Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

How to do laser diode matching in Sonet ?

Not open for further replies.


Member level 2
Aug 17, 2002
Reaction score
Trophy points
Activity points
laser diode matching

An optical guy asked me to see if there is any other way to do laser diode termination. Application is SONET OC192 DML.

Conventionally, just tline and a series resistor with laser diode to make a 50Ohm termination. This approach has the catch of dissipating most power in the resistor.
A modified version is to use a wire-wound resistor to do a complex impedance matching since normally laser diode has some parasitc cap.

The guy hope he can use quarter wavelength Tlines to transform laser diode low impendance into 50Ohm, without the extra resistor. However, I feel uncomfortable with this idea since the matching would occur only in narrow band(I would give a guess of 1G~8G). However, SONET data contains frequency component as low as 40KHz. Severe BW limitation and ISI shall appear in this approach.

Heard a company called hytek claimed to have something called "backmatching", which does not require 50Ohm termination from laser side. No details online.

Any suggestion?

Use laser diode driver IC.

Be careful, laser diodes are very nonlinear and temperature dependend. If you need maximum lifetime and power, you need feedback circuit.
Very interesting are VCSEL diodes.


See datasheed of some IC manufacture, I'm find that MAXIM have good.
It's very hard to drive laser without feedback because power depends of temperature, and only with IC you can on right way modulate laser.

I think I need clarify the problem here:
it's about termination of laser diode, not control scheme.

I understand pool_77 and goxy basically talking about the control of bias and modulation current -- usually applying dithering tone at low freq. The same idea can be applied to both EML and DML. I recalled Maxima and ADC have chips for this.

But here I'm concerned about the termination. That is, can we do a good RF termination without the extra resistor in series with laser diode?
Given broadband signal under process, I doubt it.

BTW, VCSEL is available only for (very) short reach so far.

Anyway, thanks alot for comments.

Laser Meshing on 1Ghz with TL you can find in Genesys Examples.
Also can be simulated in MWO:
PCB:FR4-Er:4.5;Los tang.:0.0004 ; Metal Tick.:1.42; Rough.:0.094 ;Height:31 1000x500 mil

Goxy, thanks for the file. It's very interesting.

Usually, the laser manufaturers make the laser in series with a resistor to make a 50Ohm dynamic impedance.

For most good drivers, they have 50Ohm output. I see some people prefer using open collector to drive laser (like the new Maxim DFB driver) in order to save power; it's kind dangerous if the trace is long enough between driver and laser, and the laser is not well matched.

From laser maker's point of view, since they can't make sure how long the trace would be and what the driver's output impedance, it's always helpful do to 50Ohm termination.

I would first try to use @DS to do 4 quarter Tlines (a low pass filter type impedance transform) simulation to see if my friend's idea would work. And then try EM simulator if necessary.

Again, thanks a lot for your effort and time.

I would keep you posted about my progress.

You are right for laser termination.
If you find some low cost high power (above 50mW) laser on 1550nm please send post here.

Not open for further replies.

Part and Inventory Search

Welcome to