Grant Muller

Coffee Roasting: Build or Buy?

For the last five years I’ve been roasting coffee in this:

I got 5 years out of my old machine, the Frankenstein Turbo Crazy, home grown of course

My Frankenstein Turbo Crazy.

But alas, after five years of operating “outside of normal temperature ranges”, this happened:

Before this happened. Half a decade of operating "outside of normal temperatures" melted the convection heater...

It seems that convection ovens weren’t meant to have their temperature limiters removed and run for hours on end. The handle, which controls the reed switch to turn the heating unit off and on, melted right off the base.

I knew this would happen someday, and I had planned to build another machine with a leaf blower and a commercial heating element, with robot arms and a positron brain. Alas, the unhappy melting happened right in the middle of rebuilding a fence, writing a bunch of software and getting through several releases at the office. I was posed with the classic question all technology households are posed with: build or buy?

A guy has got to have his home-roasted morning brew. I made a rash decision. I ordered one of these:

The Muller House gets a new coffee roaster

A GeneCafe from Sweet Maria’s.

After a week of waiting the GeneCafe finally arrived so after fulfilling my daily obligations I came home and took it for a test run on some Rwanda NKanka Kinyaga. In Benjamin Franklin style, here is how five years of DIY roaster technology stacks up to the commercial grade:

GeneCafe Frankenstein
  • Configurable Temperature
  • Configurable Time
  • Integrated cooling unit
  • Excellent chaff removal
  • Single unit
  • Easy to clean (so far)
  • Very even roast
  • Requires less monitoring
  • No coasting
  • Extremely high temperatures
  • Infinitely configurable
  • Very low cost
  • Long coast time
  • Small batch size
  • Long roast/cool times
  • Temperature
  • Very high coast
  • No integrated cooling unit; long cooling times
  • Requires constant monitoring
  • Occasionally uneven roasts
  • Poor chaff removal

The verdict. After the initial two test runs I’ll give the win to the GeneCafe. If only for the sweet analoguish knobs. We’ll see how it holds up after half a decade.

What’s the plan for my beleaguered Frankenstein Turbo Crazy? Well, I’ve always wanted an outdoor water heater…

Update: The coasting issue I complained about previously is no longer a problem. Seems I just needed to read the instructions…

Casio MG-510 Midi Guitar

Back when I documented the repair of the Casio PG-380 MIDI Guitar, I had no idea that this post was going to dominate the traffic patterns to my little home on the web. Fully 1/3 of all visitors to this site come to that post, asking questions, posting comments, and requesting repairs. One request I’ve gotten over and over is a repair on the Casio MG-510.

The Casio MG-510 is like the little brother of the Casio PG-380. The base functionality is similar, but the 510 lacks some of the extra features that the PG-380 offers. The 510 has no space for an expansion slot, and no internal synthesizer, which for most software synth users is just fine. The biggest differences you’ll notice between the 380 and the 510 are hammer on sensing and the ability to perform pitch bends. The 510 is strictly chromatic; when you bend it assumes the same pitch until you bend far enough to change notes, in which case a note off and note on message are sent and interpreted. The 380 will perform a pitch bend at even the slightest pull of the string.

The 510 and 380 share one major flaw though: the electrolytic capacitors used for the pitch envelopes. These heinous little surface mount caps tend to leak over the years, especially on the 510, leading to corrosion and in most cases total failure of the MIDI capabilities in the guitar.

I finally got around to repairing one of these guitars, and the process is so similar to the PG-380 that it would be a shame not to document it. If you’re new to this you should probably refer to the post on the PG-380 before getting started.

You will need:

  • 6 x 1 uF non-polarized electrolytic capacitors
  • 4 x 10 uF polarized electrolytic capacitors
  • 1 x 22 uF polarized electrolytic capacitor
  • 1 x 4.7 uF polarized electrolytic capacitor
  • 1 x 33 uF polarized electrolytic capacitors
  • Anything you need to unsolder old capacitors and solder on new ones

First, crack open the back and take a look at the boards:

(1 of 5)

You’ll see two double-stacked and plugged into three header cables.

Take both boards out (unlike the PG-380 you have to operate on both):

(2 of 5)


Take a look at the capacitors on both boards below:

(3 of 5)

Top of PCB 1 – C9, C18, C33: 1uF non-polarized electrolytic


(4 of 5)

Bottom of PCB 1 C42, C52, C63: 1uF non-polarized electrolytic


(5 of 5)

Top of PCB 2

  • C4, C22, C29, C12: 10 uF polarized electrolytic
  • C30: 22 uF polarized electrolytic
  • C31: 4.7 uF polarized electrolytic
  • C48: 33 uF polarized electrolytic

Basically, for both boards, replace the capacitors with the caps above using capacitors of identical value. It shouldn’t matter if you use polarized caps for the entire repair, since the frequencies are not high enough to affect response times, but use non-polarized where needed above if possible.

You will find that you have 2 "extra" caps (seems like 2 for each string plus 2). I know that one capacitor is used for CPU reset (C30), but I’m not entirely sure what the last one is for. I replaced it anyway.

Some notes:

  • The traces on the top board are very small. You might find yourself pulling them while unsoldering the old caps. Not to worry, there are plenty of places to solder the new caps.
  • Corrosion makes for crappy contacts. If you find that your caps have corroded, particularly on the lower board, you will need to sand the corrosion down with steel wool or other light abrasive until you can expose some copper to solder to. On the guitar I repaired the corrosion was severe, and I spent a lot of time scraping out leaky capacitor guts.

That’s really all there is to it. Plug the boards back into their headers, screw them back into the guitar, and adjust the trim pots as needed to calibrate the guitar again.

Ring Modulator: Prototype to Final Build in One Ridiculous Step

RingModulator-16 If you’ve been to this site before you know that I’ve been building a ring modulator for Bill Graham to use with his Rhodes for the better part of a year. If I had enough time to do it right it’d take me less than a week, it’d be stable, and I wouldn’t be worried that a 3 foot tumble would render it useless. But alas, Bill had some gigs coming up, and I wanted to put this project to rest in the interest of getting some of my time back, so I resorted to some rather ridiculous means to complete it. What follows is not to be emulated or admired, merely witnessed.

RingModulator-8 Having assembled the circuit on a breadboard in the previous post, I had a working prototype that if you pinched the alligator clips just right, would produce the effect I was looking for. I had a day or so to get this thing boxed up and stable enough to work at some gigs, not nearly enough to design the PCB, etch, reassemble, test and ship. What’s a time-starved designer to do? Box up the prototype breadboard and all, right into the oversized power supply box, wiring up the controls right to the front panel.

RingModulator-10 I started the process by first removing the breadboard strips from the substrate they were attached to. The entire breadboard wouldn’t fit assembled into the case so I simply transferred the screw hole locations from the substrate to the new case so I could attach the strips to the inside. After drilling and attaching the strips it looked something like the picture to the left.

RingModulator-11 Moving the controls from the board to the panel was easy enough. As you can see I had to settle for using the PCB mount pots I ordered expecting to mount this on a board, rather than the panel mount ones which are much easier to solder wire to. Live and learn I guess. I drilled holes for the 4 controls knobs (Frequency, Depth, Pre and Post Gain) in a row on the panel, along with two more for the input and output. I tried to keep everything on one detachable panel so that if I ever did get around to designing and etching a board I could replace the breadboard strips with it and not have to make any other modifications, I did, after all, order two of everything. I do think ahead on occasion.

RingModulator-13 With the wiring soldered to the pots I jammed the other ends of jumper wire where the pots used to reside on the board and ran a final test. Everything appeared to be working, so I wired up the power directly to the terminal strips and packaged it all together. There, done. For now.

RingModulator-17 Not that it sounds any different than the last audio samples I posted, here are some new samples from the last test run before Bill came and fetched it for a gig:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

So I’m calling this project done. I may come back to it and do it right some day, but that won’t be anytime soon…

Ring Modulator: Prototype Take Two

RingModulator-3 About a year ago I posted an article about a Ring Modulator prototype I had created using 2 audio transformers and some matched diodes. The design was beautifully simple, and I may return to it someday, but it had a number of shortcomings. The circuit I started with would have needed a preamp for my input signal and a separate oscillator. In addition, I would have probably needed some means of amplifying the output signal, and mixing the effected and un-effected signals together. I’m not quite good enough with circuits to throw all of those disparate components together on the fly, so I sought out another circuit that had this integrated into the design.

I came across this design based on the AD633 chip and a reference to a design by Roman Sowa:


I actually found several different circuits based on Roman Sowa’s design, but I liked this one. It was clear and concise, easy to read, and split the components up into easy to understand modules. You can clearly see the input stage, the oscillator (with waveform selector…another bonus), the multiplier and the output stage. I got to work in the basement prototyping this design to see how it sounded.

Most of the components I used were whatever I had on hand, with the exception of the very expensive AD633 chips (8 bucks from digikey). The pots I used were whatever linear equivalent pots I had laying around. I figured that would work for the prototype testing, if I liked how everything was turning out I could pick up the real pots as part of a second order, and design the PCB while I waited for them. The power supply is an old kit I built up about a decade ago from Craig Anderton’s book…still delivers 18 V as steadily as the day I built it. You’ll note the schematic calls for 15 V, but the TL072 and AD633 chips this circuit is based on can easily handle 18 V, so I just used what I had.

Since I had an input stage to work with this time, I tested with a guitar.

Here is just a quick run up the strings, once with modulation, once without…and an accompanying sweep of the frequency:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

I just goofed off with this next test, playing some scales and random notes with the frequency mostly held steady:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Then I played some chords, usually changing the carrier frequency after each strum:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Finally, here is an example of using a ring modulator as a seriously tremolo:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

So there we have it. Very little carrier leakage (to be resolved with tweaking the null pots), steady oscillation without the need of a separate carrier instrument, everything integrated into one circuit. Maybe I’ll toy with putting the oscillation frequency knob into a pedal…

Look out for the PCB design next, hopefully it won’t take me another year.

Casio PG-380 Midi Guitar

1218315849_5b5784b61f_o Several months ago Jason asked me if I could fix a MIDI Guitar. I didn’t have the slightest idea how to fix one, and had only speculative knowledge about how they work, so naturally I said “yeah, sure, piece of cake”. If I’d have known at the time the kind of gear lust this project would create I might have turned him down at the outset.

The Casio PG-380 is a guitar that translates the notes you pick on the strings into MIDI Notes. Think of it like a Keytar, only its ACTUALLY a guitar. You may think that the name brand somehow reduces the quality of this particular instrument, but you’d be mistaken; this baby is top of the line. It can translate amplitude, hammer-ons, and string bends with very little latency. I’d soon find out how hard to find, and how expensive, buying one of these would be.

The problem sounded simple: only the bottom two strings of the guitar were producing notes. At first I figured this must be a calibration issue or something, so I tweaked some of the pots on the board, messed with action height, etc in an effort to get the MIDI pick up to hear and translate the notes. This effort proved fruitless so I turned to the web.

How do you translate audio into midi? I had a vague idea how you could do this with envelope followers and some basic filter networks, but I wanted to understand how this thing actually worked before I could say with any certainty what was wrong with it. I looked around for a long time on the web and turned up nothing related to the technical aspects of converting the output of a guitar pickup to MIDI. In the end I relied on the premise that there must be a filter network to divide the audio by string, and a logic device to convert that analog value to a digital stream of bytes. Since 2 of the 6 strings were working, I could assume that the logic device was probably ok. I turned my attention to what I assumed was the filter network.

I cracked open the case and had a look around. I followed the traces from the pickup back to the 6 calibration pots to the series of capacitors that make up the filter network. I didn’t see anything visibly wrong so I returned to the internet to see if there were any already reported issues for the PG-380. Sure enough I came across this post, which identified a common problem as deteriorating electrolytic capacitors in the filter network. It turns out that electrolytic capacitors go ‘stale’ if left unpowered for a long stretch of time. So, just replace the caps, right? Almost.


MidiGuitar-3 I’m usually pretty reckless (or overconfident), especially with my own gear, but when it’s someone else’s very expensive stuff on the line I tend to be a little more cautious. I prefer to stick with old PCBS, with large thru hole components. Think of your grandpa’s large print books. This was a modern board with tiny surface mount components, something I’ve never dealt with before. I searched around for some techniques I could use to get these little caps off the board and settled on the “hot tweezer” method. This is essentially taking a blow torch to a pair of tweezers until they’re hot enough to melt solder, then gripping the cap and pulling it off the board. This worked for the most part, though there were some persistent ones that I ended up just jamming a soldering iron under an pulling off. That “technique” ended up being a little messy; there is a plastic separator under the caps that melted all over the place. Those tweezers came in handy for scraping that crap off.


MidiGuitar-4 As for the replacement caps, I went with the smallest long lead electrolytic capacitors I could find. I had some of these lying around already and ordered the balance from Mouser. Along with some other stuff for future projects (and posts). Replacement was easy. Cut the leads short, flux the pads, tin the soldering iron, and tack one lead in place. After tacking one lead solder the other post, then fully solder the tacked post. Just like thru hole only you’re tacking the caps on top of the board. It looks a little goofy, but not as goofy as playing a keytar…

Capacitors in place I plugged the guitar in and went to work. Whoa. I hadn’t imagined using a guitar to trigger a synthesizer would be so fun.

Here is a drone sound, with a completely unnecessary string bend at the end:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Here are some chords, which I thought the pg-380 did a pretty decent job of sensing:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

And how about a silly FM bass chord:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

I’m addicted and I have to give this thing back at some point. Looking around on the internet, these puppies go for upwards of $1500. So, if you have a less than perfect PG-380 for sale, perhaps one that needs some new capacitors, I’ll take it off your hands.

The Hybrid Drum Kit

NewDrumSetup-15 Back in June, I wrote a short piece on changing my music studio environment. At the close of that piece I mentioned a buddy of mine was selling an old Alesis DM-5 kit, and that I would soon be incorporating that into my setup. Although 6 months isn’t exactly soon, I managed to tackle the task thanks to a well-timed Christmas gift and an extended vacation.

For years I’ve been trying to create a fully hybrid drum kit consisting of both electronic and acoustic elements. I took a step in that direction with the Mandala (which is playing in even larger role in the latest kit iteration), but to complete the project I needed multiple surfaces, not just multiple zones on one surface to realize what I was looking for. I reasoned that buying a simple electronic drum kit and integrating it into my acoustic kit would do that job. I would of course need something that translated the trigger inputs into MIDI (I just can’t tolerate the stock drum brain sounds), and the Alesis DM-5 kit would do it on the cheap.

I bought the used DM-5 kit not long after my last post, but ran into a hurdle. I use a Pearl rack, but the triggers all use fairly typical 1 1/2” tube clamps. I could have just bought all new clamps for the square rack, but that’s almost as expensive as buying a new rack in the first place, plus I was angling for a more universal Gibraltar rack instead (for mounting a number of other new things in the near future). So, when the wife asked what I wanted for Christmas, I sent her a link to a Gibraltar GRS-850DBL. For those with two kick drums this is a rack that will span both, for guys like me who don’t want to tune two kick drums to each other, this is a curved rack with a left side expansion.

In a matter of hours the new rack was up with the additional triggers. Everything was wired and ready to go. I created a quick drum kit in Battery with the most irritating glitch sounds I could find in five minutes, and recorded a quick and dirty test of the system that sounds a little like this:

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

All technique and mixing aside, not a bad first outing with the hybrid drum kit. I have a few loose ends to tie up:

  • Mount my laptop directly to the rack for stability (parts on the way)
  • 5 pedals is not enough, two more will complete my feet.
  • Devise a way of “patching” triggers around, sort of like a trigger rather than audio patch bay.

For those with picture lust, here are some more images:

Inside The Korg DW-8000: Don’t Put Solder on a Battery

DW-8000-1 Recently a friend asked how hard it was to replace the CMOS battery in Korg DW-8000 keyboard. I assumed it couldn’t be that hard, looked up what kind of battery it accepted (CR2032) and said “yeah, 5 minute job”. I failed to take into account early 80s circuit construction. Sure, its no Maestro PS-1B, but I certainly discovered some “opportunities” upon cracking open the case…

Getting it open is easy enough. As with any device built before the iPod age there are far too many screws… 2 in each corner, several straight up the middle, a bunch to hold the rather flimsy keyboard tray in place. etc. NOTE: To open this thing, you need to turn it upside down, make sure you support the bottom right corner so the joystick doesn’t get smashed.

Once open, you’re probably presented with a ton of dust and 3 filthy circuit boards populated with far-from-RoHS compliant components. Dead center in the 2nd board is the battery…




which is soldered to the board.





DW-8000-7 This probably only sounds ridiculous to me, but really, who solders a consumable part directly to the board. Honestly, this kind of thing calls into question the entire circuit design. I got the dead little bastard freed from its pinholes and went to grab another battery holder. I was pretty certain I wouldn’t be able to find a holder with the same pin out, so I opted to get whatever CR2032 battery holder I could find and shoehorn it in there.


I found some suggestions on the internet for doing this, one involved adding some extra wires to the pins of the holder and running them under the new holder. After examining the board I decided a more stable replacement would be to drill another hole inline with the positive lead circuit trace. This is better explained with pictures:



Grab the tools you need. The battery holder (RS #270-009), a couple of reamers, and a tiny drill bit. You can go with just the drill bit, but if you need to widen any holes I like these little reamers.






Look for a spot around the battery circle that is still sitting on the positive lead on the reverse side of the board. Its easy to see through the board to spot the lead, and for me the hole was right next to the T in “BATT”. In this picture the hole has already been drilled.







Start drilling on the reverse side, at least enough so that the lead won’t tear when you drill through on the other side.







Mount and solder the battery holder (making sure that the polarity is correct), then insert the battery. Done.




No sweat, but certainly not a 5 minute job. The moral of this story? If you’re designing a circuit with a replaceable part (like a battery), please don’t solder it directly to the board. The amateur that has to repair it in 25 years will never thank you, but they’ll still appreciate it.

When Life Gives You Lemons, Use Tie-downs.


Last weekend Cary and I were building a brick wall for a patio. The project required toting bricks from one section of the yard to another. I couldn’t use a wheelbarrow because I needed to go up some stairs (I could have gone around but it’s twice as far). We looked everywhere for brick carriers, a little tool that lets you jam a bunch of bricks between two tongs and move them around 8-10 at a time. We literally called every hardware store in a 10 mile radius, nobody could even tell us where to find them.

We had given up, but while unloading the car I noticed my tie-down straps. You know, the nylon straps with the ratcheting lock mechanism? The things you use to keep the TV you just stole from falling out of your truck? Yeah those. These brilliant little multi-taskers came to the rescue in the form of brick totes for the day.

They probably took a little longer to use than “official” brick carriers or tongs, but I don’t have space for another tool I’ll only use once, plus this didn’t cost me anything. Just lay the straps on the ground, stack up a row of brick on them, ratchet it down and walk away. You could probably drag them around this way if you were so inclined. I wasn’t.

Mandala Meets Drumset


I’ve had several chances to play my new setup now, including the Mandala. I played a live gig (recording to come) on Friday of last week, and learned a few lessons. Now that I’ve got things under control, I started to create some basic kits in Battery for my Mandala.

The Mandala is capable of subdividing into 7 zones. For a 10 inch surface that’s a lot. It’s really great for emulating snare sounds, and extremely realistic percussion (you can effectively assign any number of samples to these zones, velocity map them, and have a “real” drum). But I don’t need that. I have plenty of drums.

My first little attempt at a custom kit for the Mandala is a riff off of the tabla kit released a while ago. I found that 3 zones per pad is my ideal number, so this little snippet features about 30 samples mapped over 3 zones. Each zone is actually only one “instrument” of course, but velocity mapping dictates that I have more than one sample per zone for a more realistic implementation. Here’s a little sample of my drums and the Mandala working together.

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.