[Pecan Pico 7] Solar powered camera tracker for balloons

[FXCoder]

Hi Sven,
Great news that I2C is working now.

About the fake UBLOX...
From my research on the subject it seems the core GPS chip in fake modules should be genuine but the remainder of the circuitry may be less than good quality or limited (like smaller flash meaning updates won't fit).

Anyway I will take a good look at the M8N based module I ordered when it finally arrives (on a slow boat from China I think).
That one is for testing and development so hopefully it will be at least useable.

I like your BBQ SMT soldering skills!

Cheers,

Bob

[FXCoder]

Hello Sven,
Well the GPS unit I ordered from Hong Kong finally arrived.

The advertised unit and the sticker on the module shows M8N.
Using u-center Messages and looking at MON:VER I get...
Software Version: 1.00 (59842)
Hardware version: 00070000
Extensions: PROTVER 14.00 PPS;SBAS:GLO:QZSS

So it looks like it is a NEO7.

Anyway putting the fake or at least misleading aspects aside the unit does work.
It will suffice for testing at least using the UART interface for now.
With a bit of fine solder work DDC (I2C) could be perhaps be made available as well.

All the prototyping pieces are arriving (more slow boats from China) and I am coding the manager frameworks and some modules as I go.
I'm starting with non-critical stuff which helps with deciphering all the ChibiOS fun.

Regards,

Bob

[A319]

Hi Bob!

I have to point out, the ublox modules may communicate with different protocols. They are all stream based and all can be used by SPI, I2C and UART. You may be familiar with the NMEA0183 protocol. But I'm used to the UBX protocol which can be used in polling mode. So you don't have to look for the stream continuously. It has also some more features which you cannot use with the NMEA protocol e.g. activating power saving modes.

Btw. Many commits ago, I was using UART instead. So you may be interested in this: https://github.com/DL7AD/pecan-stm32f42 ... c30cca5cb4

Thank's for your pull request. I haven't tried it yet. Since I'm using the packet manager, many packets (~5%) gets lost when I transmit images with 2GFSK. This is really annoying. I hope that the sync was the problem.

Sven

[A319]

Hi,

I got the new Pecan Pico 8 boards designed. I finally did the step to move everything over to KiCAD.
Many thanks to Bob who finally helped me to find errors in the final stage.

What I changed compared to Pecan Pico 7b

=> Moved the project to KiCAD:
- So I am able to produce some amazing 3D views
- KiCAD has no limitations in size, layers, pads, etc.
- KiCAD is GPL licensed, so everything is open source and free and will be open source and free in future (isn't that great?)
- KiCAD has much more features and options than Eagle has (watch this)

=> Removed several features:
- Second transceiver (There was no practical use for our flights)
- SD card (I got it working, but since all our balloons are long duration floaters, we would never get the SD card back)
- Power amplifier (We had problems to get a proper power supply for the PA chip. Since it would have taken too much effort, to fix those problems, we decided to removed it. Also there was no big interest in it.)
- Removed I2C header (No more space available)

=> New features:
- Added Mini USB connector (for flashing the STM32, serial debugging and for supplying power)
- Added MOSFET for shorting the solar cells (so we can measure the solar short current)
- Added voltage divider for measuring USB voltage
- Added voltage boost capability (1V8 bus' voltage can be boosted to 3.0V by a STM32 GPIO. USB wouldn't work at 1.8V and Si4464 may transmit at a higher output level. 1.8V still needed for power saving)
- Replaced uBlox MAX7/8 module with a uBlox EVA7M chip (this module can only receive GPS and uses less power therefore. This chip is also smaller and software compatible to the MAX7/8 module)
- Made external TCXO switchable (for power saving)
- Added special solar connection pads which allows the solar cells being connected in series (the higher voltage provides a better SPV1040 performance)

=> Other hardware changings:
- Made the board smaller overall (17.5cm^2 => 10.0cm^2)
- This design uses less components and can be assembled faster than the old board (113 required components for all features, 6 optional components)
- Added current limitation (500mA) to SPV1040 (This has been done to prevent LiPO's from being charged too quickly by an USB supply)
- Moved current sense resistors of PAC1720 (Now battery charge and discharge can be measures but not the actual need by the circuit itself. Therefore the short current of the solar cells can be measured and the battery internal resistance calculated. When shorting solar cells, there's less noise/ripple on all supply busses which is good for image quality)
- Removed drills for the 100mil SWD header (This has been done due to lack of space)
- Added bigger BOOT0 pin at the side of the PCB (which can be easily accessed by a 100mil pinheader)
- Added testspads (especially for inaccessible pins)
- Removed all LDO's and replaced it with a single LDO chip with multiple outputs
- Moved FFC connector more to the edge of the board, so the connector can be opened more easily
- Changed design of 1st pin markings
- Connected all LED's to TIM1 for dimming them with PWM
- Connected GPS receiver to a 3.3V output of the LDO rather than providing a separate switching regulator (in order to get a cleaner power supply)
- Connected PAC1720 to VBAT (PAC1720 doesn't work at 1.8V but is also fine with higher voltages)

=> Design requirements:
- No more need for 0.1mm (4mil) thin traces, but 0.15mm (6mil) which is usually the standard accuracy at many factories
- New Version requires 0.2mm drills (they are needed especially for the BGA chip)
- New Version requires PCB 4 layers

Some more FAQ

Why is it important to measure solar short current and battery resistance?
Balloons work at really low temperatures (day -5°C, night -50°C). At night the internal resistance of the battery may increase as high as 50...100Ohm. So we don't want to consume too much power since the battery voltage would drop very far. On evening it's already a bit colder, so the battery reaches 4.1V (fully charged) even at low solar power inputs. The algorithm might incorrectly assume that the battery is healthy and enough solar power is available. As the transmission starts the battery voltage will drop. To detect such situations, it's good to know the internal battery resistance and the solar short current in advance.

What is the mayor advantage of having a USB connector?
It's actually not a part I would like to fly personally (because it adds unnecessary weight). But if I give the PCB away to a friend, he can't use it very much. He may not have a ST Link Programmer and may not have a UART-USB adapter with a 1.8V-to-3.3V level converter. So it might be handy for others, to use USB instead.






Some notes about the 3D images
- The SMA antenna and SMA connector will be replaced by a single wire on a real flight (to save weight).
- The solar cells are connected like I would assemble them for a flight
- The ublox EVA7M 3D model is missing
- The GPS antenna is not yet displayed

I will soon publish the source files under GPLv3. I will probably open a new thread for this new version soon.

Regards Sven