TX is not reliable on v1 on certain configurations

[marklynch]

Initial version of this TX was only working in some installs.

This is likely to do with the TX sometimes being ~7v and sometimes closer to 12v.

Thanks to help from Skymuster

Hi there Mark, thanks so much for the extra suggestions. I spent some time today troubleshooting and happy to say that i finally GOT IT TO WORK 😁

There were a few things that I did which ultimately fixed the circuit. Firstly I managed to calculate what the likely value of the internal data bus pull-up resistor was, using a dummy resistor to pull down the data bus line and measuring the drop. I calcuated it to be ~300Ω. This means that the internal pull-up is really strong, requiring a significant effort to pull it down. To get around this I ended up shorting out R3 (i.e. removing the collector resistor altogether) which allowed the data bus to be pulled down when the transistor was properly saturated.

That led to another other problem that I had with the transistor - the GPIO output wasn't sufficient to properly saturate transistor and trigger the bus pull down. To provide provide greater current to the base of the transistor I changed R4 to 470Ω and increased R5 to 100k. This initally didnt work however I discovered a loose solder joint on the base pin of the transistor which when fixed made the circuit finally come alive!

So, to summarise, the updated values of the resistors used in my circuit were:

R1 = 100k Ω
R2 = 100k Ω
R3 = 0 Ω
R4 = 470 Ω
R5 = 100k Ω

This issue it to review and implement this in the circuit

R1 = 100k Ω ✅
R2 = 100k Ω ✅
R3 = 470Ω -> 0 Ω
R4 = 10k -> 470 Ω. (or 1KΩ)
R5 = 10k -> 100k Ω

Estimated pull up to be ~300Ω - but remeasure and confirm this.

RX
DATA ----[R1:100k]----+----> GPIO1 RX
                   |
                  [R2:100k]
                   |
                  GND

TX
DATA ----[R3:0k]---- C  NPN
                             E ---- GND
GPIO2 TX ----[R4:1k]---- B
                 |
               [R5:100k]
                 |
                GND

Improved RX circuit:

RX (with 5.1 V Zener clamp):

                                   +----> GPIO1 RX
                                   |
DATA ----[R1: 100k]----------------+
                                   |
                       +-----------+-----------+
                       |                       |
                   [R2: 100k]            [ZD1: 5.1V Zener]
                       |                       |   cathode (banded end) up
                       |                       |   anode down
                      GND                     GND

[zagnuts]

@marklynch - Not sure I have any suitable zener diodes if they are needed, but happy to do a run to Jaycar to get some and then hack my board as well to test it here..... will definitely have the resistors so can check if the TX circuit mods improves things. :-)

[marklynch]

@zagnuts the zener diode is not required - as the RX is working for you already - I'll likely add to the next build though.

[zagnuts]

Good oh, @marklynch. Well, will try swapping some resistors and will let you know. I'm assuming with R3 just need to remove the existing resistor and jumper wire it....

[marklynch]

Yep - just remove and solder right across it for R3.

[zagnuts]

OK - @marklynch I can confirm that it works for me both ways now! Sadly had no suitable SMD resistors so had to mess up the circuit board with some old school jobs, but hey there was a lot of space! Yeah, not proud but whatcha going to do. Big thanks to you & Skymuster. ;-)

[zagnuts]

Spoke too soon. Was working like a charm yesterday, but today it became mute again. Tried rebooting etc but no change - back to the way it was. The circuit board physically looks fine, so hopefully it's just a failed Aliexpress resistor so will remove/test/replace them all just in case when I can. I used a 470 on R4, so maybe I'll change that to 1k to see if that makes any difference.

update
The resistors were fine. Changed R4 to 1k and it seems to be working again - unclear if it was swapping the resistor or just have a few minutes rest did the trick, but I'm happy to move on and pretend nothing happened. ;-)

[marklynch]

Great. Glad it's working again. I believe it should be more reliable as a 1k resistor and once I test it on my setup (next week) to confirm it works there too I'll update this.

[lawther]

Hey folks - how is this progressing? Are the updated resistor values above the 'officially known good' ones?

[zagnuts]

Are the updated resistor values above the 'officially known good' ones?

Mark is away I believe, but I would suggest this is as official as possible. It definitely worked (and is still working) for me:

1. Removed the resistor in R3 and bridged it with some wire.
2. R4 - changed to 1k ohm (initially tried 470 ohm but it had intermittent issues)
3. R5 - changed to 100k ohm
   This made the TX circuit able to be heard reliably. The RX circuit I didn't need to change.

[lawther]

Thanks - good to know it's been reliable! I might see if I can modify the board I have.

[zagnuts]

FYI I used my heat gun to remove the smd resistors, then a bit of flux & tinned the pads, and put on the resistors & bridge, then some ipa to clean and all good.

[marklynch]

@lawther suggest give these a try. I haven't been able to test yet with travel, but if works for both of you then it's the best known config.

I will revisit and test on my system in a few weeks and get back and if confirms that works will update the kicad files too.

[lawther]

I gave it go (R3->bridged, R4->1K, R5->100K), and so far so good. I've only tried turning the pool/spa lights on and off, but that seems to work. I'm using Home Assistant to do the controlling. It's not instant, seems to be ~1-2 secs delay. From a quick inspection of the packet log I couldn't see any obvious errors.