A few weeks back I wrote up a blog about the Doublet Headphone Amplifier which lets two people listen to the same audio source with independent volume controls. I got so many emails from folks who were interested in building one that I decided to make another run of the printed circuit boards - and they've arrived!

If you want one of the PCBs, email me with your mailing address and I'll drop one in the mail to you at no charge Sorry, they're all spoken for.

I also had a chance to clean up the circuit diagram. You can download the schematic in pdf or flash format. The schematic is broken down into two parts, the power section and the amplifier section, as shown below in miniature.

The doublet uses a 9V battery along with two TLE2426 railsplitters to generate a virtual ground and supply rails at +/-4.5V. For power savings, only one side of the power supply and one of the op-amps is active if a single channel is on. A 11DQ03 diode keeps things from breaking if you accidentally put the battery in backwards. This diode has an extra-low forward voltage drop of 0.55V which is nice for battery based amplifier circuits. The dual power supply is based on Tangent's TLE2426 virtual ground design.

The amplifier stage starts at the input jack. The left and right audio channels are each split in two and fed into the volume potentiometers. Capacitors C3, C4, C5 and C6 act to protect the op-amp from any DC offset at the inputs. The op-amps are configured in a textbook non-inverting configuration with 14dB of gain. The amplifier circuit is based on Chu Moy's original Cmoy design.

Here is one set of parts to fill the board:

• Capacitors
• Signal capacitors 4x EPCOS B32529C5334J(0.33uF)
• Power supply capacitors: 4x Vishay 2222 150 65471(470uF)
• Resistors
• 4x Vishay RN55D1001F(1kΩ)
• 4x Vishay RN55D4021F(4kΩ)
• 2x Vishay RN55D1002F(10kΩ)
• 4x Vishay RN55D1003F(100kΩ)
• Railsplitters: 2x Texas Instruments TLE2426
• Diode: 1x International Rectifier 11DQ03
• Audio Jacks 3x Switchcraft 35RAPC4BH3
• Op-Amp Sockets 2x Areva DIP308-001BLF
• Op-Amps 2x Texas Instruments OPA2134PA
• Potentiometers 2x ALPS RK0971221Z05
• Knobs 2x Multicomp MC0616003
• Indicator LEDs 2x Avago HLMP-K150
• Battery Strap 1x SPC 8459-0674

Keep in mind that the parts I've listed here are only suggestions. Though I've listed nice precision resistors for the audiophiles in the crowd, if you have some through hole resistors already, they will work fine (especially the LED resistors which don't need high precision).

There are many nice audio op-amps which will work in this design. In fact, the doublet amplifier can serve as a nice test bed for head to head op-amp listening comparisons.

Finding the ALPS RK0971221Z05 potentiometer in small quantities at a good price can be tricky. Tangent sells them in single pieces for $3.25 each which is reasonable.

Please send feedback on the circuit and the board layout as there is definitely room for improvement in both. The original Eagle files can be downloaded for those of you who are interested in moving the design forward. A few ideas are:

• Miniaturizing the board with SMD components.
• Adapting the power supply design to run off of AA, AAA or some lithium polymer battery.
• Adding a trickle charger for use with rechargeable batteries.
• Clean up the Eagle schematic (not the one shown) with better drawings for the potentiometers.

Assembling the doublet into an Altoids tin will have to wait for another post. Just remember to insulate the bottom of the PCB from the Altoids tin!

We just had some t-shirts printed with the Octopart logo on the front and the gear on the back. They are really comfortable and we try each one of them on before giving them away. If you want a (free) t-shirt just email me and I'll send you one. Check for prices and availability by searching for "octopart t-shirt".

working on cool hardware? submit blogs to contact@octopart.com

Have you ever seen a couple sharing a single pair of ipod buds, each listening with only one ear? That was the problem I wanted to solve when I started building dual channel headphone amplifiers while I was back in grad school. After a long day of soldering krytrons and high voltage capacitors in the plasma physics lab I would come home and...solder some more.

There were two design requirements I started the project with:

1) Each listener needed an independent amplifier channel.

2) The whole thing had to fit in an Altoids tin.

It works like this: plug your ipod or DVD player into the jack on the left side of the case with a male-male 1/8" audio jack. Headphones go into either of the two output jacks in the front of the case. Each knob acts as a combined on-off switch and volume control.

The design is based on a non-inverting op-amp circuit inspired by Chu Moy's original Cmoy headphone amp. The op-amps are Burr-Brown OPA2134PA which sound GREAT. You might wonder, "Why not just use a headphone splitter?" Well, that halves the power delivered to each set of headphones and sounds lousy. Also, if the listeners prefer different volumes you're stuck. This amp can drive big fancy headphones REALLY loud with very little distortion.

Back when I started this project I thought I would get a bunch of PCBs printed and start a little business selling the built amplifiers online. I sold about 15 before I had enough of the solder fumes and got tired of eating altoids and called it quits. I still have the PCB schematics and few unused PCBs. I'm rebranding it as the Octopart Doublet Amplifier. I'll go into more depth about the design and part list in my next blog post. If anyone is interested in the leftover PCBs, let me know. When you buy the parts to stuff them, just let the distributors know that Octopart sent you.

Part 2: Schematic, Parts List and PCB

Like many particle physics experiments, ATLAS is a collection of several concentric cylinders, each cylinder being a different detector technology. At the center of all the cylinders two very high energy beams of protons are brought into collision; it's then the job of the different detectors to give some clue as to what just happened by measuring the trajectories of the particles that were produced. The inner cylinders can make very precise measurements of where a particle crosses it, but are also very expensive to build per unit area. The outer cylinders are cheaper per unit area to build, but not as precise. The pixel detector - what I work on - is the inner most layer. It's taken many hundreds of man years to get to where we are now, and all for an active area of less than 1 1/2 square meters. But the upshot is that it will be able to measure track vertices - meaning where the different particles were produced - with an accuracy of a hundred microns, or about the width of a human hair. And it can do that for thousands of tracks every 40 million times a second, all the while operating in an intense radiation field for years.

Here's how it works: each module (about size of a booklet of postage stamps; see the diagram) starts from a 15mm by 60mm piece of very high purity silicon. The silicon is divided into a grid of 48,000 little rectangles, each rectangle containing, among other things, a p/n junction diode. Then 16 integrated circuits are laid down underneath the sensor silicon. These ICs amplify and digitize the charge pulses that get created when a charged particle passes through one of the sensor diodes. Then, a 2nd type of integrated circuit - the MCC, for module control chip - collects all of the information coming from the 16 front end chips and communicates the results to 300 or so custom read-out drivers in the 'outside world'; meaning outside of the detector volume of the all the other subsystems.

The final detector is made up of almost 2000 modules, arranged on a graphite frame. Then the tricky part of snaking the low voltage cables, high voltage cables and cooling fluid pipes through the outer cylinders begins! And of course, this is to say nothing of how one actually makes sense of the data once it's on disk.

So where are we now? The final module design has been finalized for several years and the manufacturing and testing of the modules has been completed for almost a year. In that time, the modules have been loaded onto their mechanical frame and permanently connected to the first series of patch panels that provide service support for the system. In December, an attempt to run 10% of the system simultaneously was successfully completed. Final tests of the connections to the patch panels mentioned above were just completed about 1 week ago; the current plan is to lower and insert the detector inside the other layers before the end of this month. After that, it's still a lengthy road, but we hope to have the entire detector cabled and ready to read-out (still uncalibrated) by the end of year.

working on cool hardware? submit blogs to contact@octopart.com