high frequency integrator

di/dt should be called switch rate or slew rate, it ususally appears in the form of dv/dt in product parrameter table.

30m to 200mhz is a bandwide of amplifier or device. which also depend on ol or cl state. of cause the higher the rate the wider the chip has band.

large or small capacitance is depended on the intigratial time. has only a little relation with bandwide.(although the small time,the large band requred)

suggest you go to NI or TI web site eyc. to seek suitable device, or maxium etc, countless pages.

Hi thanks for your replies.Here is a more precise explanation, i hope.

I am looking to make an op-amp integrator circuit. The reason being is I am currently looking to analyse a rapid changing current fault signal from an air cored current transformer which is wrapped around a voltage cable.

The voltage in the cable is inducing a voltage in the current transformer which corresponds to the changeing current in the cable. Now there is a fault which occurs in the cable, a partial discharge, which generates a rapid currrent signal from 30MHz to 200Mhz and last appproximately 1ns. It is this fault which i am trying to measure which occurs on top of the normal 50Hz osscillating signal of a power cable..

Now i need to integrate di/dt (output of the current transformer) to get the actual current fault from this transformer so i need to use an op amp integrator and then an amplifier to amplify the signal. I know the basics of the integrator is the feedback capcitor in parrallel with a large feedback resistor with a risistor input into the negative input of the opamp, but from what i can remember there are issues with slew rate for op amps at this frequency from what i have read. But what these are i dont know exactly.

I also need a sampling rate but i do not know what that should be so that is another question. When designing the integrator what should determine the sample rate of your signal. If the signal is just 1ns long then what should you sample this signal at.

Also cnpower, what is the ol or cl state?

Thanks again.

I seem to understand a little about your thinking. This may be out of my depth. I dont think this is a simple task.

induce this sparkling burr by a small capacitor to the integrator to amplify time delay to detect it is not diffcult things, but how to build this integrator is an issue. Although its a usual way to detect instant sparkling as narrow as 1nS.

the method is you have to use two high speed switcher to switch the integrator. we call it double integrator. in otder that we can enlarge the narrow pulse to have enough time to mearue it by a cheap more device.

I want to know what is 30m --200mhz? as we know, signal frequency is different from pulse wide, a single narrow pulse will has very vey wide frequency.

I dont know your skill, thus, I suggest you'd better buy a digital oscillascope of over 500mhz band, or 2ghz to capture ths narrow pulse.

not sparkling . english term should be spike.

hehe. correct.

I wish more senior specialists could come here to help you with your integrator.

you can also go to NI, TI, Maxim,etc web site to ask for help. they may have finished moduel for use.

know the basics of the integrator is the feedback capcitor in parrallel with a large feedback resistor with a risistor input into the negative input of the opamp, but from what i ca

a true integrator need to conside many probelm like bias current, zero float etc. you neednt parallele resistor, just connect capacitor at output and negtive input and negtive connect a resistor to a switcher. enough, simple to speak. and postive to ground.

this is not good for your issue.

as to sample rate, for the 1nS, you'd better use 5G. this is also a huge project to buid up the oscillitor, you have to use pll to stable the frequency. and how to make pcb will concern transmission line theory etc. as well as interference etc. if you work in an institute ro college, you can try. if not. I think you'd better buy one avaliable on market.

btw, ol, cl = open loop, close loop, which will widely use in the op design. as well as circuit. in order to enlarge op band and make it work stable.

Hi,

as emc c stated you have an integrator in your current transformer!

Use it with a small resistor to get a useful voltage.

Make sure that you do not see much of the 50Hz voltage but only your "cable fault" signal.

"The signal is just 1ns long"??? This shall be 1ms? or what else?

Then think about the waveform of the "cable-fault-signal" how to detect and how to discriminate against similar signals from may be lightnings or else.

RHABE

1nS.

if its 1mS. thing will be very simple.

hehe, k241 hasnt come here again. we wast time.

You do not provide enough information to fully answer the question. The value of capacitor to choose in your integration circuit depends on the integration time constant of interest, which is not specified. But here's a neat trick you can easily perform if the current transducer is truly differentiating the current waveform. By which is meant, that the output of the current transducer is proportional to the time rate of change of the current flowing through it. Now if that is the case, and what you are wanting is the actual current waveform instead, this will work very well and is very simple! The output impedance of a differentiating device like that is inductive, because it is in fact a transformer. That means the output impedance increases with frequency, as does the output potential as a function of frequency. In turn, this means that if the load on the transducer is very small relative to its output impedance, the output current is totally determined by the ratio of output potential to output impedance, which is constant. You can measure this current with any device such as a voltmeter or oscilloscope across the small load, and the potential or waveform is now proportional to the original current waveform, and not its derivative. Therefore an integration has been performed. This is the principle behind many magnetometers.

I agree with most all responses. As typical, the invoker doesn'at lay out the entire field of what he is working with and where he is wanting to go with some good description. I was bothered to begin with "what size capacitor . . ?". And then he goes onward with other later. k241 needs to put it all together.

This sounds like a Rogosky coil current sensor. You might be able to buy one already made and use the amplifier and power supply. At these frequencies, op-amps and components don't behave ideal anymore. You might be better off to spend a few hundred dollars and use the work of somebody who spent the months needed to tweak the design.

partial discharge, nS or pS time measure are all difficult task at present.

I m interested and waiting for new, wonderful idea and circuit to share us.

no specialists out there who can contribute their works to show us?

Hi was away onsite last few days.

Thanks for the replys they are appreciated. The signal generated is transient and can last for between 1-3 ns. If you think of a spike of current on the normal ac current flowing lasting 1-3ns depending on the point of the signal occuring. This is in relation to the frequency of the of the partial discharge occuring. And then signal has gone as it is a sharp transient thing.

Any other questions please ask.

Regards

K241

It is a rogowski coil and as emc stated i think for the higher frequencies the rogowski itself has been used as an integrator but i think this does require a specially designed rogowski coil but i could be wrong.

Any other questions please ask.

Thank

K241