Erroneous behaviour with Nodemcu and KeybusReader - help troubleshooting

[fennemaj]

First, let me start off with how awesome this is! I used the Unlocker to help get my installer code and it worked like a charm! So happy to have control over my alarm system finally. I'm moving up to the next step which is controlling the alarm system directly, long term, through Home Assistant, however, when I tried using a nodemcu instead of the arduino uno I had used for the unlocker, I ran into issues, namely that the alarm system went haywire and the alarm started going off. I decided to step back and just do the KeybusReader sketch as a troubleshooting step and the same thing happened. So, I took the exact same breadboard circuit I was using with the nodemcu, swapped out the 33K resistors for 15K resistors and connected it up to the arduino with the KeybusReader sketch loaded, and it worked beautifully. I re-flashed the nodemcu, swapped it back and same issue, the system goes haywire, and on the serial monitor there is a bunch of stuff like this showing up:

20:24:35.988 -> 123.58: 00000101 0 10010001 00000010 00010000 11000111 00010000 10000011 00000000 01000011 [0x05] Partition 1: Ready Trouble Backlight - Stay zones open | Partition 2: disabled | Partition 3: Trouble - Unknown data: 0x83 | Partition 4: none - Unknown data: 0x43
20:24:35.988 -> 123.60: 11111111 1 11111111 11111111 11101110 11101011 10001000 01111011 00011100 11101101 [Module/0x05] Partition 3 Key: 7 Partition 4 Key: | Zone expander notification: 3 4 | Module tamper notification | Keypad zone notification | Zone label broadcast
20:24:36.088 -> 123.68: 00000000 0 10010001 00000000 00010000 11000111 00010000 11000111 00010000 11000101 [0x00] [CRC Error]
20:24:36.088 -> 123.68: 11100101 1 11111011 10111111 01101001 01011011 10011100 00110111 11101110 01001011 [Module/0x00] Unknown data
20:24:36.142 -> 123.76: 00000000 0 00000000 00000010 00010000 11000111 00010000 11000110 00010000 10000111 [0x00] [CRC Error]
2

For the Arduino I'm using pins 3,5,6, per the sketch. I connect ground to the "Aux -" since I'm powering the board with the USB for the serial monitor, and I'm using a 2n2222A that I had lying around.

For the nodemcu, I'm using pins D1, D2, and D8 per the sketch, same thing, I connect ground to the "Aux -" on the DSC since I'm powering the board with the USB, everything else is the same except the resistors. I'm just taking the leads from the breadboard and plugging them into the nodemcu instead of the Arduino.

Also, I've got a DSC 1832.

Maybe it has something to do with the ground and the AUX - setup? I don't have a voltage regulator handy, so I figured I could power things with just the onboard USB plug on the nodemcu, but maybe tying that into the DSC AUX ground is creating a challenge? If I don't plug the ground into the AUX - though, the KeybusReader doesn't work with just the yellow and green connected.

I greatly appreciate any thoughts, or troubleshooting suggestions, and thank you again for your hard work on this!

[fennemaj]

I got a chance to test some suggestions from the AVRFreak forum around changing where the ground was tied into the DSC, and it looks like I was getting the same issue whether I tied into the AUX -, the black terminal, or the GND terminals for the sensors. I did some additional troubleshooting though and it looks like the culprit ended up being the resistors in the clock and read circuits.

There were two issues.

1. One was just stupidity on my part. I didn't test the resistors before using them and one of my 33K resistors had a measured resistance of 0.00, so that didn't help.

2. The other issue is that the voltage divider circuit seemed incorrect for the voltage coming out of the DSC. The 33K resistor was too high and causing the erroneous behavior. I used the same 15K resistors I was using for the Arduino circuit and it ended up working beautifully. I'm going to monitor it to see if the slightly higher voltage creates stress, but I liked one person's suggestion of using opto-isolation between the DSC panel and the nodemcu, which would eliminate the issue I experienced, and any voltage differences between the two systems.

It took me a while to track this down last night, but I had the thought that all the erroneous behavior seemed to be "write" related, rather than "read" related, so I thought, "I wonder what would happen if I just remove the 'write' pin and leave it in read only mode?" Surprise, it worked like a charm! So I was thinking, hey I've got this nailed down to only a few components which help write data to the system. So I do a little bit more and it starts to look like the NPN transistor is the culprit. So I figure, let me cross check what I'm seeing when the nodemcu is connected, against the Arduino, since both circuits were using the transistor and if it was a problem for one, I would've expected it to be a problem for the other. I plug all the interface pins of the circuit into the Arduino, to see why the write seems to work fine with the arduino but not the nodemcu, and wouldn't you know it, it's not working with the arduino either. What the heck?? I'm scratching my head. I review the circuit and see in my haste that I accidentally used the 33K resistors for the nodemcu instead of the 15K resistors for the arduino. I plug into the 15K resistors and the issue is gone, no problems reading or writing. So I take the circuit back to the bench and test the resistors. Yeah, one of the 33K resistors was showing a resistance of 0.00, so that wasn't good. But even after swapping both out for 33K-measured, resistors, it still gave the erroneous behavior. In the end, I've just stuck with the 15K resistors which send a little less than 5v to the nodemcu and I'm monitoring to see how that goes. I think I'm going to switch over to an opto-isolated circuit to eliminate the issue with the voltage difference completely and go from there. But I wanted to share in case other's had the same challenges and were looking for some insights.

[Dilbert66]

I doubt the voltage divider is the issue here . As long as you have the correct ratio. 33/15 will give you about 5volt for arduino. You need 33/10 for the ESp as you need no more than 3.3 volt . This circuit as designed works very well both for both arduino and the esp . I suspect you have other issues with your circuit. If you accidentely used 0 ohms instead of 33k then you were then feeding 12volt into the chip since the 33k is there to drop the excess voltage which potentially killed that pin causing you more issues. Try new input pins now or a different chip.

[fennemaj]

Agreed 100%. After I found the bad resistor, I decided to switch out the nodemcu board completely because I didn't want to take chances on how badly the 12v had damaged it. I was using 15/10 to get approx 4.8v for the Arduino and 33/10 for the ESP, per the diagrams. When troubleshooting, I figured my best option was to try and limit the variables that could be causing the issue. I pulled out the 33K resistors and swapped in 15K resistors when switching the boards, everything else remained the same, the board, the wires, etc. Unfortunately I don't have a scope to measure the actual output voltage from the DSC, and it could even be that the 2N2222A transistor is not the best choice.

I also wanted to say thank you for responding and your thoughts and insights, and even a greater thank you for all your hard work on developing this. You put in a ton of work to create this and I for one, greatly appreciate it. So my thanks to you and taligentx and all the others who helped develop this!!

[Dilbert66]

The 2n2222 is perfectly fine. It's not a critical component. When you say erroneous behaviour what do you mean exactly? Is it affecting the panel itself or are you not getting any data in /out? If the panel is affected, you either have a bad ground or you are loading up one of the data lines on the panel. Either via a short or putting spurious data unto the line.. Unfortunately to debug something like that properly you would need a scope. Are you sure you are using the correct write pin from the Nodemcu? If you have the wrong pin, you would be driving the line all the time, causing panel issues. Also, ensure you have the npn wired correctly which I assume you already checked. Sometimes you just need to step back and recheck everything again and never assume something is correct.

[fennemaj]

I totally agree with you. Always best to take a step back, re-check everything multiple times, etc. I can confirm that the write pin is attached to D8, the read is from D2, the clock is from D1.

The behavior I'm seeing is, if the write pin (virtual keyboard) is disconnected, the entire thing operates as expected. It reads inputs correctly, whether it's the Uno or the ESP connected to the circuit. It's only when I bridge the collector on the NPN over to the green wire, that I get a deluge of problematic signals from the circuit to the panel. A snippet looks like this in case it's helpful:

20:24:35.988 -> 123.60: 11111111 1 11111111 11111111 11101110 11101011 10001000 01111011 00011100 11101101 [Module/0x05] Partition 3 Key: 7 Partition 4 Key: | Zone expander notification: 3 4 | Module tamper notification | Keypad zone notification | Zone label broadcast 20:24:36.088 -> 123.68: 00000000 0 10010001 00000000 00010000 11000111 00010000 11000111 00010000 11000101 [0x00] [CRC Error] 20:24:36.088 -> 123.68: 11100101 1 11111011 10111111 01101001 01011011 10011100 00110111 11101110 01001011 [Module/0x00] Unknown data 20:24:36.142 -> 123.76: 00000000 0 00000000 00000010 00010000 11000111 00010000 11000110 00010000 10000111 [0x00] [CRC Error] 2

Not sure if that's helpful, but I appreciate any thoughts you have.

[Dilbert66]

Well, obvioulsy you have something wrong with the transistor circuit or the d8 line or somewhere else with your hookup. I really can't say what it is. Good luck! Check your voltages on the d8 line and the transistor base to see what they are at least. They should be zero at idle.

[fennemaj]

Sometimes the best answer is to just start over completely from scratch and test things thoroughly again. I took everything off the board, and started with a completely different layout to keep myself from falling into the same design as before. I used a bench power supply to run a low amp, 12V signal to mimic the inputs and validated the voltages at the output pins (around 2.9V) with the 33K/10K voltage divider. I hooked everything up again with the KeybusReader sketch and it worked like a charm. I have no idea what I had crossed before, but whatever it was, starting over from scratch solved it. Thanks for all the help and troubleshooting, and thank you again for making this project in the first place!

[kghandi]

I would question your input power. DSC supplies 12V which I believe is greater than the nodeMCU max input voltage. I used this buck converter:

5 Pack Mini DC-DC Buck Converter, 3A Step Down Power Supply Module Adjustable 12-24V to 5V https://www.amazon.com/dp/B08L5XSXD9/ref=cm_sw_r_cp_api_glt_i_3BMH3AF2WWHQFD8XQ3M1?_encoding=UTF8&psc=1

It is currently unavailable, but you should be able to find others like it. With this one or some others like it you would have to cut the trace on the back for the ADJustment potentiometer and solder across the pads for 5V.

The resistors and transistors as described on the main page worked perfectly for me with the above voltage regulator and a WeMos D1 Mini NodeMCU connecting to a DSC Power Series 832 (PC5010).

[fennemaj]

It's a good thought. I'm powering the system via an external USB power supply, but when I've been doing the testing, I've been just using the USB power from my laptop where I've been monitoring the serial comms to see what was happening with the board. Long term my alarm system already has quite a few devices drawing current from the board, so I'll probably stick with the external power supply for this so I don't run up against the 600mA max for the board.