[clubjoker]

Hello, everyone!

I've been super excited about this project since I first found this forum a few months ago. It took me a while, but I finally got the electronics parts together and a few sensors to start the electronics side of the build.

I'm starting with a simple and cheap(ish!) build, intending to learn as I go, and improve for a second, more sophisticated build.

First build is planned to be:

1x TX, PCB v3.12 with
Fine Offset anemometer and wind vane (new as "spare parts" from Maplin here in the UK) to give a rough measurement of wind
Davis rain gauge, metric base-mounted version (I thought the FO gauge was just too approximate!)
(possibly fake from Far East?) SHT31 temperature humidity sensor; radiation screen still to be worked out (thinking of Davis 7714, but $$)

1x RX+, PCB v1.3C
SHT31 temperature humidity sensor
2.8" TFT
no onboard TX/relay transmission (I don't intend to build a WD unit)

In this thread, I'll write up some observations and ask questions

A little background on me: it's a long time since I "played with electronics". At about age 10, I started with making a crystal radio, stringing the antenna wire to a tree from my bedroom window, and driving a copper pipe into the ground for an earth. Then I made some circuits connected to the parallel port on my Commodore 64: a small wheeled robot, and circuits to control a train set. I have fond memories of Babani books, whcih I hope I've still got in a box somewhere. Fun times! I studied a bit more at school, a Physics degree, then 20 years of an office job and no electronics experiments, and here I am at the Weatherduino project.

I've kitted myself out with a fancy (to me) Ersa i-con nano soldering station and a few tips - a joy to use. I'm a bit out of date and rusty with this sort of thing, but really enjoying it.

[clubjoker]

I've got the TX and RX set up now on my desk. Most parts are soldered onto the PCBs, but I've left a few bits on a supporting breadboard, which I intend to mount into the cases later:
- TX LED - I'm using a common anode 2-colour LED for RF transmission indicator and fan indicator
- 9V power supply - I've rigged up a 7809 linear regulator to use for the RX+ until it's housed in a box with a plug-in switched-mode power supply
I've hard-coded the TX to send 20 degrees C and 50% humidity as not sensor attached. An SHT31 is attached to the RX+.

It all seems to be working pretty well , but not quite perfectly   

Things I've noticed, and questions which I'd be very grateful if people could give feedback on:

- I realised after (grrr!) I'd put thermal grease on the heatsink and mosfet on the TX that it probably didn't need a heatsink with the fan being switch fully on and off, so heatsink is superfluous. Hmm. I suppose it matches the 7809, but I'm thinking the less metal to interfere with the RF transmission, the better.
- The RX+'s 3.3V linear regulator (IC3) and its metal pad (heatsink I assume?) on the PCB get very hot to the touch, too hot for me to touch for long - is that normal?
- The RX+ makes a bit of ticker-ticketty noise (a bit like a spinning hard drive seek noise) when it responds to an http request over the wifi network to serve up its weather page. It will probably be muffled a bit when the RX+ is in a box, hopefully not enough to be annoying in a quiet room, but again, is this normal?

- Finally, and perhaps most importantly, if I don't touch the system for a while, the "big digit" clock mode kicks in. If I then leave it for longer, a couple of hours, say, or overnight, when I come back, the display is corrupted. The text and drawn boxes etc are reversed and rotated 90 degrees from what they should be. Then, if I touch the screen:
- sometimes it redraws the summary screen and goes back to working fine again
- other times, it draws the new screens also rotated 90 degrees and reversed text. It does still seem to be working ok otherwise, though - I can cycle through the screens and see them being updated to the TFT - it's just all scrambled on the screen.
Had anyone else seen this and can anyone suggest a cause? I'd be grateful if anyone could help.

I'm going to try placing the TX further away, and see if the screen glitch still happens.

As you'll see in the photos, the ribbon cable is about 25cm long, and runs 10cm from the TX unit (could be interfering with the serial connection to the screen?). I've wired the screen up following instructions in Wizza's build thread (thanks!). I didn't make a connection to the SDO (MISO) pin on the LCD, as per Wizza.

[clubjoker]

Another minor issue I've noticed is that the first pressure measurement from the RX+ (from the onboard BMP280) after switch on, is a glitch: 747mBar. The is updated to a more believable reading (e.g. 1010mBar) after a minute or so, I imagine on the second reading. Don't know why this is happening. I'm concerned that when I put the system into service, recording weather history, this glitch will pollute the data. Anyone else seen this? If the hardware is being slow to respond, is there a software workaround to wait a bit longer before taking the first reading, or simply reject 747mBar, or something more elegant?

[werk_ag]

Hi,

- The RX+'s 3.3V linear regulator (IC3) and its metal pad (heatsink I assume?) on the PCB get very hot to the touch, too hot for me to touch for long - is that normal?

This is a bad sign. You should try to fix it as soon as possible, don't keep the unit running until this issue is fixed. Under normal conditions the 3.3V regulator runs  barely warm.
The first thing to check it power supply voltage, make sure it is no more than 9V.
Also, make sure that you always have connected the 9V external power supply before connecting the USB cable to the Meduino board. Failing to it that, will destroy your board. Please carefully read this info:

1 - http://www.meteocercal.info/forum/Thread...een-warned

2 - WeatherDuino Pro2 PLUS Receiver - Notes about Assembling and Power On

- The RX+ makes a bit of ticker-ticketty noise (a bit like a spinning hard drive seek noise) when it responds to an http request over the wifi network to serve up its weather page. It will probably be muffled a bit when the RX+ is in a box, hopefully not enough to be annoying in a quiet room, but again, is this normal?

No, it's not normal. It should run silently.

- Finally, and perhaps most importantly, if I don't touch the system for a while, the "big digit" clock mode kicks in. If I then leave it for longer, a couple of hours, say, or overnight, when I come back, the display is corrupted. The text and drawn boxes etc are reversed and rotated 90 degrees from what they should be. Then, if I touch the screen:
- sometimes it redraws the summary screen and goes back to working fine again
- other times, it draws the new screens also rotated 90 degrees and reversed text. It does still seem to be working ok otherwise, though - I can cycle through the screens and see them being updated to the TFT - it's just all scrambled on the screen.
Had anyone else seen this and can anyone suggest a cause? I'd be grateful if anyone could help.

I'm going to try placing the TX further away, and see if the screen glitch still happens.

As you'll see in the photos, the ribbon cable is about 25cm long, and runs 10cm from the TX unit (could be interfering with the serial connection to the screen?). I've wired the screen up following instructions in Wizza's build thread (thanks!). I didn't make a connection to the SDO (MISO) pin on the LCD, as per Wizza.

The wiring to the display is good otherwise it doesn't work, however perhaps the cable is too long. Usually SPI signals doesn't like long cables, especially parallel cables. Try to reduce the length of the cable to no more than 15cm. Hope this works, but the fact that the 3.3V regulator is running hot, makes me think on a worse case. Hope I'm wrong.

[werk_ag]

Another minor issue I've noticed is that the first pressure measurement from the RX+ (from the onboard BMP280) after switch on, is a glitch: 747mBar. The is updated to a more believable reading (e.g. 1010mBar) after a minute or so, I imagine on the second reading. Don't know why this is happening. I'm concerned that when I put the system into service, recording weather history, this glitch will pollute the data. Anyone else seen this? If the hardware is being slow to respond, is there a software workaround to wait a bit longer before taking the first reading, or simply reject 747mBar, or something more elegant?

I never seen it happened, and it just should not happen. By those signs, I'm afraid that something may be wrong. Try to remember if you ever had the Meduino  powered by the USB, leaving the board without the external power supply. By the signs, it probable that the 74AHC244 chip was become damaged. Really hope I'm wrong. The first thing you should try to achieve is making that the 3.3V regulator don't warm.

Also try to run the I2C_Scanner and the FlashMemory_test utilities. The I2C_Scanner should consistently be able to detect all your I2C devices (RTC, Pressure Sensor and SHT31) at all speeds.

[clubjoker]

- The RX+'s 3.3V linear regulator (IC3) and its metal pad (heatsink I assume?) on the PCB get very hot to the touch, too hot for me to touch for long - is that normal?

This is a bad sign. You should try to fix it as soon as possible, don't keep the unit running until this issue is fixed. Under normal conditions the 3.3V regulator runs  barely warm.
The first thing to check it power supply voltage, make sure it is no more than 9V.
Also, make sure that you always have connected the 9V external power supply before connecting the USB cable to the Meduino board. Failing to it that, will destroy your board. Please carefully read this info:

1 - http://www.meteocercal.info/forum/Thread...een-warned

2 - WeatherDuino Pro2 PLUS Receiver - Notes about Assembling and Power On

Werk, thanks very much for the advice. It's good to know that the 3.3V shouldn't be hot, even though not happy news!

Supply voltage to the RX+ is measuring 8.6V on my multimeter, and jumps around a a bit - 8.5V, 8.7V, ... I wonder if the 7809 I rigged up to provide it is oscillating and not regulating properly. Hmm.

I was careful to remove the relevant diode on the Meduino before I soldered it onto the RX+ PCB and powered it up.

I'll investigate more on the power supply side (maybe try a bigger capacitor on the output of the 7809 regulator for the RX+) and report back.

[clubjoker]

Another minor issue I've noticed is that the first pressure measurement from the RX+ (from the onboard BMP280) after switch on, is a glitch: 747mBar. The is updated to a more believable reading (e.g. 1010mBar) after a minute or so, I imagine on the second reading. Don't know why this is happening. I'm concerned that when I put the system into service, recording weather history, this glitch will pollute the data. Anyone else seen this? If the hardware is being slow to respond, is there a software workaround to wait a bit longer before taking the first reading, or simply reject 747mBar, or something more elegant?

I never seen it happened, and it just should not happen. By those signs, I'm afraid that something may be wrong. Try to remember if you ever had the Meduino  powered by the USB, leaving the board without the external power supply. By the signs, it probable that the 74AHC244 chip was become damaged. Really hope I'm wrong. The first thing you should try to achieve is making that the 3.3V regulator don't warm.

Also try to run the I2C_Scanner and the FlashMemory_test utilities. The I2C_Scanner should consistently be able to detect all your I2C devices (RTC, Pressure Sensor and SHT31) at all speeds.

Thank you.


The I2C scanner utility returns this repeatedly on the serial console. Seems fine?

When the I2C scanner is running, the 3.3V regulator doesn't get hot!! What could this mean?

Code:

Scanning...
I2C device found at address 0x44  !
I2C device found at address 0x50  !
I2C device found at address 0x68  !
I2C device found at address 0x76  !
done

Scanning...
I2C device found at address 0x44  !
I2C device found at address 0x50  !
I2C device found at address 0x68  !
I2C device found at address 0x76  !
done

Scanning...
I2C device found at address 0x44  !
I2C device found at address 0x50  !
I2C device found at address 0x68  !
I2C device found at address 0x76  !
done

The flash memory test returns this:

Code:

Security register (byte 64 to 127) :
0xB, 0x4, 0xB, 0xF, 0x19, 0x22, 0x1F, 0x22, 0x0, 0x0, 0x17, 0x0, 0xFF, 0xFF, 0xF7, 0xFF, 0x30,
0x30, 0x4D, 0x32, 0x37, 0x39, 0x31, 0x36, 0x5, 0x19, 0x57, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x43,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF,
Status register :10001100
Manufacturer ID :
1F
Device ID (part 1) :
22
Device ID (part 2)  :
0
Extended Device Information String Length  :
0

What does that tell us?

Thanks. I really appreciate your help. And I'm still enjoying this

[werk_ag]

I was careful to remove the relevant diode on the Meduino before I soldered it onto the RX+ PCB and powered it up.

I'm glad to know it. So we can discard the biggest concern: a problem with the 74AHC244 chip.
Try to power the unit through a regulated 9V power supply, If I correctly remember, one user from AU have reported exactly the same problem (3.3V regulator running very hot) and he solved it by changing the mains power supply.

Discarding any problem with the 74AHC244, most probably the issue with the display is due to the length of the cable. Try to make it shorter, and please report back the results.

[werk_ag]

The I2C scanner utility returns this repeatedly on the serial console. Seems fine?

When the I2C scanner is running, the 3.3V regulator doesn't get hot!! What could this mean?

Both the results of the I2C_Scanner and FlashMemory_test are fine. So your I2C, and primary SPI bus are working correctly.

If when running the I2C scanner the 3.3V doesn't get hot, it seems to mean that the warming problem is related to the TFT display. Just for testing try to run the system with the TFT backlight OFF, then check if turning it ON, will make the regulator to become hot. It's possible that your power supply isn't drive enough power to the system

[clubjoker]

I did a test with TFT backlight off last night before I had to call it a day.

Your suggestion is right: with the backlight off the 3.3V regulator runs slightly warm, not hot. In fact, when I push the push button to turn off the backlight, while holding a finger on the regulator, I can actually feel it immediately starting to cool down.

Planned next steps:
- I've ordered a step-down buck converter, as I'd like to run the RX+ off a common 12V supply (I've got a spare from old network router etc.) and it's convenient to be able to run off a 12V lead acid battery.
- Try with a better 12V supply (through 7809 to the RX+ at 9V). I've been using a cheap switched-mode 12V supply that's rated at 1.5A, but appears to be low quality.

What current draw is expected at 9V for the Pro2+ RX, with backlight on?