Gong Drone

synaesthesia

This time no schematic yet, just the sound for now. I am trying to build a drone that creates relaxing sounds and wanted to have gongs and a bit of interference as the main elements. The whole circuit needs only 3 CMOS ICs. There can be either two main frequencies (GongDrone1) or four (GongDrone23). I am still working on the fading and random pattern section. The circuit is self-running and needs no intervention after tuning. The recording is from the bare circuit with no effects added. Let me know what you think.

GongDrone23.mp3

Description:

Filesize:

1.48 MB

Duration:

1:37

Viewed:

2533 Time(s)

GongDrone1.mp3

Description:

Filesize:

1.36 MB

Duration:

1:29

Viewed:

2325 Time(s)

umschmitt · Posted: Sun Sep 14, 2014 5:15 am Post subject:

I think you should post that schematic!

RingMad · Posted: Sun Sep 14, 2014 5:23 am Post subject:

Wow! These sound great! I'm really curious about the schematic.

-- James.

PHOBoS · Posted: Sun Sep 14, 2014 6:48 am Post subject:

nice!

3rd vote for a schematic Wink

commathe · Posted: Sun Sep 14, 2014 12:24 pm Post subject:

I have to know

DUBmatze · Posted: Sun Sep 14, 2014 1:52 pm Post subject:

yes
cool sound
post the old robot p0rn plz Wink

synaesthesia · Posted: Sun Sep 14, 2014 3:36 pm Post subject:

Ok, although I am not entirely happy with the circuit yet, I am posting the schematic of what I have on my Lunetta-Lab board plus bread board. Maybe you have a few ideas how to improve it. (Note: I put a 47u cap for the tempo oscillator in the schematic. Actually, I have a 1u and an extra divider on my board. Haven't tried a 47u cap with a 40106 yet.)

The basic idea is to use a counter to generate two frequencies and waveforms. One is from the normal counter operation. For the second frequency the counter reset is driven by a resistor-diode AND gate. There are three ratios possible, all three shown in the schematic. The 16/9 doesn't sound too good. The 8/5 (U3B) and 4/3 (U4A) ratios both sound nice. Using identical dividers tends to sounds boring in the mix. Use at least one like U3B and one like U4A.

There are two signals to control each oscillator. One is the GATE. When it goes HIGH it starts to load a large cap which starts to open a transistor. This fades the tone in. When it goes low again, the cap starts to empty and the tone fades out. A universal NPN transistor seems to work fine here, but maybe this part can be improved.

The other signal is SELECT. When LOW, the divider is disabled because the diode-resistor gate is disabled and RESET for the counter stays LOW. When it goes HIGH, the diode-resistor AND becomes active and the divider becomes active, changing the output frequency.

The result is mixed using a resistor network. I have added a 10n capacitor to ground to filter the output a bit. Room for improvement here. The 100u caps determine the speed of the fading effect. 47u works as well, but will fade in and out faster. My first recording used 47u caps and slightly different resistor values.

The gong effect occurs when the GATE and SELECT are inverse to each other and then change state. For example, if a single HIGH travels through the shift register in my schematic, or when two adjacent counter outputs would be used to drive the signals and change from 01 to 10 state. This changes the frequency and at the same time fades the tone out. You could drive these two from a counter using Q0 for GATE and Q1 for SELECT. I have chosen a shift register that I feed with a random frequency because the shift register pattern is more random than a counter. In this circuit the shift register input is simply coming from an unused output of one of the counters. Still experimenting here.

The resistors values and counter outputs use Q0 and Q2 are chosen to generate a waveform in a way so that the higher frequency part is in the upper level of the curve. This is because the transistor seems to cut the upper level more than the lower level. This way it fades the lower harmonics of the waveform in first, and the higher harmonics later. If you look at the two screen shots from the scope, you see what I mean.

I am tuning the oscillators at the beginning so that there is a harmony. For that I drive the shift register at maximum speed and feed a constant HIGH to the input. It probably would be better to set all GATEs to HIGH and all SELECTs to LOW. More room for improvement.

Two waveforms sound best when tuned properly. Three is ok, but hard to tune. Listen to the track to hear how it sounds. Four oscillators is next to impossible to tune. I am experimenting with driving two oscillators with the same clock, maybe using different divider ratios. However, this way the beat frequencies from the interference become too dominant. It would be best to drive the second oscillator with a clock that has a tiny delay to the first. Maybe this can be done with a shift register or is there a simple analog way?

GongDroneV1.mp3

Description:

Filesize:

2.19 MB

Duration:

2:24

Viewed:

2264 Time(s)

GongDroneFade.PNG

Description:

The signal starts to fade in. Yellow is the waveform at the collector of the transistor, blue is the mixed output.

Filesize:

4.27 KB

Viewed:

21194 Time(s)

GongDroneFull.PNG

Description:

The signal is at maximum level. Yellow is the waveform at the collector of the transistor, blue is the mixed output.

Filesize:

4.45 KB

Viewed:

21195 Time(s)

GongDroneV1.PNG

Description:

Filesize:

46.91 KB

Viewed:

1247 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

synaesthesia · Posted: Mon Sep 15, 2014 1:19 pm Post subject:

I was experimenting in the last two hours and understood the gong effect a bit better. It is not just the switch of the divider, but a second frequency has to be present at that moment. The effect happens if at first the two oscillators are at frequencies f1 and f2 that mix, but do not produce an interference. When one oscillator then goes to fade-out and at that moment switches it's divider so that it produces frequency f3, and the unchanged frequency f2 and the new frequency f3 do produce an interference, then you get that nice gong effect.

PHOBoS · Posted: Mon Sep 15, 2014 1:59 pm Post subject:

Thanks for posting it Very Happy

animalman · Joined: Jan 21, 2013 Posts: 18 Location: out past midnight

RingMad · Posted: Tue Sep 16, 2014 4:13 am Post subject:

Thanks for sharing the schematic! I haven't really had time to look at it or understand it yet. I don't have any 74HC595's on hand... and it appears that chip can't be used at 9V, which is usually what I use for my boxes.

-- James.

PHOBoS · Posted: Tue Sep 16, 2014 10:06 am Post subject:

synaesthesia · Posted: Tue Sep 16, 2014 2:28 pm Post subject:

If you don't have a shift register, you can use a counter as well. The pattern will not be random then however, but repeats after a while.
The schematic below is labeled V2, but actually is closer to the circuit I used for my first recording with two oscillators.

You will notice that I changed the resistor values for the counter outputs again. I just can't decide which version sounds best. So PHOBoS idea to use a pot is right on the spot. The only thing I could settle on in my experiments today is to not use a resistor on Q2. Using only Q1 and Q3 makes two distinct frequencies audible per oscillator.

I used two 4520 counters here, but of course a 4040 can be used as well, or even a 4060. C4 and C5 are optional. I am still experimenting with the filtering.

GongDroneV2.PNG

Description:

Filesize:

30.42 KB

Viewed:

677 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

GongDroneV2Waveform.PNG

Description:

The waveform with the resistor values of V2 should look like this (counting to 16)

Filesize:

6.33 KB

Viewed:

545 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

RingMad · Posted: Wed Sep 17, 2014 10:15 am Post subject:

synaesthesia · Posted: Sun Sep 21, 2014 1:33 pm Post subject:

This is version 3 of the Gong Drone. Actually, I wanted to post it already yesterday, but I realized that I made a mistake. I originally wanted to run all three outputs Q1,Q2,Q3 of each 4520 through a transistor stage to fade all frequencies in and out. Then I realized that I would always have to fade them out together when the divider is activated for the gong effect. So I decided to use analog switches from a 4066 instead to just select which outputs are mixed.

In this version the lowest frequency at Q3 is always present, the frequencies from Q1 and Q2 are randomly mixed based on the state of a shift register. I am using a 4015 this time, once because it seems to be more popular here, and also because this way I can have two 4-bit random values changing at different speeds.

The first 4015 U4A is advanced by an LFO as before. It determines the fading and division for the two 4520 oscillators. Again, the gong effect occurs when a '1' travels from Q0 to Q1 and is followed by a '0'. This fades out the oscillator output and at the same time enables the divider to switch the frequency.

The second shift register U4B is clocked by the output Q3 from U4A. This results in a slower change for the outputs that control the frequency components that are mixed. The effect from this is a slight change in the sound from the two oscillators that makes the whole circuit sound a bit more interesting.

The push button is used to reset the shift registers and at the same time open both transistors. This will enable the lowest frequency of both oscillators, making it much easier to tune the oscillator frequencies.

Only part of the circuit was tested on my Lunetta-Lab. But it was promising and so I decided to build it. For that reason the circuit this time also includes a small amplifier using a LM386. Of course, you can just pick up the output at C5 and use whatever amplifier you already have. I intend to use the bass boost of the LM386 as a low pass filter that cuts off frequencies above a few hundred Hertz. Thus the R17+C6 combination in the feedback path of U5. When listening to the result I couldn't find much difference though. I guess the small speaker I used wasn't up to play back high frequencies anyhow. Will try with a better speaker soon.

I soldered the whole circuit on a small 5x7cm PCB, but couldn't complete the wiring today. So no recording yet, sorry.

GongDroneV3.PNG

Description:

Filesize:

48.29 KB

Viewed:

781 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

synaesthesia · Posted: Mon Sep 22, 2014 11:35 am Post subject:

Ok, here is the complete PCB. Works like a charm and sounds just great. No hiss, no crackle, no hum. All of that must have been introduced by the patch cables or by unused circuits on the Lunetta-Lab board. Also, using a bigger 10cm diameter speaker makes a huge difference to the small 5cm speaker I used for the first test. The LM386 works fine as given in the schematic. Power consumption at 5V is about 10mA on average and 21mA at peak volume.

GongDroneV3PCB.PNG

Description:

Filesize:

1.52 MB

Viewed:

564 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

PHOBoS · Posted: Tue Sep 23, 2014 1:17 pm Post subject:

wow you really managed to make that very dense. I'd probably end up with something 3 times that size.
I wonder how the other side looks,.

synaesthesia · Posted: Fri Sep 26, 2014 9:16 am Post subject:

The other side looks like this...
In the tried and trusted way, using 0,5mm wire for power lines and simple connections and 0,3mm enameled wire for the remaining point-to-point connections.

Plus the KiCAD PCB file. I am using 0 Ohm resistors for the vias because these great prototype boards are double-sided and normal wire could create a short.

GongDronePCBback.PNG

Description:

Filesize:

1.69 MB

Viewed:

617 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

GongDronePCB_KiCAD.PNG

Description:

Filesize:

56.34 KB

Viewed:

525 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.