Is it possible to generate very short pulse width?

[floppy32]

Hi,

I want to generate a pulse train which had1MHz freq and 0.5ns pulse width.

Is that possible with a microcontroller. I would like to use STM32 boards but Im not sure a micro can achieve this.

 

[KlausST]

Hi,

an STM surely can generate 1MHz.
But for the 0.5ns you need external circuitry.

My first idea of using just a RC high pass to generate a spike at the (rising STM signal) edge won´t work, because the rise time of the STM output may be too slow.

Math:
1MHz equals to 1us or 1000ns.
Your signal: 0.5ns / 999.5 ns.
So you need to generate a pulse with 0.05% duty cycle.

I´m not experienced enough for 0.5ns fast pulses.

Klaus

added: Aren´t there laser pulsers with that narrow pulse width?
BTW: you did not specify waveform, voltage, current...

 

[FvM]

It's generally possible with fast logic IC series. What's the complete specification, amplitude, rise-time, load impedance?

 

[danadakk]

Avalanche pulse generation :

Nanosecond Avalanche Transistor Pulser

Transmission Line Avalanche Pulse Generator - Page 1

Transmission Line Avalanche Pulse Generator - Page 1

www.eevblog.com

Transmission line & SRD :

Understanding a sub-nanosecond pulse generator

I have been reading a couple of papers (1,2) that use a Step-Recovery Diode to generate sub-nanosecond electrical pulses, for the purposes of pulsing an LED. I am simply trying to understand the pu...

electronics.stackexchange.com

http://www.hp.woodshot.com/hprfhelp/5_downld/lit/diodelit/an918.pdf

What is accuracy of pulse width over T and V you need ?

Presumption you have scope capable of Ghz performance -

How is bandwidth related to rise time for oscilloscopes? | Tektronix

How is bandwidth related to rise time for oscilloscopes? Historically, oscilloscope frequency response tended to approximately follow the rule: Bandwidth x risetime = 0.35. This corresponds to a 1-

www.tek.com

Rise Time vs. Bandwidth and Applications

In measurement instruments such as oscilloscopes and spectrum analyzers, the maximum rise time that can be measured with reasonable amplitude accuracy is directly related to the instrument’s -3-dB bandwidth.

interferencetechnology.com

Regards, Dana.

 

[dick_freebird]

The points about instrument capability and loading should be
taken more seriously than the "how to generate" if you are
interested in the pulse width numbers stated. Any accuracy
about the PW wants even more care.

At such a short PW I might consider something like a real fast
NAND gate wired with one input fed from the output, making
a time-of-flight one-shot (PW ~ TPHL+TPLH). Follow your
uC's I/O voltage to the fastest compatible 74xx### family and
see if you can expect pulse shape fidelity based on drive and
your expected load (figure tr, tf need to be < 10% of PW, apiece?).

 

[danadakk]

One possibility -

https://datasheets.maximintegrated.com/en/ds/MAX9600-MAX9602.pdf

Build a really fast pulse generator (50ps rise time) using an ultra fast SiGe comparator. – Starlino Electronics

Build a really fast pulse generator (50ps rise time) using an ultra fast SiGe comparator. - Motivation If you purchase a new or used scope you will probably need a way to test its advertised bandwidth. One way to test the bandwidth of the scope i...

www.starlino.com

EEVblog #306 – Jim Williams Pulse Generator

EEVblog #311 – Jim Williams Pulser Followup

Regards, Dana.

 

[doraemon]

Hello!

Right out of a microcontroller: no way. With an external circuit, can be...
In order to get sub-ns pulses, everything has an importance. I'm playing with lasers in my lab, and there
are a few rules. These rules apply to lasers, and in this case, what you want is to switch the laser as fast
as possible, therefore you need the current to vary fast. I will suppose you need current.
- Keep the inductance minimal in the laser power loop (assuming it's a laser, but that would be true for
anything else. The speed of your signal is di/dt = U/L, so L has to be kept small.
- Mount your system extremely close to the chip. If you use leads, then there is a high chance that
your original 500 ps target will fly at once to 5 ns.
- Never use a via in the PCB. A via adds a lot of inductance to the circuit -> back to last step.
- If you can't avoid a via, make several vias in parallel. The additional impedance due to one via will be
divided by the number of vias.

That said, if you don't need current, it might be easier.

Dora.

Hi,

I want to generate a pulse train which had1MHz freq and 0.5ns pulse width.

Is that possible with a microcontroller. I would like to use STM32 boards but Im not sure a micro can achieve this.

 

[floppy32]

Hi,

an STM surely can generate 1MHz.
But for the 0.5ns you need external circuitry.

My first idea of using just a RC high pass to generate a spike at the (rising STM signal) edge won´t work, because the rise time of the STM output may be too slow.

Math:
1MHz equals to 1us or 1000ns.
Your signal: 0.5ns / 999.5 ns.
So you need to generate a pulse with 0.05% duty cycle.

I´m not experienced enough for 0.5ns fast pulses.

Klaus

added: Aren´t there laser pulsers with that narrow pulse width?
BTW: you did not specify waveform, voltage, current...

It's generally possible with fast logic IC series. What's the complete specification, amplitude, rise-time, load impedance?

Hello!

Right out of a microcontroller: no way. With an external circuit, can be...
In order to get sub-ns pulses, everything has an importance. I'm playing with lasers in my lab, and there
are a few rules. These rules apply to lasers, and in this case, what you want is to switch the laser as fast
as possible, therefore you need the current to vary fast. I will suppose you need current.
- Keep the inductance minimal in the laser power loop (assuming it's a laser, but that would be true for
anything else. The speed of your signal is di/dt = U/L, so L has to be kept small.
- Mount your system extremely close to the chip. If you use leads, then there is a high chance that
your original 500 ps target will fly at once to 5 ns.
- Never use a via in the PCB. A via adds a lot of inductance to the circuit -> back to last step.
- If you can't avoid a via, make several vias in parallel. The additional impedance due to one via will be
divided by the number of vias.

That said, if you don't need current, it might be easier.

Dora.

Is there

Hello!

Right out of a microcontroller: no way. With an external circuit, can be...
In order to get sub-ns pulses, everything has an importance. I'm playing with lasers in my lab, and there
are a few rules. These rules apply to lasers, and in this case, what you want is to switch the laser as fast
as possible, therefore you need the current to vary fast. I will suppose you need current.
- Keep the inductance minimal in the laser power loop (assuming it's a laser, but that would be true for
anything else. The speed of your signal is di/dt = U/L, so L has to be kept small.
- Mount your system extremely close to the chip. If you use leads, then there is a high chance that
your original 500 ps target will fly at once to 5 ns.
- Never use a via in the PCB. A via adds a lot of inductance to the circuit -> back to last step.
- If you can't avoid a via, make several vias in parallel. The additional impedance due to one via will be
divided by the number of vias.

That said, if you don't need current, it might be easier.

Dora.

Are you using a laser diode driver? What do you use to obtain short pulse width PWM? Im also fine with ad-hoc solution but not a function generator. Like I can provide pulse trigger and obtain short pulses for driving laser diode? Is there such module?

 

[FvM]

You didn't yet specify your requirements. Do you want to say laser diode driver is your application? If so, which peak current?

 

[KlausST]

Hi,

@floppy32
you might improve your posting skills
* please qoute posts only when you reply to them
* don´t double qoute the same post
* use the "preview" (icon) before pressing the [Post reply] button

Klaus

 

[floppy32]

You didn't yet specify your requirements. Do you want to say laser diode driver is your application? If so, which peak current?

I didnt specify because those complete devices were expensive I think. I thought if I can build a ultra short pulse then I can only purchase the driver. I was thinking to use a microcontroller(for repetition rate)-->fast buffer(to shorten rising edges)--> PW shortener--> LMG1020 GaN Driver like here: https://www.ti.com/lit/ug/tidue52/tidue52.pdf

Can I use some logic like the follwing:

What logic family do you recommend to obtain 0.5ns PW at the output?

 

[doraemon]

Hello!

Are you using a laser diode driver? What do you use to obtain short pulse width PWM?

It's a chip designed for lasers. But the laser takes benefit of the diode switching pulse effect,
allowing it to deliver pulses in the sub 100ps. The specs say 100 ps because of some variability
among the chips, but most of the chips do better. So yes, _some_ laser drivers can do that,
not all of them.

Dora.

 

[floppy32]

Hello!



It's a chip designed for lasers. But the laser takes benefit of the diode switching pulse effect,
allowing it to deliver pulses in the sub 100ps. The specs say 100 ps because of some variability
among the chips, but most of the chips do better. So yes, _some_ laser drivers can do that,
not all of them.

Dora.

Is that a pulsed voltage or pulses current type of driver?

 

[FvM]

I didnt specify because those complete devices were expensive I think. I thought if I can build a ultra short pulse then I can only purchase the driver.

Sorry, still unclear what you are generating the short pulse for. Respectively no answer regarding load and pulse magnitude.

Regarding suitable logic families, there's a gap between fast CMOS gates, e.g. 74LVC1G series which have difficulties to achieve below 1 ns pulse width and dedicated high speed logic which reach e.g. 100 ps pulse width but have reduced voltage swing. CMOS transistors as such can produce shorter pulses but push-pull IO standards aren't well suited to transmit it. Looking at the data sheets of logic devices supporting different logic standards (e.g. FPGA) illustrates the point.

I implemented a circuit similar to post #11 with 74LVC1G LTspice models and achieved about 0.7 ns minimal pulse width at 5V Vcc, but not less.

 

[BradtheRad]

The programmable unijunction transistor seems naturally inclined to generate brief pulses.
I don't know which is easier to work with, BJT or CMOS. (Microwave transistors might suit your purpose.)

Action is similar to an SCR. The capacitor charges slowly through R1. Imbalance grows in the transistor network. When threshold voltage is exceeded (or trigger voltage is applied), bias currents flow, turning on transistors briefly.

All components must be adjusted carefully. The theoretical simulation below uses ideal components. The 30 pico-second pulse is no doubt optimistic. A diode stands in as the load.

 

[Easy peasy]

How are you going to measure the 500pS pulse width ?

if say it has a 50pS rise and fall time - you will need some serious gear to verify ....

 

[floppy32]

How are you going to measure the 500pS pulse width ?

if say it has a 50pS rise and fall time - you will need some serious gear to verify ....

I have access to such scope in a lab. LeCroy one.

 

[Easy peasy]

Good to know - there are a number of circuits out there that the average engineer can build on the bench for these short rise times and pulse durations - good luck - and we look forward to seeing a little of the results.

 

[dick_freebird]

Still think you're best off finding the fastest logic gate that
will run off the same supply as whatever you're driving,
feed it back as a time-of-flight one-shot, buffer with one
of its chip-mates, and see what you get. If it's too fast /
narrow you can pad it out with more gates. Be sure to
give it some really tight, RF-grade decoupling caps.