I own a used Clavia ddrum4 SE drum module. It employs a fairly advanced technique to detect the position of the hi-hat; on the stand is mounted a magnet, and in the hi-hat pad is a Hall effect sensor circuit.
It’s exceedingly difficult to find working hardware for the module, so a number of years ago someone reverse engineered the circuit and posted it online. I have tried to replicate the circuit on a breadboard, with one exception: I couldn’t find a retailer selling the original sensor model, so I tried my best to find a replacement unit with characteristics that are compatible with the original. I fear I have failed, because the test circuit isn’t working. I am not an electrical engineer; I only know enough to be dangerous and I’m in over my head. Let me walk through the details:
here is the circuit that was posted:
the “508” is a Micronas HAL 508
- I substituted a TT electronics Optek OH180U
- the ddrum4 SE unit provides +4.75V on the ring of the connection jack, ground is sleeve (tip is for the piezo trigger)
- I have confirmed that simply placing a 10K pot across +4.75 and ground (ring and sleeve) lets me control the hi-hat position. It’s extremely touchy with a narrow range; if I bridge a 10K fixed in parallel and recalibrate the unit (a built-in function to set the “closed” point) it makes it slightly more responsive…anyways, the point is I have confirmed the head unit works.
- in the circuit, ddrum must have fine-tuned the 10K trim pot at the factory and then sealed it, because users’ pictures show it covered with glue. I’ve tried adjusting it all over the place and recalibrating often with no success.
- I've double-checked pinouts for everything, especially diode polarity, op-amp pins and Hall effect sensor pins too.
So my questions:
can a real electrical engineer look at the HAL 508 and the Optek OH180U and see if they’re in the ballpark? Perhaps my choice of replacement Hall sensor is just completely wrong. The 508 is a unipolar switching sensor, on typically at 18 mT, off at 16, hysteresis 2. The OH180U is also unipolar switching, on at 18mT, off at 14, hysteresis 4. I've used some fairly strong magnets...not rare-earth, but some stacked together. It should be enough to get something out of it. I tried both N and S by the way. =) Perhaps I should be making a test circuit to see if the magnets are triggering the sensor? (Links appreciated!)
I’m really confused about the circuit itself; in that there are only two leads (+4.75V and ground) provided even for the op-amp supply. I’m assuming that it’s forming a parasitic load on the voltage based upon the magnet distance—which the unit measures—but that’s its supply too. Can someone explain what this should do? I’ve tried to match the design to a number of basic op-amp circuit designs on the web, but almost all of them have a separate output, so I don’t grok this pattern.
It’s also conceivable that the circuit was reverse engineered improperly…I’ve looked at the pictures that were taken, and tried hard to trace everything to confirm correctness. The problem is, the guy didn’t take apart or damage his unit (obviously! it still worked) so it’s possible we missed a connection. Does this circuit look “really close” to something familiar, but is messed up a little bit?
I traced the original op-amp’s unit 2 circuit, and it looks like they didn’t terminate it properly—the output is wide open and in+/in- are tied to ground. I terminated my unit 2 circuit properly with 10K resistors from in+ to supply and ground, output to in- like mentioned here. Is it possible that me changing this design would influence the first circuit? I don’t think so, but I’m grasping at straws without a formal education in electronics so…
Thanks so much in advance for any help.
I forgot the link to the op-amp #2 termination I used instead of their "dangling" approach: here it is. I used version (a) because I don't have a negative supply.
responding to Scott: I was very confused about the choice of Hall effect sensor as well. The sensor data sheets clearly state that they are switched sensors, i.e. on/off. However, the Clavia documentation (and several other sources) state that there are at least 8 levels of magnet position sensor detection. It's actually also implied that they interpolate between these, most likely in software. But 8 levels is quite a bit more than 2! =)
At the bottom, I am attaching some relevant pictures of the working device someone disassembled, in case that gives anyone a Eureka! moment.
I assumed the zener diode is a pop protection for when the 1/4" stereo cable is inserted into the jack; the contacts kind of jump until the cable is all the way in.
I also assumed the capacitor between signal and ground acted like a low-pass filter, but I don't really get how it plays with the op-amp. I kind of thought you'd want a high-pass filter to smooth a noisy sensor; having a magnet move close to this circuit by means of me pushing my foot on a pedal is pretty much the definition of low frequency so why would we filter that?
responding to Russell: great questions...let me try to elucidate some things (but claim ignorance on many others!)...
The ddrum4 SE brain is powered from 120VAC; literally a two-prong US polarity plug. Inside it must have a regulated power supply of some sort...I'm reluctant to open the unit because if I break something I'm pretty much toast on fixing it. (I did open it; see below).
The connection between the hi-hat pad/controller and the brain is a tip-ring-sleeve 1/4" stereo cable. That's it. Only three wires. One is definitely common signal ground, and another is also well-known...
The way that pad trigger interfaces almost always work, is that you have a piezo device attached to a metal plate covered with rubber. The physical hit of a stick causes a voltage spike across the common ground (in my case, the sleeve) and the trigger signal wire (in my case, the tip). This signal goes through an amplifier gain stage which is fed into an ADC. The ADC is scanned at a very high rate, so the unit can take slope readings of the change in signal over time in order to produce an accurate "note on" MIDI message—which includes "velocity" or signal strength—before the piezo trigger spike actually peaks. The ADC is most likely 10-bit, since their product literature claims "more than 1000 levels of velocity" (even though MIDI is only 7-bit, internally they apparently use 10).
Additionally, ddrum specifically designed the interface such that a standard 2-wire (mono) 1/4" cable can be used on this input, in case it's just a normal pad trigger. That means that the device works fine if the ring (middle wire of 1/4" stereo) is connected to sleeve (common ground)! When I measure potential across ring to sleeve, it's +4.75 V.
But what's most puzzling is this: the Hall effect sensor circuit—with the LM358 amplifier—is both powered by and communicates its output with the same two wires...sleeve (ground) and ring (+5 V).
I'm guessing some things here—again, I'm not an electrical engineer—I'm assuming that they're probably using another ADC to measure voltage across ring and sleeve. They probably have a +5 reference voltage in a comparator circuit (or some other scaling device) to detect how far this circuit above drops the +5 it's being given. Maybe that's the trick...the circuit is supposed to drop the voltage, but only by a small amount so as not to make the op-amp fail below nominal supply voltage?
Or perhaps...forgive me if I'm talking utter nonsense here...don't resistors drop both voltage and current? So maybe the circuit acts like a power sink, consuming a variable amount of current based on the distance of the magnet? If that's the case, then maybe they're somehow using an ADC to measure the current draw of the circuit?
Without opening her up, that's pretty much all I can surmise. I know for a fact that it senses the position of the magnet with two wires, and those same two wires act like ground and supply for the op-amp that's part of the circuit. I know if you bridge that supply to ground, nothing bad happens and if it's in hi-hat mode it acts as if the magnet is infinitely far away (this could be a software mode though). I know I can get a trim pot across the supply and ground to make varying levels, although the reaction band is very narrow. I could try to estimate that range if that would be useful to someone.
It's starting to sound like I should open up the main module and try to trace the circuitry and get more information.
...and I did...it's a fair amount of surface mount stuff, traces on both sides, routing all around and then over a cable to the main board. I took pictures, but I don't think I can figure out what's happening. The hi-hat jack is J10.
The interface/trigger input panel:
The wires from the original controller:
The original circuit board:
Identifying the Hall effect sensor: