Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Waste pulse energy recovery

Status
Not open for further replies.

Swend

Full Member level 4
Joined
May 14, 2019
Messages
237
Helped
10
Reputation
20
Reaction score
9
Trophy points
18
Location
Roskilde, Denmark
Activity points
2,304
Hi friends,

I wish (actually a client does) to recover the energy of some consecutive pulses and I have only come up two half-baked ideas so far, so let me know what you think.

Here's the pulse given, measured over a restive load.
Figure_1.png
The x-axis is time in nano-seconds.

These are the constraints given:
1. The pulse amplitude and frequency can vary +/- 20% from pulse to pulse
2. The resistive load is limited to about 2 ohms
3. If required, the pulse can be made short as a few microseconds or stretched to about 300 microseconds. The average power remains the same regardless of pulse length.

I have calculated the average power to be 478KW for that particular pulse. I think that's a lot of power to recover, so my best idea is simply to dump the pulse(s) into a resistive heating element and boil some water. There is not an objection to hot water as such, but they would have better use if it could be converted into a more practical dc voltage. For the dc voltage option, my best idea is an air-core transformer - but I'm not sure how efficient that would be compared to boiling water?
 

The average power remains the same regardless of pulse length
Really? I would expect constant energy per pulse.

I presume most readers can understand the problem better if you show a circuit of the existing system from which the pulse energy shall be absorbed/recovered.
 

Really? I would expect constant energy per pulse.

All pulses of the same length have the same energy. shorter pulses have less energy, longer pulses have more energy, but for all pulses the average power is the same. Once you have decided a desired length, all pulses will be of that length. That is what I have been informed.

I presume most readers can understand the problem better if you show a circuit of the existing system from which the pulse energy shall be absorbed/recovered.

Sure I would love to see it too, would make things a lot easier, but time being I don't have it.
 

Here's the pulse given, measured over a restive load.

More clarity needed.

Voltage and current do not appear to be in phase.

You can multiply current and voltage point by point and plot on the same graph.

The applied pulse is not clearly seen. Also the phase difference does not stay constant.

Have you a decent power factor corrector in place? That can save some power on the long run.

Or, am I getting everything wrong? You are applying a rectangular pulse to a resistive load and seeing these ringing?
 

Voltage and current do not appear to be in phase.

That's because they are not in phase.

You can multiply current and voltage point by point and plot on the same graph.

Yes, that is what I did to calculate average power. And I used the absolute values because it is a resistive load.

The applied pulse is not clearly seen. Also the phase difference does not stay constant.

I don't know the applied pulse, I only know the resulting pulse as shown. Phase difference is not constant, that is correct.

Have you a decent power factor corrector in place? That can save some power on the long run.

If PF correction would be applied, I would have to do it on the resulting pulse, but as you noted it's a very erratic pulse so I would consider that too complex.

Or, am I getting everything wrong? You are applying a rectangular pulse to a resistive load and seeing these ringing?

I'm not applying anything, I just got the plotted data I have shown in a .csv file, and that is the resulting pulse. Maybe if I come up with the right idea, I will be privy to more information.
 

Yes, that is what I did to calculate average power. And I used the absolute values because it is a resistive load.

Using absolute values to calculate power is not correct.

Your case appears to have considerable source reactance. Hence any correction need to be made to the source side.

Power factor correction need to be done based on the load power and the approx frequency (of the pulse); As your load is resistive, some capacitor added to the source side can make considerable improvement.

If it is the result from a single pulse, you need to take care of the source impedance. That is my best guess.
 

Using absolute values to calculate power is not correct.

Don't blame me ;-) All the knowledge I have is from the internet and my internet friends like your good self, and such friend told me that you use absolute values if it is a resistive load as negative power is also power dissapated in the load.

In case my internet friend is incorrect then the average power is reduced to 340KW

Your case appears to have considerable source reactance. Hence any correction need to be made to the source side.

In that case there is nothing that can be done about the PF.

Power factor correction need to be done based on the load power and the approx frequency (of the pulse); As your load is resistive, some capacitor added to the source side can make considerable improvement.

If it is the result from a single pulse, you need to take care of the source impedance. That is my best guess.

That is the result of a single pulse. But source impedance/reactance is what it is, it can not be changed by me, I can only try to come up with an idea for the recovery part.
 

extending the time of the pulse is useful as it allows a wider range of power components to be used - if you advise the peak volts and current over the 300uS you mention I can supply a few suggestions.

Why did you mention an air cored Tx? do you need to isolate the output recovered energy?

- - - Updated - - -

p.s. I can see they are clearly in phase - with a rising DC offset near the end - which makes it look as if they aren't.
 

p.s. I can see they are clearly in phase - with a rising DC offset near the end - which makes it look as if they aren't.

I do not see that so clearly.

This is a typical waveform for a damped harmonic motion with two time constants.The two time constants are not even close and we can study them independently.

In my eyes, the phase difference stays approximately constant (the periods for voltage and current waveforms are same) during the time seen in the plot.

Ringing is common if the source and load impedances are not tuned. Plus I guess the source has considerable reactance (cause for ringing).
 

extending the time of the pulse is useful as it allows a wider range of power components to be used - if you advise the peak volts and current over the 300uS you mention I can supply a few suggestions.

Why did you mention an air cored Tx? do you need to isolate the output recovered energy?

I meant 'air core' as in 'something that will not saturate at that frequency (abt. 25KHz)', and I mentioned tx for two reasons, first as a impedance conversion which will allow different types of solutions, and secondly it coud bring the voltage down to something that allows for more commercially available components. isolation was not the primary concern.

p.s. I can see they are clearly in phase - with a rising DC offset near the end - which makes it look as if they aren't.

I could swear that I'm seeing about 60 degrees difference, but it doesn't really matter, it can't be changed.

- - - Updated - - -

How about the idea of attaching diodes and capacitors, to convert the waveforms into a DC supply?

Sure, as long as it doesn't exceed 2 ohms in load impedance, so I'm thinking the cap would be too small to contribute anything serious.

- - - Updated - - -

if you advise the peak volts and current over the 300uS you mention I can supply a few suggestions.

sorry, overlooked that. For a 300uS pulse the current and voltages will be pretty much as shown in the plot, except that the frequency will be proportionally lower. And everything is with expected deviation of +/- 20% from pulse to pulse .
 

as to in phase - I meant rather in synchronism, as in
constant phase difference,

so 45kV peak and 6500 Apk ... ?
 

as to in phase - I meant rather in synchronism, as in
constant phase difference,

so 45kV peak and 6500 Apk ... ?

No, only 65A peak, the current is multiplied by a factor 100 in the plot so that it is better visible.
 

Would there at least be someone that can bring some certainty to the calculation method used here?

I calculated power by multiplying absolute values of current and voltage, because I have learned that in a resistive load, the negative power is also power dissipated in the load.

But c_mitra has cast doubt on my method, so I would really like to know the authoritative answer to this?
 

I have learned that in a resistive load, the negative power is also power dissipated in the load....

If you have learnt that negative power is dissipated, you are wrong.

If the voltage and current has opposite signs, the instantaneous resistance is also negative.

In this particular case, you need to multiply point by point the voltage and current to get the overall power used in the load and finally appears as dissipation. You should do the summation over at least one complete period.

For a capacitor, the voltage and current are out of phase by 90 degree. The power dissipated is integral of sin(x)*sin(x+2*pi/4)*dx over zero to 2*pi (and this integral is zero).

It is often called the lossless component.

Positive power is consumed and negative power is produced.

If you use absolute values in the above equation you will get a non-zero result. Of course you know that an ideal capacitor does not dissipate any power.
 
  • Like
Reactions: Swend

    Swend

    Points: 2
    Helpful Answer Positive Rating
Positive power is consumed and negative power is produced.

This will be my new mantra, thank you.

If you have learnt that negative power is dissipated, you are wrong.

As I said, I learnt it from an internet friend - so now I have stricken his name from my Christmas-card list, then he can think about what he did wrong come Christmas.

In this particular case, you need to multiply point by point the voltage and current to get the overall power used in the load and finally appears as dissipation. You should do the summation over at least one complete period..

Which I did, over the whole 140 uS - and the result is 340KW.
 

Which I did, over the whole 140 uS - and the result is 340KW.

340 kW in 140 uS corresponds to about 47.6J per pulse. Because you are handling large currents and high voltages, this may (or may not) be much.

To reduce the ringing effect:

Take either the voltage or current curve and use curve fit function to get the time constants.

The curve to be fitted is a sine curve, damped by a exponential and some const term (if some DC is present). Almost all graph plotting software has curve fitting options.

The same parameters should work for the other curve (voltage or current)- but for a scale factor.

The curve fit must be 99% (or so) correlation coefficient (0.99 or so).

Find out the value for critically damped curve.

See the time constants.

Estimate the approx capacitor value.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top