Fast trapezoidal pulse (rise time less than 2ns)

Thread Starter

ilginsarican

Joined Jul 13, 2017
142
Hello,
I am working on Mil-SDT 461G/ CS115 test.
According to CS115, pulse should be like the picture below, the rise time must be less than 2ns.
I've used reed relay to get this pulse and the rise time was 2.6ns with reed relay.
What can I do to achieve a faster rise time?Any suggestions would be great.
I thought I can use mosfet and gate driver but I am not sure to be able to get 2ns rise time with mosfet.
Because I have reviewed many mosfet datasheet, the rise times were larger than 2ns.


1581876642713.png
 

crutschow

Joined Mar 14, 2008
34,285
What is the circuit load impedance?

HP used to use a mercury-wetted reed relay to get fast rise-times.
It would be difficult to build an circuit that would do that faster.

For that fast rise-time an impedance controlled layout is likely needed with a copper ground plane.
How did you construct your circuit?
 

Hymie

Joined Mar 30, 2018
1,277
The measured rise time may be limited by the capacitance of the circuit into which the relay is switching and/or the limitations of your measurement circuit/method.

Consider a source impedance of 10 ohms switching into a 0.1nF capacitance, will have a time constant of 1ns. A scope probe is likely to add around 0.03nF to the circuit.

I would recommend you confirm that your measurement method is capable of measuring a rise time of less than 0.5ns and minimise any extraneous capacitance in the circuit.
 

Thread Starter

ilginsarican

Joined Jul 13, 2017
142
What is the circuit load impedance?

HP used to use a mercury-wetted reed relay to get fast rise-times.
It would be difficult to build an circuit that would do that faster.

For that fast rise-time an impedance controlled layout is likely needed with a copper ground plane.
How did you construct your circuit?
Hello,
Actually I have a device builded for CS115 and I am trying to fix it.
The device is "Pulse Generator" on the test setup picture below.
Device has a variable tranforsmer, after transformer there is a ac dc converter circuit and reed relay to switch this DC Voltage,
The relay is controlled by a microprosessor(PIC16F877A).Repetation rate(30Hz) of signal is achieved by microprocessor.


1581952257953.png
 

Janis59

Joined Aug 21, 2017
1,834
How to get GHz range fast reaction? With multi-GHz scale transistor. You have choice between BFU910 or more powerful BFP620 (90 GHz and 60 GHz) or jfet BF998.
 

Thread Starter

ilginsarican

Joined Jul 13, 2017
142
How to get GHz range fast reaction? With multi-GHz scale transistor. You have choice between BFU910 or more powerful BFP620 (90 GHz and 60 GHz) or jfet BF998.
Sorry, I did not explain the structure clearly. As you can see the picture below the amplitude of pulse is 5Amps. In the calibration texture there is 50Ohm resistor,so to get this amplitude, at least 250V DC must be applied, so switching component( transistor ect.) should be able to withstand 250V.
1581964509731.png
 

Hymie

Joined Mar 30, 2018
1,277
I’m surprised you have a reed relay capable of switching 250Vdc at 5A (over 1kW). But you have a rise time of 2.6ns; that is a slew rate of around 75V/ns (75,000V/µs).
What is the response time of your attenuation circuit?
Switching 250Vdc, 5A at a rate of 30Hz is likely to generate significant rf emissions. Your neighbours might be complaining at the loss of TV/radio signals or other interference with their electronic equipment.
 

Thread Starter

ilginsarican

Joined Jul 13, 2017
142
I’m surprised you have a reed relay capable of switching 250Vdc at 5A (over 1kW). But you have a rise time of 2.6ns; that is a slew rate of around 75V/ns (75,000V/µs).
What is the response time of your attenuation circuit?
Switching 250Vdc, 5A at a rate of 30Hz is likely to generate significant rf emissions. Your neighbours might be complaining at the loss of TV/radio signals or other interference with their electronic equipment.
Hello again,
The part number of attenuator: 24-20-34 and I did not find any information about response time.I am testing this device in the EMI/EMC lab.
 

kubeek

Joined Sep 20, 2005
5,794
Are you sure that that 5A is developed on the 50 ohm termination resistor? I doubt that would be the case, so a more detailed schematic would be quite helplful.
 

Janis59

Joined Aug 21, 2017
1,834
a) 5 Amps at 2 nsec - sems rather Mission Impossible except the spark-gaps with ionizating steamer. We did such here at 60-ies, when the first femtosecond lasers was invented. Still one device hangs in museum part - its a vacuum filled quartz pipe with two massive copper watercooled halfballs diameter about inch and distance several centimeters. On those the ultrafast non-inductive capacitor of 50 kV and 5000 mkF is fixed on. Near by both balls stays sharp needles which at the righ time gets the 300 kV 1 mkA weak-current pulse ionisating the chamber. Thus the capacitor uses the path sensed by this streamer and then the grand pulse is running through. The di/dt was with order of damn femtoseconds.
b) 50 Ohm 3 kW resistor I have produced myself. Take the 49 or 64 or 81 or 100 identical ceramic "box" style resistors of corresponding resistance and solder betwen two larger pcb platelets with some air-gap between em. All perimeter fill with small fans except the last side of four where the air will come out, my case I used 8 pieces 50x50 mm fans. The input and output is pcb center. Cannot say this elefant may work well at GHz signals, but for 100 MHz it was working rather satisfactory.
 

Hymie

Joined Mar 30, 2018
1,277
This would appear to be the data sheet for the attenuator you are using.
https://www.apitech.com/globalasset...ns/rf-components/attenuators/fixed/wmod24.pdf

Nevertheless the circuit description (from the book) notes that it is critical that an injection probe with adequate response to high frequencies is used.

Rather than measuring the signal post attenuation, I suggest you measure the waveform at the relay contacts.
You might try injecting a GHz signal at the relay contacts and observing the response of the coaxial line & attenuator.
 
Last edited:
Top