andres ramos 14 - The Coherer Mill - by Andrés Ramos

[Andres Ramos]

1. Abstract

From my experiences with the coherer I gained a new view on this fascinating component that only is a grain of metal filings in glass a tube with electrodes at its ends. When the coherer filings are iron or nickel they are suceptible to magnetic fields. A moving magnet would cause the filings to permanently realign with the lines of magnetic force. Thus the filings were moving too. Moving filings are working like very little electrical switches establishing connection and disconnections randomly. I hoped that this entropy  would provide another source of noise for the spirits to work with. I expected the signal to be made of much spikes.

2. Mechanical assembly

It was clear to me that this project would be more a challenge in mechanical issues instead in electronic. I chose to make a rigid but simple structure made by wood.

The coherer is mounted on a wooden base and filled with handmade nickel filings. A rotating magnet is fixed on a boom that is glued to a disk with a slot in it. The disk is screwed to the shaft of a gear motor. By adjusting the screw in the slot I can adjust the distance of the magnet trajectory around the magnet.

3. Electronic circuit

You see the electronic is boring simple. Just a potentiometer for the coherer bias and the coherer in series with an AF transformer to route the signal to an audio jack.

4. Tests with the coherer mill

I ran some tests in advance to find out the optimum distance for the magnet  and bias. It turned out that the settings were not that critical. The bias was adjusted to roughly  3V. The distance between coherer and magnet is around 3,5cm.

As expected the signal was pretty spiky with it's own rhythm caused by a non continuous realigning of the metal filings.

Raw signal

My standard procedure for post processing of signals with impulse characteristic is to apply  the Paulstretch function in Audacity.  It analyzes the signal and makes an educated guess to fill in the gaps while the signal is stretched. I always apply a stretch factor of 1.6 and resolution of 0.1s

Paulstretched signal

A little denoising of 9dB is rounding up the post processing. The signal quality is not very impressive. The voices are intelligible but still have the coherer typical croaking sound. However there are almost noise free.

• "Messung, bin dafür bereit" -》"Ready for measurement " Messung bin dafür bereit.mp3
• "Mehrere werden komplett aus dem Land gehen" -》"Many will leave the country completely " Mehrere werden komplett aus dem Land gehen.mp3
• "So viele werden sich bei dir bedanken" -》 "So many will be grateful for you" Es werden so viele sich bei dir bedanken.mp3

The principle of operation was proved. The spirit reactions seem to indicate that they were trying to adapt and make measurements as a base for that. They also told me they understood the function of the magnet. One message I got was "Slow down the motor". So I was considering to make another prototype with a hand crank that would give me the opportunity to turn the magnet as slow as I wanted. The result was the following  device.

The results of this device were comparable to the previous one. The impulses were fewer but with a little more natural speech rhythm. However after applying Paulstretch the differences between both devices were negligible. I could extract lots of samples as well.

• "Muss Andres zu dir" -》 "I need to join Andres" Muss Andres zu dir.mp3
• "Schleudersitze zieSchleudersitze ziehen nicht.mp3hen nicht" -》 "Ejector seats don't pull"
• "Fernseher heikel ist" -》"Television is critical" Fernseher heikel ist.mp3

One very personal message I received referring to my dad who deceased some weeks ago. I got a message from spirits calling themselves "fathers" and they offered him to take him with them in higher realms. There were indications the leaving of the earthly realm was troublesome for him.

In the video above you can see and hear simultaneously the message.

5. Conclusions

The results are interesting  in respect of the fact that the rather slow and jumping alignments  of the nickel filings are causing not the same noise as pink or white noise in my other experiments. They're less agile thus providing a limited entropy with less degrees of freedom. Maybe this is causing less noise that is not used for voice manifestations  and thus a better signal to noise ratio.

The possibility to improve the signal quality is depending on finding better algorithms  for recovering speech from impulse signals. This is still a not satisfying situation.

Hört mich Andres.mp3