Square wave pulse (0-2.5kV)&(0-100msec)

If you've got to ask how, don't bother. 2k5V is difficult and dangerous.

How difficult and dangerous depends on a couple of factors you haven't specified.

What is your load? Is it resistive, inductive or capacitative, or some combination?

How much current are you expected to be able to deliver.

What sort of rise and fall times are you aiming for?

--
Bill Sloman, Nijmegen

Those are good questions. I recently designed a 2500V amplifier for a group in the Physics dept - they're making 100 of them, four to a PCB. Because there are so many, the power consumption is an important spec, which relates to the other specs. As built, my amplifier draws 0.3mA quiescent, and delivers +/-4mA current-limit into a 4000pF load, yielding an i/C = 1kV/ms slew rate, which gives a 2.5ms risetime for a 2500V pulse. This would also be a very quiet pulse, which is another possible spec. Because 100 are being built, the components are low-cost SMD parts for machine assembly, which could be one more of ellendort's specs. The amplifier has a gain of 250, another spec, allowing one to create their pulse complexity at low voltages using ordinary circuitry.

IIRC, we discussed my design here on s.e.d.

--
 Thanks,
    - Win

--
That zero millisecond spec is going to be tough to meet.

"John Fields" skrev i en meddelelse news: snipped-for-privacy@4ax.com...

Nah. Just spec it with an integer ;-)

"John Fields" a écrit dans le message de news: snipped-for-privacy@4ax.com...

2500V during 0ms, and OV before and after?

That's the easiest part.

--
Thanks,
Fred.

The load is resistive which vary in 2 types from my samples

-(2000 - 5000)ohms

-(20-200)ohms

My aim is to apply a constant voltage on the load samples in msec in term of pulse by using 2 electrodes with the distance of 0.1mm or 0.2mm or 0.4mm with the arching controlability The current will be limited up to 125A extremely short-lived (time constant 500usec)

Rise time: approx. 500nsec Fall time: approx. 1-10usec

Pls advice

Sounds interesting, I hope you tell us what you're working on.

A 2.5kV source delivering 125A into a 20-ohm load is delivering over 300kW, which is slightly impressive,* even if it is with a quick 500us decaying time constant. A 25uF cap charged to 2.5kV holding 78 joules could deliver such a pulse. You could switch it with a small high-voltage IGBT, such as a PowerEx QIS4506001. These are rated at 4.5kV, and can likely switch up to short 225A pulses for lab purposes (or you can use two). In my experience these high-voltage IGBTs can be made to switch in 250-500ns, with enough gate drive. The QIS4506001 is $185.19 at Richardson.

Alternately you could use an IGBT module, such as a cm400dy-50h (rated at 2.5kV 400A but able to switch 800A with 12V gate drive), but it'll cost more and be harder to get. You may need to add some parallel load resistance to get it to turn off quickly for your 5k-ohm load case. I should mention that HV IGBT modules are easier to find with 1200 to 1700-volt ratings. [hint: eBay]

• I made a 1200V pulse generator that delivered 200A, 240kW pulses with a length of 10us to 100ms, with a 250ns risetime, using two small IBGTs in parallel, and I was impressed. :-) It had a fairly serious capacitor bank that didn't sag much.

--
 Thanks,
    - Win

Hi,

As you know my load resistance vary as in 2 types of voltages and pulsing time. I was thinking maybe i should create 2 systems high voltage and low voltage. (2000-5000)ohms - up to 2.5kV - up to 10msec (20-200)ohms - up to 500V - up to 100msec

IGBT's seems expensive i thought of using cascaded switching for example SCR 2N6404 600V (I can use 5 or 6 in series) they are cheap.

Since u have previous experience in designing "1200V pulse generator that delivered 200A, 240kW pulses with a length of 10us to 100ms, with a 250ns risetime".

Appreciate if you could guidance on how to go about. Especially square wave pulsing. Perhaps some reference from your design would be very helpful.

Hi,

As you know my load resistance vary as in 2 types of voltages and pulsing time. I was thinking maybe i should create 2 systems high voltage and low voltage. (2000-5000)ohms - up to 2.5kV - up to 10msec (20-200)ohms - up to 500V - up to 100msec

IGBT's seems expensive i thought of using cascaded switching for example SCR 2N6404 600V (I can use 5 or 6 in series) they are cheap.

Since u have previous experience in designing "1200V pulse generator that delivered 200A, 240kW pulses with a length of 10us to 100ms, with a 250ns risetime".

Appreciate if you could guidance on how to go about. Especially square wave pulsing. Perhaps some reference from your design would be very helpful. Please advice.

Hi,

As you know my load resistance vary as in 2 types of voltages and pulsing time. I was thinking maybe i should create 2 systems high voltage and low voltage. (2000-5000)ohms - up to 2.5kV - up to 10msec (20-200)ohms - up to 500V - up to 100msec

IGBT's seems expensive i thought of using cascaded switching for example SCR 2N6404 600V (I can use 5 or 6 in series) they are cheap.

Since u have previous experience in designing "1200V pulse generator that delivered 200A, 240kW pulses with a length of 10us to 100ms, with a 250ns risetime".

Appreciate if you could guidance on how to go about. Especially square wave pulsing. Perhaps some reference from your design would be very helpful. Please advice.

That would make good sense. 2.5kV 1.25A max, and 500V 25A max. At those new lower current specs, I'd consider using MOSFETs.

But you're asking for a "square wave" which implies turning off the pulse at some point. Once triggered an SCR stays on until the current drops below its holding current, usually about 10mA, so your 2 to 25A pulse would continue for many time constants, gradually dying away to nothing. But if that's OK, use SCRs.

Also, series switching is a dangerous business. One part always turns on first, placing excessive voltages on the rest. The last parts to turn on can get the entire high voltage, destroying it. After that part goes, the rest quickly fail as well. Then you have to replace the whole string, and the evidence of which part caused the failure was also destroyed. There are schemes one can use to help, such as adding parallel TVS stacks.

I used IGBTs rather than SCRs, because they can be turned off. Here's a fellow who used series IGBT switching, apparently without mishap, even tho I hesitate to recommend it for the inexperienced.

Perhaps it'd be best at this point if you told us what you're working on, what you're trying to do. Details, please.

--
 Thanks,
    - Win

I need to create holes in the cell membrane for drug delivery. By using constant voltages in term of pulse this will create electric field in between 2 electrodes. The cell will be places in between this 2 electrodes as the e-field will create the holes. Parameters vary from cell to cell as in type of the pulse, time, voltage and the electrode distance.

Generally this cells are the one divided into 2 categories (2000-5000)ohms - up to 2.5kV - up to 10msec (20-200)ohms - up to 500V - up to 100msec

As for the pulse i wants to try 2 types of pulse whichever successful. Exponential decay & square wave

This is why perhaps need to design the system separately as for the types of the cells and the pulse i suppose.

As for the exponential decay i'm using RC circuit as the fixed capacitor and the resistor will give the constant time (25mfd*400ohms=10msec). But i'm still facing problem in how to truncate this pulse as in to stop this pulse at various point for different cells. (0.2msec - 10msec)

And as for the square wave i believe need to use some switching to on and off controlled by cpu but yet still not clear in the design.

Pls advice

It doesn't matter that the cell has been lethally damaged in the process?

The latter example is easy, with the most difficult being 500V and 20 ohms, which implies peak currents of 25A.

Inexpensive SCRs or IGBTs can handle the switching for the 500V case, but once you've spent the extra effort to make a 500V 25A IGBT switch, why build a second SCR switch? But you may want different power sources. For example, a 100-ohm load with a 100ms time constant would imply discharging a 1000uF capacitor. If charged to 500V for a 5A peak current that's 125J of energy.

By comparison a 500V 5A = 2.5kW, 100ms rectangular pulse takes 250J of energy. That's only twice as much, but it requires a much larger storage capacitor bank, if you want to avoid much pulse-amplitude droop. For example, limiting the droop to -10% or 50V requires C = i t / dV = 10,000uF or 10x more capacitance. If you point out your spec required 20 ohms 500V and 100ms, now you'll need 50,000uF, and if you want 5% droop you'll need 0.1F at 500V, which, just checking, stores 12.5kJ of energy. That's not very realistic, especially for a tiny little cell membrane. So it would seem you need to more carefully evaluate and specify the extremes of your operating range specs. For example, do the lower resistances, with higher currents, need the full specified durations? Perhaps an energy-related spec would be better. For example, the 12.5kJ mentioned above is 100x the 125J number, and since capacitor size and cost is roughly proportional to energy, it'll cost you 100x as much to build. A 1000,000uF 500V rated capacitor will in fact be a huge bank of many large capacitors, I know, having built just such a beast.

Have you read up on IGBTs yet, as I suggested?

--
 Thanks,
    - Win

The cell damage yes. But the controlability depends on the parameters as in voltage current and time. If this parameter is not control properly then the cell will die. After this process if the cells are not damaged to the extreme they can be recovered back with antibiotics and etc, in 24 hours time which is call re-seed or re-culture process. So damage is fine.

A 20ohm resistor will be places in series with load to limit the curent. Therefore, peak current will be approximately 12.5A

I'm trying for exponential decay pulse first. I've purchased MOSFET for switching - 2SK2996. Single switch for capacitor charging and 5 switches in series for capacitor discharge.

2000-5000 ohms load capacitor 25mfd & 400ohm parallel to the load = 10msec 20-200 ohms load Eleven parallel 390mfd/500V capacitor in parallel with load, 20 ohms resistor series with the load, With load RC will give more than 100msec

This is the spec for the process. Lower resistance with lower voltage for longer time (20-200ohms, 0-500V, 0.02-100msec) Higher resistance with higher voltage for shorter time (2000-5000ohms,

0-2500V, 0.02-10msec)

specified

Cells vary from each other. The defined spec would cover most fo the cells.(Maximum voltage & time) Voltage and time too will vary accordingly to the cells. Different cell different parameters. In general can be divided into 2 types and from there we will be able to control the parameters according to the group that the cell belongs. But this does not mean all the cells need the max voltage and the max time. Sorry for the confusion.

That doesn't match any cells I'm familiar with. If the cell-wall membrance is breached, the cell is flooded with the intra-cellular soup, with lethal ionic concentrations, and the cell dies. It doesn't recover. If other cells spring up (re-seed?) then they don't have the holes you refer to. When we want to make a holes in the cell walls, we use molecules to do the task, with known holes sizes and properties. So I really don't understand your scheme. Do you have any literature references for us?

OK, five 2sk2996 in series. I thought those were small-die RF transistors? You seem to feel you know what you're doing, so report back to use when you get it working, or if not.

--
 Thanks,
    - Win

Face it- most OPs are cracked in the brain. Antibiotics to repair cell structural damage? Too much...

Yes the process that you describe is one of the method perhaps through chemical using calcium or shotgun method. But my mean here is different method using pulses.

You may search for "electroporator"

Frankly i understand the theory but having difficulty in constructing the circuit. Just trying out. That's the reason i'm here hope to find some assitance from someone who experience in this.

I'm sorry my explanation is not in term of molecular biology and they might not be correct i just describe roughly.

OK, thanks for teaching me something new. I must say it comes as a surprise, something from another field that raises my eyebrows. The "Principle of electroporation" on page 6 of the Eppendorf Brinkmann 2510 Operating Manual,

also raised my eyebrows. A three-panel cartoon, figure 4, shows how it works: In the first panel we have a soup of DNA and bacteria, the second panel is a lightning strike, labeled 2500V 5ms, and in the third panel we see the DNA inside the bacteria. Voila. Simple -- magic.

--
 Thanks,
    - Win

:)) Yes it's a simple theory. That's exactly what i'm trying to do.