Open maholli opened 4 years ago
Getting started with the power testing. Would be useful to make a harness to break out the connector so we can easily access all those pins.
Contacted Tim to set up a time to verify antennas. Tim says that he will be ready for an antenna test after today when the first transmit test happens. Coordinating time/place (and access to +Y board).
Burn wire test scheme:
NOTE: Do NOT just try to do this by actuating the power plug. It turns out that the wire is PWMed so just turning it on will melt everyyyyyyything.
Voltage across one solar cell currently 0.7V at max sun lamp, with nothing connected.
Energy harvest circuit is tested under the lamplight, the report can be found using the link below: https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Energy_harvesting_board_testing_1111.pdf
A short summary is:
BIG DEBUGGING COMMENT for Solar Power:
First mystery: all the solar cells in line produce 0.5V, which is manifestly incorrect. They should be producing more than that.
POSSIBLE THEORIES:
FURTHER EVIDENCE:
The energy harvest board is tested under sunlight. The full report can be found using following link: https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Energy%20harvesting%20board%20testing%201112.pdf
The short summary is:
Testing in sunlight: X gives power output of 0.8 V, -Y and Z give power output of 0.2 V. When measuring +3.3V and ground on the breakout board from the flat-flex cable (in late-afternoon sunlight), power output of 0.2 V on all 3 boards.
-Y and Z diodes output 1.6 V (as expected) across diodes, X output 0.7 V across diodes
What is the output voltage of Vsolar? Not just across the diodes, but the actual voltage across the entire array of cells? This is the first thing to determine before chasing down the regulator/mppt circuit.
X panel:
-Y panel:
Z panel:
Measuring between VSolar and Gnd in sunlight, we get: Single Cell = ~1.7 V Z = ~3.2 V -Y = ~0.7 V X = ~0.5 V
So it looks like the Z cells are performing properly. Unfortunately the output from the breakout board 3.3V pin is ~0.2V, so there's still an issue on the board. Additionally, I think we misunderstood the alignment of the breakout board for previous tests - to fix this, always find the two adjacent pins that are shorted since those are the ground pins.
Also installed photosensor libraries on a raspberry Pi but uncertain if the power source for the sensors is on the board, if we have to supply the 3.3V from the Pi. Will ask at office hours today.
@cwnaught photosensor is powered by 3.3V. The energy-harvesting circuit should provide this, but the sensor will not work if you aren't able to measure 3.3V between 3V3
and GND
.
As designed, the panels should be able to handle back-feeding (providing 3.3V from another source) onto the 3.3V rail.
Yesterday, Max and I fed ~3.5V to the VSolar pad on the Z board and found that the energy harvester circuit was not working properly. Rather than outputting 3.3V, it output 0.01V (measured both from the breakout board and the 3.3V pad beneath the photosensor). The energy harvester chip was also very hot. We believe Z has an issue with the harvester circuit, and that X and Y have issues with both the cells and the harvester circuit.
Future work is going to be testing half-assembled boards, as well as comparing the board-energy harvester's response to the pre-made energy harvesters that we know work.
Recommended procedure from here on out is to test the boards as we make them to more easily isolate issues.
We found that the cells on the Z board are not connected to the MPPT circuit. This is notably also the only board that actually works. We hypothesized that something about the MPPT circuit was impacting the solar power as well.
To confirm, we unsoldered the diode on the X board, and found that afterwards, the solar circuit now provided 3.3V.
Now that we have the +Y board, we also tested that. It appears to have significant solder problems. Testing the connection, we found the same issues as the rest of the boards - everything seems correctly connected but no voltage overall. All of the voltages we did observe were substantially smaller than expected, which is quite likely a soldering issue.
More issues were found with the soldering of the +Y board - it appears that all of the solar cells are also backwards.
We desoldered the diodes on the board and it works now - provides correct voltage on all parts! The evidence continues that we suspect that the MPPT has troubles.
Working thread for documenting the burnwire testing:
Got the flat flex connector and PWM working. However, having trouble with continuity. Connectivity between flat flex connector and board is weak.
FIRST ISSUE FOUND: The batt-P net that starts at the flat-flex connector just doesn't connect to its other end on the burnwire circuit. No current flows.
Dealt with this by greenwiring in a wire directly to power.
Second problem: Even with the PWM at 100% duty cycle, when power is confirmed supplied to the Batt-P net, GND to the GND net, and the PWM to the Burn-1 net, there appears to be no current flowing across burnwire circuit.
As a sanity check, when Batt-P is shorted with VBURN, the nicrome cuts immediately.
Calling it for tonight, but summary - first issue seems to be an easily resolvable layout change. Second seems tough - need to look into the IC more to understand why it's not doing the thing, or perhaps what inputs it expects.
Calling it for tonight, but summary - first issue seems to be an easily resolvable layout change. Second seems tough - need to look into the IC more to understand why it's not doing the thing, or perhaps what inputs it expects.
Great detective work, @thwhite! You're correct, Batt_P was not properly connected. I also see why the PWM signal isn't working as intended:
The sort of MOSFET switch we're after will look something like this:
To make our circuit match the above diagram, we need to do the following:
Kill switches are working as expected
Note on the antenna test - since the antenna configuration has changed, mechanical is taking over this test and reorienting it towards just checking the new antenna config.
Acquired test data on how different gauges of Nicrome affect burnwire effectiveness!
Greenwiring mostly complete, but can't find 1MOhm resistors. Will ask around.
Update: greenwiring complete. Will return to the lab asap to test!
Update again:
Having fully greenwired the circuit, the result is not encouraging. I observed no current flowing through the nicrome even at 100% duty cycle. At this time, the board passed continuity testing - the right things were connected to each other and there appeared to be no shorts.
However, the board's integrity is failing from hours of repeated hand-soldering, and I'm existentially worried that perhaps something in an IC has gotten fried. I can attempt to use the other Z board briefly, replicate the circuit in a breadboard, or take this result as good enough.
Next step - do it in a breadboard instead with a new IC so that we know for sure that nothing got fried in the greenwiring.
Ok good news and bad news:
I green-wired as described in https://github.com/spacecraft-design-lab-2019/documentation/issues/132#issuecomment-557847274 and can properly pass current through the burn wire 🎉
The current-carrying capacity of the flat flex cables are 0.5A. However, passing 0.5A through the burn wire for >10 seconds barely makes the burn wire warm. I slowly ramped up the current and found ~1.5A will burn the fishing line immediately (wasn't using the flat flex cable to supply the power, though).
@thwhite I learned this the hard way on KickSat: this burn wire scheme is a balancing act of very small resistances. When breadboarding, although the burn wire circuit will work, we found it didn't translate well to the actual spacecraft. It's necessary to replicate the burn wire scheme as best as possible when prototyping.
looks like we soldered on one of the DNI resistors during assembly (sorry!!). Once I removed R15, the energy harvesting circuit seems to work. To reiterate, the resistors circled in red below should NOT be installed on the panel:
The energy harvest board charging voltage plot is generated. Check: https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Charging%20Voltage%20plot_2019-12-09-11-46-00.449.pdf https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Charging%20Voltage%20plot.PNG
As the figure shows, the EXT BAT provides 3.979V without connecting to the battery. After connected to the battery, energy harvest circuit is able to change the battery from 3.668V to 3.7V
Set input voltage to 3.0V by connecting 4 Diodes (100V 15A R-6) in series, EXT BAT before connects to the battery is 1.3V. After connecting the battery 3.17V. At the end of the test, the clamp which connected the battery with the meter loose so the voltage drop to 0V at the end. In this mode, the V input is lower than V-output before connected to the battery. Thus the behavior should match with the yellow color plot in Fig 22 in the reference document (bq25570). The test report can be found using following link: https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Vin_smaller%20than_Vout(EXT%20BAT)_2019-12-09-15-39-02.857.pdf
Set input voltage to 3.0V by connecting 4 Diodes (100V 15A R-6) in series, EXT BAT before connects to the battery is 1.3V. After connecting the battery 3.17V.
CORRECTION - you want to set the solar input to <3.3V NOT the external battery. This is because the energy harvesting circuit uses a boost regulator to charge the battery. Boost means it expects an input voltage less than the battery voltage (that is, when charging from low-impedance DC sources).
Set input voltage to 3.0V by connecting 4 Diodes (100V 15A R-6) in series, EXT BAT before connects to the battery is 1.3V. After connecting the battery 3.17V.
CORRECTION - you want to set the solar input to <3.3V NOT the external battery. This is because the energy harvesting circuit uses a boost regulator to charge the battery. Boost means it expects an input voltage less than the battery voltage (that is, when charging from low-impedance DC sources) Yes, the solar input is the input which is 3.3V. After connected solar input(i.e., 3V), the EXT BAT gives 1.3V as I state. After measure EXT BAT, I connected it with the 3.7V battery which I have charged in the morning and measure EXT BAT which gives 3.17V. Then the EXT BAT stay around 3.17V and increasing slowly as The plot shows: 0.007 V increasing in 15 mins.
@maholli for the burnwire testing, what's the next move? I can start trying different combinations of burnwires to see if something works at 0.5A? (I also got us some test data from a different project with tests of different styles of burnwire if that would be helpful!)
The detailed energy harvest board voltage plot testing is available using the following link: https://github.com/spacecraft-design-lab-2019/SolarActuators/blob/master/EnergyHarvesterBoardTest/Voltage%20plot_1211.pdf
Now that boards are built, it's time to check if they function! Primary goal is deciding whether or not your board functions as intended.
Primary Criteria
[ ] Does your board generate the expected open-circuit voltage (Voc)? Keep in mind there is the energy harvesting MPPT connected to the solar cells. How does the MPPT effect your Voc reading?
[x] If your board has an antenna connector, coordinate with @Timvrakas to determine antenna functionality.
[ ] If your board has a burn wire circuit, nichrome wire will need to be soldered into place and a burn-wire test scheme developed
[x] If you board has kill switches, use the continuity mode on a multimeter to confirm the high-side and low-side switches are working as intended.
[ ] Torque coils will need 0ohm resistors installed prior to testing. Resistors should be installed once a cohesive coil-winding pattern is decided.
[ ] Sun sensor verification will require getting up-to-speed with circuitpython and leveraging the existing circuitpython libraries for the TSL2561. See https://learn.adafruit.com/tsl2561 to get started. Additionally, a 10-pin FFC breakout cable will be necessary to interface with the I2C signals.
Energy Harvesting Circuit
NOTE: ❗ The energy harvesting circuit should be evaluated following the same series of checks as documented in #68.