OK, I am playing around now. I've thought of a new MIDI gizmo to add in - a cross between a real-time sampler, sequencer and arpeggiator. I am also thinking it will be a MIDI master clock as well. I am using one of my boards from the Hakked! soldering iron, with ATMega328 processor and the flourescent display interface. I added a small daughter board to accept a MIDI interface (the five pin DIN style, not a USB one) and to wire the rotary encoder to.

Works so far ..

.. now off to write some code to update internal variables based on the encoder and the VFD display, and then I'll borrow some code from very clever people on t'Internet to make it all work.

New D-Link router in da house

MIDI setup

So the setup for the studio / office / playroom looks something like this now .. the space in the bottom RH corner is the new semi-modular synth. The rest is all there, and working!

Soldering pencil update #2

Aw, I like it!

GALs ..

So anyway, was messing about with the synth boards again. I have been experimenting with Mozzi, which is a sound synth library for Arduino. I am using ATMega328 on my VCLFO board, and I now have a multi-waveform LFO that has both voltage control of the waveform as well as the speed. It's pretty easy with Mozzi to drive a PWM output.

I put a filter on the original board design so the PWM output doesn't have much residual breakthrough of the sampling frequency (outside of my hearing anyway). However I wanted to play with a real DAC rather than the PWM thing, so I have been playing with some parts.

I have an AD5668 (8 channel 16-bit SPI DAC) which is in a tiny TSSOP package. It's too expensive though, to be used everywhere. I have decided to commit that board to the digital waveform generator for arbitrary waveforms in the audio range, rather than for control. I have also tried an MPC4922, a two-channel 12 bit DAC and I might use that for the MIDI to CV function.

When I took a Roland E10 apart a while ago (wasn't worth saving, except the keyboard mechanism) I found a PCM54 16-bit single channel DAC in there. I also have a drawer full of old components from a failed programme at work - where I used to work 20 years ago - including some LF398 sample and hold modules, so that is now being prototype boarded up as a daughter board for the LFO board.

Of course integrating a linear, 16-bit parallel input DAC to an ATMega with most of its pins used isn't easy, but in the search for the sample and holds, I came across a box of GAL16V8, 20V8 and 22V10. So I spent a happy morning digging out the old ALL-11P programmer, installing WACCESS, the All-11P software under WINE on Fireball, my linux machine here, installing Lattice ispLever Classic software for programming 'obsolete' devices like the GALs using Verilog, and starting to write bits of code to turn 22V10 into an SPI interface. The result is a bit of Frankenstein's Monster creation, but it should work.

I used to have a choice of programmers, but I sold two on eBay a while ago and I trashed the two I made. There was some sentimental value to having an old prototype board with a variety of ZIF sockets on it, and a step-up SMPS, along with some MC1408 DACs to set the voltage levels, but quite frankly, Hi-Lo and their All-11P is pretty, works .. 

Soldering pencil update

Ah, seem to have a bit of a problem with driving the Triac.

It seems that these logic triacs are a bit finicky when it comes to working sensibly. Even though I don't need the isolation of an opto-isolator with a diac, I have decided that is what I am going to do - stops all that tedious messing about. So the schematic now looks like this ..

.. and I should update this blog in a wee while with the results of this change.

.. ta da.

More soldering pencil fun ..

I was playing with the Hakked Weller soldering pencil, and decided that I really need to leave it alone and play on another prototyping board. You see, the thing is that there isn't a lot of room inside the Weller case for experimentation, so now it's working OK, I want to build another one for playing with.

I've got a few NEC VFD displays from a retail project I was working on, and I wanted to join that together with an Arduino base. So, I made up a nice PCB for it. Another nineteen dorrar down the drain.

Schematic ..

Just a regular 328P processor layout, with a 26-pin connector for the VFD; a filter with a gain of about 23 for the PTC element in the solder pencil, a pot to set the desired temperature and a driver for a logic triac. The whole thing runs off a 12-0-12 2A transformer, and I have included a switching PSU for the +5v, and a linear regulator for the -5v. Actually there is a -15v supply hiding near the triac driver, since the triac gate is driven by current flow, and the current flow needs to be between MT1 and the gate so the 'ground' for the driver needs to float about 15v below the voltage on the MT1 gate.

And now I am testing out the result .. so far, so good but the triac driver is causing some concerns. I mean it works, but it's messy. I have a real opto triac driver coming from China soon. I can't bring myself to paying UK prices, when they are for one off, twice the price of five coming from China. AND I have to add postage to the UK one.

More bits of the synth turn up ..

Work has unfortunately got in the way of play lately .. and then the weather has picked up nicely. It's a glorious sunny morning this Sunday, the London Marathon is on, the US Masters later and I spent a chunk of last night putting a schematic of a VCO into Eagle, and crafting a PCB.

So far I have received two boards from Itead Studios:

1) the multifunction VCLFO / ADSR board, and that is going well. I have a second style of DAC - an MPC4922 somewhere to try out - one that isn't £20 a go, but needs an external reference. The small daughter board idea for the DAC should serve its purpose well there.

2) is a dual VCA, 3 channel mixer, noise and sample and hold board. It's amazing what you can squeeze into 100 square centimeters. Quite pleased with it so far - only a couple of niggles on capacitor leg spacing. I must stop assuming that all caps have 2.5mm spaced legs (a mistake I haven't repeated on the VCO)

And so, before putting some clothes on, while the drill battery is charging I thought I would share the VCO design ..

First, the schematic. It's a well-known design based on Ken Smith's excellent VCO. I was going to borrow more of MFOS but this is a great starting point. It's probably not the only VCO I am going to cut a PCB for.

Again, I have stuck to the same format as the other two boards - slighly less than 100mm on each side, and Itead provide 10 (or 11 apparently) PCBs, double sided, thru-hole plated and silk screened in your choice of colour (as long as its green) for $19 US, including shipping. Can't compete with it, except on time, as it does take 2-3 weeks to turn up once the order is placed.

They're here ...!

Oh yummy.

Nice control of centering of drill holes, good evening tinning .. 10 boards for $19.90. Ridiculous.

This is the combined LFO/ADSR/DAC/MIDI2CV board based on a 328P from Atmel. The DAC is an AD5668 which is *tiny* .. You'll see the pads for the 328P mid way on the left, the AD5668 TSSOP header mid right, some bits and bobs - I decided to add x2 gain on the DAC outputs so you could do a full 10v swing if you feel the need.

I added a filter for the PWM outputs - top left. So I can play with that as well!

This is going to keep me busy!  
.. and we have life .. the DAC board caused a bit of fun since I changed my mind, and rather than just driving it from a few pins at random, I decided I could use the built in SPI functions of the Atmel processor - but then three pins were wired wrong, so it got fixed on the DAC board.

Getting one's Synth PCBs fabbed over in China ..

Hello,

Please kindly note that this order has not been complete.
We have been pushing factory to make boards asap and will try to send them out this week.

Sorry for the inconvenience caused.
Best regards,

Ah. Better stop running to the door every time someone brings a package then. I've only been waiting since 7th January for these boards .... *sigh*. I think I am going to try another Chinese fab site for the next boards. The first design is for a VCLFO/ADSR board, though I have had grand ideas about adding a bit of USB MIDI to it as well.

Onkyo phun

Sorting out a small issue with the lounge .. currently I have a fine Onkyo home cinema amp, but it only has two HDMI ports. I want to move the XBox out of my office to there, so grandson has somewhere to play his games. The two ports are used by the Virgin Media Tivo box, and an Apple TV2 (suitably hacked).

I found another Onkyo on eBay, with four HDMI inputs. The seller says it won't power on - and he's right, now I have bought it and it's been delivered, it doesn't - the red light remains obstinately on - showing 'standby' , and flashes a bit if you press the power button.

I had a quick squizz around with a multimeter, and found that the right front channel's output transistors are shorted all ways. This is not uncommon - if one goes, the other one goes shortly afterwards, unable to cope with more than 100 volts across it. So, to begin with I clipped them out, and tried again. Ta da, amp powers up and seems to respond to controls. The requisite Toshiba A1941 / C5198 ordered, and hopefully that will suffice.

Well it kind of does.

After much faff,

I got to the point of soldering in new devices. I checked around the locality for other transistors looking sick, and put the entire string of spaghetti back in the box, and powered up again. It stays on! Just check the bias current using the two test pins .. front left shows 9mV, and the new front right shows .. 0? Hmmm .. so both transistors in the output stage are off. Time for a schematic, and luckily Onkyo amps are well documented on t'internet, and all service manuals available. However, with the thing boxed up, it's well nigh impossible to probe anything.

They don't make anything to last any more ..

My precious Cambridge P60 amp died at the weekend, but fortunately it died in a way that was quite familiar. There was a sudden buzz coming from both speakers (so it's something common to both amps) .. a drone at 100Hz (so you know it is something connected to the power supply, after the rectifier), unaffected by volume controls, tone controls, balance (so it's not the pre-amp) .. signal still sounds through the hum (so it's not a catastrophic failure of the power transistors) .. ah. The smoothing capacitors have gone to Silicon Heaven.

Maplins are open on a Sunday, so Wifey and I popped in on our way somewhere else, and I bought myself some very nice looking 4700uF 63v electrolytics. All this before taking the amp apart; that is confidence for you.

Fortunately (and as Virgil wrote,  audentes fortuna iuvat -  fortune favours the reckless) when I did take it apart, I found some lovely 1970s vintage Plessey capacitors in there, and one had decided to pop. And leak. And make a huge mess. I can't really complain about their needing replacing after only 40 years.

Now if someone could tell me how to get the front panel re-printed .. mind you it is scarcely a stock P60; the power switch has failed so the line cord is wired direct, and I don't like DIN speaker sockets, so a quad of 4mm sockets is wired onto a small board dangling out the back. It still sounds magnificent, when driving a pair of Kef Concerto speakers of a similar vintage. The set-up sounds particularly good when playing electronic music.

Hakkedpad completion ..

Well my friends at aphroditestore have turned up with the goods, and have delivered their tiny Leonardo-compatible board. It's only about 2cm x 3.5cm, and has a nice 32U4 processor on it, a micro-USB and enough I/O for the hakkedpad. It's also small enough to hide inside the case, so no messy dangly bits ..

The designers of the Launchpad thoughtfully left some diamond-shaped pads to make it easy to solder to .. 

This made it easy to wire in the 'Pro Micro' Leonardo board .. I had to change the arduino code a bit, since the ports don't quite line up with the Leonardo, but the arcore extension loaded fine, Renoise spotted I had added a USB MIDI device ..

Time to stick the tiny board onto a couple of sponge pads, and wire in a short micro USB cable to the existing USB connector on the Launchpad - that way, the 5v serves the existing logic on the motherboard, and powers the micro board as well. It also means I can avoid trying to solder anything onto the micro USB connector which are too damnably small to see. Even with the surgical loupes.

And, ta-da! Hakkedpad in action ..

Hakkedpad !

I was bidding on that auction site again, and came across a Novation Launchpad. Owner said that he'd had a bit of an accident with power supplies and it had fried all of his stuff, so he was selling it dead.

Well I was delighted to win the auction for the Launchpad for not very many pennies, and when it arrived, all other projects went on hold for a while. I found that it was mostly alive except for the CPU, which had decided to depart this life. So what do you do?

Hack in an AVR instead!

The launchpad is an array of 80 buttons, and 160 LEDs, two per key. It's not touch sensitive, but all the same ... 

There are five 74HC164, and three 74HC165 inside. 164s (serial to parallel) drive the LEDs, and 165s read the state of the keys. The matrix is broken into four columns, each 40 LEDs (and 20 keys) long; you clock the LED states out, and clock the key states in at the same time.

So, I desoldered the CPU from the board, and wired in an Arduino Leonardo - since it was what I have lying around. The eventual target is a 32u4 micro board from my Chinese friends, when it comes.

The Leonardo happens also to have embedded USB, and there is a lovely lump of hack for the Arduino core called arcore, which happens to implement a simple USB MIDI interface. I found the programmers' guide to the Launchpad, so I could code up what was expected inbound and outbound, and a few days later (well, I have to do work as well, c'mon now) I have a reasonable emulation of a Launchpad. The refresh rate is only 53Hz, now I have implemented the three shades of green and red by PWM. I have a legitimate copy of Renoise for Linux (and Mac), and I wanted to use the Duplex mapper, which has lots of built in maps for Launchpad, but they do use all of the shades of colours.

It also has the benefit of a much faster MIDI implementation than the original - though maybe no faster than a Launchpad S, which uses a newer engine. Anyhow, I don't seem to need double buffering or anything like that to make the update performance OK .. it updates the entire button array in one refresh time, so as far as the user is concerned, it's instant. 

If you check the top RH corner, you'll see the dead CPU I removed from it ..

.. and now, I have improved the code a bit, so the refresh rate is now over 70Hz, and the colours still look good. I am going to call it done for the time being, and maybe revisit it in the future. For now, I am going to play with it ... !