Look Ma, No Wires!

I’m blogging after aeons! Been really busy; work is such :-/ It’s not like I haven’t been tinkering at all, it’s just that I’ve been tinkering less, and haven’t had the time to document it.

A few weeks ago I decided that I needed a Bluetooth-to-audio adapter for my home (which has a decent but dated Sony home theatre) and car (which has an old stereo system). So this vacation I decided to make one.

Logitech’s Bluetooth Speaker Adapter has got a 3.5mm socket as well as RCA outputs

I was looking for an end result similar to Logitech’s Bluetooth Speaker Adapter, available on Amazon for INR1,100 (~USD16) as on Sep, ‘15. There are predominantly two types of Bluetooth audio adapters available online – one, which outputs analog audio through a 3.5mm jack, and connects to your sound system’s AUX/RCA input. The second, slightly more interesting version (to my mind, at least), plugs in to the USB port of more modern mp3-capable sound systems. The adapter ‘appears’ as a regular USB flash drive with an MP3 file on it. The audio streamed over bluetooth is constantly written and buffered to the ‘mp3 file’. Pretty ingenious!

As with most of my projects, I turned to the internet to do a little pre-hacking research. I took a look at a few DIY hacks – google “hack car stereo bluetooth”, and you’ll get links to several people who’ve given new life to the old cassette deck in their dad’s rickety jalopy.

Pretty straightforward: http://www.instructables.com/id/Add-bluetooth-to-your-car-stereo/?ALLSTEPS

A bluetooth speaker modified to do the business: http://www.instructables.com/id/Bluetooth-Speaker-Hack-Home-Theater-Streaming/?ALLSTEPS

And for those people still rocking cassette tapes in 2015 (my mom, for example), this guy made a ‘bluetooth cassette’: http://car-mods.wonderhowto.com/how-to/hack-your-cars-cassette-deck-into-wireless-bluetooth-music-player-0139843/

I was initially thinking of modifying the BT earpiece I use with my phone, however that would limit me to a mono output only (there are earpieces which have stereo outputs, so these too, may be good contenders for a hack). I had an old Corseca/Byte brand on-ear headphones lying around. The ear pads were crumbling, and the sound quality was absolute rubbish. But it was still working, and I was pretty certain that the poor audio was only due to the low cost sound drivers. So naturally, I took a screwdriver to the headphones, hehe.

Not sure if the brand is ‘Corseca’ or ‘Byte’

Disassembly was rather painless – just a few screws and plastic retention clips. The electronics were crammed into the plastic cover over one ear, and the battery was stuck with double sided tape in the other cover. The cover which had the electronics also had control buttons on the other side – FF, REW, Vol Up, Vol Down, Call Answer.

Teardown time!

No surprises about the main chip being from CSR (now acquired by Qualcomm). CSR is a major in BT consumer electronics ICs, and you’d find CSR chips in everything from keyboards, to headsets, to smart microwaves. The main chip is a CSR57F68 A2DP streaming audio chip. The chip is fully integrated, and has two onboard differential audio outputs, button handling capabilities (debounce, etc), GPIOs, and of course, the BT radio itself.

 
The electronics were on a double-sided board – one side with the chips and support circuitry (Left), and the other side with buttons and status LEDs (Right)

The TSSOP chip is an L24C32, 32kbit I2C EEPROM, which probably stores config data, descriptor strings, etc.

I then desoldered the sound driver and battery wires.

  
Desoldering the mini-USB charging socket, and replaced it with a micro-USB connector for convenience. The micro-USB’s pin pitch didn’t fit the board perfectly, and the connector itself had a different footprint, so I had to use mod-wires

I also removed the mini-USB charging connector, and soldered-on a micro-USB connector (side rant: just like everyone else, I get totally frustrated pulling out ten different cables to charge ten different devices. With most manufacturers now standardizing around the micro-USB connector, things have become a bit easier, but Apple still stubbornly pushes its weird proprietary connectors, each one different from the previous).

 
The covers, before (Left) and after (Right) modification

My idea was to use the two covers – one from each ear – to form a sort of ‘case’ for the final product. I used my Dremel to trim away the plastic standoffs for the screws on the plastic covers. I then fashioned some new screw inserts out of acrylic. The inserts were hot glued into place. Small holes were drilled into the inserts and the bottom cover to allow for the screws.

 
The acrylic inserts/standoffs for screws, and the 3.5mm audio connector

I had a 3.5mm audio socket (probably scavenged from an old motherboard or CD drive) in my junk box, which I also hot glued in place. Wires were then used to connect the audio outputs from the board to the socket. I took care to connect the left and right channels to the correct legs on the socket. The negative output of each differential pair was soldered to the ground pin of the socket.

Connecting the wrong output to ground would result in an out-of-phase output from the speakers. While this may not be such a big deal with headphones, it would result in sound ‘cancelling’ itself out (primarily poor bass response) if I used the BT adapter with speakers. The AudioCheck website has cool online audio files that help you determine if your speakers are connected correctly.

 
 
The finished project – screws on the sides, and the PINK (!!) audio socket sticking out – not the prettiest

A small slot had to be cut in the bottom cover to accommodate the audio jack. The covers are held together by three screws. The final device is not the prettiest, but it’s functional and works really well.

   
Charging through the hacked-on micro-USB connector

I Got Spinners Dawg!

I am an engineer. By profession, by hobby and therefore, by passion. And being an engineer requires me to browse through PDFs several pages long, or plow through code confusing enough to make your eyes pop out. I may do this just to look cool, but all the same, I have to do it. Using a mouse scroll wheel with its annoying ‘click’ makes my index finger sore pretty quick. So then I switch to my middle finger. And pretty soon I find that I can’t give someone the bird.

Griffin’s $40 PowerMate

Now the simple solution would be to buy myself a jog wheel / spinner. Give the jog wheel a slight twist and its momentum would keep it gliding over pages and pages of the boring PDF. But I’m an engineer. So I decided to build myself one (obviously the phrase “Why buy a cow when milk is so cheap” does not apply to engineers).

The inspiration for the build was not really ‘direct’. I never really got down to building one until I watched one of Dave’s excellent teardown videos. I was actually googling how VCR heads have a helical scan and then opened up a myriad other links, this instructable being one of them.

Osgeld has done a pretty fine job with his jog wheel

I had torn down a VCR about a year ago and mined it for parts, so I already had the basic stuff at hand. I also had an old USB mouse picked up from a dorm trash can. The tactile buttons had gone bust, but the chip itself was pretty useable.

 
The crusty old VCR head. Oxidized and rusted

Some more googling led me to these tutorials:
http://www.instructables.com/id/Computer-scroll-wheel-bearings-version/step5/Finished-future-improvements/
http://www.instructables.com/id/Desktop-Scroll-Wheel-and-Volume-Control/

Disassembly of the head was was relatively easy. I’m just going to explain it through pictures.

 
A brushless motor drives the head assembly

 
The magnetic pickups are clearly visible in the image on the right

 
The rotary transofrmer

[Digression - the pickups feed the signal to the coils of a rotary transformer. The transformer helps couple a signal across two rotating elements without the use of mechanical elements like slip rings. I’ve never seen a rotary transformer before, so this was the most interesting part of the teardown. The transformer ‘core’ was glued down. A blast of hot air from my heat gun helped dislodge it.]

My mouse chip was an all-in-one solution with the USB comms, the optical sensor, and the click/scroll encoders integrated onto one chip. This is good since I was able to make my circuitry compact. Some mice (mouses?) distribute USB and sensor duties across two separate chips. The particular chip that I used was the A1198. Googling for the chip’s pinout or datasheet proved useless. So I used a multimeter to reverse engineer the PCB. An MSPaint pinout is below; hopefully someone else will find it useful.

Vreg gives out approx 3.3v All V should be connected to Vreg. Resistors pulldown the quadrature inputs Q0, Q1 to Q_pulldwn. If a mechanical encoder is used, then the common pin of the quadrature encoder should go to Vusb. Xtal is a 12MHz ceramic resonator.

The pin spacing for the chip was a bit odd. Luckily the pins weren’t staggered like you find on some sensors. I had some 2mm pitch veroboard, which, after wiggling the pins a tad, worked just fine. I used the ceramic oscillator from the original mouse, and sprinkled a few SMD capacitors which I had scavenged from old electronics.

 
Left: IC pins splayed out
Right: I drilled a small hole for the aperture of the optical sensor to sit in

The original mouse used a mechanical quadrature encoder for the scroll wheel. I converted this into an optical encoder by using some cheap IR LEDs and phototransistors. The resistor-phototransistor-combo’s output swung from 0.1v to 3.3v. I found that I didn’t even need to use a comparator on the sensor output since the signal was fairly ‘clean’. The encoder wheel was designed and printed using shimniok’s Wheel Encoder Generator.

  
Left: The first sensor board I made used one IR LED and two phototransistors ground flat. This didn’t work well at all – very poor resolving power.
Centre: On the second board I used SG-105 photoreflective sensors. They’re tiny and low profile compared to my earlier sensors.
Right: The back of the SG-105 board. SMD resistors used to bias the opto-elements; 220ohm on the LEDs, 470kohm from the phototransistor collector to Vusb.

For the base I used a Canon lens holder given to me by a photographer friend. A little bit of drilling here and there made it an excellent base.

 
A Cannon lens cover was used as the base.

 
Left: The completely soldered ‘mouse’ board was stuck with some double sided foam tape in the Canon lens holder.
Right: The sensor board was fixed using a screw and a spacer. The USB cable was secured with a dot of hot glue.

 
Testing the sensors with the help of my third hand and a DSO. The signal is typical of a quadrature encoder and is quite clean, as you can see in the trace.

Overall I’m really happy with the way this jog/spinner wheel has turned out. No more sore fingers, and zipping over pages in a PDF, or lines of code, or through frames of a movie has become easy and super fun! Infact I’ve used it extensively while editing this very document :-) One possible addition at a later date would be to add a push button at the bottom of the shaft so that the whole spinner can be ‘pushed-to-click’.

Edit: I’ve uploaded a video to Youtube

Jog Wheel