high frequency pulse generator?

[kydong]

I want to generate a rectangular pulse with width from 2ns to 10ns (perfect 5ns). PRI is about 10 (micro-second)
The amplitude of the pulse is about 5v to 20v or depend, the amplitude is not importance.
How can I make them ?
Are there any IC do its requirement?
Thaks you all!

 

[alexxx]

Hi!

I have played with XR2206 a long time ago when I was a student for my final year project. I remember it was a bit expensive though. I can't say I remember a lot of its functionality but please take a look, it might suit you.

 

[kydong]

Thanks for four help. But it's quite not suitable, this IC only make the pulse with fmax=1Mz, I need the much higher this frequency...How can I solve this?

 

[FvM]

Some parameters missing form your specification:
- pulse rise- and falltime
- pulse generator output impedance
- intended load

XR2206 is surely unable to generate fast pulses in the intended range
Fast logic ICs can possibly achieve the pulse parameters (depending on additional requirements) but no output voltage above 5 V

 

[aliraza786]

if you need a regtangular wave then (if i am not wrong)...use 555 and take output from the 6th pin...mean from cpacitor the wave generating there are very similar to the triangular wave....

 

[kydong]

Some parameters missing form your specification:
- pulse rise- and falltime
- pulse generator output impedance
- intended load

XR2206 is surely unable to generate fast pulses in the intended range
Fast logic ICs can possibly achieve the pulse parameters (depending on additional requirements) but no output voltage above 5 V

Thanks for your advice.
I need generate the pulse with some addition information such: pulse rise <1.5ns, fall time < 1.5ns, pulse width 5 ns. PRI is about 10 micro seconds. I dont care about output impedance so much since I tend using a power amplifier following the pulse generator
Could you give me more ideas?

---------- Post added at 15:46 ---------- Previous post was at 15:43 ----------

if you need a regtangular wave then (if i am not wrong)...use 555 and take output from the 6th pin...mean from cpacitor the wave generating there are very similar to the triangular wave....

You mean I use Pulse Generator Based on Avalanche Transistor ? I see that in this circuit, there is a IC 555, but I am dubious about the experimence when using IC 55 to generate a such high frequency pulse like that,
any idea?

 

[biff44]

Well, you can not use a function generator chip...too slow.

There are a lot of ways to do it, but one way with little intrigue would be some high speed gates:



If you really need 5V, you could use a broadband amplifier. If you want to vary pulsewidth, you can vary the line lengths, or add a shunt tuning cap to ground across one of the delay lines.

Input square wave at 50 khz would give you the 10 us PRI.

 

[kydong]

Well, you can not use a function generator chip...too slow.

There are a lot of ways to do it, but one way with little intrigue would be some high speed gates:




If you really need 5V, you could use a broadband amplifier. If you want to vary pulsewidth, you can vary the line lengths, or add a shunt tuning cap to ground across one of the delay lines.

Input square wave at 50 khz would give you the 10 us PRI.

thanks for your idea.
I have some questions, what is the kind of the line here, just normal line? How can I calculate the length of line to control the time delay ? or just do it and justify the length of the line to generate suitable pulse width? I want to simulate it first, what's the soft I can use? Proteus?? is it ok?
You've ever tried it before? it really generate the pulse with very short width? (5 nano-second)
Thanks in advance. waiting for your reply

 

[biff44]

For want of a better answer, I would just make the two lines 50 ohms microstrip.

One has to be physically longer than the other by the 5 nS. In calculating the width of the microstrip, you can probably also figure out the effective dielectric constant. Energy will propagate down the line at the speed of light as slowed down by that effective dielectric constant Eeff

X = V t when x is the extra lenght of one of the 2 lines, V is speed of light in the media, t is 5 ns.

V= (3 x 10^8 meters/S)/√(Er)

Er will probably be in the 2 - 3 range.

 

[kydong]

Thanks so much, but I want to simulate it first, what the software I should use? I found out that Proteus can not do it, may be a high frequency software?

 

[FvM]

As far as I'm aware of is Proteus including regular SPICE analysis capabilties and will be able to analyze the circuit. It doesn't provide an EM solver to calculate e.g. microstrips. But they can be modelled as SPICE transmission lines.

Apart from Proteus, you can refer to a recent SPICE variant, e.g. LTSpice.

 

[kydong]

How about using a L,C element and placed it after one NOT gate? I thiink that will be ok, a L,C element play a role like delay line transmssion ? it is OK?

 

[volker_muehlhaus]

It's really simple with a delay line. Propagation speed in typical coax cables (RG58, RG174 etc) is ~20cm/ns, so that you need 1m cable length for a delay of 5ns.

 

[biff44]

the delay line made of transmission line will be very broadband, and therefore your leading and falling edges will be preserved. If you use a L-C network, the lowpass nature will round over the leading/falling edges. You will end up with a noisy output circuit, since the trigger point for the output gate is sloppy now.

But yes, you could just use an LC time delay network. I would try for one with the group delay you need, but without the lowpass roll-off.