One of the things blogs do that I find incredibly useful is taking a look inside consumer products. Things like Charles Guan‘s Beyond Unboxing, series, AvE’s BOLTR videos, and EEVBlog and Stefan Gotteswinter‘s teardowns provide insight into the inner workings of that device which is only possible by having the product in hand and taking it apart. It helps educate readers on how they work, and informs them how good quality they are, how easy they are to hack, and ultimately whether to buy them.
So here’s some interesting internal photos I’ve collected over the past couple years:
First up, some pneumatic components. This 3V1-06 is a cheap Chinese knockoff of the original, which seems to be made by Airtac. They’re available on ebay under $5, and are rated for 10 bar. G1/8″ (BSPP) threads on the Common and two Normally Closed ports, #10-32 threads on the Normally Open port on top, 1/4″ spade terminals for 12V positive, negative, and ground. It’s an unbalanced direct acting solenoid poppet valve, which means that the bore is very small (1.2mm) so that the solenoid can overcome the pressure, and therefore the flow rate is fairly low at 140 L/min at 7 bar.
Next, the 3V1-06’s competitor, the Clippard EV series. This is the 2-way EV-2-24 variant I found in a box at Noisebridge, but if I were to buy one, I’d get the 3-way EVO-3-12 ($30 from Clippard). This one uses a steel diaphragm shaped poppet with flex joints instead of a piston, and the solenoid is axial flux instead of radial flux, with a flat coil instead of a hollow one. Other than that it’s pretty similar to the other one, an unbalanced direct acting solenoid valve with 10-32 ports. If it were the 3-way variant, it would have a hole in the center of the coil for the Normally Open port for the top of the diaphragm to seal against when it moves up. Clippard lists the poppet at an amazingly tiny 0.007″ of travel. It’s a bit more compact than the 3V1-06, but the airflow is much lower, 17 L/min at 7 bar.
The last pneumatic component I have is another ebay knockoff of an Airtac part, the HSV-15 manual slide valve. They come in G1/4″ through G1/2″ variants, meaning they very conveniently screw onto the back of QEV valves like the BQE-02 and BQE-04 for a simple semi auto air gun:
Unfortunately my QEV is too tight to be disassembled, but the slide valve is quite simple. Only two parts, the body and the sleeve. The sleeve has two O-rings that seal against the outside of the body, which has a wall in the middle blocking the air, and a set of holes on either side of the wall. When the sleeve is forward, both sets of holes are between the O-rings, allowing air to flow between the body and the sleeve, and move between the two sides. When you pull the sleeve back, the front O-ring passes behind the front set of holes. The rear chamber is now sealed, and the front chamber is vented to the atmosphere. Simple, and effective. Just don’t leave the front O-ring sitting on top of the front set of holes or you’ll continuously vent air.
Here’s a great set of drawings from dewey-1 on Spudfiles for this slide valve, and the QE-02 QEV:
The other thing I have some decent photos of are brushless motors and controllers.
Here is a now-discontinued Hobbyking XK2860-B-2700KV, a 4 pole brushless inrunner. As you can see, it has an iron stator which the wires are wound around, pretty standard.
But I was very confused when I opened up this also-now-discontinued B28-57-15L 2 pole inrunner, to see no iron core at all. The copper windings are arranged in a caret (^) pattern, and are held in place with nothing but epoxy. This apparently allows for more copper mass, allowing more power to be put through the motor with less heat, but at the sacrifice of low-end efficiency and torque because there’s no iron core to direct the magnetic fields. They also operate more smoothly at very low speed, because there are no iron slots to create cogging. You can use this property to tell whether a motor is ironless or not without opening it up.
I really wish manufacturers would specify which type each inrunner is, but they do not. It is up to us hobbyists and reviewers to document which motors are which. Every motor in the XK series that I’ve used has had an iron core, with this B series motor being the only ironless motor I’ve encountered.
This motor is also a 2-pole motor, meaning for the rotor they can use a hollow cylinder of Neodymium, radially magnetized. For 4-pole motors, it’s impossible to make them from a single magnet, so they use 4 rectangular magnets attached to a square steel body, and held on with a carbon fiber wrap against the centrifugal force.
For speed controllers, the cheap and medium sized option is the Red Brick 70A. I have a number of the v2 iteration, first seen September 2015. The v3, which is almost the same with one pair of pins swapped, was first seen February 2017. Here’s a RB70A-Opto and RB60A-Opto side by side, and they appear to be absolutely identical, except for the label. They later seemed to drop the 60A from stock, which makes sense. Both have 18x Ubiq-Semi M3006D N-channel DPAK FETs, rated at 30V 5.5mOhm. The input capacitors are three Chongx 470uF 35V 105C, the linear regulator for the onboard logic is the venerable 7805, and the buck regulator on the non-opto versions is the almost as well known LM2596S. Contrary to the name, the “opto” versions, like most “opto” ESCs, don’t actually have an optocoupler on the input, they just lack the buck voltage regulator. They have a single board for both signal and power.
Most importantly, they both have an Atmega8 microcontroller and run the amazing Simonk firmware (another topic for a future post). However my pull request with the rb70a3 config file hasn’t been accepted yet, cough it’s been 3 months Simon, so you need to grab it from my copy of the repo. These also don’t come with a bootloader on them, so you’ll need the incredibly handy pogo pin tool and USBasp for the first time you program it, or just buy one with Simonk already loaded.
I can’t seem to find my pictures of them, but another solid speed controller option is the F-60A. Separate power and signal circuit boards means it takes up more space, but they use 18 of the better quality Ubiq M3016D, which has 4 mOhm Rds(on) instead of 5.5mOhm.
My favorite, however, is the ZTW Spider 60A. Though it is a double board design, it’s significantly smaller than the other two, because of its higher quality International Rectifier H8318 FETs, with a 3.1 mOhm Rds(on) rating. For some reason they only populate 18 of the FETs, even though the board has spots for 24. It’s also a little strange in that the auxiliary voltage output is 12V instead of 5V, but I just cut that off. Input capacitors are a pair of 35V 330uF Rubycons. Unfortunately they’re out of stock, but it appears that you can buy the boards with a different firmware under the Aerostar brand. Aerostar also makes an 80A version, which would appear to be the same board with all 24 FETs populated.
Well that’s it for now, I’ll do this again when I have some more interesting things to take apart.