts in an ESD safe moisture barrier bag.

Basically, our gyroscopes use a silicon (MEMS) ring which is setup and controlled to vibrate in a precise manner. When the gyro is rotated, the resonance pattern changes; the way it changes being proportional to the rotation rate applied to the gyro. Electronics around the ring control the resonance of the ring and also sense the motion of the ring.

When the gyro is subjected to changes of temperature, the bias and scale factor of the gyroscope can change.

We therefore provide two outputs which can be used to sense the change of temperature.

A temperature sensor is included to sense the temperature of the electronics within the gyroscope.

The ring's resonating frequency is also sensed and output as a digital signal. The frequency of the ring is proportional to the temperature of the ring. The frequency of the ring is proportional to the temperature of the ring. This ring frequency is nominally 14 KHz, with the FRQ signal being two times this frequency, that is , nominally 28 KHz. The frequency changes with temperature at -0.76 Hz/degC.

By subjecting the gyroscope to a changing temperature, it is possible to measure the errors (bias and scale factor) of the gyroscope against the frequency output and the temperature sensor output. Using look up tables or fitting polynomials to these errors, it is possible to compensate for the errors, by subtracting the derived error from the output of the gyroscope.

The temperature sensors will be more responsive to changes in temperature of the environment than frequency because of the longer thermal path to the MEMS ring. Equally, in a highly (angular) dynamic environment where the ring will be heated by the nulling action of the secondary loop, the frequency output will be more responsive. In a fairly stable environment, where the temperature across the whole of the gyroscope is stable, then both methods are equivalent.

In general terms, the simplest method for temperature compensation is to use the on-board temperature sensor as the first step. This can be regarded as the primary (coarse) thermal error correction. A further refinement of thermal compensation can be achieved using the ring frequency (FRQ) since this is a measure of the temperature of the ring.

At normal room temperature (+25degC) operation, the FRQ signal is between 27.4kHz and 28.6kHz. The ring gets 'stiffer' as the temperature drops and thus the frequency will increase, and vice versa. The temperature coefficient of the ring is between -0.82 and - 0.70Hz/degC (nominally -0.76Hz/degC). So, if the value of FRQ drops by, say, 7.6Hz then you can assume the ring temperature has increased by 10degC to +35degC [-7.6 / -0.76 = +10degC]. The silicon ring is supported on a glass substrate surrounded by an inert gas inside a sealed metal can. So it is quite well insulated, thus there is a lag between a change in the ambient temperature and the temperature (and thus frequency) of the ring. The temperature sensor on the board reacts quicker to the ambient temperature fluctuations.

The basic inertial data output from DMU30 is angular rate (deg/s) and acceleration (g).  Delta Theta and Delta Velocity outputs can be considered as a re-scaling of these basic data outputs.   Provision and use of these values is an industry standard expected and used by many of our customers.

Delta Theta is in degrees and represents the degrees rotated over the sampling time of 5.0ms (or 1/200th of a second).  For example, if the IMU axis is rotating at 200deg/s, Delta Theta will be 1 deg.

Similarly, Delta Velocity is a re-scaling from g to m/s, representing the change in velocity change over the same 5ms.  The output in m/s is equal to the acceleration in g multiplied by 0.04903325 (divide by 200 and multiply by 9.80665, the latter being a universal measure of g).

24-bit sigma delta analogue to digital convertors are used within DMU30. Compensation is carried out using single precision floating point mathematics. The DMU30 output message is also output as single precision floating point numbers. So the mathematical resolution is very small indeed.

In practical terms, the minimum resolvable change is dominated by the sensor noise. With averaging, this can be improved to a point where the minimum resolvable change is dictated by the bias instability derived from Allan Deviation plots. For the gyro channels this is below 0.1 deg/hr and for the accelerometers, around 10μg.