Making variable-width pulses with an EOM

I have an EOM I am using to generate pulses from a CW laser. I would like to get pulses in the 0.5 ns to 2 ns range, but I am finding that most benchtop function generators or pulse generators usually die out in the 5-10 ns range. RF function generators that I have seen are over my budget. Since the optical signal mirrors the electrical signal, it needs to be straight up and down - not a switch turning on an RF source, as is shown in a "" tutorial.

Meanwhile, everyday people do optical communications at 2.5Gb/s and 10 Gb/s, among other rates. Surely there must be some relatively inexpensive electrical sources. I've found various links for XFP and SFP modules. Though these have lasers built in, does anyone know of demo boards or evaluation kits for these types of things that will generate the pulses? Do these modules actually generate the pulses, or just have the drive amplifier and receiver circuitry?

I can amplify them after with an RF amp we already have, so output voltage is not important. OEM modules would be fine, as I can supply them with regulated DC voltage if necessary. If anyone knows of a cheap pulse source, I would really appreciate it.


Reply to
Loading thread data ...

If your budget want to keep you in small k$ range, you might not have much luck. As an electrical engineer, I'm left wondering about your specific requirements.

Do you want continuously variable pulse widths?

Do you need 15 ps transition times or is 500 ps rise/fall fine for pulses down to 500 ps?

Do you need to generate the pulses through extenal electrical control or are you looking for pushbutton operation?

How fast do you need to generate a second pulse of the same width? ...of a different width?

Do you want the width on a dial where you don't necessarily know the actual width as you turn the dial but get "longer" and "shorter" pulses as expected or do you want to specify the width with digits?

I've seen arbitrary waveform/pulse generators that can deliver narrow pulses but I'm worried they're out of your budget as well. The Tektronix Data Timing Generators, for instance,

formatting link
have a pulse width minimum of 290 ps with 5 ps resolution. The ~$25k price tag is a bit harsh if all you want is pulse shaping from another data (trigger) source. A quick google search for "pulse generator" came up with hits from places like Avtech ( that have specs along the lines of what you need at more moderate pricing (I clicked on one instrument, $5k).

So. What do you really need?

Reply to

Thanks for your response. I'm a grad student, and so the budget is under $1K. Thanks for the link to Avtech. They do have OEM modules in the $2K range which would work, and there is currently one on an E- Bay auction. I also found a demonstration board for an Analog Devices PLL that operates at 1.2 GHz. That would give ~1 ns pulses, though I would want to drop the rep rate to somewhere in the range of

10KHz-1MHz. I anticipate taking whatever signal I get and putting it in the RF amp we already have, so anything above 100mV (into 50 Ohms) would probably work.

I am thinking there should be a number of devices that work out there. Gigabit ethernet seems to be fairly common these days. While I havn't studied ethernet, I would imagine they would have to supply pulses around 1 ns in width. There is a great deal of optical equipment built for 10 Gb/s and 2.5 Gb/s. These have to use some kind of electrical-to-optical modulation. It seems to me there should be some way to stick an SMA connector on one of these, and adjust the bit pattern to send a single 1 every few microseconds, or change pulse widths by sending a succession of NRZ 1's. Unfortunately, I come from the optical side, and I'm just trying to pick up the electronics on the fly.


Reply to

The comm systems you mention don't communicate with pulses but with coded NRZ data. You drive the laser (or EOM) until you get a transition to not drive (or to drive less).

If you want pulses to occur on demand with specific widths and low overall duty cycle, you're no longer working in the communications modes.

Reply to

For anyone still interested in the topic, I did find a circuit that will do ns pulses. Even better, you can probably get the IC's as free samples if you are aligned with a company. It uses the LT1721 Comparator from Linear. Here is a link:

formatting link
So, the good news is that it is cheap and readily available. The bad news is that it would require soldering. RF speed (for the ns transitions) scares me when it comes to my soldering ability, as does surface mount components. I think this circuit, despite it's simplicity in components and acquisition, may be too difficult for me to construct without ringing like a tuning fork. S

o, I am still looking for an evaluation board-type kit, but at least there is hope. And, I hope anyone else that might struggle with ns generation now has another option they can look at.

Reply to

The circuit comes back to my original questions on your needs. While the circuit can produce short pulses, you're limited in retrigger and tunability. Is 30 ns between pulses fine? The circuit suggests the delay is adjusted by changing the "variable delay" resistor while leaving the threshold at the fixed level. I would have thought a fixed, different delay in place of the "variable delay" would work with a potentiometer-adjusted threshold.

The minimum pulse width is limited by the output gate's capabilities. You'd get similar performance from one higher-drive 74AC08 or 74AC11008 with two of the gates used to buffer your trigger signal: one buffer feeds the final AND stage directly, the other feeds the AND stage through a length of coax you cut to get the delay you want. Both signals should have similar terminations to achieve similar levels for both paths and keep the reflections out of the coax. With a 5V supply and two 100 ohm resistors for

50 ohm cable (or two 150 ohm resistors for 75 ohm cable) connected as 5v - R1 - termination point - R2 - gnd you'd get the swing you need with the available drive current. Soldering is still needed. This approach is only an extension of the cuircuit you referenced.

Point is - you can achieve your goals in many ways. If your goals are generic, we can only guide you to generic solutions.

I'd ask again: what do you want? It might be something someone like me could spend $15 at digikey and give you an ugly board with beautiful single-purpose output pulses.

If you look again at my original list, you could tell us what's important and what's not.

- John_H

Reply to


First, thanks for your feedback. While I'm comfortable soldering DC circuits and circuits up to the MHz level, I know enough about RF to want to avoid my soldering ability getting in the way of a clean signal. It seems to me that it would be easier to solder a short on a coax without causing problems, so a delay line with AND gates may be something I could do. Maxim had an interesting app note on how to prototype with surface mount components and a copper plane board, so perhaps that is not as difficult a thing as I previously thought.

Here is a more detailed description of what is important. The circuit needs to supply pulses that range from 1 ns to ~10 ns, preferably adjustable between runs. It is important to hit the 1 ns, not the 10 ns, as we already have a benchtop function generator that can provide

10 ns pulses.

The circuit will modulate a CW optical signal, which will then go into an optical amplifier system operated near saturation for stability. Because of that, it will amplify small optical signals more than large signals, such as would be caused by ringing or other electronic effects.

For optical reasons, we would prefer the PRF to be 10KHz-1MHz, adjustable between operation, which would be a temporal pulse spacing of 1-100 microseconds. This is obviously much larger than the 1 ns pulse width, but perhaps could be triggered by an external function generator. I also think that drift in the non-pulsed time periods may be suppressed somewhat by adjusting the DC offset, which can be done externally. I'm not even that worried about whether the repetition rate stays constant, as long as it stays within about 50%.

The pulse width could be either square, trapazoidal, triangular, or Gaussian. Due to the optical amplification, it will end up somewhat Gaussian in the end anyway. Peak voltage could be anywhere from 100 mV to a few volts into the 50 Ohm load of the amplifier. I am able to provide adjustable regulated DC voltages for power in the tens of volts and up to an amp, which should be plenty.

So, hopefully that should characterize pretty much everything about the system I am looking for. That would be nice if it were just $15 digi-key parts, as long as I could construct it without destroying the output.


Reply to
Matt-5589 Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.