TM220A table top pick and place overview
via Dangerous Prototypes
A pick and place is a machine that puts electronic components onto a circuit board that has been coated with solder paste. To complete the prototype you just place the populated PCB into a reflow oven. We hope it will speed up production of one-offs and single prototypes.
The NeoDen TM220A is a table top pick and place designed and manufactured in China. Most PnPs are huge machines that take up a room, but this fits nicely in the workshop. It doesn’t require a separate compressor, it has a noisy internal vacuum pump that provides suction for lifting parts.
Up to 15 reels of components can be loaded. 12 x 8mm, 2 x 12mm, and 1 x 16mm. A tray at the front holds larger components like chips. The bigger TM240A that has twice as many reels and costs $1000 more. For us the TM220A is the ideal size. More importantly, it’s light enough to carry up two flights of steep stairs into the workshop.
Flip the switch to start the machine. You’re greeted by a short start-up sequence. The menu is separated into 3 tabs: ‘Tasks’, ‘Manual’, and ‘Setting’. The Tasks tab shows a list of placement files that can be run. The Manual tab has buttons to manually control and test most hardware.
The Settings menu controls the language (English or Chinese), speed, calibration, etc. In the video run at the lowest 10% speed setting. The higher settings the table and throws components around, after the video it’ll be moved to a workbench anchored in the corner of two walls.
You’ll need a manufacturer-supplied password to save any of these setting permanently. Our machine was supplied in Chinese mode, and we got the code to change it without difficulty. We have heard of cases where people were unable to get it though.
The most important thing for our workflow is to get the placement data out of Eagle as quickly as possible. It doesn’t make sense to spend an hour programming the machine to do a single prototype.
We’re working on a ULP to dump Eagle board files for the TM220A/TM240A with a few clicks. This is key to making the machine useful in our situation. Run the ULP, assign a reel (or no reel) to each component on the BOM, then export the placement file. Placement files are written to an SD card, the card is then stuck in a reader on the control panel of the TM220.
The TM220A has two placement heads. They can place two of a single part in one movement, or each can be fitted with a nozzle for a different size part. Our ULP doesn’t take advantage of this yet, but it could be added in the future.
Placement files are simple plaintext CSV files with lists of reels, parts to place, and position information. The wiki has more information on the format of the placement file and ULP.
The TM220A doesn’t have a vision system, it relies entirely on calibration and positioning. This means the board needs to be aligned flush with the machine’s coordinate center. In the manual control menu a button turns on a laser sight that shows the alignment of the PCB in the vice.
Parts are advanced by the pick and place head. It moves to the reel and a solenoid controlled needle pushes into the holes on the side of the reel. The entire pick and place head moves outwards to pull the part. The feed distance is defined per reel in the configuration file.
While the pick and place head advances the reel one part, a set of friction wheels grab excess film from the part reels.
After the reel advances the head drops down and picks it up with vacuum and rotates it to the correct position. The head moves to the part’s location on the board and then drops it by removing the vacuum.
This is an example of the working screen showing the part’s reel location, coordinates on the board, rotation, height (which is usually 0), whether to skip it, and the description on the board. This is all defined by the format shown on the wiki.
Conclusions
We’re withholding judgement until our easy-export ULP is working. Our goal is to save time by quickly placing common parts on single prototypes. There’s a lot more room for error and hand adjustment than on a manufacturing run, even if it isn’t perfect for high volume work it will still probably be a useful tool for us.
The value of a top and bottom vision system that compensates for misalignment is immediately apparent. This machine depends entirely on calibration and registration. Maybe someone will develop an open source add-on for these cheap machines.
As always, we caution against buying a pick and place to manufacture your first open hardware project, especially a cheap machine. Many small startups regret the time and effort invested get fairly mediocre results. Running a production line is a whole additional job. If you’re doing your first hundred or thousand board we recommend contacting local assembly houses. You don’t have to go to China, there’s assembly places everywhere, including the US.