[ARCHIVED THREAD] - Electronics Question (Page 1 of 2)
Posted: 5/13/2004 5:30:09 PM EDT
|
I need to create an 8kW burst for <.001 seconds. Probably 200V and 40amp. How would you suggest that I go about doing this? Gated? No, I can't provide any more information about the purpose, because I don't know it. I just got told to figure it out. I can probably update tomorrow with more info. Edit to add: budget is not a problem. Thanks |
|
it is a lot of current. However, I'm pretty sure that those two parameters are required for the application. [I'm still trying to get more info on the application] What i'm looking for is a description of a setup with whatever components are required to get that burst. |
A 12V one farad capacitor is about the diameter of a coke/beer can, and the length of a can and a half. Got one in the garage somewhere... I have a couple 115V 480mF caps somewhere, they're fatter than a coke can, but shorter. Jonathan |
It depends on the voltage it can take. You can get a 1 farad cap that is smaller than a coke can. And remember, I'm from Texas so everything is a coke. 
|
|
If I remember(sorry, it has been 10 years), you can use a semiconductor hockey-puck like switch that is activated with intense light - like from a laser burst. But, for your requirements, I'll bet you can get away with a really nice FET and switch it on for your pulse time. here is something I just Googled: http://pulsedpower.usc.edu/newpage/muri/cp3muri.shtml Now on to the basics. Power is energy per unit time. That is Joules per second. A kW is kJ/s. For your time of 0.001 seconds, you need to dispense 8J x 2. Under ideal situations, the source impedance must equal the load impedance which is why you need twice the power - 50% dissipates at the source and 50% dissipates at the load. Energy in a capacitor is 0.5CV^2. Defibrillator capacitors, used to restart the heart, are fairly large and may work in this case. Now the next trick is to have a low source and load impedance to dissipate quickly, AND take into consideration the RC time constant for the discharge rate. Also make sure the capacitor can handle sucha a rapid discharge. What is the rise-time requirement of your pulse? (ps: don't forget to include the loading of the FET on the circuit) Also, if you are sending this down a long enough line (see your rise time) then you might want to be concerned with impedance matching, etc... edited - Oops, "minor math error" |
|
He wants 8000 watts of power (80 100 watt lightbulbs) lit full, for 1 thousandth of a second. The other guys are telling him that most likely, a capacitor (an electon storage device, used in electric circuits to quickly deliver power, like a battery than can completely drain in .1 sex) is the easiest (maybe) way to do it. The others are using ohms law and a standard power formula to determine how much voltage (think pressure in a water hose). current, (think flow in a water hose) and resistance is necessary to get 8000watts for a very short period of time. Like an ignition source. TXL |
|
If you use capacitors they MUST be able to handle that fast of a discharge. You can short them but the transient response will probably be larger than .001 seconds. The large ones for car stereo systems are not (from what I have seen anyway). My personal vote is for the
Check the response time on the transistor. |
We probably need to know a little more, like: Must the pulse be less than .001 seconds, does it repeat at regular intervals, is the waveform important, how accurate does the output have to be, what is the power source? Knowing what the application is would be nice.
|
|
Philadelphia Experiment? Ok, for exactly 1/1000th second burst: You can use a crystal oscillator to generate a square wave at 1Khz (you could even use a mean $0.30 555LC timer). TTL or CMOS logic have no problem with this, so you need a counter and then a couple logic chips to create the desired time. (Actually, you should clock it at 10kHz and go for 10 steps for accuracy). To interface the logic, just use the good old transistor pulldown driving a MOSFET (I am assuming this is DC? Otherwise a TRIAC solution would be needed). There are many MOSFETS Rated to 200V and Extremely high current, go 20% over in current handling. Depending on Duty Cycle (how often do you need to generate this pulse?) you could "under rate" the MOSFET. Use a seperate power supply to run the clock/logic than for the pulse, or have an extremely well filtered/stabilized power supply on the logic side. Need a schematic?  -- Edited to add, if you play with Microchip PIC Controllers, all of this could pretty much be done with one logic chip and support components (crystal oscillator, power supply, MOSFET Trigger external). The assembly language to control these is relatively simple. I forgot to ask, is this an AC or a DC pulse? DC is the easy one, AC is difficult, depending on frequency of the AC (i.e, are you dumping a microsecond pulse of RF???) |
Showoff. |
|
Yeah, knowing to application would help. Other wise just use a cap, 1 ms really is not that fast so you can getaway with a fairly cheap cap. Ceramic or Tantalum would be best for higher frequency. You would just a transformer to charge it. The bad thing about this is that the voltage and current will drop off exponentially when you try to discharge it. That could be a problem if you need a more constant voltage. How is this signal supposed to be triggered? What is the duty cycle? |
|
The reason for using a capacitor is simple. You can easily regulate the energy output. Your pulse duration will be a fixed time based on your load, etc, and all you would have to do it turn it on and then reset it whenever you need to. If you notice, the V^2 term of the energy formula.....are you stuck with 200VDC? You could go with a high voltage doorknob capacitor, or even create your own capacitor with glass and aluminum plates, for example. With a higher voltage, you can use a smaller capacitor and then your RC time constant could easily be SHORTER. on another note: FETs do have "high" (even 0.05 ohms will be a problem) resistance values, even when fully on. This may cause problems. Also, you want a super fast rise time and max out Vgs on the FET. (There are optical FETs available, often used as isolated relays and can handle the high voltages.) The voltage rating of the FET may not satisfy your requirementsm even at 200VDC. You may want to go isolated anyway since the control sircuit is probably at a much lower voltage anyway. You could always make a super fancy solenoid contact switch .... or figure out how a detonator works. (what exactly are you making????) Also, use a dual interlocked switch design where you charge the capacitor and then dump the energy independently. |
|
Detonation...lol. No, nothing quite that extreme. I was wondering if i would see that question ouit of someone; it was the first thing i thought of. I've seen some good starting points in here. No it does not need to be on a regular basis, in fact, fairly infrequently. Didn't want to go with a PIC/PLC setup but may do so. MOre later, off to go meet with the co worker to get more info. Thanks, Nick |
??? Are you referring to using a Vircator as opposed to a Flux Compression Generator ? Using a Vircator as a High Power Microwave device will indeed couple it's output to the target more efficiently, but it's much harder to build, and you're gonna need a ton of energy either way. |
This is our version. I captioned this picture "Moments before Dave realized we should not have put the ladder on wet ground" 
|
|
paraphrased from the other guy: 'When we were discussing yesterday what would happen to our samples of TiNi when taken to the maximum threshold before disintegration of the <1mg sample, ... , Today Dr. G ran some numbers and came up with an approximation of 40amps, 200V, for 10^-5 seconds [theoretically 8000 watts, but the duration is actual]. The idea being to measure physical changes during that pulse of energy.' So, again, what would you guys recommend to get this 10 microsecond jolt of energy with maximum value of the above indicated 8kW @ 200V and 40amps? PIC? or...? Is that short of a discharge time possible? I'd have to go dredge up the formula for length of time of a capacitor discharge, but I think that shorting them might almost take too long. He's never done a PIC, and neither have I, but I'd like to learn. There are other factors that may require me to increase the kWage because of dissipation, as mentioned by others above. Edit to add: We can assume the sample to be 10 ohms and mass no greater than 1 mg. Also, does anyone know where I can get flat Aluminum wafers? Not just machined flat, but totally distortion free - optically flat. |
|
Is this what you are making? popularmechanics.com/science/military/2001/9/e-bomb/index.phtml |
| To get a clean burst of 8kW, for the entire 10 microseconds, you'll need a larger reserve than 8kW to discharge. Can the burst exceed 8kW? Does it have to be that powerful for the duration of the pulse? How important is accuracy? Also, if it's into a 10 ohm load, that changes things drastically as opposed to a short or 0 ohm load. In fact, at those current levels, every milli-ohm has a pretty big influence. Can the voltage be higher and the current lower? This is a pretty vague challenge! |
|
Based on what I know, I'll offer the usual Engineer's cop-out of we need more information. Specifically, what kind of waveform is needed for this pulse? What are the tolerances (ie is it okay if the pulse starts at say 205 volts and ends at 195 volts)? What power source is available, and what kind of load is being driven? You can't really spec it out without knowing these things. I don't think a raw cap discharge would do it, since that would generate an exponential decline instead of a sharp cutoff (unless that's okay, given the waveform requirements/purpose of this whole thing). The cap might be needed if your power source couldn't handle the current pulse by itself. I don't know that much about logic circuits, but if you want a fairly steady 200V for the whole pulse, I presume you'll need a power transistor, like a MOSFET maybe, to control the power. Then, you'll need something to control the MOSFET. This is where my knowledge gets really shaky, but you could probably do it with either a clock and a couple of simple logic chips, or a whole microprocessor. That'll get you a nice square wave (at least I think, depending on the properties of the MOSFET). Getting other waveforms is harder. |
Square is ideal, however, i understand the limitations associated with obtaining such a waveform with so much energy, such a short duration, and small scale. --Regarding the exponential dropoff of a raw C discharge, that would work, if it decayed to a low enough level quick enough.
Yes. I understand the 8kW to be theoretical now. The burst, and therefore inputs, will have to increase. The prof seemed pretty set on using something in the neighborhood of 200VDC and 40A, but if the situation dictates, we can vary it. Brasspile, I like your thinking on the PIC. Taking into account the amount of resistance and the 10microsecond duration, is it still doable that way? I've never done one, but I'm sure i can figure it out. Using DC power. |
Most PICs will do 20-40MHz. This should be no problem for your spec of 10ms. You can get some that even have their own built in oscillator, making them completely self contained. The 18F1320 is one of these and also features the large instruction set. You could probably get away with less for a lot of things, but the 18F1320 will give you a lot of room to play around. |
|
is it 0.001 seconds (1ms)? 8kW for 1ms = 8.0J dissipated at the target load or 10 us? 8kW for 10us = 0.08J dissipated at the target load for basic energy in a capacitor - that is 4uF at 200V now, you can jack up the voltage, still using your 200V source, and charge a capacitor. Let's say you jacked it up to 2000V. The capacitor will be 40nF. What most EEs say about CSs - Leave the analog, power, RF, firmware/DSP, etc to us and you stick to PC programming. he What is your target impedance? (aka the load?) How far will the pulse travel from the source to the load? Are there any special cable requirements? |
I dont know what you plan to do with the aluminum wafers - maybe make a capacitor? I have used aluminum printing plates to make capacitors. They are very flat, but once, when I cut mine, they warped, so that capacitor ended up vibrating, but it still worked. |
What are the electrical properties of this TiNi sample? Get out an ohm meter and measure the resistance of the sample from end to end. Is it Titanium and Nickle? If so it must be an alloy, right? |
|
If you just need energy transfer, then maybe electrical is not the way to do it....What about the absorption spectrum of the sample? What wavelength of light does it absorb really well? Longwave infrared? If so, use a CO2 laser and zap the fuck out of it with a nice pulse! And this is only for the 1mg sample? Because if you had a 1kg sample, there would need to be a hell of a lot more energy for any of your attempted methods. |