White Paper

High-Precision Sine/Cosine Interpolation

For speed control or positioning, drive controllers today demand high-

resolution magnetic or optical position sensors which require special integrated
circuits for sensor signal conditioning and sine/cosine-to-digital conversion.
This white paper describes the methods and challenges of the “interpolation”
using sine/cosine-digital conversion (S/D conversion), it discusses sensor-
related measuring errors as well as their compensation, and it illustrates the
latest chip solutions and their selection.

Contents:

1. Methods of Sine/Cosine-to-Digital Conversion

1.1 Flash Conversion
1.2 Vector-Tracking Conversion
1.3 SAR Conversion with Sample-and-Hold Stage
1.4 Continuous-Sampling A/D Conversion
1.5 Interpolation Components in Comparison
2. Measuring Errors with Examples
2.1 Concepts for Signal Conditioning
3. Summary
4. Literature

©2014 iC-Haus GmbH Page 1/17 www.ichaus.com

White Paper

1. Methods of Sine/Cosine-to-Digital Conversion

High-precision magnetic and optical sensors [1] provide the angle or length
information coded in a 90 degree shifted sine and cosine signal. The interpola-
tion is responsible for the non-linear A/D conversion to transform sine/cosine
signals into angle steps (see figure 1) that are displayed either incrementally as
so-called quadrature signal, or as absolute data word which represents the sine
signal’s phase angle.

Figure 1: Angle conversion via “interpolation”

The non-linear conversion function is usually the arc tangent, so that the phase
angle PHI can directly be determined from the sine and cosine voltage.

Multiple A/D conversion concepts can be applied:

- a flash conversion, such as the one in iC-NV, which uses many individual
comparators;

- a vector-tracking conversion, such as in iC-NQC and iC-MQF, in which

only a few comparators control a counter upwards or downwards in or-
der to first pick up, and then track the input angle;

- a SAR conversion (such as in iC-MR), which is similar in essence to the

vector-tracking conversion, but which holds the input signal until the
counter value has approached;

©2014 iC-Haus GmbH Page 2/17 www.ichaus.com

White Paper

- as well as with a linear A/D converter (such as in iC-TW8), which digitaliz-

es the sine and cosine signals separately, and subsequently calculates the
angle.

Fully integrated magnetic and optical single-chip encoders, such as iC-MU or iC-
LNB, use vector-tracking conversion to offer position data in real time [1, 2].

1.1 Flash Conversion

Figure 2 shows a flash conversion with many individual comparators, which

each switch at different tangent function thresholds. At least one comparator is
involved per bit of angle resolution, which implies a considerable hardware
effort, and accordingly demands a lot of chip space – unless precision circuits
are renounced. Therefore, this concept is only compatible with a relatively low
resolution and only if the requirements for precision are not too high.

Figure 2: Flash Conversion

The fast conversion is advantageous: the comparators work parallel and switch
almost simultaneously. Due to the development of switching spikes during
settling, a patented edge distance control that creates equalization is applied.

©2014 iC-Haus GmbH Page 3/17 www.ichaus.com

White Paper

Successive edges coming too close are shifted, so that an actually countable
output signal is generated – the circuit functions as a filter, however, undis-
turbed input signals pass without a delay, i.e. this filter function shows no
latency.

The flash conversion does not require sampling. Thus, the generated quadra-
ture signals feature an “analog” jitter behavior as these are not aligned to any
clock signal – which is ideal for speed controls. Typical applications which
benefit from this type of converter are optical and magnetic motor encoders.

1.2 Vector-Tracking Conversion

The vector-tracking conversion is used for higher resolutions (see figure 3). It
has one primary comparator which controls a counter upwards and down-
wards. The digital counter value feeds a D/A converter which produces an
analog tangent signal. This tangent signal is mixed with the cosine, which then
produces a sine signal – afterwards, sine is compared to sine.

Figure 3: Vector-Tracking Conversion

When the system is settled, the counter contains the phase angle and tracks
every input change step by step, or rather bit by bit – jumps are not possible. It

©2014 iC-Haus GmbH Page 4/17 www.ichaus.com

White Paper

is advantageous that this system as well functions virtually clock-free, and that
it is activated only by an input change – the latency period is relatively short.

Since only one comparator is required, its design can be made for precision. An
additional advantage regarding the accuracy is that a potential circuit offset
error would affect all switching points the same way – comparable to a hyste-
resis. The tracking converter’s incremental output signals jitter analog as well.
A clock alignment is not visible until the adjustable limitation of the maximum
tracking rate is reached, e.g. through a glitch at the input signals.

Due to its real-time characteristics as well as its high resolution, this type of
converter is preferred for linear position measuring systems.

1.3 SAR Conversion with Sample-and-Hold Stage

For absolute measuring systems that do not have to output incremental

signals, sampling converters as shown in figure 4 are suitable. The SAR convert-
er works similarly to the vector-tracking converter, with the exception that the
approximation register reaches the adjacent phase angle much faster because
its steps can be larger and it does not have to track bit by bit.

Figure 4: SAR Converter with Sample-and-Hold Stage

©2014 iC-Haus GmbH Page 5/17 www.ichaus.com

White Paper

Triggered by the external data request, the input signals are frozen up through
a sample-and-hold circuit. In this system, the analog settling time primarily
determines how fast and with what precision the conversion can be executed.

This type of converter can usually be found in motor control systems and
inverters which process analog encoder signals – or in position encoders, if very
high angle resolutions are required.

1.4 Continuous Sampling A/D Conversion

The typical approach: iC-TW8 uses continuously operating linear A/D convert-
ers (figure 5) and calculates the phase angle afterwards. The advantage here is
its digital signal processing: Signal errors can be deducted either one-time by
pushing a button to ease initial calibration, or permanently by automatic
functions to compensate sensor drift.

Figure 5: Sampling A/D Conversion

Signal filtering now makes it possible to reach resolutions which exceed the
actually available A/D converter resolution. The synthetically produced incre-
mental output signals show a perfect duty factor of 50 % and are nearly free of
jitter. However, due to signal processing a constant latency time of a few

©2014 iC-Haus GmbH Page 6/17 www.ichaus.com

White Paper

microseconds is introduced, which may need to be considered in control

systems.

Primary target applications are high-resolution linear length gauges, as well as

rotary encoder systems which benefit from the supplied automatic signal
correction.

1.5 Interpolation Devices in Comparison

Needless to say, the application determines which converter type is suitable:

The tracking converters iC-NQC and iC-MQF persuade with real-time, the
minimum latency of less than 250 ns is mainly determined by analog path
runtimes.

With the sampling converters iC-MR and iC-TW8 the required settling time to
the measuring value is decisive (see table 1) and can therefore limit the possi-
ble sampling rate. While iC-MR resolves the angle position with 13 bit within 2
µs, the continuously operating iC-TW8 requires 6 samples in total and 24 µs in
order to renew its position data. On the other hand, if the speed is constant, iC-
TW8 can reduce the existing latency to approximately 4 µs via an adjustable
digital filter. As common for a resolver evaluation, the output position catches
up the actual input angle – though in a significantly shorter time period.

Table 1: Conversion Characteristics

©2014 iC-Haus GmbH Page 7/17 www.ichaus.com

White Paper

Besides the resolution, the accuracy must be considered as well, which is not
only related to the A/D converter core’s quality, but also to the dimensioning of
the signal conditioning. Each D/A converter that takes corrective actions in the
signaling path requires chip space, and consequently causes costs – an optimi-
zation task for the circuit designer. The comparison of the devices in table 2
shows that the converter iC-MQF provides a lower resolution than iC-NQC.
However, it shows a higher precision due to a more finely graduated signal
conditioning.

Safety-oriented encoder systems demand additional functions: the device iC-

MR possesses special diagnostic functions, e.g. for signal and temperature
monitoring, for memory checks, as well as for error simulation. For controller
communication, one parallel interface as well as various serial interfaces are
available. The configurable position data output via BiSS C can be carried out
with life-cycle counting and an extended 16 bit CRC.

Table 2: Operational Characteristics

©2014 iC-Haus GmbH Page 8/17 www.ichaus.com

White Paper

2. Measuring Errors with Examples

The following example in figure 6 of a magnetic pole wheel scanning using MR

sensors shows which measuring errors need to be considered, if necessary.

Figure 6: Application Example with Error Sources

Potential error sources can be:

- an imprecisely magnetized measuring target

- signal errors of the MR sensor regarding offset and amplitude

- sin/cos phase error through imprecise sensor alignment

- signal errors through false or insufficient conditioning

- measuring error through imprecise conversion

Without counteractions the interpolation result will be faulty and incremental

output signals will stand out due to an excessively large jitter. The position
jitter in consequence of mechanic angle alteration is certainly acceptable; jitter
due to the measuring system, on the other hand, is not – unfortunately, a
distinction and allocation is not possible.

©2014 iC-Haus GmbH Page 9/17 www.ichaus.com

White Paper

Therefore, a precise knowledge of potential error sources is essential. The

formula for angle calculation shows which signal errors need to be considered:

Formula: Angle Calculation via Arc Tangent Function

Relevant error sources are offset voltages, deviations from the ideal phase
difference, differences between the sine and cosine amplitude, and possibly
also distortions of the wave shape through harmonic contents. Thus, it would
be important to know if all signal errors always need to be “conditioned” or if it
is allowed to neglect.

Three case examples for estimating the required conditioning accuracy:

• magnetic, on-axis, 1 CPR: for 0.1° (12 bit) accuracy:

signal errors < 0.2 % required (@ 200 Hz)

• magnetic, off-axis (32 pp), 64 CPR: for 0.1° (12 bit) accuracy:
signal errors< 12.8 % required (@ 12 kHz)

• optical, off-axis, 2048 CPR: for 20” (16 bit) accuracy:

signal errors < 22 % required (@ 400 kHz)

©2014 iC-Haus GmbH Page 10/17 www.ichaus.com

White Paper

Example 1: If an angle accuracy of 0.1° is expected mechanically (12

bit/revolution) by an on-axis Hall sensor system which supplies only one sine
period per revolution, it can be deduced that each signal error must be lower
than 0.2 %.
This precise conditioning is certainly achievable, though the manual adjustment
is very time-consuming and presents a possible challenge for the available
measuring equipment. See tools at http://www.ichaus.de/tools.

Suitable devices are: iC-NQC, iC-TW8, iC-MR.

Example 2: While sampling a magnetic pole wheel via MR sensors the required
interpolation depth, and technically also the requirements for signal precision
are being reduced. Nonetheless, a more accurate conditioning will be preferred
depending on how precisely the measuring target has been magnetized.
The input frequency rises with the number of poles – which is not a problem
for vector-tracking converters due to the reduced interpolation factor.

Suitable devices are: iC-TW2, iC-MQ, iC-NQC, iC-TW8.

Example 3: When talking about an optical encoder system with e.g. 2048 sine
periods per revolution, which should be resolved more finely, the requirements
for the signal conditioning do not seem to be too high. However, the scale error
generally already reaches the maximum permissible measuring error, so that
an additional signal conditioning error cannot be tolerated (see table 3).
Therefore, and because of the high input frequency, significantly higher de-
mands can be made on the interpolation circuit. Sampling components such as
iC-MR are required.

Table 3: Angle Error Depending On Calibration

©2014 iC-Haus GmbH Page 11/17 www.ichaus.com

White Paper

2.1 Concepts for Signal Conditioning

In order to achieve a good interpolation result, sensor signals need to be

conditioned [3]. Components iC-MQF and iC-MR apply an analog front end
(AFE, see figure 7) for signal conditioning, which is adjusted via multiple DACs.
In contrast, iC-TW8 applies a self-adjusting digital signal correction.

AFE for Signal Conditioning

Figure 7: Analog Front End for Signal Conditioning

Precise instrumentation amplifiers offer a coarse amplification for signal

adaption and balance signal differences through fine adjusters. A further aspect
is the offset correction in the front end via D/A converters, which can track
their correction signal-dependently. The front end can measure the DC compo-
nent in the signal, as well as the sensor supply and use it for reference. Addi-
tionally, a current controller provides stable conditions e.g. by powering the
MR bridge or the LED of optical systems. The advantage here is that, if adjusted
at room temperature, the calibration accuracy is maintained even for tempera-
ture changes.

©2014 iC-Haus GmbH Page 12/17 www.ichaus.com

White Paper

Key Features:

• integrated current/voltage conversion and voltage dividers

• offset-corrected instrumentation amplifiers
• separately adjustable coarse and fine gain
• sensor drift compensation through tracked offset references
• signal stabilization through regulated sensor supply (to sum value
or Lissajous figure)

Digital Signal Correction

In the analog path, iC-TW8 only possesses coarse adjusters for gain and offset
in order to get the input signals to a favorable range for A/D converters (see
figure 8).

Figure 8: PGA Front End with A/D Conversion and Digital Signal Correction

Accordingly, merely digital, calculated signal corrections occur. An elaborate

drift monitoring evaluates deviations from the factory calibration and can be
configured for alarms. The angle position is calculated via CORDIC algorithm.

Key Features:

• adjustable coarse gain (6 to 45 dB, 3 dB/ step)

• adjustable analog offset correction (100 mV/ step)
• digital offset and offset drift correction (244 µV/step)
• digital compensation of gain mismatching (0.02 %/ step)
• digital phase correction (0.056°/ step)

©2014 iC-Haus GmbH Page 13/17 www.ichaus.com

White Paper

Concept Strengths

Both concepts demonstrate strengths: After power on, corrections in the

analog signal path settle already in standstill of the measurement system, in
that the sensor power supply is adjusted to obtain the signal conditions at the
time of calibration. There are no additional delay times in the signal path, fast
interpolation results are possible. For initial factory calibration, measuring
equipment is required which may need to be automated.

The digital correction uses the existing movement either initially to define its
best suitable static adaption, or permanently for a dynamic drift compensation
in the application. Measuring equipment for calibration is not necessary,
recalibration in the field can either occur automatically or by pushing a button.
This is beneficial for modular systems which are installed by the customer.

The following table 4 shows a cross-comparison of the devices with regard to

the implemented correction possibilities.

Table 4: Cross-Comparison of Signal Conditioning

©2014 iC-Haus GmbH Page 14/17 www.ichaus.com

White Paper

Devices Feature Overview

iC-NQC 13-bit Signal Conditioning Interpolator IC

 Real-time incremental
 BiSS absolute with period
 BiSS slave BP1, SSI

iC-MQF Programable 12-bit Sin/Cos Interpolator IC with RS422 Driver

 Real-time decimal incremental

 RS422 fail-safe
 Controlled sensor supply

©2014 iC-Haus GmbH Page 15/17 www.ichaus.com

White Paper

iC-MR 13-bit S&H Sin/Cos Interpolator with Controller Interfaces

 BiSS or embedded
 Singleturn & multi-turn
processing
 Safety monitoring features

Key Features:
Fast S&H interpolation: 2 µs,
precision signal conditioning,
source control output (ACO), 1 Vpp
line driver output, parallel 8-bit µC
interface, serial I/O (BiSS/SSI, SPI),
I2C, 12-bit ADC (temper. sensing),
safety features

iC-TW8 16-bit Sin/Cos Interpolator with Auto-Calibration

 Self-calibrating
once/permanent
 Perfect incremental signals

Key Features:
250 ksps, 16-bit, constant latency
(24 µs), lag retrieval to 4 µs (servo
loop), binary/decimal x0.25 to
x16384, post-AB divider [1/1 to
1/32], fin 125 kHz, A/B/Z 8 MHz,
min. edge spacing tMTD 31 ns ,
automatic offset, gain, phase,
push-button calibration, distortion
compensation via LUT, signal
quality monitoring, setup by pins,
I2C, SPI, 3.3V (15 mA), 5 V

©2014 iC-Haus GmbH Page 16/17 www.ichaus.com

White Paper

3. Summary

As demonstrated through the different methods for S/D conversion, respec-

tively for interpolation, multiple important criteria have to be considered in
order to select the optimal solution. The current selection table [4] including
the newest iC solutions can also be downloaded online.

4. Literature

[1] Encoder Technologies in Comparison: Magnetic vs. Optical,

Elektronik 10/2012
[2] 18 Bit Absolut Encoder-IC, Elektronik Industrie 03/2012
[3] Easy Conditioning and Safe Transfer of Sensor Signals, Elektronik Indus-
trie 4/2010
[4] Product Selector Interpolator IC

About iC-Haus
iC-Haus GmbH is a leading, independent German manufacturer of standard iCs
(ASSP) and customized ASiC semiconductor solutions with worldwide represen-
tation. For more than 25 years the company has been active in the design,
production, and sales of application-specific iCs for industrial, automotive, and
medical applications.
The iC-Haus cell libraries in CMOS, bipolar, and BCD technologies are specifical-
ly suited to realize the design of sensor, laser/opto, and actuator ASiCs,
amongst others. The iCs are assembled in standard plastic packages or using
the iC-Haus chip-on-board technology to manufacture complete microsystems,
multichip modules, and optoBGA/QFN in conjunction with sensors.
Further information is available at http://www.ichaus.com.

©2014 iC-Haus GmbH Page 17/17 www.ichaus.com