ubx-cfg

ESP-GPS Testing

Choosing the right constellations, satellite limits, and logging rates is a complex, but crucial decision.

It would be good to identify a shortlist of optimal configurations, based on M10 and ESP32 clock speeds.

Static testing provides a controlled environment to test devices systematically, and is very easy to perform.

Be aware that the +/- values in the software may mislead you… see earlier test comparing 5 Hz and 10 Hz.

This battery of tests has been inspired by the plan for fundamentals testing, but is specific to the ESP-GPS.

Overview

There are several phases planned, which can each be split up over a number of days:

Compare both of the 2 constellation configurations - 6 hours
Test whether GPS + GLONASS + Galileo improves accuracy - 6 hours
Test whether GPS + Galileo + BeiDou B1C improves accuracy - 6 hours
Compare both of the 3 constellation configurations - 6 hours
Compare 4 constellations against 3 constellations - 6 hours

The final outcome should be an optimal configuration for the existing ESP-GPS devices.

Note: 16 Hz is not a divisor of 1000, but it is closely related to 4 Hz and 8 Hz.

Approach

Test Duration

Individual tests should record for a total of 2 hours
Ideally, ESP-32 devices should switch roles after 1 hour
2 x Mini Motions @ 5 Hz to be included in all tests

Configuration

M10 clocked at 192 MHz - high performance
Baud rate of 115,200
ESP32 devices clocked according to the logging rate
- 80 MHz for 8 Hz
- 160 MHz to 16 Hz
- 240 MHz for 20 Hz

Data Analysis

Acquisition times
- Time taken to acquire all of the signals
Cumulative results in GPS Speedreader
- Speed - 1 hour averages
- Total distance - derived from speed data
Percentiles from individual values - 50, 68, 95, 99.7
- Speed over Ground (SOG)
- Speed Accuracy (sAcc)
- Horizontal Dilution of Precision (HDOP)
Dropped frames
- Indication of CPU bandwidth being exceeded

Phase 1

Compare both of the 2 constellation configurations:

Test	ESP-GPS #1	ESP-GPS #2	Rate
1.1	GPS + GLONASS	GPS + Galileo	10 Hz
1.2	GPS + GLONASS	GPS + Galileo	16 Hz
1.3	GPS + GLONASS	GPS + Galileo	20 Hz

Total duration = 6 hours, excluding analysis

Notes:

These logging rates are for high performance mode - i.e. M10 CPU clocked at 192 MHz.
The viable logging rates for the default M10 clock speed would be 5 Hz, 8 Hz, and 10 Hz.

Expectations

I suspect that GLONASS will have the least favourable results, but it needs some actual proof.

The 20 Hz tests might result in some dropped frames, demonstrating that the CPU bandwidth has been exceeded.

Dropped frames can potential be avoided by reducing the maximum number of satellites from the default of 32.

Phase 2

Test whether GPS + GLONASS + Galileo improves accuracy:

Test	ESP-GPS #1	ESP-GPS #2	Rate
2.1	GPS + GLONASS + Galileo	GPS + GLONASS	10 Hz
2.2	GPS + GLONASS + Galileo	GPS + GLONASS	16 Hz
2.3	GPS + GLONASS + Galileo @ 16 Hz	GPS + GLONASS @ 20 Hz	-

Total duration = 6 hours, excluding analysis

Notes:

These logging rates are for high performance mode - i.e. M10 CPU clocked at 192 MHz.
The default M10 clock speed theoretically limits GPS + GLONASS + Galileo to 6 Hz.

Expectations

The addition of Galileo should improve the accuracy of the results.

The 16 Hz and 20 Hz tests might result in some dropped frames, demonstrating that the CPU bandwidth has been exceeded.

Dropped frames can potential be avoided by reducing the maximum number of satellites from the default of 32.

Phase 3

Test whether GPS + Galileo + BeiDou B1C improves accuracy:

Test	ESP-GPS #1	ESP-GPS #2	Rate
3.1	GPS + Galileo + BeiDou B1C	GPS + Galileo	10 Hz
3.2	GPS + Galileo + BeiDou B1C	GPS + Galileo	16 Hz
3.3	GPS + Galileo + BeiDou B1C @ 16 Hz	GPS + Galileo @ 20 Hz	-

Total duration = 6 hours, excluding analysis

Notes:

These logging rates are for high performance mode - i.e. M10 CPU clocked at 192 MHz.
The default M10 clock speed theoretically limits GPS + Galileo + BeiDou B1C to 8 Hz.

Expectations

The addition of BeiDou B1C should improve the accuracy of the results.

The 16 Hz and 20 Hz tests might result in some dropped frames, demonstrating that the CPU bandwidth has been exceeded.

Dropped frames can potential be avoided by reducing the maximum number of satellites from the default of 32.

Phase 4

Compare both of the 3 constellation configurations:

Test	ESP-GPS #1	ESP-GPS #2	Rate
4.1	GPS + Galileo + GLONASS	GPS + Galileo + BeiDou B1C	8 Hz
4.2	GPS + Galileo + GLONASS	GPS + Galileo + BeiDou B1C	10 Hz
4.3	GPS + Galileo + GLONASS	GPS + Galileo + BeiDou B1C	16 Hz

Total duration = 6 hours, excluding analysis

Notes:

These logging rates are for high performance mode - i.e. M10 CPU clocked at 192 MHz.
The default M10 clock speed theoretically limits GPS + GLONASS + Galileo to 6 Hz.
The assumption is that GPS + Galileo + BeiDou B1C would be a better option, supporting 8 Hz.

Expectations

I suspect that GLONASS will have the least favourable results, but it needs some actual proof.

The 16 Hz tests might result in some dropped frames, demonstrating that the CPU bandwidth has been exceeded.

Dropped frames can potential be avoided by reducing the maximum number of satellites from the default of 32.

Phase 5

Compare 4 constellations against the best 3 constellations, assumed to be GPS + Galileo + BeiDou B1C :

Test	ESP-GPS #1	ESP-GPS #2	Rate
5.1	GPS + Galileo + BeiDou B1C + GLONASS	GPS + Galileo + BeiDou B1C	8 Hz
5.2	GPS + Galileo + BeiDou B1C + GLONASS	GPS + Galileo + BeiDou B1C	10 Hz
5.3	GPS + Galileo + BeiDou B1C + GLONASS @ 10 Hz	GPS + Galileo + BeiDou B1C @ 16 Hz	-

Notes:

These logging rates are for high performance mode - i.e. M10 CPU clocked at 192 MHz.
The default M10 clock speed theoretically limits GPS + Galileo + BeiDou B1C + GLONASS to 4 Hz.
The assumption is that GPS + Galileo + BeiDou B1C would be a better option, supporting 8 Hz.

Total duration = 6 hours, excluding analysis

Expectations

I suspect that 4 systems will show negligible benefit, because the maximum number of satellites is limited to 32.

The 10 Hz and 16 Hz tests might result in some dropped frames, demonstrating that the CPU bandwidth has been exceeded.

Dropped frames can potential be avoided by reducing the maximum number of satellites from the default of 32.

Summary

The intended outcome of this investigation will be the optimal configurations, based on M10 and ESP32 clock speeds.

For example:

	M10 @ 128 MHz	M10 @ 192 MHz
ESP32 @ 80 MHz	GPS + Galileo + BeiDou B1C @ 8 Hz	-
ESP32 @ 160 MHz	-	GPS + Galileo + BeiDou B1C @ 16 Hz

In order to prevent dropped frames, maximum numbers of satellites will almost certainly need to be restricted.

Different logging rates can be offered to anyone who wants smaller files, e.g. GPS + Galileo + BeiDou B1C @ 4 Hz.

Future refinements to the ESP-GPS configuration may also benefit from the findings of fundamentals testing.

Useful Tools

Use GNSS View with mask angle of 5 degrees to ascertain the expected number of satellites for each constellation.

Ignore the square satellites (geo-stationary) and focus only on the round ones (MEO).

This site is open source. Improve this page.