I want to address the unreliable head detection that I and others have.
To do so, I need copies of the circuit diagrams around the detection mechanism.
I have been thinking about the issue and I expect the circuit that a pull up resistor and that the loop through the pogo pins pulls to ground. If the resistance of the pull down loop is too high (bad connection) then the line will stay high by the pull up resistor. I think that is how it works, but I am guessing. It is the logical design.
Does anyone have access to that part of the circuit?
Also. Is there a 5 V & GND power rail available on the pogo head?
I am thinking that we might be able to retro fit an optical sensor but I need power and details of the current detection circuit.
@mvillion indeed the detection is not very reliable. Also having some struggles from time to time. Instead of using an optical sensor a magnetic sensor could might be also an option.
You are right. A magnetic sensor would be a better choice and very reliably and there is one I can think of that needs no power. I have not seen them in years but I am sure they would still be around. Great suggestion. I will look into this.
I founds them and have ordered some. I am going to play around with them and see if I can find a way to retrofit them into the existing UTM. This would be a HUGE headache solved. It is one of the things that made we walk away for a year. (and have a sulk). Reliability is paramount if something is to keep my attention. LOL
What is need is the schematic for the controller board - the main PCB. I found the listed schematic as a PDF however I am fairly certain that it is not accurate.I would be extremely surprised if the main chip was connected directly to the pogo pins without any supporting infrastructure such as pull up resistors or short circuit protection. Possible, but surprised.
Here are some actions that you can take if the FarmBot fails to verify that a tool is mounted.
The best thing that I have done is cleaning the pogo-pins and adding the jumper link to the top of the tool head. Also sometimes the pogo pins just need to be replaced. They get old and the springs can wear out.
I had a similar problem- I used a meter to measure between the 2 points and found that they were not always making electrical connection. If I pushed it slightly it would make. I ended up putting washers under each pin which put more pressure on the connection and it has been fine ever since.
I have had this repeating fault from new. It is not reliable and cannot be trusted to swap heads, which sort of defeats the point of it. I cannot afford to swap out a head and loose watering capability when I am away (for example).
I have just realised that magnetic switches are not likely to work as there are already magnets in the UTM. Bugger. Back to optical I go…
Each Farmbot device type and version (e.g. ‘Genesis’, ‘v1.5’) has its own BOM and cloud folder for the Farmduino schematics . . which ‘model’ bot have you ?
I have written a very successful LUA script that attempts to load the head 5 times, putting it back if the connection is not found. It works extremely well. I will release a final version in time but there is a story on this.
I changed the magnet arrangements in the UTM head and redesigned the magnet arrangement on the tool head and raised the screws to make better connection with the Pogo pins. It now snaps into place with a satisfying click.
I wrote a soak test script which was fun. 100 head mounts and dismounts. This is where I discovered the find_home() LUA script is coded backwards performing X,Y then Z instead of Z,Y,then X as per the documentation. I found this out after finding damaged plants as the head was dragged across the garden bed as I had a find_home() every 5 mounts.
The script ran successfully 70 times then suffered X axis stall (Watching the log files). I found that one of the wheels on the bottom of the gantry had shattered, so I never make it to 100 successful sequential mounts and dismounts. I still think 70 is a new record for me. The combination of the new magnet arrangement and LUA code that retrys in a very nice way really made the difference.
I have wondered, are there lots of people with this unreliable head detection issue or is it only a few that suffer.
For those who are interested, here is a version of the code I have written. I will be publishing it on its own, when I have completed 100 successful tests.
-- === CONFIGURATION ===
local safe_z = 0 -- Safe Z height before mount attempt
local max_retries = 5 -- Max number of mount attempts
local fail_x = 0 -- Fallback position if all retries fail
local fail_y = 1206
local debug = true
-- === READ TOOL FROM VARIABLE (Tool type only) ===
local tool_input = variable("tool_to_mount")
if not tool_input or not tool_input.tool_id then
toast("ERROR: No valid tool slot provided")
emergency_lock()
return
end
-- Extract the actual tool’s ID from the ToolSlot table
local requested_tool_id = tool_input.tool_id
local slot_x = tool_input.x
local slot_y = tool_input.y
local slot_z = tool_input.z
if not (slot_x and slot_y and slot_z) then
toast("ERROR: Tool slot coords invalid")
emergency_lock()
return
end
-- === PRE-CHECK: IS THE REQUESTED TOOL ALREADY MOUNTED? ===
local raw = get_device("mounted_tool_id")
local mounted_tool_id = (type(raw) == "table") and raw.value or raw
if mounted_tool_id then
if mounted_tool_id == requested_tool_id then
toast("Requested tool already mounted.")
return
else
toast("Other tool mounted. Dismounting...")
dismount_tool()
wait(1000)
end
end
-- === FUNCTION: FORCE TOOL BACK INTO SLOT TO RESET PHYSICAL STATE ===
function force_unmount_tool()
toast("Forcing tool into slot...")
move_absolute(slot_x, slot_y, safe_z)
wait(500)
move_absolute(slot_x, slot_y, slot_z)
wait(500)
move_absolute(slot_x, slot_y, 100)
wait(500)
end
-- === FUNCTION: MOUNT TOOL WITH SAFE RETRIES ===
function mount_tool_safely()
local attempts = 0
local mounted = false
while attempts < max_retries do
attempts = attempts + 1
if debug then
toast("Mount attempt " .. tostring(attempts))
end
local pos = get_position()
if pos and pos.x and pos.y then
move_absolute(pos.x, pos.y, safe_z)
else
toast("ERROR: Unable to get pos")
emergency_lock()
return
end
mount_tool(tool_input)
wait(500)
if verify_tool() then
toast("Tool mounted OK")
mounted = true
break
else
toast("Fail " .. tostring(attempts))
wait(1000)
if attempts < max_retries then
force_unmount_tool()
wait(1000)
end
end
end
if not mounted then
toast("Failed after " .. tostring(max_retries))
move_absolute(fail_x, fail_y, safe_z)
wait(500)
toast("Emergency stop")
emergency_lock()
end
end
-- === EXECUTE ===
mount_tool_safely()
