The Groaner

Pointless Electronics PE-002 · Pick it up and it groans · A tilt switch, a recording module, a speaker and a metal box · By Onion Madder · Mess O' Pedals

It sits on the desk doing nothing. Pick it up and it groans at you. Put it down and it groans again. That is the entire behavior, and there is no control on the outside of the box to change it, stop it, or apologize for it. It is the second Pointless Electronics device, and it has a product page.

There is no circuit board in this build. I designed one, drew it, drilled it and got most of the way through soldering it before working out I did not need it - more on that below, because the mistake is more useful than the build. What actually goes in the box is a recording module, a tilt switch, three cells and a speaker, and the whole thing is a handful of solder joints.

It takes about an hour, most of which is the grille.

Before you start

tap to expand / collapse

Record first, swap the resistor second. This is the only order that works, and getting it backwards is the one mistake in the build that looks like a fault. The reason is in The 470K Trick below - read that section before you touch the iron.

Drill before you solder. Aluminum swarf is conductive and it goes everywhere. The grille and the speaker mounting get done first, then blow the box out, then start on the electronics.

The box is a conductor. A bare module rattling around inside a metal enclosure is a short waiting to happen. Tape it, sleeve it, or glue it down before the lid goes on.

What you'll need

tap to expand / collapse

1× ISD1820 recording module - spot it: small blue board with a black blob or DIP chip, a microphone can, and three tactile buttons marked REC, P-E and P-L. Records about ten seconds through its own mic. The ISD1820 guide has the whole board mapped if you want it
1× tilt switch (SW1) - spot it: two legs, and it makes and stays made when you tip it past an angle. If it only clicks briefly when you tap the bench and then goes quiet, that is a vibration switch - a different part with different manners. Both work here, see Things To Try
1× 470K resistor (R) - spot it: yellow, violet, yellow, then gold. This replaces the ~100K already on the module and it is the whole sound of the thing
1× 4 Ω speaker, 5W - spot it: two terminals, no polarity worth worrying about at this level. The module drives it straight off its SP pads; there is no amp in this box. 36-45mm suits the medium grille
3× AA and a holder - spot it: three cells, not two. 4.5V sits comfortably inside the module's 3-5V window and stays there as they drain. Two cells start at 3.0V and fall off the bottom within a week
1× 125B enclosure - spot it: 122 × 66 × 40mm cast aluminum with a screw-on backplate. Mine is Tayda's winked silver. The grille goes in the front, the face of the box

The whole circuit

There is no more of it than this. The module records and plays; the tilt switch is a button gravity presses; the speaker hangs straight off the module's SP pads.

battery + ────────────────┬── module VCC └── SW1 (tilt) ── module P-E battery - ──────────────────── module GND module SP+ / SP- ───────────── speaker (4 ohm, hot-glued behind the grille)

Steps 1–5 Record the groan

Do this on the bench, before anything is mounted and before the resistor is changed. Keep it short - the resistor swap is going to stretch it to about twice as long, and a groan that runs on for fifteen seconds stops being funny about eleven seconds in. The module drives the speaker on its own, here and in the box.

Record the groan
Step	Do
1	Connect the battery holder to the module's VCC and GND pads.
2	Connect the speaker to SP+ and SP−.
3	Hold the REC button down and groan into the microphone. Two seconds is plenty. Let go.
4	Press P-E. It should play your groan back at normal speed and normal pitch.
5	Groan again until you like one. Each recording overwrites the last.

The ISD1820 module held between finger and thumb, before anything is changed: a round electret microphone at the top left, a row of six small surface-mount parts along the top labeled R2, R5, D1, C1, R1 and R4, an empty 14-pin socket marked U1 in the middle, a two-by-six header on the left labeled VCC, GND, FT, P-L, P-E and REC, a white two-pin speaker connector marked SP1 on the right, and three push buttons along the bottom labeled REC, PLAYE and PLAYL, the REC one with an orange cap. — The module as it comes, chip out of its socket. The row of parts along the top is where the resistor lives; the header on the left is where the battery and the tilt switch will go.

Steps 6–10 Drop it down

Drop it down
Step	Do
6	Find the oscillator resistor on the module - the ~100K one, not the LED's. It sets the sample rate.
7	Desolder it and fit the 470K in its place.
8	Desolder the LED, D1, while the iron is hot. This module goes inside everything I have ever used it in, and if a project wants the LED for any reason I break it out to the panel. The Groaner does not: the basic electronic version of a groan tube has no indicator, it simply plays when you flip it, and I wanted the box as simple as possible.
9	Press P-E. Your groan is now about half speed, an octave down, and has lost all its top end.
10	Write “REC 100K / PLAY 470K” on a scrap of tape and stick it inside the box. Future you will need this.

The same module with the chip now in its socket and the microphone removed, leaving two bare solder pads inside the MIC circle. In the row of parts along the top the rightmost position, R4, is empty, two bare pads where the resistor was. The white speaker connector is gone too, its pads bare beside SP1. — Step 7, half done: R4, the oscillator resistor (it was marked 104, which is 100K), desoldered off the top row. The mic and the speaker connector have come off as well.

The module with a through-hole resistor soldered onto the R4 pads, its body standing up past the top edge of the board on bent leads. The D1 position in the same row is now empty as well. White and black speaker wires are soldered straight to the SP1 pads at the right. — The 470K in R4's place, on its leads, standing clear of the board. D1, the module's LED, is off the board too, and the speaker wires go straight to the SP1 pads.

Steps 11–14 Wire the tilt switch

Wire the tilt switch
Step	Do
11	Find the two pads of the P-E button on the module.
12	Solder one leg of the tilt switch to one of them.
13	Solder the other leg to the other one. The switch is now wired in parallel with the button, and no polarity is involved.
14	Tip the module by hand. It should groan once per tip, all the way through, with no stutter.

The solder side of the module, blue with the letters MH. A small glass tilt switch is soldered on this side, its two leads bent to reach two pads near the top center. Green and black wires are soldered to two pads on the left, and white and black wires leave from the bottom right. — Steps 12 and 13 from underneath: the tilt switch soldered straight across the P-E button's two pads on the solder side. The green and black leads are the battery, the white and black the speaker.

On a dark desk, a black holder with three AA batteries on the left, the ISD1820 module in the middle with the 470K standing up off its top edge, and a small square-framed speaker on the right, all joined by short wires. — Step 14: the whole thing running on the bench before it goes near a box. Three AAs in a holder, the module, the speaker on its leads. Tip it and it groans.

Steps 15–21 Box it

The grille comes first, then the speaker goes on the back of the grille, then everything gets wrapped and screwed down.

Box it
Step	Do
15	Drill the speaker grille in the front of the box - see Boxing It Up.
16	Blow every last chip of swarf out of the box.
17	Hot-glue the speaker to the inside of the face over the grille, all the way around the rim, wired straight from the module's SP+ and SP− pads.
18	Tape or sleeve the module so no bare copper can touch the case.
19	Mount the tilt switch so it sits just inside its threshold with the box flat on a desk - putty first, glue once you like the angle.
20	Fit the batteries, put everything in, screw the backplate down.
21	Pick it up.

A cast aluminum 125B box held open side up in one hand, a pink keyboard behind it. Twenty-five holes are drilled through its face in a diamond: one hole at the top, then rows of three, five, seven, five, three and one. Four corner bosses with screw holes, nothing else in the box. — Step 15 done: the square grid drilled with its corners left out, which makes a diamond of 25 holes, seen from inside the empty box.

The same box with a round speaker face down over the holes, its magnet reading 4 ohm 5 W, a thick bead of amber hot glue all the way around its rim onto the box floor. Its white and black leads run up to the ISD1820 module, which rests on the box edge on its wires. — Step 17: the 4 Ω 5W speaker hot-glued over the grille, rim sealed all the way around, wired straight off the module's SP pads.

The box with everything inside: a black holder with three AA batteries down the left, the glued speaker in the middle with a blob of hot glue on its magnet, and the ISD1820 module on the right with its buttons and header showing and green, black, white and black wires between them. A brass center punch lies beside the box. — Step 20, just before the backplate: three AAs, the speaker, the module, and the center punch that did the holes.

The finished Groaner held upright in one hand: a cast aluminum 125B in Tayda's winked silver finish with a diamond of 25 drilled holes in the middle of its face and nothing else on it. — Step 21. The final build: a square grid of holes in the face, the corners dropped, and nothing else on the outside.

The off-board wiring

Pad	Goes to
VCC	Battery +, straight in. 4.5V from three cells. There is no power switch.
GND	Battery −.
P-E	One leg of the tilt switch. The other leg goes to VCC. In other words, straight across the P-E button's own two pads.
SP+ / SP−	The speaker, straight off the module, hot-glued behind the grille in the front. There is no amp between them.
REC	Nothing. The onboard button is the only thing that ever touches it, and only before the lid goes on.

If it doesn't work

Troubleshooting
Symptom	Check
Nothing at all.	The cells. Then that you actually recorded something.
The onboard buttons work but tilting does nothing.	Meter across the tilt switch's two legs and tip it. If it never makes, the part is dead - and if it appears to have one leg, the other one snapped flush at the seal and it was dead before you bought it.
It groans at normal pitch.	You re-recorded with the 470K fitted. Put the 100K back, record, refit the 470K.
The groan stutters or restarts.	Your switch is a vibration switch, not a tilt switch. Either live with it or put the one-shot board back in.
It groans constantly, or whenever anyone walks past.	The tilt switch is mounted right on its threshold. Rotate its mount a few degrees so it sits clear of the trip point at rest.
Thin, papery, no weight to it.	The speaker isn't sealed to the plate. Air is going around the back of the cone and canceling the low end. Hot glue all the way around the rim.
Quiet, or the groan clips into a buzz.	There is no amp and no volume to turn up - the module drives the speaker on its own. If it buzzes, the cells are tired.
Worked on the bench, dead once boxed.	The module is shorting against the case. Take it out, wrap it, put it back.

How it works, which is barely at all

The module's P-E button does exactly one thing: it connects the P-E pin to the positive rail. The chip sees that pin go high and plays the whole recording. That is the entire trigger mechanism.

So a tilt switch wired across that button is also a button. It's just one that gravity presses. Tip the box, the contact closes, P-E sees the positive rail, the groan starts. Tip it back, it closes again on the other edge, and it groans again.

The 470K trick

This is the whole sound of the box and it is one resistor.

The ISD1820's oscillator resistor sets its sample rate, and the sample rate governs both recording and playback. Stock it's about 100K. Put 470K there instead and the chip runs about half speed. The datasheet's resistor table stops at 160K, and past it the oscillator stops following the resistor: 470K is not 4.7 times slower, it is about twice. Mine was recorded in one second and groans for two.

The pitch drop lives in the mismatch, not in the resistor. Record at 100K, play back at 470K, and playback runs slow - the tape-speed trick, done in hardware. If you ever re-record with the 470K still fitted, it records slow and plays slow, the two cancel, and your groan comes back at normal pitch sounding like nothing happened. That is not a fault and nothing is broken. It is the resistor doing exactly what it says.

The reason it ends up sounding like a groan tube rather than like a slowed-down person is that sample rate takes bandwidth with it. Dropping the rate doesn't just lower the pitch, it throws away the top end. What comes out has no consonants and nothing above about a kilohertz - which is precisely what a groan tube is, acoustically: a spring, a membrane, and no treble.

470K is where I stopped. Lower values give you a shorter, higher, clearer complaint; higher ones go so slow they stop reading as a voice at all. It is worth clipping a few in before you commit, because this resistor is the only creative decision in the entire project.

The board I didn't need

I built a 555 monostable for this. A tilt switch chatters, I reasoned, and every bounce would restart playback and chop the groan into a stutter - so the switch would trigger a one-shot, the one-shot would hold its output high for a couple of seconds, and the groan would be guaranteed to play through.

Then I wired the switch straight to P-E to test it and it worked perfectly first time.

A decent tilt switch closes once and holds. It isn't a ball rattling in a tube. There was no chatter to clean up, so there was nothing for the board to do. I had designed a fix and then gone looking for the problem, which is the wrong way around and cost me an evening.

The board wasn't wasted, mind. What it actually is - a 555 monostable with an edge-triggered input - is “any filthy contact in, one clean pulse out”, and that's a permanently useful thing to have on a shelf. It debounces bend points, it turns a footswitch into a fixed-length gate, and it'll pulse a motor. It just has nothing to do with this box. The version that did get finished is the One Shot on the stripboard shelf: 9 by 18, a TLC555, drawn in Copper Bottom, with its layout file and its own page.

If you take one thing from this build, take that: wire the simple version up and see it fail before you design the thing that fixes it. It takes two minutes and it is occasionally humiliating.

Boxing it up

One grille and nothing else. It goes in the front, the face of the box, because that is where it looks right. Nothing else goes on the outside - no knob, no LED, no label, nothing to operate. That's deliberate. A control implies an operator; a box with nothing on it but a grille implies a creature.

Don't eyeball the holes. There is a 1:1 grille drill template for the 125B. Three sizes on the sheet - 23mm, 33mm and 43mm - all on 3mm holes at a 5.24mm pitch, which leaves a 2.2mm web everywhere. Page one is the 122 by 66 outline drawn true size with the medium grille centered - the sheet says backplate, and the face is the same outline, so it goes on the front the same way: tape it over the face and punch straight through. Page two is the sizes as cut-out patches.
↓ Or the simple grid: 7 by 7 on two axes, centered on the face - drill the center, the 12 / 3 / 6 / 9 of each ring, then the rest, and stop at any ring (PDF)
Print at 100% / “Actual size”, never “Fit to page” - check the bar measures exactly 100mm before you drill anything. Center-punch every dot before any drill touches the metal, and pilot every hole at 2mm.

There is no power switch. It is on the whole time the cells are in, and an ISD1820 idles at about a microamp between groans, so that is the cells' shelf life, not the box's. Keep every hole at least 12mm in from an edge - the corner bosses and the lid lip live in there.

The arithmetic behind 50,000 groans. Three AAs hold about 2,500mAh. The datasheet gives the chip 1 µA in standby, 10 at worst, and 15mA playing with no speaker on it; the speaker adds whatever a 4 Ω cone takes from an output rated for 8, so call a groan 50 to 100mA. Mine was recorded in one second and plays for about two, which is 0.03 to 0.06mAh a groan, or 45,000 to 90,000 groans from fresh cells. Over ten years the cells lose more to sitting than to groaning - up to 900mAh to standby at the worst case, another 500 or so to alkaline self-discharge - so the ten years is what gets it, with 20,000 to 45,000 groans still in hand when it does. Unmeasured: the groan current. A meter in series with one cell during a groan pins the whole sum, and the page says 50,000 until then.

Things to try

Use a vibration switch instead. Then it complains about things that happen near it - a door slamming, a book dropped on the desk, someone walking heavily past - rather than things you did to it. More paranoid, and a genuinely different object. You'll want the one-shot board for this one, because those really do chatter.
Change the resistor. It's the only voice control the box has. 220K for a shorter, clearer moan; 1M if you want something that barely resolves as a sound.
Record something that isn't a groan. A sigh. A single tut. Your own name, said with disappointment. The mechanism doesn't care and the object changes completely.
Put a pot on the oscillator resistor and bring it out to the panel - 100K pot with 20K in series. I left it off on purpose, because a knob turns a creature back into an appliance, but it's your box.
Weight it. A slug of something in the bottom. People trust heavy objects, and a heavy object that groans when you lift it is doing more work than a light one.

Credit where it isn't due: the ISD1820 module is a commodity part and the trick of pulling its oscillator resistor about is well-trodden ground among people who bend toys. The box, the arrangement, the 470K, the template and the words are mine. Build it, bend it, make it yours.