modulateable laser diode drivers

Mike Noone schrieb:

It is standard task in CD and DVD drives to modulate laser diodes. When asked for something specific, you can look at Sam's laser faq:

I have used the IC_Haus chips for low power applications and the IXYS MOSFETs and gate drivers if more power is required.

Richard

Well, at least this year they're asking their homework questions at the beginning of the semester, rather than at the beginning of finals week. ;-)

Cheers! Rich

The rise time is a function of the bandwidth. The old rule of thumb is

Are you going to pulse the laser, or do the sine method?

Is this hacking, or a project for money? Hopefully it is hacking as I would like to see the results of what you come up with.

Hello Mike,

Analog Devices offers laser diode driver chips. ADN2841 and things like that. Most are under $20 but if that's too much you'd have to roll your own. Basically a current source where current is steered into ground until you want a laser pulse. Now the steering transistor lets go for a few nanoseconds or whatever pulse length you'd like, then closes again.

As you have recognized the current through the laser needs to be limited and regulated. There is a photodiode feedback on the laser diodes for that, usually. Stabilization is not a trivial tasks if you are dealing with occasional pulsing. Best to read up on that stuff before a few hundred Dollars go poof.

Usually you'd also need a controller for the thermo-electric cooler. AD and others carry TEC chips for that, same price range. Unfortunetaly it all goes towards PWM which isn't so cool should phase noise be critical in your case (but can be done).

hmm. we have a system at work that uses an inline high frequency transformer to modulate the HV line going to the laser. its a simple basic laser and the HV supply is pasted through a round core a few turns, on the other side is a pulse amplifier we use to TX a signal.. it only modulates the laser at a small percentage due to the fact the simple laser unit wasn't designed for a large modulated signal like this in this manner. at the other end, it simply goes into a detector with slow AGC so that the AM signal can be detected. it wasn't designed by us but it works great. i don't think a 100 mhz could be done using the technique above, well not on the lasers we are using..

Well, at least in this post you're jumping to incorrect assumptions immediately, instead of asking questions first to find out the right answer.

-Mike

You are talking about modulation, and not 100% on/off driving, right? The way I do that is quite simple. I use a standard moderate-speed light-sensor feedback laser-diode current-source controller, and add a modulation network at the laser diode. I operate the laser at about half current, as appropriate.

The laser impedance is very low at these currents, a few ohms. The low impedance allows me to treat this node like a summing junction, so that adding a 50-ohm resistor in series with an ac-coupling cap from the coax cable carrying the RF modulation provides both the RF signal to the diode, and well terminates the 50-ohm coax. Using small dimensions, I have been able to modulate low-cost CD-ROM lasers up to 200MHz in this fashion, and done a bit differently in a quick lash up, up to 1.5GHz (I may have modulated the laser successfully well above that, but the photodiode I was using wimped out).

From my Jan 10 2005 s.e.d. post, "The 664nm red DVD laser was an Hitachi HL6504FM and my detector was an Optek OPF480 PIN diode, biased at -100V, measured into a 25-ohm load (double-end termination) with an HP network analyzer.

"Components were 1206 SMT hand-soldered with zero-distance spacing. Bias-tees were 12GHz-bandwidth Picosecond Pulse Labs. A first attempt at a PCB stripline replacement only goes to 600MHz so far."

. bias-tee laser . ________| |______,---------, . ________|-||-+--|______(-50R-|>|-' . 50-ohm |____X__| \\\\ mirror PD bias . coax | \\\\| optics 100V . 50R //| & etc | . | // __|____ . laser // ________| X |___50-ohm . current ,-|>|----+-)_______|--+-||-|____ coax . supply +-||-50R-' | coax |_______| term. . '----------' bias-tee

Jamie Morken answered, "Cool! That looks like a neat circuit, I am buying that same photodiode from Digikey (part#: 365-1029-ND)"

Comment: In the latter approach, the bias-T isolates the 50-ohm coax RF signal from the low-frequency current source, all on the 50-ohm line. But, making a good bias-T can be tricky, and using the low-Z at the laser diode instead should be an easier approach, perhaps not good to 1.5GHz, but certainly to 200-400MHz.

Hi Mr. Hill - Ideally, I would like to be able to drive the diode with any waveform. For now, my plan is to pulse it with a square wave at about 20mhz, with approximately a 2 or 4% duty cycle (pulse width = 1 or 2ns). This signal will be going into an Acam TDC (Time to Digital Converter) chip, probabaly the TDC-GPX

Do you think your circuit would be well suited for my application? I must admit that I am not as good at analog circuit design as I would like to be. (strength is really in digital embedded systems)

Thanks,

-Mike

I am currently planning on pulsing it with a 1-2ns pulse, every 50ns or so. These numbers can be changed if necessary, but the important thing is the rising edge time, as it will be starting a time to digital converter and then the returning pulse will be stopping it.

The project is for a robot I've been working on for the last 15 months or so. I want to give it the ability to avoid obstacles and map out areas. If you're interested, there are some pictures, schematics (outdated and with some mistakes), and some really old videos of it here:

Thanks,

-Mike

Analog's offerings look quite good. I especially like the ADN2525, with a 24ps rise time! Unfortunately, I can't find it for sale in single quantites, nor do they seem to sample it. I plan on contacting them on Monday - but being a student that probabaly won't go too far. Some of the others look pretty good too - I especially like the ADN2870, which is available for about $10 in single quantities which is completely reasonable.

I have been thinking about going the TEC route. I think initially I will skip that as it is added cost and complexity, and I don't think it will be terribly important as I don't particuarly care about having the same power. Does that sound like a logical approach?

Thanks!

-Mike

For moderate speed on off switching a common way is to just use a transistor to drive it or possibly a few very fast cmos gates in parallel.

You need to limit the current of course, a fixed resistor is simple but a bit limited, an easy way to do this at high speed is to provide a constant current source then divert this away from the Laser with a diode conected to the cmos gates or a transistor.

If you have low duty rate im not sure how you would implement optical power feedback.

For high speed sinewave modulation as Win has sugested its a good idea to use a DC bias source and ac couple the modulation signal.

Colin =^.^=

Looks nice, did you ever measure the capacitance of the laser diode ? The data sheet I found for that diode didnt mention capacitance.

Ive managed to modulate a cheap pointer type laser diode @ 1ghz with a simple single transistor driver, originaly designed for 10mhz, although it was quite easy to get it to work at 100mhz, it required an ever increasing pull up current to get it to 1ghz, one problem is the capacitance of the diode im using, rlt6305, wich measures at over 50pf wich is surprisingly high, coupled with the lead inductance even with the diode next to the driver transistor must be quite a few nh.

When I finally got it to work at 1ghz I measured the RF voltage accros the diode at the ends of the lead as >3vp-p but at the base of the diode case it was less than 1vp-p.

My next layout im cropping off the laser diode leads and soldering the case directly to the ground plane, Im now also using a dc bias and AC coupling the RF, using a chain of MMIC and a big Gaas Fet.

I try and avoid transmision lines if at all possible as its often hard to wideband match the terminations without lossy pads etc. The oscillator is at the start of the MMIC chain and part of the voltage accross the laser is easily fed back to the PLL via transmision line.

Im not sure if the limiting factor is the capacitance or the internal speed of the Laser wich makes it necessary to drive the laser with a larger p-p voltage at these speeds, wich then makes the capacitance more of a problem, maybe driving the diode off harder doesnt speed up the laser much and I just need a better laser.

Anyone know of any faster lasers (5mw visible)?

Colin =^.^=

Well, the above 50-ohm setup went to 1.5GHz, so that works out to no more than 2pF. The laser dies are very small, and fragile, too.

Hmm, I suppose that can work if the laser is mounted right there.

If I had presented my circuit with a 0.5ns-risetime pulse, I imagine it would have responded just fine. But remember, I was prebiasing the laser, which is important. I think you can bias it at low light levels, say 10 or 20% of its nominal output level, but for fastest response I'd guess it has to be above the lasing threshold.

Ive measured 3 different types of laser diode and they come out at 30-60pf, I measured a varactor diode too just to make sure my LCR meter could measure this accuratly and it was spot on. I found it quite surprising it was so high and still not sure if I beleive it as it can work to 1ghz.

I found a data sheet for the diode you mention from alldatasheet.com wich shows the modulation response and its dropping very fast at 1.5ghz, but they dont specify the test set up.

Im wondering if I should switch to a vcesl as ive heard people mention a capacitance as low as 2pf, and theyr specd at 2.5gb/s but there doesnt seem to be any visible ones readily available.

It helped with reducing inteference too, wich is mainly why I did it in the first place as my received light level is very low.

Colin =^.^=

That's correct, I shouldn't have used 50 ohms, sorry. In fact the dynamic resistance could be considerably under 1 ohm.

If you measured the capacitance at 0V or with a reverse bias, you probably got a lower value than it has when operating.

My measurements were so much faster than the datasheet, about 3x, I suspect they in fact were suffering from slow detector response.

just found a few here

some go upto 155mhz but some of the datasheets with internal circuits give you ideas on how to build your own.

Colin =^.^=

You are going to need to sweep the laser, much like a bar code reader. My gut feeling is you are going to have difficulty measuring short distances with a pulse technique. Have you investigated clock jitter? Worse yet, from the avi, I wouldn't call that smooth motion, so the crystal will have microphonic tendancies. That is, vibrating the crystal will modulate it. You can average out the readings I suppose to get around the crystal vibration.

Videos are of the robot's first steps. There were countless bugs in the code at that point, and motion was sloppy. If you look carefully in the video you can actually see that one of the legs turns off during certain movements - just becomes a complete gimp leg. Was a major software bug that I became aware of after taking that video.

Anyways - I was planning on having the robot stop, and then scan, then start again. What clock jitter are you referring to?

Thanks,

-Mike