Solar Power Supply

Message

Ton_vN · #1 Post by **Ton_vN** » 26 Feb 2019, 08:53

DIY is nice & interesting (for technical people) and may save money, but

sometimes one simply is not willing (or not having the time) to make the effort ......

In that perspective looking for a ready-made solar power supply to 'feed' my ESP8266s.
Basic requirements:
- suitable for operation outdoors, either in open air, or in box with transparant cover
- incl. battery, to have power supply in the night and during less favourable light conditions
- battery capacity of > 2000mAh
- unattended operation, with simulateaous charging of the battery while power supply to the device
- protection against 'deep discharge'

Anyone experience with such kind of Solar Power Supply at affordable price?

Would this device fit the requirement, with an attached solar panel with micro-USB-connector?
Or would this version be the more 'general' solution (also with an attached solar panel with micro-USB-connector?

Sektflasche · #2 Post by **Sektflasche** » 26 Feb 2019, 12:36

Hi,

if you want to build it by you self, I used this project: https://www.instructables.com/id/Solar- ... r-Station/.

Ton_vN · #3 Post by **Ton_vN** » 28 Feb 2019, 21:06

Thanks @Sektflasche

Instructive reading!

Unchanged WEMOS_Pro + TP4056 + Panel + LiOn-battery seems good solution for my first application:
read-out of light levels.

No need to work at night, just need to start at dawn, bridge some darker periods during daytime, and fade out in the evening.
In that perspective perhaps a more powerful solar panel will be an improvement, for earlier 'wake-up' in the morning, and to have some spare/reserve capacity in the darker season, in combination with a slightly bigger battery.

Vic20 · #4 Post by **Vic20** » 05 Mar 2019, 19:43

Hi

I enjoy the challenge of hacking existing solar devices to fit ESP8266 projects inside.

My latest uses one of these https://www.amazon.co.uk/dp/B079QFVFJC

It uses a protected Li-Ion 18650 so all you need is a 3.3v LDO regulator to drive the 18650
Room inside for a BME280 and light sensor to make a fully waterproof solar powered weather station and still have the working security light.

With the ESP waking every 10 minutes to upload to thingspeak, the device has been happily running and keeping the battery fully charged for 3 months now.

grovkillen · #5 Post by **grovkillen** » 05 Mar 2019, 19:50

I guess you could monitor the PIR as well? Thanks for sharing!

TD-er · #6 Post by **TD-er** » 06 Mar 2019, 12:12

Please note that the last build has some options to allow for "CPU eco mode" (sorry, could not think of a better name), which does reduce power requirements of the CPU significantly.
It does have a few drawbacks, but for that we have the documentation

https://espeasy.readthedocs.io/en/lates ... l#advanced

Ton_vN · #7 Post by **Ton_vN** » 09 May 2019, 13:49

In the perspective of independent setup to link WEMOS-PCB with battery 18650/3100mAh and with Solar Panel 5V/5W (all purchased through AlieX), experimenting with a battery-shield for WEMOS D1Pro , but that just raises questions ......

Regardless whether Solar Supply or grid-oriented supply is used, it looks as if the battery is not being charged.
Assumption is that the shield (like TP4056) operates separate from the WEMOS_PCB, but is that true?
Because it is kind of UPS, for which it is not uncommon to handshake with their 'client'.
Related experiences with this shield?

Ton_vN · #8 Post by **Ton_vN** » 09 May 2019, 14:09

@TD-er

The description of 'CPU Eco mode' seems to indicate that the CPU is 'taking a nap' if not requested to do something.

Considering 'ECO-mode' I would have a different look at this aspect.

Perhaps it is already available somewhere as specific software, but not found:
why not install a scheduler in the ESPEasy-software, which is controlling the following aspects (as Example)?
1) forcefully put in deep sleep the processor for predefined times, set in a menu of ESPEasy
That can be e.g. not operating during specific hours of the day, or for a specific time per hour (set as start & stop times).
2) during 'awake'-period, the processor has to be alerted by specific call, and until that time is half-dormant,

more-or-less 'taking a nap', listening only for the alert-signal. After alert, it is active for a preset period to perform a few scheduled functions.
2) if needed during 'sleeping-time' [e.g. to change the menu mentioned under 1) ], wake-up the processor by the Reset-button to open a window of 30 minutes

Domosapiens · #9 Post by **Domosapiens** » 10 May 2019, 23:21

Ton_vN wrote: ↑09 May 2019, 13:49 Related experiences with this shield?

Yep, ...my experience
charging the battery via USB does not work when the Wemos is not switched on.

TD-er · #10 Post by **TD-er** » 11 May 2019, 21:56

@Ton_vN

We already have deep sleep in ESPeasy.
But that only can be set to a specific interval.
At boot (wake from deep sleep), the wifi connection is initiated and after that the rules may be processed.
So at this moment it is not possible to set some proper time schedule to wake.
Without WiFi or GPS, the unit simply has no knowledge of time.

What we can do, is to add a check for the battery voltage when waking from deep sleep and then consider just to go to sleep again if the voltage is too low.
But that should be a new feature. It isn't implemented yet.

Ton_vN · #11 Post by **Ton_vN** » 12 May 2019, 16:07

@Domosapiens

Your message puzzles .....

If the Battery Shield gets PowerSupply through it's USB-interface, then the WEMOS is On, which is OK.
But the question is, why is the BatteryShield not charging the battery, parallel to the WEMOS?

1) If the BatteryShield is autonomous, do I need to set 'something' on the shield? What?
Only thing detected is link J1 which may be set to get 1A-power instead of normal 0.5A
2) If the BatteryShield in some way is dependent on the WEMOS, how/what?

Ton_vN · #12 Post by **Ton_vN** » 12 May 2019, 16:12

@TDer

Your response triggers me to reconsider.

You are right:
'deep sleep' is really deep sleep, and no wake-up other than a basic timer deep down in the processor, in no way from the outside world.

Perhaps the idea I suggested is more 'light sleep', reducing the activity of the firmware to a small scheduler-kernel, enabling a defined kick to start, either from the outside world, or from an internal, programmable timer, not so deep down as the timer for 'deep sleep'.

Domosapiens · #13 Post by **Domosapiens** » 12 May 2019, 21:01

Ton,
In my case: the BatteryShield is charging the battery, parallel to the WEMOS
When I charge the battery through the shield, the Wemos starts (see also schematics)

Battery full? green LED.
Charging ... red LED

Situation 1:
Wemos On, running on battery (4.17V), plug-in USB on shield, Red light goes on.

Situation 2:
Wemos Off, plug-in USB on shield, Wemos starts, Green light on (4.35V)

Situation 3:
Wemos On, running on battery (4.52V), plug-in USB on shield, Red light goes on.
After a few minutes, Red becomes Green (now 4.34V)

Charging the battery via USB does not work when the Wemos was not switched on.
Something the TP5400 decides based on 5V Vin versus 5V Vout versus Vbat ???
(did not study the datasheet)

Seen this ?:
https://wiki.wemos.cc/products:d1_mini_ ... ery_shield
Schematics
https://wiki.wemos.cc/_media/products:d ... v1.3.0.pdf?

For a solar panel experiment, I used this step-up booster:
https://nl.aliexpress.com/item/5V-2-1A- ... 4918y1IVvX
in this way

Code: Select all

Solar-----Booster------Wemos
             |
          Battery

Ton_vN · #14 Post by **Ton_vN** » 12 May 2019, 22:33

@Domosapiens

Reading your messge, apparently I have Situation 0:
USB-Power to shield => red LED => power to WEMOS & no Power to Battery

From your message, I get the impression that my test-setup might be the problem.
In this setup the feed to the BatteryShield through it's USB-interface has been tried by means of several 'original' devices providing USB PowerSupply.
Apparently the output of all these devices always is below the internal threshold of the BatteryShield to enable charging of the battery.

At least I cannot otherwise explain that de LED is red, but the charging voltage for the battery is (a too low) 0.45V.

ToDo: check the USB power supplies .....
And find a more powerful one (or at least one with more voltage).
Puzzled, because you generally should not feed through a USB-interface with a higher voltage than 5V.

Domosapiens · #15 Post by **Domosapiens** » 13 May 2019, 00:17

A few streets away .... at Ikea .... 2Amp USB charger.

Just to make sure ..
You need to feed the battery shield USB plug .... not the Wemos plug

LED is red, but the charging voltage for the battery is (a too low) 0.45V.

Sounds like a dead board, ... a dead battery, ...polarity reversal .....

Ton_vN · #16 Post by **Ton_vN** » 13 May 2019, 13:39

Indeed, IKEA is near, as well as few other possible suppliers.

Connections & Polarities all

multiply tested as OK.

Tend to agree with conclusion:
Because Battery + WEMOS 'behave' OK => BatteryShield seems DOA for at least the battery charging circuit.
If no battery attached the output from the shield is still not more than 0.45V

Time to switch from BatteryShield to separate Protection Board TP4056 in this Setup.

Domosapiens · #17 Post by **Domosapiens** » 13 May 2019, 21:17

Ton,
that protection board is lacking a booster.
Thus low voltage from the solar panel will not be used without the booster.

Ton_vN · #18 Post by **Ton_vN** » 15 May 2019, 10:30

@Domosapiens

Understood, but with normal daylight the solar panel should sufficiently charge the battery to provide supply during next night and some more time.
My first application will be for a setup with light-sensors, which anyway hardly require attention & power during the night .....

First test without booster:
if needed, insert later.

Raises the related question how to simply reduce the WEMOS' measuring rate & processing during the night, to save energy.

Wiki · #19 Post by **Wiki** » 15 May 2019, 10:41

Booster makes no sense because if the solarpanel delivers not enough power for the TP4056 it will not deliver enough power for a booster either.

Ton_vN · #20 Post by **Ton_vN** » 15 May 2019, 10:54

Possibly this booster is better suited to feed the battery-charger, but what is the probablity that the output of the solar panel exceeds 2,5V, but not reaches 5V?

Just a trade-off:
wait some time vs. extra circuit

Wiki · #21 Post by **Wiki** » 15 May 2019, 11:11

These kind of boosters are consuming a lot of power even in idle. Better use a 6V rated solar panel and just attach the TP4056.

If possible use the deep sleep, use the eco option of the current ESP Easy (Tools/Advanced section) and apply some modifications like described here: viewtopic.php?p=32183#p32183

Using deep sleep f.i. 30 minutes, awake time a few seconds will give you a running time of up to 80 days just with a 3000 mAh 18650.

chemmex · #22 Post by **chemmex** » 16 May 2019, 10:08

Even better will be using switched charger such as TP5000. Even though it doesn't have MPTT, it can catch low sun condition provided the panel OC voltage is between 5-6V

Ton_vN · #23 Post by **Ton_vN** » 18 May 2019, 09:42

Making use of 1 extra component perhaps this is a simple(r) way to get better power supply & charging:
1) power source = 1* solar panel slightly oversized with output >7V, or put 2* 5V solar panel in series
2) feed panel output to cheap 3-leg power circuit like 7805-regulator
3) feed regulator output to charger board TP4056

1) and 2) in combination provide a strong & stable 5VDC-input much earlier & longer than just 1 5VDC solar panel
=> in 3) the battery as buffer is needed shorter time => more capacity to bridge gaps.

Blunt · #24 Post by **Blunt** » 12 Jun 2019, 05:14

There are many types of solar power, but it is not necessarily necessary for solar lights. On the contrary, solar panels are more important.

Ton_vN · #25 Post by **Ton_vN** » 03 Aug 2019, 13:35

Simplicity & proven design is best:
for the power&processor-setup went for the configuration described in https://www.instructables.com/id/Solar- ... r-Station/
For light-measurement my WEMOS+ESPEasy interfaces 1*BH1750+2*TSL2561+1*ML8511

It works, but as improvement without 'hardware solution', for conservation of energy I would need to implement 'sleep'-mode:
in the present, 'simple' setup the 3 sensors + the periodic communication by ESPEasy quickly 'eat' the energy, causing 'holes' in the timeline.

Have a second solar panel laying near the lightmetersetup, during sunny conditions powering the fan of my PWS for active ventilation:
will have a surplus during daytime.

to be checked whether it can provide a simple helping hand to the lightmetersetup in the way described here.

dynamicdave · #26 Post by **dynamicdave** » 03 Aug 2019, 20:36

Here's a link to a solar-powered weather station one of my IoT Club students has been building.

It includes a TP4056 to handle charging a Li-Ion battery from a solar-panel.

The project uses a Wemos D1 Mini and ESP-Easy of course!!!

https://discourse.nodered.org/t/solar-p ... tion/12499

grovkillen · #27 Post by **grovkillen** » 03 Aug 2019, 20:45

Looks good Dave! Thanks!

Ton_vN · #28 Post by **Ton_vN** » 18 Aug 2019, 12:28

This information which is further explained in this documentation describes various levels of 'sleep' for WEMOS, from 'No' to 'Deep'.
With an application script.

Now just the challenge how to make a stand-alone implementation (with ESPEasy?) that
1) if daytime, then
2) every hour (or comparable time),
a) the WEMOS wakes up
b) performs a measurement cycle for the connected sensors
c) uploads the measured values to it's controller
d) takes a nap again
3) else deep sleep

TD-er · #29 Post by **TD-er** » 18 Aug 2019, 13:11

A simple check to the ADC value at boot and enter deep sleep again would already help a lot I guess.
This would not be time based and making all kind of assumptions, but just base its decision on the battery voltage.

Well there should be some checks also to see if the range is plausible etc. (or else it would not start again if the wire to the voltage to be measured got loose)

Ton_vN · #30 Post by **Ton_vN** » 18 Aug 2019, 13:18

@TDer

Complication of my intended application is that interface A0 is occupied by the UV-sensor.
On the other hand under ESPEasy's SysInfo I remember to have seen a check on VCC:
is that alternatively usable for battery-checking after boot?

TD-er · #31 Post by **TD-er** » 18 Aug 2019, 17:31

Ton_vN wrote: ↑18 Aug 2019, 13:18 @TDer

Complication of my intended application is that interface A0 is occupied by the UV-sensor.
On the other hand under ESPEasy's SysInfo I remember to have seen a check on VCC:
is that alternatively usable for battery-checking after boot?

Hmm, not really sure. You should have some switch then to toggle between sources to measure.
I guess changing the UV sensor to one that communicates via I2C is easier then.

Ton_vN · #32 Post by **Ton_vN** » 18 Aug 2019, 21:36

Hooking an analogue UV-sensor like ML8511 or GUVA12 to WEMOS'A0 is easy.
Otherwise, the 'Minor' aspect to be solved is that the standard catalogue of devices in ESPEasy (upto/incl. R147) does not have an I2C-controlled UV-sensor.
Candidates with I2C-interface would be SI1145, or better VEML7x.
Is such sensor included in one of the later versions of ESPEasy?
;-( Building a new plugin seems hard work ..........

Looking at alternative sleep-control functions just wondering how to realise a simple external trigger for 'Light-sleep'.
Less power-saving than 'Deep sleep', an ESP-power-decrease to 10% seems interesting enough.

However, for 'System-approach' the power-aspect of an ESP-node must be investigated Node-wide, because it should not just cover the ESP8266, but also the various interfacing sensor-PCBs: in my configuration 2*TSL2561+1*BH1750+1*ML8511
Looking at that volume, probably a different kind of power-control layout must be found.

TD-er · #33 Post by **TD-er** » 19 Aug 2019, 20:22

> 2*TSL2561+1*BH1750+1*ML8511
The first two look quite familiar to me, so I can almost 100% certain telling you they are supported. Although I am not sure the TSL2561 is supported with multiple instances of the same sensor.
The last one is not that familiar, so I don't think it is supported yet (almost 100% certain)

Ton_vN · #34 Post by **Ton_vN** » 21 Aug 2019, 08:37

@TD-er

All indicated PCBs are supported by ESPEasy in combination:
- Lightsensor BH1750 has selection-option for 2 I2C-addresses
- Lightsensor TSL2561 has selection of 2 other I2C-addresses
- for both sensortypes the output of ESPEasy is (general) luminosity.
- UV-sensor ML8511 has analogue output and hence 'universally' fits to A0/ADC of WEMOS/ESPEasy.
For the ML8511-signal either in 'Rules', or in the receiving software, a scaling must be implemented for the translation from Voltage to UVI.

Considering solar supply just guessing how much power is needed.

Waiting for components, have built a provisional PV-supply recycled from solar garden lamps.

'Appearance' will never win a prize, but the output power fills a first need:
# 5 cells in series => 0V ~14VDC, related to lightlevel & load
# a regulator 7809 controls output to a 12V fan, while a regulator 7806 feeds the buffered power supply for an ESP8266
# available current is limited, and not much beyond feed of a fan plus something extra, often seeing that output of the 7806 is too low.

: PVPower_from_recycling; FanPower_recycled [640x480].jpg (88.06 KiB) Viewed 156789 times

TD-er · #35 Post by **TD-er** » 22 Aug 2019, 10:34

- TSL2561 has selection of 2 other I2C-addresses

Being able to set the I2C address is no guarantee it will be usable with multiple instances of the same plugin in ESPeasy.

The thing is, most plugins declare global variables to store intermediate data of that sensor. So if you run 2 (or more) instances of that sensor on the same node, this data may cause issues if used by "the other" instance of the plugin.

I've been working on adding support for multiple instances of plugins, but only a handful of them have been adapted to support multiple instances.

Ton_vN · #36 Post by **Ton_vN** » 23 Aug 2019, 11:56

Emperically demonstrated in my setup that 1*BH1750+2*TSL2561+1*ML8511 (at A0) functionally cooperate at 1 ESPEasy-Node without problem.
The TSL2561s implemented as 2 separate devices.

TD-er · #37 Post by **TD-er** » 23 Aug 2019, 13:03

Ton_vN wrote: ↑23 Aug 2019, 11:56 Emperically demonstrated in my setup that 1*BH1750+2*TSL2561+1*ML8511 (at A0) functionally cooperate at 1 ESPEasy-Node without problem.
The TSL2561s implemented as 2 separate devices.

I've looked at the code and I do believe it should work indeed.
It is using a global variable for the I2C address, but that one is updated every time a READ function is called and all communication is done in one streak without calling other plugins in between.
So indeed it should be working with multiple instances.

Ton_vN · #38 Post by **Ton_vN** » 25 Aug 2019, 12:43

Sometimes 'little things' block or help.

Made a setup according to this diagram.
But apparently the margins of my combination of solar panel plus SI-diode resulted in out-of-range (low) input to the controller:
many gaps in power supply to the WEMOS.
Replacement of the silicium diode by 2 parallel germanium diodes improved the power feed:
the forward voltage over the GE-diodes is lower than the forward voltage over the SI-diode (nominally 0.3V vs nominally 0.6V)
=> on purpose the 2 GE-diodes parallel to reduce their resistance
=> (slightly) higher voltage to the Controller, now within capture range.

Still the solar power supply is not really reliable ...........

TD-er · #39 Post by **TD-er** » 25 Aug 2019, 20:23

Ton_vN wrote: ↑25 Aug 2019, 12:43 [...]
=> on purpose the 2 GE-diodes parallel to reduce their resistance
[...]

The only reason you may want to do this, is to allow for higher currents.
The voltage drop over a diode will not change, no matter how many you place in parallel. (except for the effects of current related voltage dop)
A diode has a fixed voltage drop and it depends on the type of diode (like you also mentioned) how much it is. But it is not comparable to a voltage drop over a resistor.

TD-er · #40 Post by **TD-er** » 25 Aug 2019, 20:36

Oh and looking at the diagram you linked

This is not going to work.
A solar panel which can deliver 120 mA, will drop its voltage when you try to draw 120 mA from it and the sunlight is not intense enough.
Very likely this board does not limit the current, so it will try to get as much as possible and thus always pull the voltage too low.

I don't know what this charging board does with low input voltages.
Does it just stop charging and always use an up-converter to make 5V from the Li-Ion battery?
Or will it go below 5V and does it not output anything for a while until the up-converter does start?

What you could do, is try to add a zener diode or some MPPT circuit in front of (or instead of) the charging board to get 5V or nothing, but with the maximum possible output current.
I think you probably also should have some capacitors near the Wemos board and I guess you should have them on the 3V3 line.
If you have them on the 5V line, you may disturb the charging board's functioning.

Shardan · #41 Post by **Shardan** » 26 Aug 2019, 14:33

I'm wondering if a Buck/burst converter type might work...

A small converter like this:
https://www.aliexpress.com/item/3263962 ... 4c4dwlzGg5
Note it has two IC's and can convert up and down.

I've checked several powerbanks, these are simply a LiIon or LiPo battery with a charger circuit inside.
Characteristic behavior of all: The loading current is heavily depending on voltage.
At 4.9V they nearly don't load at all, at 5V they pull 300....600 mA depending on Type.
At 5.1V the current goes up to 400 mA (smallest) and 1.1A (biggest, 20.000 mAh).

So using this converter should have a simple effect: If the panel cannot deliver full power even with this converter module
voltage will go down... still the converter tries to keep up to 5V.
As loading current depends on voltage it should adjust to the max the solar panel can deliver at given luminosity.

Could that work?

Waiting for some delivery to test

Maybe such a converter might work with the shown diagram too?

dynamicdave · #42 Post by **dynamicdave** » 26 Aug 2019, 20:11

I bought these solar panels from Banggood for one of my students who built a solar-powered weather station earlier this year.

https://www.banggood.com/5_5V-1W-180mA- ... rehouse=CN

He used pretty well the same circuit as yours except he didn't have a diode in the positive wire from the solar panel (just a direct wire).
I think I read somewhere that the TP4056 has an on-board diode to prevent reverse energy going back into the solar panel (at night).
So putting in your external diode will create a forward voltage drop across the diode - which is not good news as it reduces the efficiency of the circuit.

One thing you need to check is... does the TP4056 have a battery protecion circuit which is located at top-right of your diagram.
Your TP4056 appears to have such a safety mechanism - but there are some devices being sold on-line that don't - just BE CAREFUL.

The battery he used was a TR 14500 3.7V Li-Ion battery rated at 2500mAh, made by GIF (about the same size as standard AA battery).

His circuit uses a Wemos D1 Mini (running ESP Easy) that goes into 'deepsleep' for 10 to 15 minutes, wakes-up, takes some measurements, publishes them via MQTT to Node-RED then goes back to sleep.

His system has been working for over three-weeks.

Here's a link to the project if you want to read about it.

https://discourse.nodered.org/t/solar-p ... tion/12499

Ton_vN · #43 Post by **Ton_vN** » 27 Aug 2019, 12:11

@TD-er
The setup was OK during testing with a 'normal' USB Power Supply at the place of the solar panel, and therefore suspect that the 'problem' is somewhere at the TP-board.
In the line from the solar panel the diode is fitted with the intention to protect against reverse currents [regardless whether the TP-board has a protection-diode or not]

1) Hint to check on presence of battery protection circuit is good one.
2) 5V-Zener over the entry of the TP4056-circuit may also have protective value against too high input.

Presently sunny/blue sky every day and therefore sufficient charging generally is not a critical aspect, but input > 5V and/or over-charging of the battery is something different, requiring such protection.

Wiki · #44 Post by **Wiki** » 27 Aug 2019, 13:11

I am running an outdoor weather sensor equipped with a 6V 100mA solar panel outlined exactly as the above shown diagram, except the diode. It is performing pretty well. A 18650 battery, 2800 mAh, is sufficient to feed the board for app. five-seven weeks without any sunlight (depending on the chosen deep sleep configuration) using some hardware modification I summarized here: viewtopic.php?f=4&t=6840&p=38161#p37892

The TP4056 boards with seperate connectors for the battery as shown in the diagram above are protecting the battery agains overcurrent (3A usually), overvoltage (4.25V usually) and undervoltage (2.5V usually). As dynamicdave already posted there are TP4056 boards on the market which only have an overvoltage protection against the power input. These ones are equipped only with two output connectors, so the battery and the load are connected in parallel. These boards are only suitable for pure charging purpose and should not be used in a project like the one we are talking about here if the battery isn't equipped with its own protection circuit.

Ton_vN · #45 Post by **Ton_vN** » 27 Aug 2019, 17:23

'Lessons learned'

Checking/applying the hints from the previous messages:
1) my TP-boards have the layout as shown, which might mean that battery-protection etc. is present
2) Voltage to the TP-board shows fluctuations, but Power from the TP-board to the WEMOS seems always OK to feed the 'bare' WEMOS
(due to the work of the internal power controller?)
3) the sensor-PCBs had earlier been checked&tested to work on direct 'real' USB-Power (anticipating long interface lines)
4) after test 3) had connected the sensor-PCBs to the output of the TP-board
=> in practise that is not a good idea = sensors sometimes/unpredictably 'striking' = direct power from the TP-board less suitable for the sensor-PCBs?
5) as result of 4) the sensor-PCBs now connected to the 3V3-pin at WEMOS-PCB = using the WEMOS' power controller as buffer
=> operation of the sensor-PCBs seems OK regardless of low voltage

From 4) and 5) perhaps the conclusion that for long interface lines the application of an extra stepdownconverter to 3V3 provides a good power feed near such sensor-PCB?
Seems better than connecting long lines to the WEMOS's supply of 3V3.

TD-er · #46 Post by **TD-er** » 27 Aug 2019, 23:04

I assume these boards also have a step-up converter to make 5V.
These work as a switching power supply, so there is some square wave involved and some coils.
I think it may not be a bad idea to add some filtering to the generated output from these boards, like a small capacitor and a bigger one.
You never know what those signals may do to the sensors or the ESP board.

Ton_vN · #47 Post by **Ton_vN** » 28 Aug 2019, 10:43

For power the analogue technology has some value (as demonstrated again).
Stepdown is easy enough, therefore reconsidering the buffering & smoothing effects of diodes, condensators and analogue power controllers to provide smooth & clean power to the application electronics.

This experience raises the following question:
has somebody seen a technical solution to recycle a solar lamp with single battery into a 5V-power supply for WEMOs etc.?
Will certainly require an upconverter followed (in the context of this recent experience) by a 'smoothing circuit'.

TD-er · #48 Post by **TD-er** » 28 Aug 2019, 11:56

Not really experience in this particular case, but I do have experience with step-up converters releasing their magic smoke.
What you really should take into account is the current drawn by an empty capacitor.
Strictly speaking, an empty capacitor should be considered a short circuit.
Depending on the capacity, charging current, etc. this short circuit state can be very short.
But step-up converters cannot handle these short circuit events very well. (at least the cheap ones without protection, sense resistors etc.)

So what you should consider is adding a (low Ohm) resistor in series with the load to limit the charging current of the capacitor.
Still keep in mind the dissipation in the resistor when the capacitor is being charged. For a short time, the resistor can handle a lot more than the rated power (1/4 Watt resistor can handle 10's of Watt for milliseconds) but just to be sure you may want to place 2 in parallel to halve the current and thus only handle 1/4th of the power per resistor.

But I think there are others here on the forum way more experienced in this field.

Ton_vN · #49 Post by **Ton_vN** » 28 Aug 2019, 12:30

Agreed.

That implies some calculations to determine the layout of the combination of resistor & condensator at the output of the stepup converter:
1) during initial charging of the condensator at the output, the max. current should be acceptable for the converter
=> max. size of the resistor, relative to the size of the condensator
2) under nominal load of WEMOS+Sensors, the resulting voltage on the condensator should be approx. 4.5V to comfortably stay within the input range of the WEMOS controller.
Considering that WEMOS plus it's peripherals usually will not 'eat' much power (with the charged condensator taking the burden at the peaks during WiFi-transmission), an average consumption current could be < 15mA.
=> min. size of the resistor

Sensible approach?
Any ideas for a practical, 'safe' setup?

Wiki · #50 Post by **Wiki** » 28 Aug 2019, 13:26

Solar lamp recycling:
The pcb of this tiny solar lamps mostly consist of a small chip like the QX5252F or similar, an inductor, a capacitor and a diode. The chip leads the current of the solar panel during day time directly unregulated to the battery, during night time the power of the battery to a small, pwm-driven boost circuit to lift the voltage to a sufficient level for driving the single led, taking the sunlight resp. solar voltage as indicator.

So the pcb itself of the solar lamp is pretty useless for our applications. But unsoldering the pcb and soldering three solar panels themselves in series directly to the TP4056 should give somewhat 6V and a kind of charging current, I assume 50-60 mA at full sunlight. Not very much, but depending on the deep sleep configuration (e.g. 5 secs awake/30 mins sleep) of the ESP it might be sufficient to serve as an useable power supply.

Additionally: I am driving my "mobile" devices by feeding the 3.2 V - 4.2 V of the battery directly to the 5V connector of the Wemos - with a replaced voltage regulator. Works perfectly.

Who is online