Need help with 1000 VDC high speed switching circuit

[eeprom]

Hello,
I need to create a rotating electric field using DC. In summary, imagine 8 vertical steel rods arranged in a circle, with a 9th steel rod in the middle of the circle. Like a small circular fence with a rod in the middle of it. I need to energize each of the outer rods at 1000VDC with the middle pole at DC common potential. The rods have to be energized one at a time for 5mS at a time. There will be almost no current flow. The purpose of this is to create an electric field.

So, how does one create a high voltage high speed switching apparatus that is safe? Any ideas?

 

[MacGyverS2000]

First thing that comes to mind is a DC motor moving the rods you need...



Though "high-voltage" and "high-speed" rarely go together with "safe"... I guess it depends upon your definition of "safe", then.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Comcokid]

5msec for switching electronics is not high speed.

Now, if you can handle about 0.2 to 0.3 msec delay, and you're not concerned with your rods possibly becomming shorted, then there are available solid-state relays that can do the job. These relays are in dip packages, but this will still be easier than switching with N-channel MOSFETS, and designing a gate-drive that will be above ground by over 1000V.

Look at Panasonic AQV258 or AQV259, 1000V and 1500V 30mA and 20mA devices. Theses are photo-MOS devices. The control side is basically a LED.

On the output of these I would put a resistor to ground to discharge the rods. Otherwise, your rods will be capacitors that just charge-up to 1000V and never discharge before becomming energized again. You may want to measure the capacitance of the rod to size the resistor for the discharge.

If the rods are really, really long, or very close to the common rod, then the capacitance could be high enough to create a discharge issue, depending upon your actual application and rate-of-discharge needed. In that case, you may need to use additional relays (with a resistor to limit max current below the relay rating) to discharge the rods.

 

[eeprom]

Thanks for the note on the panasonic MOSFETS. I looked at some similar chips on digikey yesterday. I am stumpted. How can you have a standard sized dip package with 1500V between pins. I'm trying to keep all my rods 1 cm apart, and I'm pretty sure they are going to arc. How can you separate 1500V by 0.1 inches?

Thanks also for the note on the resistor on the rods. I had not thought of that.

EE

 

[VEBill]

...I'm pretty sure they are going to arc...

1000 volts, 1 cm, shouldn't.

 

[Skogsgurra]

Remember, up to 1 kV is still LV (Low Voltage) in motors.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[eeprom]

I thought 600V was the cut off for low voltage.

 

[djs]

Tongue in cheek how about a rotating bar the comes in contact with the barrs in the outer circule? Similiar to a communator on a DC motor except that the brush rotates not the communator.

 

[MacGyverS2000]

Given the OP's description of his desired setup, a moving motor-type machine is what I envisioned. Others here have suggested more "solid-state" solutions, but without knowing what the OP is really after, I chose to err on the safe side and stay close to his idea of a moving machine.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[eeprom]

I've got an old distributor cap and rotor from a truck. I'll connect the rods just like spark plugs, and I'll drive the rotor with a small motor/vfd. In theory this will work.

 

[Skogsgurra]

If you look inside the distributor, you will see a gap between rotor brass and each plug connection. I don't think that the 1 kV can jump reliably across that gap. You may need to add a conductive material to the rotor.

LV, MV and HV - it probably depends on where you are. The rule here (EU) is that all machines up to 1 kV are LV machines. May be different where you are.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Comcokid]

The voltage isolation of DIP packages has sometimes been an issue when you consider board contamination, age, etc. In the past it was always an issue in UL design for optoisolators where spacing had to be 0.25", and by the time you drilled a hole and added pads for a 300 mil dip you could be under the minimun.

If you're concerned, use a conformal coating. The coating could be no more than an appropriate RTV brushed over the part and on the opposite side of the board where the leads come through. You could also bend the leads outwards and make the dip into a gull-wing surface mount device.

 

[VEBill]

"...appropriate RTV..."

Appropriate means non-acidic. If it smells like vinegar - don't use it on electronics.

 

[eeprom]

I already have the distributor/rotor plan in motion. It should be running by Monday. So if that doesn't work, and I can't make it work, then I will try the solid state approach. I think the idea of coating the PCB with an insulator/sealant is an excellent idea. I have some film used for sealing photovoltaic assemblies. It is a dupont material (PV5300) with isolation something like 2.5x10^16 ohm-cm resistivity. It is very easy to apply, and would work well for the RTV.

So apparently those little Panasonic dips can actually work. It's hard to believe. But they are only a few dollars each. I think it would be worth $150 to find out.

Thanks for your suggestions.

EE

 

[OperaHouse]

Have you thought about a resistor to drain off the charge? A charged bar will not just go to zero when the contact is removed. That would allow for another solution, If the capacitance is low enough each bar chould be charged by a pull up resistor to 1000V. A simple FET then could just short out the inactive bars,

 

[itsmoked]

More common than "smeared something" is to have little smiles routed across the no-mans-lands areas. This removes the variability of anything smeared.

Keith Cress
kcress -

http://www.flaminsystems.com