Info from this week's video meeting.
20251120 ROS Lawn Tractor Automation meeting - Length 1:02:07 This video: https://youtu.be/CXP1BMELopM
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20251120
======
Here is a sexy Black Friday purchase....an AGM battery for my RTK Base Station.
*Thread Reply:* My new AGM battery weighs 64 lbs. That was a surprise. After a day and night without any charging and running my RTK Base station the voltage has moved from ~12.7V (the shipped state of charge) to ~12.6V. I will brush the snow off the solar panel today to see if it will now charge up. Temp in the shed is showing ~32 degrees and the outside air is ~27.
Info from this week's video meeting.
20251204 ROS Lawn Tractor Automation meeting - Length 33:18 This video: https://youtu.be/p9PhudztJRg
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20251204
======
I have tweaked my base station monitoring again. I wanted the ESP32 to log the actual time, in readable format. (ESP32 code) To do that I calculate the current local (i.e. Eastern) time on my RPi and then via a serial print command send the message "SETTIME <epoch adjusted time>" so the ESP32 knows the actual time of day. (RPi code) The ESP32 doesn't actually 'know' the current time, it is simply tracking the elapsed seconds and formats the date adding the elapsed seconds to the time from the SETTIME command in the function "String getCurrentTimestamp()" on the ESP32. I pass SETTIME once per day at the same time I pull the ,csv down from the ESP32. (script triggered by cron job) Supposedly the crystal on the ESP32 can keep time accuracy to ~ 1 sec/day.
I also reduced the fields I was logging to minimize the load storage and IO load on my SD card.
Here is the link for the Waveshare camera
https://www.waveshare.com/wiki/IMX219-83StereoCamera
https://developer.nvidia.com/embedded/learn/get-started-jetson-orin-nano-devkit
Info from this week's video meeting.
20251211 ROS Lawn Tractor Automation meeting - Length 32:11 This video: https://youtu.be/MF3anrR7Dn0
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20251211
======
I'm in Texas and figuring out a Base station for here. I have my ComNav GPS from previous days. Thinking about connecting an ESP32 and pushing the correction data to the web. Not sure if the ComNav RTCM correction messages I'm currently outputting (1004, 1006, 1008, 1012 and 1033) will be enough to get a Fix for my F9P. In the process of re-learning programming an ESP32 I put a .h header in my include folder in PlatformIO to pull data about the ESP32. That's been useful to confirm exactly which board I have and also helps with version control being able to show what script is running on the ESP32 and when it was compiled.
Well I have my ComNav K706 GPS (2018 vintage) outputting the same RTCM messages my Pittsburgh F9P base station is. The reason that is important is hopefully the dual F9P's I have on my tractor electronics board will be able to achieve RTK Fix using that correction data while I'm here in Texas. That will be a tomorrow test, plus I have to program my ESP32, which is getting the data from the COM port on the GPS, to push it to UDP port 6001, which is where my rover/tractor is programmed to pull it from. I've not done that part yet. I do have my python version from my Pittsburgh base station as a guide.
Attached is a screen shot of the ESP32 web page used for testing to see what data the ESP32 is seeing. The Novatel-like configuration statements are below.
UNLOGALL COM2
LOG COM2 RTCM1005B ONTIME 10 (Stationary reference station position, every 10 seconds)
LOG COM2 RTCM1074B ONTIME 1 (GPS observations)
LOG COM2 RTCM1084B ONTIME 1 (GLONASS observations)
LOG COM2 RTCM1094B ONTIME 1 (Galileo observations – you already have this)
LOG COM2 RTCM1230B ONTIME 10 (GLONASS code-phase biases, every 10 seconds)
SAVECONFIG
Seems ESP32's don't like hardcoding the IP address in the script. Seems if DHCP does it, they are more likely to not refuse connections from clients. I'm using the library ESPmDNS.h which facilitates me being able to access the ESP32 from the rover/tractor RPi using 'HOST = "rtcm-esp32.local"'. To get to the HTTP page I need to use a network scanner to pull the IP address. In my case, 192.168.1.95
ESP32 test code: ```#include <WiFi.h>
include <ESPmDNS.h>
include <WebServer.h>
include <algorithm> // For std::remove used to clean up disconnected connections at ~line 101
const char* ssid = "mySSID"; const char* password = "myPassword";
WebServer httpServer(80); WiFiServer tcpServer(6009); std::vector<WiFiClient> tcpClients;
unsigned long lastSendTime = 0; unsigned long counter = 0; const unsigned long sendInterval = 2000; // Send every 2 seconds
void handleRoot() { String html = "<html><body><h1>ESP32 mDNS Test</h1>" "<p>mDNS hostname: rtcm-esp32.local</p>" "<p>Current IP: " + WiFi.localIP().toString() + "</p>" "<p>TCP test server running on port 6009</p>" "<p>Connect a client to see 'Hello world' + counter every 2 seconds</p>" "</body></html>"; httpServer.send(200, "text/html", html); }
void setup() { Serial.begin(115200);
WiFi.begin(ssid, password); Serial.print("Connecting to WiFi"); while (WiFi.status() != WL_CONNECTED) { delay(500); Serial.print("."); } Serial.println("\nWiFi connected!"); Serial.print("IP address: "); Serial.println(WiFi.localIP());
// Start mDNS if (MDNS.begin("rtcm-esp32")) { Serial.println("mDNS started: rtcm-esp32.local"); } else { Serial.println("mDNS failed to start"); }
// Web server httpServer.on("/", handleRoot); httpServer.begin(); Serial.println("HTTP server started on port 80");
// TCP server tcpServer.begin(); tcpServer.setNoDelay(true); Serial.println("TCP server started on port 6009"); }
void loop() { // Handle web requests httpServer.handleClient();
// Accept new TCP clients WiFiClient newClient = tcpServer.available(); if (newClient) { if (newClient.connected()) { tcpClients.push_back(newClient); Serial.printf("New TCP client connected (total: %d)\n", tcpClients.size()); } }
// Send data to all connected clients every 2 seconds if (millis() - lastSendTime >= sendInterval) { if (!tcpClients.empty()) { String message = "Hello world from ESP32! Counter: " + String(counter++) + "\r\n";
std::vector<WiFiClient> disconnected;
for (auto& client : tcpClients) {
if (client.connected()) {
client.write(message.c_str(), message.length());
} else {
disconnected.push_back(client);
}
}
// Clean up disconnected clients
for (const auto& disc : disconnected) {
tcpClients.erase(std::remove(tcpClients.begin(), tcpClients.end(), disc), tcpClients.end());
}
Serial.printf("Sent message to %d client(s)\n", tcpClients.size());
}
lastSendTime = millis();
}
// Small delay to avoid watchdog issues
delay(10);
}
Python code to confirm it is able to receive data on port 6009 (port number arbitrary)
"""
Minimal Python TCP client to test mDNS resolution and connection
Connects to rtcm-esp32.local:6009 and prints received data
"""
import socket import time
HOST = "rtcm-esp32.local" # mDNS hostname PORT = 6009
def main(): print(f"Attempting to connect to {HOST}:{PORT} ...") sock = socket.socket(socket.AFINET, socket.SOCKSTREAM) try: sock.connect((HOST, PORT)) print(f"Connected successfully using mDNS resolution!") sock.settimeout(5) # For non-blocking read with timeout
while True:
try:
data = sock.recv(1024)
if not data:
print("Connection closed by server.")
break
print("Received:", data.decode().strip())
except socket.timeout:
continue # No data yet, keep looping
except Exception as e:
print("Error receiving data:", e)
break
except Exception as e:
print(f"Connection failed: {e}")
print(" - Check that the ESP32 is running the sketch and on the same network")
print(" - On Raspberry Pi, ensure avahi-daemon is installed/running: sudo apt install avahi-daemon")
finally:
sock.close()
if name == "main": while True: # Auto-reconnect loop main() print("Reconnecting in 5 seconds...") time.sleep(5)```
Everything has to come off and get worked around the new Jetson. The mounting board is beginning to look like Swiss cheese. Maybe a new board is due. Did get JetPack flashed on a 128 gb card. Won’t know if everything works until next week.
Info from this week's video meeting.
20251218 ROS Lawn Tractor Automation meeting - Length 41:26 This video: https://youtu.be/yOnEY03DSco
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20251218
======
I have RTK Fix with the help of my Texas base station which consists of my older ComNav GPS feeding my ESP32 which puts the RTCM corrections on the network using Wi-Fi.
Yeah that wiring job was poor so I have to go back and fix it on my LED light tower.
@JeffS just what your bot needs to do stairs!
https://www.facebook.com/reel/1191512589215688/?fs=e&fs=e
It may be innocent but you never know these days! https://apple.news/Ax8IXXmnARwWL7GRAk1BN0Q
*Thread Reply:* Inside the pump house there is a Franklin Electric control box that is triggered by the pressure switch and sends power to the pump that is below ground at ~200 feet. An electrician, to get 110 volt for the light, connected a wire (black) to L1 and a wire to ground (white)
Info from this week's video meeting.
20260101 ROS Lawn Tractor Automation meeting - Length 1:04:32 This video: https://youtu.be/cuojyTsqRnE
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20260101
======
My journey on resurrecting a 20 year old oscilloscope. I now know the BIOS freezes before any applications start. The problem is not an old HDD. Two theories on fixing that. One, easier, but unlikely, the CPU is overheating due to heatsink paste deterioration. Or two, more likely based on my symptoms, the capacitors on the motherboard power rail, have deteriorated and as they warm up they provide unstable power which causes the BIOS to freeze. So I need to pull out the cards and remove the motherboard. Then remove the capacitors.
Before removing the cards: After removing the cards: Six of the caps that I have removed and ordered replacements from Mouser:
*Thread Reply:* Before I changed the battery I received a CMOS error. After a new battery that error went away. The scope does beep at ~.5 Hz for ~60 seconds. I don't know what that signifies. It might be that is stops when the BIOS freezes or its a normal startup process.
To be more clear about its behavior, as it is booting I can open the BIOS menu. After ~60 seconds it becomes non responsive. As an example I can go into the BIOS menu and use the up/down arrows to move around the menu. After it freezes, that ability stops.
Since I bought a CF Card and adapter to replace the HDD, I decided to try and load FreeDOS on to the CF Card. I relearned how to burn a CD and booted to the CD running FreeDos. The motherboard started the process and got to 4% before hanging.
I'm hoping my issue is old capacitors not delivering smooth power. Replacements arrive Tues/Wed. A guy at https://groups.io/g/HP-Agilent-Keysight-equipment/topic/117057746
suggested it could be an overheating CPU. I also ordered thermal paste and pads to cover that base too.
Just fyi. https://weeklyrobotics.com/weekly-robotics-342
I'm using https://ntfy.sh/app to receive a daily notice of how many times the ESP32 booted. The flow is the ESP32 at the pump boots when the pump comes on, it radios the other ESP32 over LoRa of the event, that ESP32 makes an entry in the DB SQLite on the connected RPi which once a day sends an ntfy notice. There is also a process that runs on the RPi to check if the pump comes on three times in an hour. That would signal a problem. My next upgrade will be to move off the SD Card and use an SSD instead. That will be the next visit to Texas.
I was able to get FreeDOS installed on a compact flash and it had 1980's games
Info from this week's video meeting.
20260108 ROS Lawn Tractor Automation meeting - Length 1:10:53 This video: https://youtu.be/Uy2iT6DxPFw
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20260108
======
*Thread Reply:* This is a close up of the IC alignment. The 'magic' I see is the top of integrated circuit is aligned with the bottom of the magnets.
Info from this week's video meeting.
20260115 ROS Lawn Tractor Automation meeting - Length 49:56 This video: https://youtu.be/xJ2DGuKfuiU
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20260115
======
Info from this week's video meeting.
20260129 ROS Lawn Tractor Automation meeting - Length 48:42 This video: https://youtu.be/0xZxMygjXQY
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20260129
======
Changed motor controllers from sawtooth to mmd10 and eliminated 1 Pca board. Next is all battery power setup!
Info from this week's video meeting.
20260205 ROS Lawn Tractor Automation meeting - Length 48:08 This video: https://youtu.be/3pVQcfCzLsE
Go to the HTML index to get the description for this meeting: http://sampson-jeff.com/RosAgriculture/LawnTractorMeetingNotes.htm#20260205
======
'ROI" stands for region of interest in my code where I monitor if my water pump light comes on. It is subtle, but over time my Logitech USB camera has dropped down about 6 pixels below my initial setup. The red arrow is the placement of my light. This sagging is another reason why monitoring using the electrical circuit is a less fragile solution. Anyway, I now have a routine to run between 11pm and 6pm to calculate and update the x, y coordinates for the brightest spot. Seems python has a function for "centroid calculation". I can then set the other regions relative to the light position. These are used to compare brightness levels to detect when the light has actually come on. Like I said, its fragile.
*Thread Reply:* Not perfect yet, but better. Last night the program self-adjusted to move the area of interest higher by 5 pixels.
Some progress, We have keyboard control for drive motors and Logi 720p camera calibrated servos. Now the real work starts.
*Thread Reply:* It's really just for a test. I can't imagine this pump would stand up to much abuse. If I can get a few hundred gallons a day I would be happy. I will put it in a 4" pipe and try and keep the cold water flowing over the motor to keep it as cool as possible. At least that is what I'm thinking.
Long video, but interesting to see how he replaced the proprietary engine ignition system with an off the shelf ABS sensor and HEI unit. https://www.youtube.com/watch?v=I5cpMHBEngs
This is a photo of a communications receptacle on a MEP-805B generator. Another side project to read the status messages using an RPi. First step is to DIY a plug.
