!!!How to build a programmable High Speed Pulse Generator ?

[tjukb]

avalanche pulse generator

Hi All，

I have been working in building a programmable High Speed Pulse Generator circuit to test the Xsister(MOSFET,IGBT)'s swithing time（HSPG is used to drive the DUT）,
but I am failing to get the pulse in the nanosecond reigon.
The fastest rise time I get is around 8ns(the fall time is about 50ns),
and the waveshape quality is bad when the output load（50ohm）is added at the output.

The necessary specification information is below:

1. output voltage is ±30V(0.1V Step),programmable;
2. pulse width is 0.1us -1000us,also programmable,PRF(pulse repete frequency) is about 1Khz orso;
3. both the rise time and fall time of the pulse is less than 1ns(the less the better);
4. the max direct current load of the pulse is about ±2A(so the Rloadmax = 30 /2A =15ohm);

According to the specs above, avalanche xsistor and SDR maybe not fits the design, so what about the mercury-wetted type of relay,is it can work?And how about the PCB layout, is there anything should pay more attention to?

so can you give me some suggestion and working circuits that I can play with?

thanks in advance!

 

[tjukb]

speed pulse generator

anyone there？

 

[FvM]

how to build a pulse generator

Some specifications are more or less contradicting. I don't think, that you'll suceed when trying to merge them into a single instrument.

ns or sub-ns rise time is only meaningful with matched impedance. Typically, it would be used with 50 ohm load. So, at least the current specification doesn't apply. You have to know, if the voltage and pulse width specification in conjunction with sub-ns rise-time is essential for your application. If so, it gonna be difficult, I think.

I must admit, that I didn't follow the scientific papers in this field during the last two decades. Looking at the available commercial pulse generators seems to indicate, that it's still complicated. State-of-the-art pulse generators are using custom IC output amplifiers, I doubt if you achieve similar specs with discrete devices (e.g. RF power BJT or MOS transistors).

For sub-ns at higher voltages, avalanche BJT and mercury wetted reed relays are still interesting candidates, I think. Unfortunately, both have only fast make action. So you have to use delay lines or some (very limited) push-pull design option. Delay lines are only useful up to a few 10 ns pulse width, avalanche transistors also won't stand the dissipated power with longer pulses, because they have a considerable voltage drop. Mercury relays have a ms switching delay with µs jitter, which is unacceptable for any application, that needs synchronized pulse generation.

I think, you should rather target to a 2 - 5 ns risetime, that's achievable with fast MOSFET or BJT.

 

[tjukb]

very thin pulse generator

Hi FvM，

thank you for your patiently and meticulously reply.

yes, it's realy a hardwork to complete my SPEC with discrete devices (e.g. RF power BJT or MOS transistors).

so can you give me some advice that how to accomplish the Pulse Generator in 2 - 5 ns risetime?

How about mosfet driver or RF BJT / RF MOSFET?
there are some MOSFET Driver such as TC1412,MAX5054,DRF1200,
and some RF MOSFET divice such as ARF473 and ARF520 .
so in your opinion, which one above is better for my application?

thank you for your timely apply again.

Some specifications are more or less contradicting. I don't think, that you'll suceed when trying to merge them into a single instrument.

ns or sub-ns rise time is only meaningful with matched impedance. Typically, it would be used with 50 ohm load. So, at least the current specification doesn't apply. You have to know, if the voltage and pulse width specification in conjunction with sub-ns rise-time is essential for your application. If so, it gonna be difficult, I think.

I must admit, that I didn't follow the scientific papers in this field during the last two decades. Looking at the available commercial pulse generators seems to indicate, that it's still complicated. State-of-the-art pulse generators are using custom IC output amplifiers, I doubt if you achieve similar specs with discrete devices (e.g. RF power BJT or MOS transistors).

For sub-ns at higher voltages, avalanche BJT and mercury wetted reed relays are still interesting candidates, I think. Unfortunately, both have only fast make action. So you have to use delay lines or some (very limited) push-pull design option. Delay lines are only useful up to a few 10 ns pulse width, avalanche transistors also won't stand the dissipated power with longer pulses, because they have a considerable voltage drop. Mercury relays have a ms switching delay with µs jitter, which is unacceptable for any application, that needs synchronized pulse generation.

I think, you should rather target to a 2 - 5 ns risetime, that's achievable with fast MOSFET or BJT.

 

[FvM]

build pulse generator

Personally I would try with RF power devices, either BJT or MOS in unsaturated (current source) operation and a real output impedance. As a voltage rating of 80 or 100 V is required, the device choice is most likely rather limited. A several 100 watts power dissipation of the output stage must be accepted, unfortunately.

Saturated operation would allow considerably lower disspipation, but don't achieve the intended speed or a good waveform, I fear.