1
\$\begingroup\$

Consider a 1:1 transformer.

Short DC pulses are applied to the primary (input) coil. Once it's positive current pulse, once it's negative current. It's not strict alternating current (i.e. two successive positive pulses are possible). I want to block negative current on the secondary coil without using any diodes.

Is it possible at all?

I thought about two possible solutions:

  • connect secondary (output) coil to the solenoid densely wound around a permanent magnet. This way the current flowing in one of the directions will see the resistance, because it will be harder for it to build magnetic field that is in opposition to the magnet's field. Is it the right thinking?
  • connect secondary (output) coil to DC voltage source so there will be constant DC current in the coil. Positive induced voltage will pass, adding to the existing DC bias. Negative induced voltage will be blocked by the opposite current - or - will it subtract from the existing DC bias?

I'm looking for a way to pass positive pulses through the transformer and block the negative ones without using any semiconductors.

Clarification

Duration of all consecutive pulses is constant. Input voltage is also constant, between 1-5V, but it may drop after being passed through many transformers connected in series.

"Transformers" are supposed to be microfabricated (printed) on PCB as pairs of inducively coupled microcoils. Many of them will be connected in series. It's currently not possible for me to microfabricate semiconductor elements, including any diodes (only copper wires are possible). Also, any vacuum tubes will be too big to use them.

One possibility is to use microwires that are routed to one or few external components, but I don't want to mount an external diode for every microcoil, because there may be many of them.

\$\endgroup\$
4
  • \$\begingroup\$ I'm sure you have your reasons, but that sounds a little silly to me. \$\endgroup\$
    – Ignacio Vazquez-Abrams
    Jun 11, 2013 at 2:59
  • 8
    \$\begingroup\$ It's a terrible question. The restrictions on the valid solutions are silly and unrealistic. Nobody would actually design a real system like that. And you didn't include any information on the nature of the pulses (voltage, duration frequency, etc.). Even so, it is fun to think of the answers. Just don't make a habit of this. (Note: I have zero authority.) \$\endgroup\$
    – user3624
    Jun 11, 2013 at 4:18
  • \$\begingroup\$ Does a vacuum tube diode count as a semiconductor? \$\endgroup\$
    – Anindo Ghosh
    Jun 11, 2013 at 5:35
  • \$\begingroup\$ @AnindoGhosh: yes \$\endgroup\$
    – PSz
    Jun 11, 2013 at 8:52

4 Answers 4

Reset to default
1
\$\begingroup\$

Long ago before diodes, DC was made from AC by synchronous commutation. This usually meant a motor spinning at the AC frequency with contacts driven by the motor shaft so that connections from the input AC to the output DC were effectively flipped twice per line cycle so that the output was always presented with a positive or zero voltage.

You could possibly do something similar with relays switching on and off at the right times to capture only the positive spikes of your pulses.

However, why not use diodes? That is the obvious and simple way to fulfill your other requirements.

\$\endgroup\$
1
  • \$\begingroup\$ Well, I was trying to find other, possibly unusual answer and to check if it's possible at all - that's why I asked this question. But now I see there's no point to search any more. I will try to use small SMD diodes. The only thing I'm afraid of is the voltage drop on each of the diodes, which are connected in series through the "transformers". \$\endgroup\$
    – PSz
    Jun 11, 2013 at 13:21
3
\$\begingroup\$

Wire it to a DC motor. Put a ratchet on the shaft so the motor can only turn in one direction. Then connect the shaft to another DC motor that is being used as a generator.

That'll do it, but it will be sloppy. No more sloppy than the original requirements, however.

\$\endgroup\$
2
  • \$\begingroup\$ Can you make an electric motor without using any semiconductors? I agree that the requirements do sound ridiculous. \$\endgroup\$
    – user
    Jun 11, 2013 at 8:30
  • 1
    \$\begingroup\$ @MichaelKjörling Yes, you absolutely can make a motor without semiconductors in it. In fact, most motors are that way. Of course, all commercial motors were designed on computers but there are no semiconductors inside the motor. youtube.com/watch?v=xbCN3EnYfWU \$\endgroup\$
    – user3624
    Jun 11, 2013 at 13:39
3
\$\begingroup\$

I'm looking for a way to pass positive pulses through the transformer and block the negative ones without using any semiconductors.

\$\endgroup\$
2
  • \$\begingroup\$ The OP also says "without any diodes," so who knows. \$\endgroup\$
    – Bitrex
    Jun 11, 2013 at 3:33
  • \$\begingroup\$ +1 for creativity, but what if there is a long string of positive pulses with no negative ones. The positive DC bias of the input will be removed by the transformer and then you'll chop off the bottoms of the positive pulses. I think the OP hasn't given enough info about the pulses to do this "properly". \$\endgroup\$
    – user3624
    Jun 11, 2013 at 3:52
2
\$\begingroup\$

What you're looking for is known as a magnetic amplifier. they use dc controlled current coils that bias the magnetic core into different regimes of operation. They have the characteristics of handling very high power and being mechanically very robust.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.