-
Pad Construction Matters
12/15/2015 at 05:17 • 0 commentsI was doing some tests today with different pad designs, and I found that one is far ahead of the others. It is pretty simple, and only changes how the foam is laid out.
Previously, I had just sandwiched the piezo elements between approximately equal sized / shaped / thickness of closed cell foam. This worked alright, but there was some issues with differing sensitivities depending on where you hit the pad - if you hit right above the sensor it would sometimes be (much) louder than if you hit near the edge.
The change was to reduce the thickness of the bottom part, and decrease the width while increasing the height of the top part.
Pictures explain better than I can:
The bottom part of the foam should be about 1/4" thick, and the same size (or slightly larger) than the piezo element itself.
The top part of the foam should be about 3/4" thick, and pyramid or cone shaped.
Since the bearing edge of my drums are 3/4" thick, if you were to look at it from the size before putting the head on, it should stick up higher than the rim by about 1/4" or so.
Once the drum head is on, tighten the rim down so that the head is taught. The foam cone will compress down to be flush with the head.
You also do not need to put the sensor in the middle of the drum, as I had done previously. On these new pads, I am finding that I can hit the head anywhere and it sounds the same. I have thus moved the sensor up to the top 1/3 of the pad:
I am now in the process of changing all my existing pads to the new design.
Also, a teaser for the coming week - the hi hat pedal design is coming along nicely, and I hope to have it finished sometime this week if I can find any time to work on it. Keep posted for logs on that...
Cheers
-
Hot air reworking and a demo video!
12/06/2015 at 04:45 • 1 commentIn my design, I pass the piezo signals through a voltage divider prior to running the signal through the first op amp for level holding; this serves a double purpose of bringing the (potentially tens of volts) signal down closer to 3.3v logic levels. Originally I was using a 1k in series with a 200 ohm resistor to ground for the divider, for about 1/6 the signal. In my breadboard testing that proved to be correct, but in real life (with better drum pads) the signal was reduced too much, so the signal to noise ratio of the input was horrible. I decided to change the 200 ohm resistors out for another 1k, and tests showed that this would work much better. However the problem remained of how to actually do it. Trying to desolder with my iron made a mess of the beautiful board when just doing one resistor (of 12 total), so I knew I needed to do something else.
I was reluctant to use hot air to do the rework; my hot air 'station' is a $25 Canadian Tire heat gun used for stripping paint, and the aiming precision is... poor. I needed some way to remove the resistors without melting the through hole pins and damaging the other components.
The solution was to wrap most of the board in tin foil, leaving only a slit still visible. I was nervous when doing it, but it turned out wonderful. Below is a picture (after the old resistors had been removed, and I had replace them with new 1k ones with solder paste, but before I reflowed the solder):
Now that the voltage divider passes through higher signals, the board is much more accurate. You can see the video below for a demonstration. The camera's microphone input is hooked up directly to the line out of Drum Master, so you can get a good idea of the latency (essentially nil). The microphone that I was talking into was plugged into the line in on Drum Master, but since plugging a mic into line in doesn't give the best volume, it is a bit difficult to hear me talking... turn on subtitles to help understand what I am saying. Please excuse my horrible drumming skills, it has been many years since I have played (and I was pretty bad back then, too!) ;-)
-
Bass Pedal Mount
11/29/2015 at 03:24 • 0 commentsI finished the bass pedal mounting hardware today. This lets you attach a normal bass pedal in the right position. It is very simple, consisting of just a metal rod stuck in the middle of a wooden piece, with another wooden piece mounted at a right angle which attaches to the main kit frame.
A picture is worth a thousand words, so read on for the four thousand word explanation:
-
Controller Assembled
11/28/2015 at 17:36 • 6 commentsLast night I finished putting the controller together, in the enclosure, with all the wires soldered up. Not much to say... pictures are the interesting bit here. See below for more Drum Mastery Goodness!
Display and encoder connected:
Pads' RCA jacks connected:
Looking forward towards the display:
Looking backwards towards the RCA jacks:
Enclosure mounted on drum frame:
Closeup of RCA jacks with pads connected:
-
Faceplates Arrived
11/24/2015 at 01:44 • 0 commentsThe Ponoko faceplates arrived today. They were slightly too wide for the case I had made on the weekend, so I had to trim them slightly, but it looks like it will work out just fine.
Some pictures of the case (before sanding, glueing, and finishing) are below:
-
It's Alive!
11/14/2015 at 18:59 • 5 commentsI just finished soldering up the second Drum Master board (the first having gone the way of all the earth, due to repeated rework attempts to fix all the stupid problems I made the first time). Lo and behold, the thing actually works!
First up I have a video showing some of the features:
In addition, I have some pictures of me placing components prior to soldering, a closeup of the soldered board, and the entire board (with and without the Teensy 3.1 + Audio board):
-
Extra Flash Memory Working
11/08/2015 at 20:05 • 0 commentsThe Teensy Audio board has space for an 8 pin SPI flash memory chip; this is much faster than playing directly from the SD card, and thus allows for more concurrent samples to be played at once. (From testing, playing from the SD card allows for about 2 sample polyphony, whereas the SPI chip is around 12+). The only problem is that the 8 pin footprint limits the capacity of the chip; the largest that I have seen is the 16MB Winbond W25Q128FVSIG chip.
Since 16MB is too small for a complete set of samples, I had designed support for a larger Spansion S25FL512SAGMFI011 chip onto the Drum Master board; this gives me 64MB, which should be more than enough for a complete set of samples. (You can then switch between samples by loading samples off the SD card onto the SPI chip).
I had not tried to use the new chip until today; I was having too many problems with the analog part of the circuit. However, today I figured that I should verify the other parts, prior to ordering new parts for the analog bits (if I need to buy other components, I would rather do it at the same time and not pay shipping twice).
However, for the life of me I couldn't get the large SPI chip to work. The logic analyser showed what looked like a floating line on the CS pin; I was getting all sorts of noise.
(This is my testing setup... wires, wires, everywhere!)
Finally I thought to pull up a schematic of the Teensy Audio board.
Turns out that the pin I was using for the large SPI flash memory CS was pin 15, which was allocated for the audio board's volume control pot. I had not soldered a pot on, so I figured it was fine. However I had missed the 0.1uF capacitor which was included on the board.
In case anyone is following along at home, the capacitor you need to remove is this one:
A quick desolder job and the 64MB SPI flash chip is now working perfectly.
-
A Diode is a Diode is not a Diode
11/07/2015 at 02:31 • 7 commentsSo I finally figured out what was wrong with the PCBs... turns out the schottky diodes I was using don't work properly for this application. They were the one component which was different between my breadboard circuit and the PCB. On the breadboard I was using a through hole BAT48 from my junk bin, and it worked fine; after soldering the SMD diodes that I had used on the board onto some pins so that I could use it in the breadboard, I was able to reproduce the same errors that I saw on my PCB.
I have no idea why it is not working; I am not a diode expert, but I don't see anything in the datasheet which jumps out at me.
The bad diodes' datasheet is http://www.mouser.com/ds/2/302/PMEG3010ER-353425.pdf .
The good diodes' datasheet is http://www.st.com/web/en/resource/technical/document/datasheet/CD00000768.pdf . (In addition to BAT48, I tried some other diodes in my junk bin, including BAT46 and some random, thick 1.5A power diodes I bought years ago and don't remember what model they are... all of them work.)
If anyone has an insight on what is different between these two diodes that could cause the difference, please let me know. The circuit in question is a simple peak hold circuit using an op amp, diode, and capacitor:
Input is connected (via some other components) to the piezo which gives an oscilating waveform; output should be the max voltage of the input wave.
When using BAT48 as the diode, it works perfectly. When using PMEG3010ER it doesn't (the output pretty much just mirrors the input, so you see a quickly oscillating waveform).
Oh well... regardless of the root cause, I now know what I need to do to get it working. I guess another Digikey order is coming up soon... ;-)
Cheers
-
Something is not right
10/24/2015 at 16:37 • 0 commentsSomething is wrong with the boards, but I can't for the life of me figure out what it is.
The issue appears to be in the peak hold circuitry. This is where the peak of the waveform is held at a constant level for downstream amplification and sampling. THe circuit is quite simple, and consists of an op amp, schottkey diode, capacitor, with a couple of resistors to help condition the piezo signal before going into the opamp.
When protoyping, everything worked fine with this. Adjusting the capacitor value would affect the falloff rate of the peak; a 0.1uF would take a couple of seconds to go down, and a 10nF would take maybe half a second or so. The design used 10nF caps, as this was plenty of time to do the ADC sampling.
Well, after I got things soldered up, I plugged it in, and I couldn't get a signal for the life of me. Some troubleshooting found a few solder problems, which I fixed. I was now seeing the signal pass through, but the peak hold was not working - the waveform looked very similar to my original Drum Master board, where the pizeo was rectified and low pass filtered, but only with a diode + RC network, no active hold circuit.
Discouraged, I re-implemented the circuit on a breadboard, to make sure I had done it right. It worked there.
I then tried soldering up just the peak hold circuit on another copy of the board, in case there was some soldering issue or something that I had not seen... nada. Still the same results.
At this point I am at a loss of what to do. I can't think of any rational explanation for why it works on the breadboard and not on the real circuit. I have repeatedly checked continuity between critical points, checked the component values, checked for shorts, etc. The only differences I can think of are that I am using a different diode on the SMD version, which has slightly less forward voltage, but given how it is configured in the op amp circuit this should not have any impact. I suppose that parasitic capacitance on the breadboard could be making things different, although normally that would cause problems, not make things work. Furthermore, the design checks out in Falstad's simulator, and I would not expect this to be the case if it were a very finicky circuit which relied on parasitic capacitance.
If anyone has any thoughts on what could be wrong, please feel free to let me know. I am getting frustrated; it's been a whole week fighting these problems, and I am pretty much out of thoughts on what could be wrong.
(Some of my ugliest soldering - it looked good originally, but during troubleshooting I tried changing the values of a number of resistors + capacitors, and my re-work skills are not quite up to par...)
Until next time...
Cheers
-
Boards Arrived!
10/16/2015 at 19:08 • 0 commentsThe boards showed up on Wednesday; almost two months to the day since it shipped, and more than two months since the order was placed. Oh well, boats are slow, I guess.
I have yet to test the board's continuity, let alone solder it, but a brief inspection looks good. I hope that I will find enough time to do some soldering this weekend... I'll post another log when things are getting closer to completion.
Cheers