FAQs AS50/51/52xx

For quick troubleshooting, check the following:

1. In 5V-mode: is the 5V supply stable ? Is the 3.3V supply stable ? is there a buffer capacitor (1..10µF) on pin VDD3V3?
2. In 3.3V-mode: are the supply pins VDD5V and VDD3V3 tied together and connected to a stable 3.3V supply voltage?
3. is the magnetic field in range (check with MagIncn, MagDecn or MagRngn outputs)?
4. is the PROG pin tied to VSS at startup (to avoid accidental switching to alignment mode – this is for testing purposes only)
5. check the PWM output, if available. Does it linearly increase with angle?

In most cases, checking the above suggestions solved the problem. If this still does not help, try to connect your encoder to any AS504x demoboard and test your setup with the demoboard software by selecting your assembly as an external encoder

(see AS504x demoboard operation manuals for further information)

Yes, you can! You don’t necessarily need to calibrate or program the chip, you can use it as-is. The default (unprogrammed) settings should be sufficient for the majority of applications.

At power-up, the chip runs through a couple of compensation loops and sets the OCF (offset compensation finished) – status bit. If only the incremental output is used, the outputs A=B=Index will all remain high until the chip has completed power-up (CSn must be low). If CSn is high during power-up, the outputs A=B=Index will be high until power-up has completed AND CSn is pulled to low.

The status of the hardware pins is as follows:

• MagINCn: not defined
• MagDECn: not defined
• A_LSB_U: high (see text)
• B_Dir_V: high (see text)
• Index_W: high (see text)
• PWM_LSB: standard PWM signal, pulse width is invalid until power up time (tPwrUp) is completed

The startup time variation over temperature or supply voltage is low. The specified tPwrUp (20 / 50 / 80ms) is valid over the full temperature range.

If short startup is a concern, you can poll the OCF status bit in the serial bitstream of the SSI interface. As soon as OCF is high, startup has completed and the chip can be used.

Incremental output: available on AS5040 and -35: The incremental output is usually used for medium to high speed applications, requiring incremental information only. At power-up, the absolute angle position is not available. To get the absolute position, the encoder first needs to move to the home (or zero-) position where an index pulse is generated and then obtain the absolute position information by counting pulses when moving away from this reference.
In contrast, the AS5040/-43/-45 encoders provide absolute output information; the angle position is available immediately after power-up. No homing of the device is necessary.

To read the absolute position, the user can select from several types of output formats, depending on which fits best to the particular application:

Serial output (SSI): available at AS5040,-43 and -45: this output is for fast data transmission, it requires 3 signals: Chip Select, Clock and Data out.
This type of output can also be used to read multiple devices over a single 3-wire interface in a so-called daisy chain mode.

Analog output: available at AS5043: this “classic” type of output has been in use for decades. This output requires only one wire (Analog out) and the output voltage is proportional to the rotational angle of the magnet. It can be used for example as a contactless replacement of a potentiometer.

PWM output: available on AS5040 and -45: similarly to the analog output, this output requires only one wire (PWM out) for signal transmission. The absolute angle information however lies in the time domain (pulse width) and not in the magnitude (voltage) of the signal, making it very robust in electrically noisy environment.

UVW outputs: available on AS5040: Brushless DC motors typically use 3 conventional Hall switches, mounted on a PCB near the end of the magnetic rotor to provide the commutation information for the motor controller. The AS5040 can mimic these signals and replace the 3 Hall switches. In this mode the PWM output changes to an incremental pulse output and the UVW information is available in addition to the absolute serial output. Consequently, the AS5040 is both an absolute and incremental encoder as well as a commutation switch at the same time. The ability to provide absolute information and the zero position programming allows for optimized commutation achieving maximum torque.

The specified driving capability in the datasheet refers to the VOL and VOH levels. You can sink/source more current, but you may not be able to maintain the VOH and VOL levels any more.

You don’t necessarily have to connect the PROG pin to VSS. You may also leave it open. However, if you have a long wire or cable connected to this pin it may pick up crosstalk signals and be not be permanently at low any more. This increases the risk that the chip may switch unintendedly to alignment mode (PROG = high while CLK = falling edge). That’s why it is recommended to ground this pin when it is not used for programming.

This is an internal value, it should be used as a reference point. Remember that the airgap, strength and size of the magnet also has an influence. The value of 32 refers to a typical airgap of about 1mm in the middle of the recommended range (~0.5 – 1.8mm) with the recommended magnet (NdFeB Æ6mm x 2.5mm MN-35H). Ideally the magnet is centered, when the difference between minimum and maximum reading over a full turn is as small as possible, regardless of the actual value.

Note: the limit of 32 applies to the AS5040/43, the AS5045 has a limit of 4x32 = 128, as it has four times the resolution as the AS5040/43.

First of all, accuracy is not to be confused with resolution. In fact the two issues are not necessarily related with each other.
Resolution is the number of steps per revolution (e.g. 1024, 4096)

Accuracy is the worst case deviation of the indicated angle versus the actual angle (e.g. +/-0.5°) over a full turn.
There is an uncalibrated and a calibrated accuracy. Calibration is a time and cost consuming process, as you need to measure your encoder assembly against a known precision reference
(e.g. a ≥13-bit optical encoder).

Therefore, our aim for the design of the AS5000-encoders was to get the accuracy of the encoder as good as possible without calibration. The uncalibrated accuracy of our encoders, in other words the accuracy as they come “off the shelf” is specified better than +/- 0.5° with centered magnet. That is sufficient in most cases.

Of course you may still add some external calibration, where the calibration parameters are stored in an external memory or microcontroller to further enhance the accuracy of the encoder. In this case the accuracy is essentially limited by the accuracy of your calibration system.
The maximum achievable accuracy is limited by the transition noise of the encoder (= jitter). Per default, the AS5040 has a transition noise (rms, 1 sigma) of 0.12°, the AS5043 and -45 both have 0.06° (fast mode) and 0.03° (slow mode). In each case, the transition noise may further be improved by digital filtering (averaging).

Now, let’s assume we did a perfect calibration and our theoretical accuracy is +/-0°.

What are the additional limitations?

• a) Quantization Error:
  This is the error caused by the finite step size of the ADC, or in other words, due to the digital step response of the encoder, you can only be accurate within +/- ½ LSB

  For the 10-bit AS5040/43, this means +/- 0.175°
  For the 12-bit AS5045 , this means +/- 0.044°
  You can’t go lower than that, even with perfect calibration

• b) Transition noise
  Every ADC has a transition noise, this is the jitter between two adjacent readings.

  Transition noise is about 1LSB peak-peak
  ~0.35° p-p for AS5035, and -40 and
  ~0.08° p-p for AS5043 and -45 in slow mode

  This value is on top of the Quantization error

  Transition noise can be reduced by limiting the bandwidth; that is, reducing the effective sampling rate through digital averaging.

As shown in Q9:”What is the best accuracy you can achieve?” with good calibration and eventual digital averaging, you could get to an accuracy of ~0.1° using a 12-bit encoder (AS5045). However, this value does not include

• c) Temperature
  There is an additional error caused by temperature changes. We can compensate the temperature effects of the magnet on-chip, that’s why we don’t need to store any magnet temperature coefficients (as is required with linear Hall sensors) but the temperature of the chip itself effects the accuracy too.

  The AS5000-series are already designed for superior behaviour over temperature as well, but like every analog circuit, you can never completely ignore this influence. Tests have shown that there is hardly any influence in “normal” operating temperature conditions. Only at the extremes, especially around -40°C there may be cases where the accuracy changes by ~+/- 0.4°, but there is no general rule which temperature is “good” and which is “bad” over all batches and process corners !

  This is also reflected in the datasheet, as the accuracy (parameter INL) is +/-0.5° at room temp. and +/-0.9° over the full range includes that +/-0.4° additional error over temperature.

  To get rid of this additional error, the temperature drift of the sensor must be evaluated. This may be a time consuming task and could in practice only be carried out with a limited number of temperatures (e.g minimum, maximum operating temperature and room temperature).

The answer to that question is “Yes and No”.

It’s NO because it doesn’t make much sense to calibrate the chip without a magnet. Calibration only makes sense for a completed assembly, where a given magnet at a given (mis)alignment at a given temperature is calibrated together with the chip as a unit. Therefore it can’t be done at chip level.

It is YES, because the AS5000-series offer what you may call a “1-point-calibration”, which is the “Zero Position Programming”. Once you have programmed the Zero position, you are very accurate and repeatable at this spot; in fact within +/-1/2 LSB. Again, it can only be done at the final assembly. The Zero Position programming itself must still be done by the user, but the Data is stored on the chip. This approach is useful as there may be applications where the sensor is used as a stand-alone unit without a controller (e.g for incremental, PWM or analog output applications).

The absolute error of the magnetic rotary encoder depends on several factors, mainly on amplitude and offset matching of the Sin- and Cos- paths inside the chip. The phase error between Sin- and Cos- signal paths also has an influence in the total error, but due parallel signal processing, it can be neglected for the AS5000-series encoders. The error curve contains a fundamental (1f) function and a second order (2f) function.

This pattern repeats with every rotation, but is not consistent from one part to another.

Therefore, the curve shown in the datasheet is (as noted) only an example of an error curve. Since the Offset and Gain matching will differ from one part to another, the error curve will also have a different shape.

The repeatability is +/- ½ LSB. This counts for both 10-bit and 12-bit encoders. In other words, the AS5040 has a repeatability of +/-0.17° and the AS5045 a repeatability of +/-0.04°
This is independent of magnet placement and temperature. At a given position (aligned or misaligned) and at a given temperature, the repeatability is +/-1/2 LSB.

The requirement however is that the magnetic field should be in range. A weak magnetic field causes more noise and consequently the repeatability may be poorer.

The repeatability does not change with temperature, but the INL (or accuracy) does. The INL can change by +/-0.4° over the temperature range of -40° to +125°.

This means for example that if you don’t move the magnet but sweep the temperature, the indicated angle may change by +/-0.4°.

For a given temperature however, the INL is again repeatable and can be compensated by calibration quite well.

That depends on the type of output you are using:
For the PWM output or analog output, you may connect CSn permanently to VSS,
for the incremental output, CSn must be tied to VSS (at least for a short pulse) and
for the serial output, CSn is used to latch new measurement data, it must be set according to the timing diagram in the datasheet(s).

No. The encoder chip measures the angle of the magnet constantly with a sampling rate of 10.42kHz (AS5040, AS5043/45 in fast mode) or 2.61kHz (AS5043/45 in slow mode) and updates an internal register after each measurement. A falling edge of CSn latches the data to the serial shift register. This means that after CSn has been brought low, you don’t have to wait for a measurement to be completed. You can start reading data immediately after 500ns (tclkFE).

This is in principle possible for annual volumes >1M. Such parts would have to get a separate marking and part number in order to avoid mix-and-match with existing ASSP’s. For the customer it has to be determined whether it is worth the extra expense for documentation, part qualification, inventory, etc…

Practice has shown that in most cases, programming the chip requires a complete assembly (e.g. for zero position programming) and that this step can be easily integrated in the regular production test. The powerful SDK (software development kit) allows very quick and simple implementation of the programming sequence into the user’s test software.

Vibration should be no problem. One of the big benefits when using the AS5000-series magnetic rotary encoder IC’s is that they are manufactured in a standard CMOS process.

There are no additional post processing and there are no mechanically moving components in the IC package.

That’s why you can assume that it can withstand the same vibration as any other “regular” IC.
Your biggest concern in such applications is not the rotary encoder IC itself, it is the other factors that may be a problem when large vibrations occur. These are:

• components with a large mass, such as electrolytic capacitors with radial connectors or the mounting of the sensor magnet in the rotating shaft.
• Both may fall off when they are not properly mounted after an extensive exposure to vibrations.

The Encoder IC package (SSOP-16) does not have a large mass and it is soldered with many pins, so there should be danger of breaking off due to vibration.

You may also examine whether the alignment of the magnet to the IC is affected by vibration. It is recommended to keep the center of the magnet within a radius of 0.25mm from the center of the IC.

The encoders have a typical power consumption of 16mA. If your application allows that you don’t have to permanently read the encoder, you may use the AS5040/43/46 in a pulsed mode. The sequence could be as follows:

• Apply power to the chip (5.5V or 3.3V). The current consumption will be around 16mA
• Wait until the chip has completed a measurement. You can either wait for tpwrup (20 / 50 / 80ms; see datasheets) to expire or you can poll the OCF status bit in the serial data. When OCF is set, the angular data is valid
• Disconnect the power from the chip. The current consumption drops to 0mA
• Wait for a predetermined period of time (depending on the application)
• Repeat the cycle

The total current consumption will then be

Where:

• I_avg = average current consumption
• I_nom = nominal current consumption in on-state (typ. 16mA; max. 21mA)
• t_on = duration, while the sensor is active (min. 20 / 50 / 80 ms; depending on product and mode)
• t_off = duration, while the sensor is switched off (depending on application)

The recommended placement of the magnet is that is that the center of rotation, the center of the magnet and the center of the IC are in one vertical line. The parameters specified in the datasheet (unless otherwise noted) include a horizontal misplacement of the chip within a radius of 0.25mm to the IC package center.

This includes the max. +-0.235mm (mis-)placement of the silicon die within the plastic package. Therefore, if the magnet can be aligned to the silicon die (e.g. by using the alignment mode or when mounting a bare die), the misplacement tolerance is wider, namely a radius of 0.485mm.

A misalignment of the magnet in X- and Y- direction will influence the accuracy (actual position versus indicated position). If the center of the magnet is centered over the IC package center, the full accuracy (+-0.5 degrees) can be guaranteed. This figure comes from the chip internal matching of amplitude, phase and offset of the SIN and COS channels and includes the placement tolerance of the silicon chip in the plastic package.

If the misalignment is out of this range, the chip will still operate normally, but with reduced accuracy.

In practical case, the maximum INL over +/-0.25mm radius misalignment and over temperature is +/-1.4°. Taking into account a +/-0.4° drift over the full temperature range of -40°….+125°C, the accuracy will be +/-1.0° over the misalignment range of a +/-0.25mm radius at room temperature.

Unfortunately, the AS50xx rotary encoders cannot be mounted off-axis. The center of the rotation axis must be over the center of the IC, as shown in the product presentations and datasheets.

This is because the encoder is an absolute measurement system using vertically oriented magnetic fields and the chip must “see” the complete magnet as a whole in order to determine the exact absolute position.

The vertical distances of 0.5 to 1.8mm refer to our reference magnet(s), NdFeB N35H, d=6mm h=2.5mm. This distance marks the limits at which the magnet is in the “green” range of 45…75mT.

You can move this range by using stronger or weaker magnets. In practice, distances up to ~5mm vertical gap can be obtained for the “green” magnetic range.

The chips will continue to operate, even if the magnetic field is out of range. However, due to higher noise at weak magnetic fields, the jitter will be increased (= the min/max reading at a stationary position of the magnet). For absolute measurements, this jitter may be reduced by averaging several subsequent readings. The digital serial output as well as the PWM output will always be active, independent of the magnetic field strength.

The AS5043 analog output however will be turned off in the “red” range. This safety feature may be disabled by OTP programming (see AS5043 datasheet)

You need a 2-pole magnet, diametrically (top view: left and right) magnetized, with the rotational axis centered over the center of the IC. Magnets with the poles located at the top and bottom or multipole ring magnets cannot be used. We recommend rare earth magnets, such as Neodymium-Iron-Boron (NdFeB) or Samarium-Cobalt (SmCo) magnets.

The mounting of the magnet in a shaft depends on the material the shaft is made of. If it is non-magnetic, the magnet can be simply inserted directly into the shaft. Usually, this is done by pressing the magnet into a tight centered hole in the shaft.

Since there is usually no adjustment of the magnet rotating angle required (it can be done by software with zero-position programming), a cylindrical magnet is easiest to handle. Sometimes, glue is used too, but only in combination with a tight fit of the magnet in the hole.

For magnetic shafts (e.g. iron shafts), the magnetic field lines of the inserted magnet will be “short circuited” by the ferromagnetic material of the shaft. This weakens the magnet and requires that it must be moved closer to the IC. The better approach in such case is that the magnet is not directly inserted into the shaft, but rather separated by a non-magnetic material (e.g. plastic, copper, etc..) from the magnetic shaft. In other words, the non-magnetic separator is pressed into the magnetic shaft and the magnet is pressed into the separator. The separator should be thick enough to keep the magnet away from the magnetic shaft. A separation distance of >3mm is adequate. The larger the separation, the less influence from the shaft can be expected.

Another consideration is the ambient temperature range. The user must take care that magnet is sufficiently held in place, considering the expansion and contraction of the shaft over temperature.

Magnetic field lines in different rotor materials:

• left: magnet directly in magnetic rotor shaft: not recommended; field lines are concentrated in the shaft
• middle: magnet directly in non-magnetic shaft: recommended; no influence of rotor material
• right: magnet in magnetic shaft, separated by non-magnetic spacer: recommended

With AS5040, the LIN and COF alarms will only be set when the magnetic field is too strong.
These bits will not be set when the magnetic field is too weak.

AS5045 can be programmed such that the LIN bit will be set both when the magnetic field is either too strong or too weak AS5043 has the LIN alarm enabled by default for both too strong and too weak fields.

The COF alarm will only be set when the magnetic field is too strong due to exceptionally strong external magnetic fields. It does not occur in normal operation with the AMS reference magnet(s).

When no magnetic field is present, the Hall Sensors only pick up noise. In this case, the encoder will display random information. However, the out-of-range state is clearly indicated by the MagInc-MagDec status bits.

The digital output is intentionally NOT disabled in such case, as some customers are using the encoders also in a weak magnetic field environment, sometimes with additional external digital filtering (averaging) and can still use the sensor ICs under these extreme conditions with acceptable results.

If the magnetic field is within 45…75mT, no alarm bit is set. This is the “green” range, at which the AS5035/40/43/45 should be operated. With the recommended magnet (NdFeB Æ6mm x 2.5mm N-35H), this magnetic range corresponds to an airgap between chip surface and magnet surface of about 0.5 to 1.8mm.

If the magnetic field is below 45mT or above 75mT, both status bits MagInc and MagDec will be set. The range where the MagInc+MagDec alarm bits are set, but the LIN alarm is not set is referred to as the “yellow” range. You can still operate the AS5040 with good performance in the yellow range.

The LIN (Linearity) – Alarm is set at about >135mT on all encoders. In addition, the AS5043 and -45 will also set the LIN alarm when the magnetic field is too weak (when the airgap between magnet and IC is >~2.8mm). The range where both MagInc+MagDec bits as well as the LIN alarm bit are set is referred to as the “red” range. In the red range, a larger linearity error must be expected, although the chip can still be used.

The COF (Cordic Overflow) is set at a magnetic field over ~165mT. Once the COF alarm is set, the angular data is invalid. The OCF bit will not be set in normal operation, it would only be set if the magnetic field is much stronger than normal.

You can resolve these alarms by bringing the magnet within the recommended airgap range and –if possible- remove any external magnetic fields.

The “red” range (see Q29:At which magnetic field levels do I get a MagInc/-Dec, LIN- and COF-Alarm?) does not distinguish between too weak and too strong magnet. However, there are several ways to tell the difference:

1) check fluctuations of the SPI interface:
If the magnetic field is weak, there will be more noise, resulting in a fluctuating angle information at the SPI interface without magnet, the fluctuation will be highest. If the magnetic field is strong, the SPI angle information will be stable within +-1/2 digit.

2) chose a magnet that is not too strong or far enough away so it never reaches the “too high” alarm.
It must be verified by the customer that with this setup also does not generate a “too low” alarm in normal operation, only in a failure case.

3) switch to alignment mode:
Without a magnet, the alignment mode reading will be close to zero (only noise). However, the reading must be near zero over a full turn of the magnet, as there may be situations where a misaligned magnet also reads close to zero at certain angles.

5) Check the incremental outputs in step/direction mode
Similarly to 1) above, a method can be used that does not require the SPI interface at all. Furthermore this method can be used also while the motor is spinning:
I) put the AS5040 in “Step / Direction” incremental mode, either permanently by user OTP programming or temporarily

II) Monitor the incremental output signal B_Dir_V.
In “Step/Dir” mode, this signal indicates the direction of rotation. For clockwise operation, it is constantly high, for counterclockwise operation, is is constantly low. When the magnet is not rotating, it remains at the last setting before it stopped.

In normal operation, with a magnet properly attached, the signal only changes state when the direction of rotation is reversed.

Without a magnet however, the encoder will pick up only noise and the incremental outputs will be fluctuating.

Consequently, when the incremental output signal B_Dir_V changes state repeatedly over a short period of time (e .g within a few milliseconds), this is a clear indication that no magnet is present or the magnetic field is very weak.

This test can also be used while the rotor is rotating, as the signal B_Dir_V will always be constantly high or low, depending on rotation direction. if the motor shaft holding the magnet is rotating in one direction and the signal B_Dir_V is fluctuating, this is a clear signal that the magnet is broken or has fallen off.

The magnet must be positioned with the neutral zone between the poles in parallel to the long side of the IC (parallel to the rows of pins) with the north pole facing pins 1-8 and the south pole facing pins 9-16.

The insensitivity to external magnetic fields has several reasons. The major ones are the use of lateral Hall elements and a differential measurement technique. The lateral Hall elements are only sensitive to magnetic fields perpendicular to the chip’s surface. They are not sensitive to magnetic fields in the horizontal plane.

The differential measurement senses only the difference in the magnetic field of opposite Hall sensor pairs. External DC magnetic fields will influence the absolute magnetic field, but not the differential field that is detected by the Hall sensor pairs. In addition, the sensitivity of the Hall sensor path was chosen not too sensitive, which results also in less sensitivity to external magnetic fields. Since the encoder operates in the near field of the magnet, the (desired) field of the sensor magnet is already relatively strong at the chip’s surface and cannot be easily disturbed by external fields.

In practice, the total magnetic field (permanent magnet + external disturbing magnetic fields) should not exceed 80mT in order to comply with the performance specified in the datasheet. The chip will continue to operate well beyond this range, but due to saturation effects, the output linearity may decrease.

Magnetic shielding is normally not required, as the AS5040 compensates the influences of external magnetic fields (see Q32:How can a magnetic encoder be insensitive to external magnetic fields?).

In cases were extreme external magnetic fields are present and high accuracy of the sensor is required it is certainly a good idea to provide magnetic shielding by using ferromagnetic sheet metal, for example.

The AS5000- family of magnetic rotary encoder ICs are very robust towards external magnetic fields, even to strong fields as from brushless DC motors. Since the magnet + encoder assembly is mounted at the end of the rotor shaft, there is usually a distance of one or more centimeters between the encoder IC and the BLDC rotor magnet. In practice, special calibration or magnetic shielding will not be required.

Ferromagnetic materials (such as Iron, Cobalt or Nickel) will act as a field-concentrator and transfer most of the magnetic field into the material. Thus they add as a block for the magnetic field and should be avoided. Paramagnetic materials, such as Aluminum or diamagnetic materials, such as Copper weaken the magnetic field only slightly or hardly at all. These materials can be used between magnet and encoder IC.

So it depends on the magnetic properties of the material in discussion. Stainless steel is usually paramagnetic and can therefore be used between sensing magnet and encoder IC without problems. However, some stainless steel compositions are magnetic and should therefore be avoided.

As a general rule, if you can manage to get the proper magnetic field strength at the surface of the chip (Status indicators MagInc = MagDec = 00), you can put any material between magnet and IC. For weak magnetic materials, you may have to use a stronger magnet in order to get the proper magnetic field (45mT....75mT) .

The chip will continue to operate even outside the recommended field strength, but with reduced accuracy.

Another issue that must be considered is speed. Magnets rotating at high speed generate eddy currents in magnetic materials which in turn may distort the magnetic field. This effect can reduce the accuracy.

The AS5035/40/43/45 Hall sensors are arranged in a circle of 1.1mm radius. Consequently, only the vertical magnetic field around this circle is of interest. At diametrically magnetized magnets, the magnetic field is zero at the center and increases linearly towards each pole. It usually reaches a maximum close to the edge of the magnet. Consequently, smaller magnets have a stronger magnetic field at the 1.1mm radius from the center than larger magnets.

This means that larger magnets must be either stronger or moved closer to the chip in order to reach the recommended magnetic field strength.
On the other hand, larger magnets allow a larger area of misalignment since the linear region at which the magnetic field increases linearly with distance from the center is larger too.

Non – circular magnets, e.g square shaped magnets may be used as well, but they are only as good as their smallest dimension. For example a square 4mm magnet would be only as good as a round 4mm magnet, but use 1.27 times more space.

The thickness of the magnet has less influence in the performance of the encoder. Thicker magnets allow more airgap between encoder and magnet.

No! The AS5035/40/43/45 is produced in a regular CMOS process without an additional field concentrator. Even very strong magnetic fields will not damage the chip.

We recommend a 6mm diameter rare earth magnet (NdFeB or SmCo), as it is a good compromise between size, vertical distance and freedom of horizontal misalignment.

You may also use smaller magnets (down to 3mm) but you lose the freedom of misalignment. A larger diameter magnet allows more horizontal misalignment than a smaller one, but also requires a stronger magnetic field to maintain a strong differential signal. The chip will still work, even when the magnet is misaligned, but the accuracy will decrease.

No, the hysteresis is only enabled in incremental mode to avoid jittering between two positions when the magnet is stationary over a transition point.
For absolute mode, a hysteresis can be introduced in software by the user, if required.

This effect will be noticed especially when the magnet is stationary at a transition point between two values. Due to the transition noise, the absolute output reading may flicker. You can reduce the transition noise and hence the flickering by averaging the readings. The more samples you average, the less flickering will occur, but it will also slow down your maximum speed.

You can also get rid of the flickering by introducing a hysteresis by software.
A similar effect where the angular data may be unstable over several degrees occurs when the 3.3V power supply is unstable, e.g. when the VDD3V3 pin is not buffered by a capacitor (see datasheet for proper connection of the supply pins in 5V and 3.3V – mode).

You may also refer to the Package Drawings and Markings section at the end of the datasheet(s). The package thickness (Symbol A2) is. 1.73mm +/- 0.05mm, so the surface of the die is 1.73 - 0.576 = 1.154mm away from the bottom of the package.

With the recommended magnet (N35H NdFeB Æ6mm x 2.5mm ), the recommended airgap at which the MagInc/MagDec outputs are “green” (in range) is about 0.5 - 1.8mm from the top surface, which is about 1.08 - 2.38mm from the chip surface.

Translated to the backside of the chip, this means that the recommended airgap for a magnet at the backside of the chip is (1.08 to 2.38mm) - 1.154mm = ≤1.23mm.

Note that the indicated rotational direction is reversed when you mount the magnet at the backside of the chip. You can either correct for this externally by software or you can do it internally by setting the CCW-bit in the OTP register which reverses the indication of rotational direction.

As short lines, the maximum clock rate is 1MHz, which allows over 50.000 angle readings per second. At long lines however, the clock rate may be reduced due to capacitive and inductive impedances in the cable.

See Application note AN5040-10: Data transmission over long wires.

Yes, you can! You can connect all DO and CLK signals in parallel and select individual encoders with a low signal at pin CSn.

In daisy chain mode, PROG is connected to DO of the next encoder. While CSn is high, DO switches to tristate and the internal pulldown-resistor at pin PROG will pull the tristate DO output to LOW. So PROG is LOW when CSn goes low. It will not switch to alignment mode as this will require that PROG is high at the falling edge of CSn.

To avoid that a HIGH Parity-bit can switch the next device in the chain to Alignment mode, the parity is always followed by a LOW signal (see daisy chain timing diagram in datasheet)

Yes. In daisy chain mode, the angular data from each encoder in the chain is latched synchronously with the falling edge of CSn. The latched data will be frozen until the next falling edge of CSn. So it does not matter at which speed the data is clocked into the microcontroller in daisy chain mode. The received data represents a synchronous “snapshot” of the angle from each device in the chain.

There is a slight irregularity, though: since the sensors run on an internal oscillator, that is not synchronized with CSn, the non-symmetry between the measurement of each sensor can lies within one sampling period (96µs or 384µs)

The incremental outputs are locked at power-up while pin CSn is high. This feature allows external controllers to enable the chip only when they are ready to receive data. To start sending pulses, pin CSn must be pulled low (either for a short pulse or permanently). Pin CSn may also be permanently tied to VSS at power-up. In this case, the AS5035/40 will generate pulses, as soon as the internal startup is completed.

The period and phase jitter that can be observed at the incremental outputs is due to the overall noise in the system. The sources of noise are the Hall Elements, the Preamplifier and the ADC. This voltage noise is translated by the Cordic DSP into angle noise, or so-called transition noise.

The transition noise is 0.12° rms (1 sigma). A value of 3 sigma (0.36°) includes statistically 99.73% of all readings. The transition noise is always 0.12° rms, independent of the selected incremental resolution. Consequently, the relative period and phase jitter will be reduced with lower resolution.

See also Q57:Why is the period and phase jitter reduced at higher speeds?

The incremental pulses are generated from the 1024 absolute positions of the 10-bit system. You need 4 absolute steps to decode a 2-channel incremental signal:
Example:

The incremental pulses repeat with every 4th absolute code: e.g. codes 0,4,8,12,… produce the same incremental signal (A=0 B=0). Therefore the number of quadrature pulses is: number of absolute steps / 4

To decode a single channel incremental signal (as in step/direction mode) , only two absolute steps are required:

The single-channel incremental pulse repeats with every 2nd absolute code: e.g. codes 0,2,4,6,… produce the same incremental signal (A=0). Therefore the number of single channel pulses is: number of absolute steps / 2

A misalignment of the magnet only affects the accuracy, not the resolution of the encoder IC. You will always get 1024 absolute positions and “no missing pulses”, even when the magnet is not perfectly centered.

When the AS5040 incremental outputs are compared to a real-time optical reference encoder at high speed, it may seem that the encoder does not deliver the correct number of pulses.

These „missing“ pulses can be attributed to the propagation delay:
In incremental mode, the delay is 192µs. This means, that there will be a correct number of pulses, but it may take up to 192 µs until the “missing” pulses are synthesized by the interpolator. Therefore, when triggering the pulse counter with a real time reference it may appear that pulses are missing, but they are actually just delayed by 192µs behind.

The number of pulses within 192µs can be calculated as
(rpm * 360 / 60) * 192µs * (pulses_per_rev / 360) = rpm* 0.00327 (for 1024 pulses)
The best way to determine whether there are missing pulses or not is to count he number of pulses from one (AS5040-) index pulse to the next. These pulses must always be the same, independent of speed up to 10.000 rpm and more.

Typically, a speed measurement can be made by measuring the pulse width of an incremental pulse. However, due to the transition noise, a single pulse will not give adequate accuracy. Therefore this kind of measurement requires averaging of more than one pulse sample. The more averaging is used, the lower will be the transition noise and hence the jitter of the pulse width. An averaging of four samples will reduce the noise by 6dB or 50%
There are several ways to accomplish this:

• Measure several pulses and perform a digital averaging. This type of measurement is useful when the measured speed changes over a wide range, as it always uses a defined number of pulses for averaging
• Count the pulses in a given time window. This type of measurement is very simple and easy to accomplish, however it may have to be adapted depending on speed, e.g. at slow speeds there may be too few pulses in a given time window or the window may be too long for high speeds
• Measure the frequency of the incremental pulses, e.g by using an F/V converter.

Incremental jitter noise is defined as the variation in width of an incremental pulse at a constant rotation speed :

Jitter noise = (pw_max -pw_min ) / pw_nom
Pw_max /-min = maximum/minimum pulse width delivered by the system
Pw_nom = nominal pulse width defined by the number of pulses per revolution and by the rotating speed

To reduce the jitter noise, try to use a strong magnetic field, as this will improve the signal/noise ratio of the Hall sensor signal.

Another way is to reduce the incremental resolution. A reduction of resolution by 1 bit reduces the relative jitter noise by 50%, e.g. an AS5040 programmed to 9-bit (128ppr) resolution will have only half of the jitter noise (in %) as the same chip with 10-bit (256ppr) resolution. The absolute jitter noise certainly remains the same (0.12° rms), independent of resolution, but reducing of resolution by 1 bit doubles the width of the pulses (pwnom) and hence the relative jitter noise (pwmax-pwmin ) / pwnom is reduced by 50%.

Unfortunately it would not work. The AS50xx principle requires a 2-pole magnet as you need to be able to detect only one unique position per turn. A multipole magnet would have more than one identical images over a full turn.

The PWM base frequency of the AS5040 is factory trimmed to 976Hz +/-5%. = 927….1024Hz ,
which is equal to a total period of 976µs ….1079µs for one PWM signal (pulse ton + pause toff ).

The PWM base frequency of the AS5045 is factory trimmed to 244Hz +/-5%. = 232….256Hz, which is equal to a total period of ms 3903µs….4314µs for one PWM signal (pulse ton + pause toff ).

Optionally, the AS5045 PWM base frequency can be OTP-programmed for half that frequency (double the total period)

If you use only the PWM pulse width (ton) to determine the angle of the magnet, your reading will only be as accurate as the trimming of the PWM frequency, essentially +-5% at room temperature, +-10% over the full temperature range.

However, by measuring the full duty cycle (measure both ton and toff) you can cancel these tolerances and get the same high accuracy as the absolute output over the serial interface:

and

If the PWM output is used to generate an analog output, it can be obtained by adding a simple analog lowpass filter to the PWM output (see AS5040/45 datasheets). The jitter noise can be eliminated by designing an adequate low cutoff frequency for the filter and by using higher order filters.

The AS5035/40 has an internal sampling rate of 10.42kHz, meaning it gets a new absolute reading every 96µs. At 1024 positions per revolution, the maximum speed is 1024 positions x 96µs = 98.3ms per revolution. Or, 10.18 revolutions per second, which is 610 revolutions per minute (rpm).

• Absolute Mode (serial interface)
  The AS5040 can still be used at speeds much higher than 610rpm,but it will miss some absolute positions from one reading to the next, so the actual number of readings per revolution will decrease:

  the number of samples per revolution (n) is:
  
  

  

  rpm= rotational speed (revolutions per minute)

• Incremental Mode
  The built-in Interpolator can determine exactly how many positions it missed since the last reading and re- synthesizes the missing pulses. This way the system guarantees no missing pulses up to 10,000 rpm and higher. However, it requires two sampling periods until the synthesized waveform is regenerated. Therefore the output delay is 192µs.

The period and phase jitter of 0.12° rms (see Q47:Why is the incremental pulse width different even at constant rotating speed) always relates to the minimum step size of 0.35° (10-bit LSB). At high speeds, the magnet will travel more than one step between two consecutive measurements. The interpolator will synthesize the missing pulses with equally spaced pulses and the total phase jitter of 0.12° rms is divided by the number of steps that have been traveled in total.

Example: assume the magnet moved from position 1 to position 7 from one reading to the next. It jumped 6 steps ahead. You need 4 steps to generate one quadrature AB pulse, so the interpolator will synthesize 1½ incremental pulses. The total jitter per period is then reduced by a factor of 1,5.

Yes, you can! In BLDC-mode, the UVW signals are available on pins
3 (A_LSB_U),
4 (B_Dir_V) and
6 (Index_W).
An incremental pulse output with 512 ppr is available on pin 12 (PWM_LSB)

and the absolute serial interface is still available on pins 9 (DO), 10 (CLK) and 11 (CSn)

The AS5043 has a built-in error detection for the analog output, which switches the analog output off when the magnetic field is in the red range (see Q29: At which magnetic field levels do I get a MagInc/-Dec, LIN- and COF-Alarm?).

It is possible to disable this feature by OTP programming. In this case the analog output will always be on, independent of the magnetic field strength.

The output stage of the AS5043 is an OP-AMP where the inverting input is available on a pin (FB). This pin must be connected to feedback resistors from the analog output pin to form a non-inverting amplifier.

Without any feedback, the OP-AMP is used in open loop mode and will fluctuate or be stuck at the supply rails (VDD5V or VSS). By default, these resistors must be applied externally (see datasheet). As a programming option, the feedback can be switched to a fixed internal gain. In this case, no external resistors are required.

You can use the Zero-Position-Programming feature of the AS5000-series magnetic rotary encoders not only to program the Zero- or 0000-Position at a defined mechanical position, but you can also define any output value for a given mechanical position. This is done by adding an offset to the zero position register.

Simply, to program the output for zero at a given mechanical position, you perform the following action:

• a) Assemble the sensor; the orientation of the magnet is not critical
• b) Move the magnet to the mechanical position for which you want to set the output to zero
• c) Read the angle of the magnet at this position
• d) Write and program the value, obtained at step c) into the OTP zero position register (see programming section in the data sheet)
• e) Reboot the sensor by cycling the power. The digital output will now show 0000 at this mechanical position

Now, let’s assume you don’t want the sensor to read 0000 at your mechanical position, but rather it should be at the mid point between minimum and maximum reading. For example, you have an AS5043 with analog output in 90° mode and you want the output to be at VDD / 2 at your mechanical stop position:

Repeat the steps a)…c) :

• a) Assemble the sensor; the orientation of the magnet is not critical
• b) Move the magnet to the mechanical position for which you want to set the output to VDD/2
• c) Read the angle of the magnet at this position

Now you need to add an offset to this value. The offset is the angle from zero to your desired value. In case of the example above, an output voltage of VDD / 2 in 90°-mode would require a rotation of 45° or a digital angle of 128 steps (1024 steps would be 360°).

Consequently, you add 128 to the value obtained in step c). If this sum exceeds 1024, subtract 1024 from the sum.

• d) Now write this result into the OTP zero position register and set the AS5043 to 90°-mode by setting the OTP bits OR1/OR0 to 10. You can program both settings in one step (see programming section in the data sheet)
• e) Reboot the sensor by cycling the power. The digital output will now show 0128 at this mechanical position and the analog output will be at VDD / 2 (in 90°-mode)

Due to the sampling rate of 10.4kHz / 2.6kHz you won’t be able to get 4096 readings per revolution at high rotating speeds. However, each reading you get has 12-bit resolution.

For high speeds, it is recommended to use the incremental outputs available on AS5035 and AS5040, as they can be processed much easier (simply by counting pulses rather than processing high speed serial data).

The interpolator circuit used for the incremental outputs also guarantees “no missing pulses” even at high rotating speeds.