262Views19Replies

Author Options:

Is it possible to make a specialized spark gap device that operates similarly to a transitor? Answered

I have a serious issue with utilizing an H-bridge to switch polarity of high voltage electromagnets. The voltage is roughly 110 DC rectified from mains. Every mosfet I use has exploded when I tried to switch the mosfets. I tried using relays but they keep arcing and tripping my breaker. I would very much like to know if it was possible to make a spark gap that works similar to a NPN transistor.

Discussions

0
None
dsanco

3 years ago

You are looking for a vacuume tube. Generally more robust than the silicon counter parts. They have a grid separating two terminals to control current passage between them.

0
None
steveastrouk

7 years ago

You're killing the things because you aren't handling the inductice kick-back properly.

Correct H bridge operation is

1       2
   -----
3       4

1.) 1 And 4 on
2,) 1 and 2 on
3.) 3 and 2 on
4.) 3 and 4 on.

2-3 happens ONLY when I =0 and 4-1 happens only when I=0

You CAN try putting VERY big diodes across each mosfet or relay contact. When I do it, I use a whacking big bridge rectifier for 4 nice, heatsinked diodes.

Steve

0
None
50ul84n3steveastrouk

Reply 7 years ago

I believe that the nature of my problem is being a little skewed. You seem to be under the assumption that I am having trouble with the kickback. What I really have a problem with is that during the switching of relay 1 and 4 to 2 and 3 the resistances of the airgaps between the contacts is less than 60 ohms thus allowing the electricity to short circuit from relay 1 to relay 3. The real question I wanted input on is that if a spark gap were to be made at 1.25 the distance the arc can occur, and a plate were placed halfway between the anode and cathode would applying a charge to the plate allow the arc to jump between the 2 terminals through the plate?

0
None
steveastrouk50ul84n3

Reply 7 years ago

That is PRECISELY when the inductive kick will happen, and is clearly happening. The inductors do not want to change state !

0
None
50ul84n3steveastrouk

Reply 7 years ago

They are already rectified and so therefore can not send a reverse voltage through the positive leads. There is also a diverter that I use to divert the kickback to ground but it only works if the relays can reach a resistance of >60 ohms before an arc is started. I also tested the circuit without the electromagnets and still get the same result so if you want to still claim it's a problem with the coils go ahead, but if my problem were that simple I would have gotten the answer from wiki.

0
None
steveastrouk50ul84n3

Reply 7 years ago

Post your circuit. Where is this "rectification" acoss the coils ?

Steve

0
None
50ul84n3steveastrouk

Reply 7 years ago

Here is my H-bridge, and it's not worth the breath to chide me for using paint.

H-bridge.jpg
0
None
steveastrouk50ul84n3

Reply 7 years ago

As I suspected, the diodes are wrong. If you left the diodes in when you tested without the magnet, I am not surprised you blew the mosfets ! Which Mosfet did you use ? Did you check the Vg max ?

Steve

0
None
50ul84n3steveastrouk

Reply 7 years ago

the Vg max was somewhere close to 200v but I never used the mosfets with this exact circuit. I ran a switching test to double check the bias, when the mosfet exploded I chose to use relays instead. I know exactly where my problem is and the question I posted was whether a pseudo-vacuum tube / spark gap transistor were possible. I don't like the idea of using MOSFETs primarily because there will be absolutely zero heat dissapation capacity along with the circuit operating in a high ambient heat environment.

0
None
steveastrouk50ul84n3

Reply 7 years ago

If you know your exact problem, you don't need an answer. I am telling you that the diodes, as drawn, are connected wrongly, and won't work. You are only switching an electromagnet, and not igniting a nuclear weapon.

Steve

0
None
50ul84n3steveastrouk

Reply 7 years ago

What would you suggest I do to correct the arcing of my relays? You are incorrect about the diodes being wrong simply because the electromagnets work until the relays are triggered then relays flash a bright green and my breaker trips. As to using spark gap transistor idea I do need the precision and modular scalability. Providing that data is favorable the system will need to be scaled to function in the Kv range with megahertz switching speeds.

0
None
steveastrouk50ul84n3

Reply 7 years ago

Its also occurred to me that you are assuming that one set of contacts is fully released before the others close - that's not true with relays, unless you are VERY careful to design delays in. So you are crow-barring your supply that way, as well as adding sustaining arcs, because the diodes are wrong.

Steve

0
None
steveastrouk50ul84n3

Reply 7 years ago

You're telling me that everything is right. I am telling you, from the benefit of 25 years working with power electronics, that something is wrong.

Here is the correct layout. Your diodes need to be big and fast.

H bridge with relays.JPG
0
None
kelseymh50ul84n3

Reply 7 years ago

Paint is perfectly fine for line drawings! However, it is worth our breath to chide you for not cropping the image, making all of your annotations utterly illegible.

0
None
NachoMahma

7 years ago

.  Keep in mind that an inductor (eg, electromagnet, transformer, &c) will have a large inrush current.
.  There will also be a large spike when the coil is de-energized that needs to be clamped.

0
None
50ul84n3NachoMahma

Reply 7 years ago

I have already designed a circuit to divert back current to ground but the problem resides in the fact that the arcing in my relays is creating a path that has less resistance than the rest of the circuit so it bypasses my circuit.

0
None
kelseymh

7 years ago

Have you tried using solid-state relays rather than reed-type? I would think that 230V 90A might be sufficient to not "explode."

0
None
50ul84n3kelseymh

Reply 7 years ago

I would have already opted to use these if it was only within my price range. 4 of these would set me back at least $200 which is not very cost efficient considering that this is for a simple proof of concept project and not a fully funded production model.