For years I’ve used a normal pair of tweezers to pick up small parts. It works fine but in the quest for further improvements I looked to buy a vacuum pick-up (also known as vacuum tweezer) tool to assist in this process but I didn’t see anything suitable. Some were low-cost, others were expensive. At reasonable cost I could not find an appropriate tool with the features I wanted.
I could have stayed with the manual tweezers but deep down all of us as engineers want to continue to optimize and in that spirit it was decided to build a custom solution and have some computer integration functionality too. Check out the 10-minute video here for more detail (first 4 minutes are the technical stuff, then a demo of it in action followed by an assembly video).
The end result is useful for picking-and-placing small surface mount technology (SMT) components but could also have uses in other industries such as jewellery making (for picking up small stones/beads).
This blog post documents the steps in the hope that others find it useful too.
What are the requirements?
For usefulness the tool needs to be able to consistently and accurately pick and place parts otherwise it is less useful than tweezers. It must be at least as consistent and accurate as tweezers and must improve the user’s pick and place speed by a significant margin compared to normal manual tweezers to make it worthwhile to construct this project.
Increased throughput (i.e. to pick and place a certain number of components within a shorter time) will be achieved through several features I’d not seen on commercial vacuum tweezer tools. I’ve only started using this project and time will tell if the features are useful or not.
Another requirement was that the tool must be reasonably reliable and not break down often or at least be based on low cost and easy-to-replace parts (such as a replacement tip if the original one falls off and is lost).
Although the tool needs to be cost-effective compared to commercial off-the-shelf vacuum tweezers, there has to be a balance of reliability/maintainability and features and it would be nice to use this tool for commercial prototyping as well as hobbyist use.
The mechanical aspects of the design involved a bit of experimentation but the final result is repeatable using easily obtainable parts. Some vacuum tweezer systems rely on a compressed air source but I didn’t have this so for now this project uses a built-in electric pump from ebay (it remains to be seen if it is a reliable component but it can be swapped out at a later stage if it turns out that additional reliability is needed in this area).
The pump has two ports, one for suction (vacuum) and the other for blowing. Such pumps are likely also used for aerating small fish tanks so that could be an option too although I didn’t open one to confirm. Both the suction and blowing ports are used in this project. Diaphragm pumps make a noise but could be placed under the desk (with a remote LCD screen) if it got irritating at ear level. Some attempts were made to minimise vibration being transmitted in the mounting of the pump, and the sound was significantly reduced but more could be done in this area. It makes far less noise than my hot air tool but this tool needs to be used for hours on end. The noise is diminished immensely by placing it on a rug, so I need to spend more time investigating pump mounting methods.
If two users sitting nearby required vacuum tweezers (i.e. to assemble PCBs simultaneously) then the cost and the overall noise level can be further reduced by using a ‘double head’ pump rather than constructing two units. I went with a single head pump for now but the project could be extended into a double unit if later desired.
To control the air flow a valve was used; this has three ports and much like a single pole double throw (SPDT) switch the valve will connect one of two tubing connections to a third connection under control of an electrical signal. The particular valve used will block the first tubing connection when the valve is activated.
In theory this sounds straightforward but in practice the blowing port on the pump won’t have any air flow when the vacuum port is blocked. For the vacuum pen attachment, we want to be able to switch from ‘suck’ to ‘blow’. A sustained blow operation is unwanted, since that would blow away the components. Instead what was tried was a short controlled burst of air. To achieve that, a small reservoir was constructed which gets pumped up (i.e. increasing pressure) during the component picking vacuum (suck) cycle. It was named Mr. Sneeze. When the valve switches to blow mode, the pressure built up in Mr. Sneeze is used to eject the component from the end of the vacuum tweezers.
Without doing calculations I resorted to trial-and-error to establish the reservoir size. I accidentally over-sized it (by about 500%!) by using the nearest container at hand (an aluminium can with airtight screw-cap) which was about the size of a C sized battery and so I then filled the inside with a piece of wood dowel (painted in PVA to prevent wood fibre getting inside the pneumatics) to resize to an appropriate reservoir air volume; this was easier than scrapping it and finding a smaller container. If I was building it again I would just use a smaller container.
To control the pressure a regulator is used. Even without the regulator anything bad is unlikely to happen (the low cost pump will probably leak and the tubing will pop off if there was any extreme pressure build-up) but the regulator is inexpensive to add and provides more control over the behaviour when releasing components from the vacuum.
A home-made filter could be added into the design although I didn’t do this for the prototype; it could be wise to have some thin filter material before the vacuum tweezer connection port in case any dirt or tiny components get sucked into the tubing!
To operate the valve a foot-switch is used. The switch is wired to a board so the microcontroller has ultimate control The has an on-board USB interface Software was written so that whenever the switch is double or triple pressed a message is sent over the USB interface(the looks like a USB UART to the connected PC
When a double- or triple-press message event occurs, appropriate PC software will then send the ASCII encoded text to display on the on-board LCD. The reason for this is to increase user efficiency; the conveniently-placed LCD display can be used to indicate the component reference and value for the next part that needs to be picked and placed. Once the part has been placed the user can perform a double-press on the foot-switch to advance the display onto the next component that needs to be placed.
The idea could be extended even further to automatically highlight where on the board the next component needs to go (perhaps display it on the computer screen). It would be nice to see an EAGLE user language program (ULP) for this.
If the user makes an error the display could be reverted with a triple-press. There are also buttons on the unit that could be optionally used.
The vacuum tweezer end of the design was based around easy-to-obtain syringe dispensing tips (nozzles). They come in different colour-coded diameters. The brown one is perfect for 0603 sized parts and larger, but the yellow one is great with 0402 and 0603 parts and even some ICs. The yellow one will therefore be used the most. The tips can be swapped out quickly because they are twist-on/twist-off.
Vacuum Pen Assembly
To make the vacuum pen a plastic bottle intended for eyeliner was used. I liked it because it has two flattened sides so that if the handle is placed down and then subsequently picked up, it can very quickly be held in the same exact orientation as before, which is critical when the dispensing tip has been bent at an angle in the desire for good ergonomics. The brush end was removed with pliers and the stem was found to be hollow so all that was required was to chop the end off for the tubing to pass through. Throughout the design of this project thin 4mm tubing was used (2.5mm inside diameter).
Another benefit of the plastic bottle method is that there can be lots of room so it would be quite easy in a future upgrade to build a version with a white LED inside it (and run thin wires inside the tubing) to supply illumination. It was considered a nice-to-have feature but not essential to this first vacuum pen prototype.
An alternative approach to creating a vacuum pen could be to 3D-print it based on any custom design.
As mentioned the nozzle part of the design was based around syringe dispensing tips with a standard fitting known as a Luer fit. A Luer connector adapter was used to connect the tubing (it is a push-fit into the tubing). One end takes the tubing, and the other end allows for a Luer-compatible dispensing tip to twist on.
The front panel has the display and a few buttons. They were mounted such that they could be easily unscrewed for adjustments/repositioning. A tubing coupler was also fitted so that the vacuum tweezer pen assembly can be disconnected easily.
The rear panel hosts the connection to the foot switch and the adjustment screw for the pressure regulator. During first use the regulator is adjusted until the components can be easily released without flying off, and then the locking screw is used to fix the setting.
When assembled the micro-USB interfaces from the FRDM board are accessible from the rear panel.
This was an off-the-shelf part; it internally contains just a microswitch. It was wired to a RJ45 network cable because they are low cost. I used pins 1 and 2, wired to the normally-open connections of the internal microswitch.
Building the Pneumatics
The three main components, the pump, the valve and Mr. Sneeze are screwed down to the base. For the pump, some thin foam pads were used and they helped reduce the sound. Rubber feet on the underside of the case helped too.
Building the circuit
As mentioned earlier the heart of the control system is the FRDM board. Using such an off-the-shelf board simplifies the construction. The remainder circuitry was fitted onto a small shield PCB and mounted with long spacers above Mr. Sneeze to pack everything into the case easily.
For a bit of noise immunity, the design uses an optocoupler as part of the interface circuitry to the foot switch. This isn’t mandatory but I had the part and and it was convenient to use in the design.
There is also a bit of flexibility in the circuit to cater for different voltage pumps and valves to allow experimentation. The valve uses less power than the pump so I chose to use a DC-DC converter (marked U3 in the schematic) to power the valve if the voltage to the pump is lower than the valve requires. If you know precisely which pump you wish to use then the DC-DC converter is not needed (it can be replaced with a wire link) and the valve can be chosen to match the voltage of the pump.
The vacuum tweezer system described here uses easy-to-obtain parts and the overall design is maintainable and adjustable because all major components are removable including the vacuum tweezer nozzle. The design has some features aimed to improve component pick-and-place speed.
It remains to be seen how reliable it is, but over intermittent tests it seems to work really well, I’ve been pleasantly surprised. I’m still practicing how to use it most effectively. I also have a quieter pump now so I’ll try to swap that in sometime too, although the existing one doesn't sound so bad currently; I may change my mind after a few hundred components placed : )
Next Steps / Ideas for Improvements
I plan to write additional software to help port bill of material (BoM) content from popular CAD packages (primarily just EAGLE for now) into the device.
If I was designing a new revision of this project, I would also add a small speaker that could be used to provide an audible alert under software control.
Another thing that could be done to improve the design would be to fit a small white LED to the end of the tweezers (there is room) and to use one of the LED wires as an earth connection touching the metal part of the nozzle to eliminate any potential risk of static build-up (I believe the risk is low since the flow rate is so low). For now, one could just periodically touch the metal nozzle to achieve this (a short beep as a reminder could be implemented).
If there is interest, I’ll create a parts list and PCB file for this project since it would be nice to collaborate with others on improvements to this project.