Friday, August 15, 2025

Elimag Watch Demagnetizer Analyzed

 The ELIMAG Watch Master demagnetizer from about 1958.

 
In previous posts to this blog (here and here) I wrote about my experience with various watch demagnetizers. I found this vintage Elimag worked the best for me and I was very curious to learn more about how it works.
 
Here are a couple pictures of my unit, note the handy instructions printed on the bottom. 


 
I am guessing ELIMAG is an abbreviated contraction of ELIminate - MAGnetism. The ELIMAG was invented and patented by Frank Dostal, the chief electronics engineer at American Time Products. Frank obtained many patents related to watch making, precision timing and other devices. American Time Products eventually became a division of Bulova Watch Company. 
 

What's inside the ELIMAG?   

 
First I should caution the reader:
This is an old device, constructed long before current safety standards. 
DO NOT OPEN THIS UNIT. 
DO NOT PLUG IT IN. 
DO NOT USE IT. 
It operates at extremely high voltages and can be very dangerous. 
Internal voltages may exceed 675 Volts DC!
 
If one was going to open it (and you shouldn't unless you are an expert with high voltage electronics) you might consider doing the following:
  1. UNPLUG THE UNIT!
  2. PUSH THE ACTIVATION BUTTON SEVERAL TIMES TO ATTEMPT TO DISCHARGE THE CAPACITOR INSIDE! Note this may not work if the switch is broken, the lamps is broken, or the circuitry is broken. Proceed at your own risk: DANGER!
  3. SHORT ACROSS THE LARGE CAPACITOR TERMINALS WITH AN INSULATED HIGH VOLTAGE RATED TOOL!

Here are some pictures of the inside of my unit:




Schematic 


Here is the schematic from Frank Dostal's US patent #2,838,720. Note this does not exactly match the actual circuit inside my unit. 

Below is a schematic I reverse engineered from my ELIMAG. It is not surprising that it does not perfectly match the patent as products often undergo changes as they are optimized for production. An inrush limiting resistor R1 has been added and the bulb P19 has been moved in the circuit so that it no longer shows that the unit is powered but instead indicates with a flash when the device is activated. The resistor  R2 has also been added to limit voltage and current through the lamp.
On the left is the AC source (the AC power plug for 120VAC RMS), It feeds the device through a resistor, R1, that limits inrush current. 
The dashed box, 20, contains a voltage quadrupler circuit that I have seen called a "Dickson Voltage Multiplier". When the switch is in the default parked position, the large P18 capacitor is charged. Pushing the momentary switch dumps that charge to the P6 solenoid inductor and the P19 neon lamp. 
I initially was confused about the source voltage and the multiplier. Note that US power of 120V RMS at 60 Hz is actually about 170V peak and 340V peak to peak. 
For the multiplier: 4 x 170V, minus the diode drops of about 1V each, results in about 675V to charge the large P18 capacitor.
P16 exists to limit the current that charges the large P18 capacitor and reduces any possible spikes that could result from actuating the switch.

Here is the netlist I constructed from this schematic:

The solenoid specs

If I am reading the patent correctly, the coil described there has an outside diameter of 1-3/4". The coil is wrapped around a cardboard cylinder for an unspecified internal diameter that I estimate at about 1-1/4" because the cardboard cylinder fits over a core with a diameter of 1". The height of the coil is also unspecified but likely around 3/8" as it must fit in a recess 1/2" deep. The coil is shown fitting into a compressed iron powder block with an integral post so that only the top side of the coil is exposed. 

This is much different from the solenoid found in my unit. These are my approximate measurements of the solenoid:

Core: solid iron-powder cylinder, Ø25 mm × 8 mm long

Coil: ID 25 mm, OD 55 mm → radial build 15 mm, axial length 8 mm

Wire: AWG 26 Cu, bare ≈ 0.405 mm; with thin enamel OD ≈ 0.44–0.48 mm

I should note that the value of the solenoid inductor is a very rough estimate. I just gave ChatGPT these roughly measured dimensions and it estimated that the solenoid would be ~11–17 mH (depends mainly on exact wire OD/turns and the mix’s μ) and have a resistance (20 °C): ~8.3–10.3 Ω.

Simulation of Operation 

My unit takes about 15 seconds to reach a full charge of about 490V. Note that ChatGPT estimates the full charge should be approximately 675V. I suspect the capacitor in my unit is leaking and limiting the maximum charge. The rectifiers also look to be in rough shape with some evident corrosion.

When you plug the unit in, the capacitor charges. There is no on/off switch to disconnect the power source, be wary!

Here is a chart from chatGPT showing the theoretical capacitor charging. It estimates the capacitor could get to 675+ V DC! 

My unit actually only reaches about 490V and takes about 15 seconds to get there. 

Pressing the button on the unit operates the demagnetizer and flashes the neon light.

Assuming a full charge of 675V, here is the resulting voltage across the solenoid when the button is pressed:

I measure the actual charge across the capacitor in my unit. It reached 490 volts after about 15 seconds. Using 490 V in the simulation, this is the resulting voltage across the solenoid when the button is pressed:

The simulations show the voltage oscillates (rings) at about 670 Hz.

I don't have an oscilloscope available, so I cannot verify the accuracy of these simulations. Remember that I just estimated the specifications of the inductor! But these results are still interesting and somewhat match Frank Dostal's claims in the patent:

"On the passage of the current through coil 6, the oscillatory circuit comprising coil 6 and capacitor 18 will oscillate for a short interval, of perhaps 1/20 second "

"the voltage multiplier 20 shown in Figure 4 is shown as a voltage quadrupler, and hence for a line voltage of 110-120 volts at 60 C. P. S., will result in a charging voltage of some 440 to 600 volts for the capacitor 18"

"l have furthermore determined that the frequency of the oscillatory circuit, coil 6 and capacitor 18, is not critical although I prefer to have such frequency above 200 cycles per second to eliminate all danger of injury to parts of the watch movement. Winding the coil 6 with enameled copper wire of number 26 gauge and to the dimensions above given, and with capacitor 18 of some 4 uf., the frequency is some 500 C. P. S. and is most satisfactory. I have also found that a time interval of some five seconds between successive operations of the switch 17 is advisable to give the capacitor 18 ample opportunity to be recharged after a discharge." 

"It will be noted that the demagnetizing operation of a watch with the instant demagnetizer is simple in that, the demagnetizer having been connected to the supply line, the watch is placed, preferably with its face down as above stated, on the sheet 5 and the button 17 pressed and then released. Complete demagnetization is thus a matter of 1/20 of a second or so"
 

Safety of this Vintage Demagnetizer

To be clear, by modern standards, THIS VINTAGE UNIT IS UNSAFE!

I am not an expert but I see several significant problems. An expert would certainly find more.

  • The unit has no ground.
  • The unit has external metal parts, the switch and bulb housing, that could become energized by failures inside the unit.
  • There is no power switch to disconnect the unit from mains power.
  • The large capacitor charges to dangerous voltages.
  • The large capacitor charges whenever the unit is plugged in and has no way to discharge if the unit is unplugged.  (My particular unit does discharge slowly, probably due to capacitor leakage.)

If you have one of these demagnetizers, I would suggest you only keep it as a museum piece, unplugged. If you do use it, make sure you unplug it after every use and push the button afterwards to discharge the capacitor.

For more Information

There is a great thread about demagnetizers on the National Association of Watch and Clock Collectors forum here

Friday, August 8, 2025

More on Watch Demagnetizers - Strength Matters.

After a recent trip, I found my new Hanhart was gaining about 40 seconds a day. I checked it with a magnometer app and a compass and it had become magnetized.

I did have a couple cheap plastic "demagnetizers" and I have used them successfully on my Omega (see my previous post).
But I cannot get them to work on the Hanhart. Maybe the Hanhart is too well shielded, as it is specified and marketed as "antimagnetic".

So I brought the watch to two watchmakers... both didn't want to touch it because it was under warranty. That was strange to me and I essentially begged them to put it on their demagnetizers and they refused. Maybe they just didn't want any liability for an attempted repair, especially knowing the watch was still under warranty.

So I went to eBay.
First I bought a nice vintage South Bend unit:



Testing the South Bend with a small screwdriver, it was CLEARLY more powerful than the cheap chinese units I had tried. So I put the Hanhart through it many times... and sadly it didn't help.

Somehow, I remained convinced the watch was magnetized, as it's symptoms perfectly matched that. No beat error, good amplitude, but 40+ seconds a day too fast. I still believed the "antimagnetic" shielding of the watch was just making it harder to demagnetize.



A bit of research showed that push-button "instant" demagnetizers are significantly more powerful that the others. So I searched for and found an old Elimag that was claimed to still be working:



The Elimag is claimed to be very powerful, as it uses a voltage quadrupler circuit, meaning it charges to more than 400 Volts. I pulled the crown out (stopping the watch) and put the Hanhart on the Elimag face down and pushed the button. For the heck of it, I did it again, and then on the side crown up and crown down.

And then tested the watch on my timegrapher.... FIXED! +3s/d face up! Yay!
Watch is working perfectly.



Beware of magnetism. ;-)

Tuesday, April 1, 2025

Watch Demagnetizer Testing and Demonstration (also works with small tools)

 My newly serviced mechanical watch suddenly gained half a minute overnight. So I checked it and found the timegrapher showed more than +20 seconds a day.

A quick google said the likely cause was magnetism and a compass did show it was slightly magnetized.

So I bought two cheap watch demagnetizers on Amazon. A blue "Tui Ci Qi" unit for about US$13 and a white "HUAXING" unit like a mini MRI machine for about US$20.


I put both of these devices on a power meter, and found the blue one used about 7.6W and the while one about 23.3W. And then I played with both, magnetizing and demagnetizing a small screwdriver. In my opinion the white unit worked better, but it could be the technique. 

For the blue unit, most recommend putting the watch/tool on the pad, turning the unit on and then slowly lifting the watch/tool up and away from the pad for more than 24" (60cm) before turning the unit off. But some recommend rubbing the watch/tool back and forth and around on the pad. I tried both with my magnetized screwdriver with variable results.

For the white unit, most recommend putting the watch/tool in the unit's hole, turning the unit on and then pulling the watch/tool out and away from the unit for more than 24" (60cm) before turning the unit off. However, some people recommend passing the watch completely through the unit from one side to the other.

For either unit, multiple passes could be used to possibly improve the results. I found the white unit consistently demagnetized my screwdriver in one pass.

I made a video showing a demonstration of the white unit:

I did disassemble both units, and both are VERY crudely made. I highly recommend only plugging these in when using them and unplugging them when done! Also, the manufacturers recommend only uses them for a very short time before letting them cool off. The white unit says to use it for 1 minute maximum.

Interior photos for those that are curious.



Oh, after a pass through the white demagnetizer the performance of my watch is restored. ;-)