M8C module. The application note also explains the potential impact on GNSS
CAM-M8
TCXO-to-crystal migration guide
Application note
Abstract
This document provides options and guidelines for migrating TCXO-based CAMcrystal-based CAMperformance and other possible hardware/firmware concerns.
UBX- 21004953 - R01 Document information Page 2 of 16
C1-Public
, including but not limited to, with respect to the accuracy, correctness, reliability and fitness for a particular
-blox.com.
-blox AG.
-blox.
-blox assumes no liability for its use. No warranty, either express or
parties of this document or any part thereof is only
-blox at any time without notice. For the most recent
CAM-M8 - Application note
Contents
Document information ................................................................................................................................ 2
2.2 Power requirements ................................................................................................................................... 5
2.4 Temperature ................................................................................................................................................ 7
2.5 RF front-end design and ground plane ................................................................................................... 7
Related documentation ........................................................................................................................... 15
Revision history .......................................................................................................................................... 15
UBX- 21004953 - R01 Contents Page 3 of 16
C1-Public
CAM-M8 - Application note
1 Introduction
This application note describes the migration procedure from TCXO-based CAM-M8Q module to
crystal-based CAM-M8C module.
The CAM-M8Q uses a TCXO, while the CAM-M8C uses a crystal. This small difference in the internal
oscillator leads to some considerations described in this document. For example, the frequency
tolerance of the crystal is wider than that of TCXO. This means that the receiver must search over a
wider range of frequencies, which will extend the time-to-first-fix (TTFF) especially in weak signal
conditions.
In addition, the crystal’s frequency is highly sensitive to temperature-variant environments.
Therefore, the operating temperature, as well as the heat dissipating systems on the board need to
be taken into consideration.
Nevertheless, with proper adjustments and design guidelines, crystal-based GNSS receivers can
achieve very similar performance to a TCXO-based solution and are thus worth considering as an
alternative to many applications.
☞ This document is still under development. New or additional information (e.g. test data) might
be added in the future.
UBX- 21004953 - R01 Introduction Page 4 of 16
C1-Public
CAM-M8 - Application note
Front-end
Same
Same
2 Migration guidelines
The CAM-M8Q and CAM-M8C have a different type of internal oscillator, and a different source for
real-time clock (RTC).
This section provides details on aspects that need consideration during CAM-M8Q to CAM-M8C
migration.
2.1 CAM-M8(Q/C) comparison
The table below summarizes the specifications to be considered during the migration.
Table 1: CAM-M8Q to CAM-M8C migration comparison (default mode: GPS & GLONASS including QZSS, SBAS)
Dynamic Tracking (dBm) -167 -164
TTFF (sec)
1
2
3
Recommended Highly recommended
Same Same
30 105
15 100
Same
Same
☞ When migrating to crystal-based CAM-M8C module, make sure the receiver is not operated in
Galileo-only mode. Crystal variants are not suitable for Galileo-only operation due to worse
performance (TTFF, sensitivity).
2.2 Power requirements
The crystal-based CAM-M8C allows a wider voltage supply range. This is due to the lower voltage
required by the crystal. Nevertheless, the products have overlapping operational voltage ranges and
similar current consumption.
The table below shows the expected current draw of CAM-M8Q and CAM-M8C. More information is
available in the CAM-M8 Data Sheet [1].
1
Mainly for applications where the GNSS module is under thermal activity on the board.
2
Voltage supply = 3.0 V and single crystal feature enabled on CAM-M8C.
3
Cold and hot start under good GNSS visibility and using power levels of -130 dBm.
As shown in Table 2, CAM-M8C module consumes a bit higher current than CAM-M8Q during the
acquisition time (Icc Acquisition). For applications with limited external battery capacity, be aware
that under weak signal conditions, the crystal-based CAM-M8C has longer TTFF compared to the
TCXO-based CAM-M8Q (refer to test results presented in Figure 2).
Note that for those applications that wait for the initialization message to start operation at the
startup, the delta time may vary when migrating to CAM-M8C. The variation, in the order of 100 ms,
occurs especially when the voltage ramp is slow, and the BBR memory is not maintained (no external
backup supply).
Section 2.3 explains why the crystal-based CAM-M8C has higher hardware and software backup
current compared to TCXO-based CAM-M8Q.
Contact u-blox technical support if the current consumption aspects mentioned above affect your
application.
2.3 Real-time clock (RTC)
The CAM-M8C is the optimized variant for cost-sensitive applications. One of the reasons is the
absence of the RTC (32 KHz oscillator) compared to the TCXO-based CAM-M8Q.
In the CAM-M8Q, the RTC will provide a reference time during the off times, considerably decreasing
the time-to-first-fix when switching on again. For that, the V_BCKP pin needs to be supplied with an
external battery that keeps the RTC alive during the off times.
However, the CAM-M8C compensates for it by using the crystal signal as the RTC. For that, the
crystal needs to be powered during the hardware and software backup modes, resulting in higher
current consumption compared to the TCXO version. The difference in the backup current for each
product can be seen in Table 2 and is relevant for battery-powered devices.
Using the crystal as an RTC feature is called “single-crystal”, and by default it is enabled in the
CAM-M8C. It can be permanently disabled by sending the following command:
4
Use this figure to dimension the maximum current capability of power supply. Measurement of this parameter is with 1 Hz
bandwidth.
5
Simulated GNSS constellation using power level of -130 dBm. VCC= 3 V.
6
Average current from startup until the first fix.
7
Use this figure to determine the required battery capacity.
☞ Once the disable command is sent, it cannot be reversed.
2.4 Temperature
The frequency drift for crystals and TCXO oscillators is very dependent on the ambient temperature.
Although the receiver can correct such offset, it is recommended to avoid quick temperature changes.
As a brief explanation, a GNSS receiver can track satellite signals up to a certain high dynamic value,
which is defined as Delta frequency/ Delta time (Δf/Δt). As a result, a temperature change in a very
short time at the crystal will end in a very high dynamic, in the worst scenario losing phase lock.
Although both crystal and TCXO are highly sensitive to any quick temperature changes, due to the
wider frequency range of crystals compared to TCXO, special attention is needed for crystal-based
designs.
If it is possible that the receiver is placed under these conditions, it is highly recommended to isolate
the module by thermally minimizing the thermal conduction over the PCB and place the module far
from fans or other components with quick body temperature changes that can increase the board
and ambient temperature. Adding elements for heat dissipation between the receiver and other
elements as well as increasing the surface contact area of the board around stabilizes the
temperature.
The effect of the temperature on the crystal can be seen iFigure 1 below. u-blox crystal-based
modules can easily re-adjust the frequency drift for normal operation. It is important to mention that
all crystal oscillators qualified by u-blox pass extensive tests to ensure such smooth frequency drift
over full operation temperature range (-40 to +85 °C).
Figure 1: Temperature effect on various crystal-based modules
2.5 RF front-end design and ground plane
CAM-M8Q and CAM-M8C concurrent GNSS modules are designed with an integrated GNSS chip
antenna and SAW+LNA as RF front-end, SAW filter providing better jamming immunity and LNA for
signal amplification.
Consequently, both CAM modules have the same system noise figure and out-of-band immunity with
same signal reception quality.
As described in the CAM-M8 Hardware Integration Manual [2], the CAM-M8Q/C modules offer the
option to switch to an external active antenna. Using an external active antenna can ensure good
signal reception, and therefore mitigate the potential performance degradation caused by migrating
to crystal-based CAM-M8C solution.
If the application uses mainly the integrated chip antenna, it is very important to follow u-blox’s
recommendations about placement of the CAM-M8 module on the PCB, and the required ground
plane design for antenna gain and radiation pattern described in the CAM-M8 Hardware Integration
Manual [2]. The shape and dimensions of such a ground plane plays very important role on system’s
C/N0 values of the GNSS signals, and thus could impact the overall device performance.
Note that these recommendations become even more important when switching from the TCXObased CAM-M8Q to the crystal-based CAM-M8C.
2.6 Performance
2.6.1 Startup sensitivity and TTFF
Sensitivity tests are carried out in a lab environment, where most of the parameters can be controlled
and adjusted one by one. This also includes GNSS signal power levels. Due to the integrated GNSS
chip antenna in the CAM-M8Q/C modules, for test repeatability in a controlled environment, an
external RF signal is used for CAM-M8 tests, feeding GNSS signal directly to the RF_IN pin on the
GNSS chip.
Crystal-based GNSS receivers are characterized as having a longer time to synchronize with GNSS
signals. The effect is more visible when the signals are weak and the GNSS visibility is poor.
Such behavior can be seen in Figure 2, where the times-to-fix on a crystal-based M8 chip become
longer than those of the TCXO-based chip as the GNSS signal power drops.
☞ Note that the values in the horizontal axis are not linear. If all levels were present at the
horizontal axis, the curve would be plain until -140 dBm, where it would increase exponentially
with weaker signals.
Figure 2: TTFF vs. signal power in dBm and equivalent C/N0 inside parenthesis for crystal and TCXO variants during cold
8
starts
(default mode: GPS & GLONASS including QZSS, SBAS)
8
Results obtained on our test sites using a good LNA in front and an attenuator to decrease power level.
In general, a strong signal will give the shortest time-to-first-fix. At room temperature (+25 °C), the
TTFF differences between the TCXO variant (orange line in Figure 2 ) and the crystal variant (blue line)
grow as the GNSS signal levels drop. Figure 2 shows that under a strong signal environment, the TTFF
is very similar for both TCXO and crystal products.
The GNSS signal power levels above 43 dBHz (-130 dBm) are considered as strong signals. The cold
start results in Figure 2 show that the TTFF numbers are still very close to each other even at weaker
signal condition of 33 dBHz (-140 dBm). Such Carrier-to-Noise ratio (C/N0) levels should be achievable
with good open-sky visibility (best to have the satellite at the Zenith) with a proper PCB layout design.
For most crystal-based GNSS receivers, TTFFs degrade with weak signals and at the limits of the
operating temperatures (e.g. -40 and +85 °C), as mentioned in section 2.4. As an example, a receiver
which starts at -35 °C will gradually increase the crystal temperature due to both components’
proximity (self-heating), which results in an increase of the clock drift during the acquisition of the
GNSS signals. Nevertheless, the crystal variants have showed a very good behavior under those
temperatures, as shown in Figure 3. The deviation associated with the operating temperature is not
relevant when GNSS signals are strong enough.
Figure 3: TTFF vs. signal power for crystal and TCXO variants during cold starts at +25, -35, and +85 °C (default mode: GPS
and GLONASS including QZSS, SBAS)
Figure 3 shows that at the extreme operating temperatures (-35 °C and +85 °C), both crystal and
TCXO variants performed worse, especially in a weak signals environment.
As a summary, the longer TTFFs due to the crystal’s wider drift and the extreme operating
temperature may cause devices to have lower C/N0 values in the existing design. Proper ground plane
design is important to mitigate such possible performance degradation, as described in section 2.5
and the CAM-M8 integration manual [2].
Road tests shows real behavior in dynamic scenarios. The road tests allow measuring the position
accuracy delivered by the receiver. The accuracy, calculated as the offset to the real position, is
showed in error percentiles for 2D and 3D coordinates.
Three different road tests have been carried out for both crystal and TCXO variants. The goal of these
tests is to assess the impact of different signal power levels and to see if the degradation is similar.
As mentioned earlier, for CAM-M8 chip antenna modules, the GNSS signal is fed directly to the RF_IN
pin of the chip, skipping the on-board RF front-end components (SAW and LNA) of CAM-M8Q and
CAM-M8C. The differences seen for crystal and TCXO variants in the following figures can be
extrapolated to the differences to be expected on the CAM-M8Q/C variants with different C/N0
signals.
☞ The C/N0 value in the following figures and tables is the median of all GNSS signals used for
tracking along the test.
☞ The test results are based on limited samples and should be considered as a reference.
2.6.2.1 Rural areas with good GNSS visibility
The test in a rural area is characterized as having excellent GNSS visibility most of the time,
alternating with weak signal areas where there are trees and small houses along the road.
Figure 4 shows the error percentiles for both TCXO and crystal variants at two different signals power
levels. One is very strong with an average C/N0 of 44.7 dBHz, and the second one with 31.6 dBHz (13
dBHz lower). As shown in Figure 4 and Table 3, the position accuracy of TCXO and crystal variants is
very similar under both scenarios, the same for the degradation (Δerror/ Δsignal attenuation).
Figure 4: Position error in meters for crystal and TCXO in percentiles at 31.6 and 44.7 dBHz in rural areas (default mode: GPS
and GLONASS including QZSS, SBAS)
Table 3: Position error in percentiles for crystal and TCXO variants at different signal power levels in rural areas (default
mode: GPS and GLONASS including QZSS, SBAS)
31.65 dBHz
M8_Crystal
Strong signals
44.75 dBHz
M8_TCXO
44.8 dBHz
M8_Crystal
Note that GNSS signals around 32 dBHz are in line with the threshold defined in section 2.6.1, where
TTFF numbers of both TCXO and crystal variants are still very close to each other even at the weaker
signal condition of 33 dBHz (-140 dBm).
The rural road test results further confirmed that crystal-based u-blox module can achieve very good
position accuracy under strong/good signal condition. It even reaches a similar accuracy level
compared to a TCXO-based solution in most cases under weak signal scenario.
2.6.2.2 Urban canyon areas
As with the rural area test, urban canyon road rests take place in good signal (36.5 dBHz) and in weak
signal (29.5 dBHz) scenarios. As expected, in the urban canyon environment, the errors in the position
increase for both TCXO and crystal versions as shown in Figure 5 and Table 4.
Figure 5: Position error in meters for crystal and TCXO in percentiles at 29.5 and 36.5 dBHz in urban canyon (default mode:
GPS and GLONASS including QZSS, SBAS)
Table 4: Position error in percentiles for crystal and TCXO variants at different signal power levels in urban canyon scenario
(default mode: GPS & GLONASS including QZSS, SBAS)
29.65 dBHz
M8_Crystal
Good signals
36.55 dBHz
M8_TCXO
36.55 dBHz
M8_Crystal
Note that although the position errors of both TCXO and crystal variants are very big in an urban
canyon area, such performance is normal and expected for all standard precision receivers under such
particularly challenging environment. Based on position accuracy comparison shown in Table 4, the
position accuracy is similar between crystal and TCXO-based solutions, independent of the signal
power levels.
The real track used in urban canyon test is shown in Figure 6 below.
Figure 6: Scenario used for “Urban canyon” to compare performance between TCXO and crystal variants
Finally, a highway scenario has been used in the road test, under good GNSS signal and weak signal
conditions. In this case, the receiver calculates a position where conditions change rapidly on a
highway due to the car speed. Figure 7 captures a part of the drive and gives a good representation of
the test conditions.
Figure 7: Part of the “Highway” scenario used and track of the receivers
The higher speed is more challenging for GNSS receivers due to the tracking loops. The highway
scenario means the tracking is more difficult. Thus, the degradation of the signal levels has stronger
influence on the position accuracy as shown in Figure 8 and Table 5.
Figure 8: Position error in meters for crystal and TCXO in percentiles at 28.5 and 42 dBHz in a highway test (default mode:
GPS and GLONASS including QZSS, SBAS)
Table 5: Position error in percentiles for crystal and TCXO variants at different signal power levels in highway scenario
(default mode: GPS and GLONASS including QZSS, SBAS)
28.5 dBHz
M8_Crystal
Good signals
41.95 dBHz
M8_TCXO
42 dBHz
M8_Crystal
Highway test results demonstrate once again that u-blox crystal-based receivers have very similar
position accuracy compared to the TCXO-based variant under both weak and good GNSS signal
condition on highway. It is also clear that the highway weak signal scenarios cause worse position
accuracy of the GNSS receiver, independent of the TCXO or crystal oscillator.
3 Conclusion
From startup sensitivity and TTFF test (section 2.6.1) and road tests (section 2.6.2), we can see that
when GNSS signals levels are good (above 35 dBHz), TTFF and position accuracy are not much
impacted when migrating from TCXO-based CAM-M8Q to crystal-based CAM-M8C. In this case,
migration to CAM-M8C will not significantly impact the overall GNSS performance.
However, under weak GNSS signals scenario, TTFF and tracking sensitivity of crystal-based
CAM-M8C are a bit worse than those of the TCXO-based CAM-M8Q. It is important to notice that the
performance of the CAM-M8Q/C chip antenna modules depend very much on the PCB layout and
ground plane dimensions. These two parameters will have huge impact on the C/N0 values of devices
using CAM-M8 modules, thus special care is needed during CAM-M8Q to CAM-M8C migration.
For designs using an external battery as the main power supply, the time-to-first-fix position after
powering off or entering power save mode needs to be considered as well. If the signals are weak or
there is limited satellite visibility, the acquisition periods will increase, and consequently reduce the
battery lifetime. If, in addition, the application uses the HW and SW backup modes, current
consumption becomes even more relevant when switching to crystal-based CAM-M8C module.
Contact u-blox technical support team in case you need further guidelines and recommendations
during the CAM-M8Q to CAM-M8C migration.
To sum up, CAM-M8C is a good crystal-based solution for applications where optional external active
antenna is in use or operation with a weak signal is not necessary.
UBX- 21004953 - R01 Conclusion Page 14 of 16
C1-Public