A very simple, but useful PCB. A power supply for the Muffsy Relay Input Selector (or anything else that can be powered by 5V DC/1A):

The MeanWell IRM-05-5 takes an AC input between 85 and 265 volts, or a DC input between 120 and 370 volts and converts it to 5 volts DC at a maximum of 1 ampere.

This open source PCB project created in Eagle lets you add input and output cabling, as well as providing mounting holes for the PSU. Download it here:

Github repository for the Eagle project files

The project contains the schematic, the board, and the gerber files (and also the .cam file to generate your own gerbers).

I'd really like a stereo input selector with relays, but those things are hard to find! (Apart from some of dubious quality on the *Bay)

Better make my own then. This one's got a custom footprint for an ESP32 devkit module, but it can also be controlled with a rotary switch.

I decided to use five Panasonic TQ-2 relays. The ESP32 module and relays are powered separately, power ground and signal ground are separated to avoid injecting any clicks, pops or noise into the audio channels.

The whole project is open source, free to use as you wish. Eagle project files, gerbers, the Eagle library for the ESP32 module and BoM are all available on the hackaday.io project page.

And, wouldn't you know, Hackaday presented this little side project on their blog!

I've been getting my new hobby room ready, and I knew I needed some kind of stereo system in there. Preferably one that didn't annoy others in the house, so I ended up ordering PCBs from Nwavguy's gerber files. Yup, I'm gonna use headphones. :)

The components that I didn't have at home were ordered from DigiKey, and I ended up doing quite a few component substitutions. The cabinet + front and back panels were ordered from Headnhifi.

I did spend a lot of time identifying the right components. There are a lot of different resistor values, and they have to be cross referenced with the BoM. Of course, it didn't help that I have a box with all E24 resistor values that aren't sorted in any way.

Once the sorting of components was done, it didn't actually take long to solder the thing. I'm really happy I got the recommended enclosure and panels. This thing looks really great.

Initial impressions are really good. My main problem (until the Sennheiser HD 600 headphones arrive) is that I only have an old pair of Koss PortaPro available. The foam around the earpieces is really old, and creates a cloud of tiny black dust...

Here's the finished Inverse RIAA. I haven't done extensive testing yet, but it works as promised AND it measures exactly the same on both channels. I'm really happy with the result.

Get the gerbers for this project here: invRIAA.zip

If it's going to be called accurate, you'd better get two identical channels. Here's how the invRIAA fares:

In order to construct this board, I built the vacuum pickup tool:

The lead free paste was applied using a stencil, which is extremely convenient. To do the actual soldering, I used this reflow heater bought cheap on eBay (Nope, I don't have space for a reflow oven...). The only negative with this heater is that the LCD back light is more on the front, so you need to tilt it to see what it says.

You really don't want to go with a normal heat gun for this. They push about 600 liters of air per minute, which will blow away all the components. This one does about 30 liters/minute, and it's adjustable both on the tool itself and by replacing the nozzles.

I even bought an Atmega based transistor tester as a kit, and it contained three SMD components. Here's how that turned out:

This is my first venture into SMD components, so I'm going to have some fun with tools, solder paste, stencils and hot air. :)

Get the gerbers for this side project here: invRIAA.zip

An inverse RIAA circuit has been on my wish list for a while, as it makes testing frequency response so much easier. But how, you might ask?

Well, the first problem you'll encounter when trying to test a phono stage is the fact that few signal generators produce a signal level low enough. If they do, they're often not very accurate at those levels.

The second, and most pressing problem, is that the phono stage applies equalization to your signal. Input a 50 Hz signal at 5 mV, and the output will be almost 5 mV. Input a 20 kHz signal at 5 mV, and the output will barely be measurable. You won't be able to see the frequency response from the readings, without doing a whole lot of conversions (and taking DMM/scope tolerances/misreadings into consideration).

What's needed is something that takes your signal and turns it into what you'd find on a record. That's your inverse RIAA equalizer.

Feeding your signal through an inverse RIAA equalizer, and then through a phono stage, will create a flat output at all frequencies. If this inverse RIAA equalizer is sufficiently accurate, it can be used to measure the accuracy of your phono stage. It's got the added benefit of bringing a higher input signal down to cartridge level.

I got Hagerman's inverse RIAA filter, and while it's a nice little device, I wanted to get one with better accuracy. That's achieved with a lot more components to even out their tolerances.

Not one of my designs, this is the Accurate Inverse RIAA from HIFISonix. I decided to make a stereo version. As I already have lots of screw terminals and DIP switches, I decided to use them too.

There's a lot of components in there, which is why it's SMD (the board size is approx. 8 x 5 cm). All SMD components are 1206 size, so it should be manageable to solder them in. I thought I'd use a hot air gun for the soldering.

I still have to get the boards manufactured, and I need components for it. Not sure how long it'll take, but I'll definitely let you know when it's done.

UPDATE (2017-03-09):

All components have arrived, and a few boards have been ordered. I also sprung for a cheap hot air soldering "pen" on eBay, and a stencil for the SMD-components on the board.

I managed to get the board size down to 84 mm (83.98 mm to be exact) x 51 mm. 84 mm width is what's needed for the board to fit into one of those B0905 enclosures. :)

UPDATE (2017-03-10):

Here's the final layout. The caps have all been changed from 1206 to 0805, since I only found them in that size at a decent tolerance of 2%.

The attenuation for MM is -44 dB, and for MC it's -68 dB.

I got a vacuum "tweezer" that was completely useless for these small components, so I ordered a fish tank air pump to make a real vacuum pickup tool based on this video:

UPDATE (2017-03-16):

The boards and stencil are here:

Here's another project with freely available Eagle project files.

There was a strange little thing that appeared in Danish magazine Ny Elektronik (New Electronics) in 1989. A very simple preamplifier that used two L63 tubes in Class A with no feedback and it operated pretty much badly out of spec. (It was supposed to though, that was the whole angle of the article.) It was called The Bastard, and gathered quite a following (it got the name because it was a hybrid. The phono stage used transistors for better SNR), and somebody suggested I should try it.

Once I had some suitable tubes (6S2S, NOS, shipped from Smolensk in Russia), I knew I could go ahead with the project. I decided to skip the phono stage, since I already have some of those, and did only the line stage. ;)

A few iterations of the PCB drawings later, and it was time to order some boards as well.

I called mine The BSTRD:

The original Bastard, while probably sounding very good, did not perform that well (it had up to 6% THD+N). Some other Danish guys took a second look at it, and made a couple of changes. First, they bumped the operational voltage from 37V to 80V, to get it into spec. Then they tamed the gain and improved the performance by adding some feedback.

The result was this circuit which has a THD+N of 0.185% and a gain of 2.3:

The unregulated PSU in the original article wouldn't work for this revised version, so I made my own. It's actually two power supplies, both regulated, that delivers 78V/0.7A and 6.3V/1.5A. The 6.3V is for the filament heater, so it doesn't need a lot of filtering as long as it can deliver the required current. The 78V features a voltage quadrupler and uses an RC filter for better smoothing.

Measuring the PSU with load shows 0.00 mV AC on both DC outputs. I'm satisfied with that, although I haven't checked out the noise on the scope.

The BSTRD has been built and tested, and it sounds pretty darn sweet.

Note that because of the voltage quadrupler on the 78V side, you ABSOLUTELY NEED TO USE TWO TRANSFORMERS.

Since the filament heater draws 300 mA of current (a total of 600 mA for two tubes), the LT1086 and its heat sink get really hot. I would recommend using 6 to 9V AC, anything higher than that would probably require moving the regulator off the board and fit a larger heat sink.

All project files can be found here: https://hackaday.io/project/16944-the-muffsy-bstrd-valve-preamp

The Bill of Materials is available here: https://hackaday.io/project/16944-the-muffsy-bstrd-valve-preamp/log/49757-components-bill-of-materials

Here's a little side project, and it comes with Eagle Project files so you can order, etch and modify it anyway you like. All project files are available on this Hackaday.io project page: https://hackaday.io/project/18624-muffsy-constant-current-led-tester

LEDs seem to accumulate. They're scattered around, some don't work, some are blown, most of them are transparent so there's no way of knowing which color they are.

In addition to just testing my LEDs, I'd also like to adjust the brightness and find the forward voltage drop. All using something that won't break any more LEDs, no matter what I do.

I've been working with the LM317 (and others) voltage regulator, and noticed that it can be used as a constant current source. LEDs are driven by current, not voltage, so this is a great advantage in several ways:

• I don't have to worry about how high the input voltage is
• I don't have to take the forward voltage drop is
• Limiting the current ensures that the LEDs are not destroyed if the polarity is wrong
• The current and voltage can be measured and used in any circuit

Before I get to the circuit, here's how to use the LED-tester to decide the resistor value needed for a certain brightness:

• Place an amperemeter in series with the LED in the LED tester. This will give you the current. Let's say it's 10 mA.
• Place a voltmeter in parallel with the LED in the LED tester. This will give you the forward voltage drop. Let's say it's 1.8V
• You want to place this LED, at this brightness in a circuit with 15V.
• The voltage is your circuit voltage minus the forward voltage drop: 15V - 1.8V = 13.2V
• You can now find the resistor value using Ohm's law, R = U / I: R = 13.2V / 0.01 A = 1.320 ohms
• A 1k3 ohm resistor will be suitable for this LED when it's powered by 15V.

The circuit works as follows:

• Power it with a 9V battery
• The LM317 will maintain 1.25V between the Output and the Adjust pins
• Place resistors between these two pins, and you can regulate the current
• Use one fixed resistor to set the maximum current
• Use a potentiometer to adjust the current from maximum to minimum, LOG or LIN
• Use a suitable connector for the LED under test, 5/5.08 mm screw terminals will fit directly on the board
• Place a voltmeter in parallel with the LED under test, this will give the forward voltage drop
• Place an amperemeter in series with the LED under test, this will give the current

Oh, there's always one more thing. The LM317 will stop regulating under 2-3 mA. Since high intensity LEDs often can require less than 1 mA to reach the desired brightness, we'd like to go down to below those 2-3 mA. 

To make sure there's always enough current drawn from the voltage regulator, place a power-on LED from its output to GND. In this circuit, it draws approx 4 mA.

Here's the PCB layout. Using a LOG pot will give better response within the low current region, a LIN pot gives better response at higher current.

The dimensions of the board above are in millimeters. The Eagle project files are available here: https://hackaday.io/project/18624-muffsy-constant-current-led-tester