Developing HV DC Pulses

<snip>

What you are dong is electrophoresis

If your oil is dispersed in water into really small droplets each drop of oil ends with a small electric charge - the water in the vicinity of the droplet organises itelf into an electric double layer, and an electric field in the water moves the oil droplets one way and the water the other.

You need to know how much current you need to feed in to keep the droplets moving - the droplet charge ends up flowing into the electrodes.

If you want a pulsed electric field with relatively long pulses - 10msec or longer - it makes sense to use something like Cockroft-Walton circuit

to build up a high DC voltage using a relatively fast-switching AC source and small capacitors over a few msec, then turn it off again and let it decay before you build it up again.

Getting a long high voltage pulses directly means big and expensive capacitors and inductors. It doesn't make sense.

It's sort of in the ball-park the high voltage inverters Baxandall Class-D oscillator was developed to drive

Bill Sloman, Sydney

What is the load like electrically? How do you connect to the emulsion?

So far, there is no connection to the emulsion, it is just an electric field. Here's a quick drawing showing the setup.

There is some evidence that people use electrodes (grids) in the emulsion, and that will probably be tested at some point. But, as yet I don't have info on what voltage or current to expect with electrodes in solution.

I'm hoping get my son to do a capacitance/resistance test of the tube and emulsion this week. He has other job duties and they need to be done. I'm throwing lots of ideas at him and just waiting to see what he decides to do. MIkek

OK, no liquid conduction, just a bit of capacitance.

A simple flyback circuit would work: a DC supply, a mosfet or SiC fet, an inductor, maybe a diode or two. Possibly two fets.

My Pockels Cell driver is a sort of flyback that recovers the energy from the capacitive load.

I've been thinking about a similar case, driving pulses into a pretty hunky capacitive load with minimal parts and dissipation, but bigger c, lower v, and higher f.

What's a reasonable pulse width and frequency?

That would dissipate a lot of power in the liquid, maybe boil it.

Does ionic conduction have time lag?

But would that allow any pulse width adjustment?

One article references 8Hz as the optimal frequency, but no info on pulse width. I'm suspecting 8Hz is not really the optimum, but had more to do with the pulse width they had at that frequency, but, I don't know how they developed their pulse. I also have a feeling different oils will require different frequency, pulse width and voltage. On the other hand, I'm way over my head on both the electronics and the physics of water/Oil droplets. I'm winging it as best I can. I just read coalescence of (water) molecules only happens during rise and fall time of the pulse, so maybe short pulses are fine, but it is a physical movement so it takes time. (I don't understand why on fall time!)

I have now idea about the voltage or current with electrode on an emulsion.

So, yes, I don't get it.

As always energy use is to be minimized in commercial use.

So the line going from the transformer through the top of the centrifuge tube is not a wire? That's an electrode in the emulsion.

It will conduct, probably a lot. It has a large portion of water, no? Remember an emulsion is tiny drops of oil in the water. Even if it is only half water, that's enough to complete a circuit. The centrifuge tube forms a capacitor, between the emulsion and the aluminum tube, with the centrifuge tube as the dielectric. In this case, the separation is 100% due to the conduction of the AC current through the emulsion.

You might do better using a higher frequency, rather than a higher voltage. But then your half wave rectification tests might argue against that. Not sure how you might easily generate higher frequencies than 60 Hz. Can you just interrupt the circuit with a transistor, or a tube?

Are you talking about two electrodes in the emulsion? If you mean your current set up, the capacitance will be low and the resistance probably inconsequential. I'm saying the RC filter formed will be dominated by the impedance of the capacitance at 60 Hz.

Good luck.

How did they determine that?

I wouldn't worry about any details as yet. To optimize anything, you first need to characterize and understand the issues. Then you can worry about energy and such. Seems to me, you are a long way from having any idea what is happening or how it might be optimized.

You got me curious enough to do a search on electrostatic oil water emulsion separation, finding this:

Table of contents links don't work, scroll down and expand for some discussions of theory and equipment, tiny images can be expanded. Apparently if the emulsion contains less than 4% water it will be nominally non-conductive, and this method appears to be in use only for removing water from crude oil in combination with other methods at the moment. It seems like it might work in other applications if at least one electrode is insulated, probably also in combination with other methods. Wavy plate coalescer with every other plate insulated perhaps? Results no doubt heavily dependent on the nature of the emulsion to be separated.

Glen

I suspect the reason for pulsing is that you can excite water ions to avalanche, and want to have some 'off' time to allow excess ions to recombine. Not sure what the rate is, but it depends only on water's relaxation, which should be possible to look up. I suspect you want to wait milliseconds, not nanoseconds... and pH control is going to be important.

The reason for DC is that you want an upper field limit (because the velocity produced is moderated by viscosity limits and subsequent heating) for modest acceleration of the droplets. That's going to depend on droplet size distribution. A good supply of known distribution goo (big jar of mayonnaise?) will be useful. You can make your own mayo, without the vinegar or lemon juice.

A good vacuum tube circuit can handle a kilovolt of switching, and is easy to drive; for experimenting, in the likelihood of runaway currents, that's be better than solid state for this. A variable HV supply, big resistor, and tube to chop the voltage down to zero... should work.

Uhm, it is an electrode centered of the glass tube in the emulsion, made of 12gauge copper wire.

Well, that was a new thought I had while napping (because I didn't sleep last night thinking about this) I used a fullwave bridge, a halfwave bridge is kinda like a DC pulse, i.e. it give a rest time between electric fields. I'm not clear what the rest time does, but it is easy enough to make it a half wave DC.

Hmm, I found a Cylindrical capacitor calculator, on memory of tube size the vessel without emulsion, >air dielectric<, is 2.5pf to 3.5pf. Water has a permittivity of about 80 (depending), oils, 2.2 to 2.9, But, 70% oil, 30% water emulsion? I hope to find out some time this week.

Mikek

No idea, Quote; "In the presence of a pulsed DC field, coalescence can be observed only during rising and falling edges while no coalescence occurs in the middle of the pulse width" reference to a paper with Chinese all over it but researchers name is Taylor.

I followed up on the journal, Chemical Engineering Research & Design, Vol.74, No.5, 526-540, 1996, the index stops at Vol.74 No.8, doesn't go back to No. 5. Oh, found it would be behind Elsevier pay scheme even if available.

:-) Not concerned, just noting that lots of power is not good.

I have not seen that reference, the section, "Emulsion-Treating Equipment" does go into (electrodes) grids submersed in the emulsion. Thanks, Mikek

I have a few MOTs salvaged, but we can run the neon sign transformer on a variac and get up to 9000Vac output and I can put a half-wave voltage doubler on it. Also, it is self limited to 30ma of current. Thanks, Mikek

<snip>

Ions in water don't move all that fast.

<snip>

I have worked on HV pulsers for field asymmetric ion mobility spectroscopy (FAIMS) that used resonant coils to produce the first few harmonics of a DC waveform. While this work was not published, here is another approach (which I tried, but didn't work):

?repid=rep1&type=pdf&doi=398cb8bdd459203c8358449135f8e33473c15b59

Thanks for that, I'm don't really understand what it is doing, and I think the high side low side driver gets complicated. Although later down the line it might be worth having someone more knowledgeable look into that, It's in the file. Thanks.

I was shown this much simpler approach and want some feedback on it.

using these 1400V transistors, My concern, would improper layout add an inductance that could over voltage the devices at shut off? The resistive voltage divider, How much voltage can you put across a resistor? Will a 1W withstand a higher voltage the a 1/2 W? (Standard resistors) I understand there are HV resistors. Thanks, Mikek

<snip>

Read the data sheets.500V for an axial lead resistor is pretty typical,

Mostly - bigger parts have longer paths from one end to the other.

There are, The search engines I looked at didn't pick them out, but any rep should be able to tell you about them. The ones I saw from time to had the resistive element on the inside of a relatively long - couple of cm - glass tube. Sinking them in transformer oil or sulphur hexafluoride helps, but hardly anybody does.

The only example I ever ran into was in a 600kV transmission electron microscope, built for the Melbourne University Physics Department in the 1960s by Alex Strojnik from Croatia (his son seem to have done something similar in Arizona more recently).

Photomultipler power supplies tend to use regular axial lead resistors, but they don't go much over one or two kV.

You can easily get a single 1000 volt mosfet. A passive (resistive) pullup would be all you need to drive a small capacitive load at low duty cycle. So, given a HV dc supply and a pulse generator, 555 or something, you need two parts.

If an exotic HV resistor is a hassle, series a few 1-cent axials.

Adding active pullup is easy too. 1 or 2 more parts.

He does seem to want to get up to 9 or 10kV. A 1kV MOSFET isn't all that interesting in that context.

Since his shortest pulse is 10msec, it's not offering him anything that he actually needs. Do try to answer the actual question rather than the one you'd have preferred him to have asked.

But mount them on teflon insulated stand-offs.

But not at 9kV.