Author Topic: Laser Diode driver - kHz modulation & 1% duty cycles  (Read 8652 times)

0 Members and 1 Guest are viewing this topic.

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Q
« Reply #25 on: April 26, 2019, 01:01:49 am »
(1)  Is there anything in this circuit that strikes you as faulty or terribly incorrect?  Are any of my assumptions flawed or incorrect?

When the output is off, the external capacitor used for frequency compensation is charging up.  This is called integrator windup and it results in the slow recovery.

Quote
(2) Is there a better way to go about compensating this OPA for a better transient response and reduce the ringing on the rising edge?  Can I do this empirically as before?  I do not have access to a VNA to do fancy stuff like the loop gain/phase measurement ...

The configuration of Q1 is a major problem; it is adding uncontrolled voltage gain within the feedback loop.  Reduce the value of the series base resistor so that it only suppresses parasitic oscillation and add a low value resistor in series with the emitter to stabilize Q1's transconductance.

With Q1 fixed, the external feedback capacitor should not even be required removing the slow ramp up.

Quote
(3) Why does the OPA seem to quickly slew to 500mV, but then exhibit that slow ramp-up to threshold?

I suspect this comes from the operational amplifier recovering from saturation of its internal circuits.

Quote
(4) Can this integrated OPA do what I am asking of it, given its GBWP and slew rate?

The topology is the problem and not the speed of the operational amplifier.  A faster amplifier will not recover from integrator windup any more quickly.

Quote
(5) Any other ways to improve this circuit by adding or deleted external elements?

I might use a different circuit configuration.  Replacing the operation amplifier with an operational transconductance amplifier would be one option but they are not common.  (1) This provides a direct current output so Q1 is no longer needed and frequency compensation is considerably simplified.

There is a trick I would at least evaluate though.  Ground the operational amplifier's output through a low value resistor to a low impedance source, and then use the positive supply pin to drive the laser diode.  This has the effect of replacing Q1 with the high side transistor of the operational amplifier's output stage and the output resistance now controls the transconductance.

U111 in the first example shown below gives some idea of what this looks like.  The power supply pins become current outputs, the normal inputs stay high impedance inputs, and the output becomes a low impedance input.

The second example below shows the same idea but the output of the operational amplifier is being used for local feedback.  Essentially this circuit configuration is an operational amplifier with a current feedback amplifier added to the output.

Quote
(6) Would an NMOS work better here, if I were somehow able to get the circuit to not oscillate?

A bipolar junction transistor is actually a better choice here because of low and predictable threshold voltage.

Quote
(7) Eventually I would like to get this circuit to work up to 10kHz modulation rates with 1% min duty cycles.  Would that be possible given this OPA?

Absolutely, the operational amplifier is more than fast enough.

(1) You could use a 723 voltage regulator; it has one.
 
The following users thanked this post: smoothVTer

Offline smoothVTerTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: Laser Diode driver - kHz modulation & 1% duty cycles
« Reply #26 on: April 26, 2019, 02:28:25 pm »
not sure if it's been said/linked already but the usual constant-current circuit used for sciency-stuff is:
Libbrecth-Hall http://www.submm.caltech.edu/kids_html/DesignLog/DesignLog179/MillerMUSICReadoutDocs/HEMT%20Power%20Supply/Libbrecht%20and%20Hall,%20A%20Low%20Noise%20High%20Speed%20Diode%20Laser%20Current%20Controller.pdf
Erickson https://arxiv.org/abs/0805.0015
Seck https://arxiv.org/abs/1604.00374

I have not seen this paper despite googleresearching for a couple of months now.  A lot to digest in this paper!  Thank you, I hope to gleam some insight from this. 
 

Offline smoothVTerTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: Q
« Reply #27 on: April 26, 2019, 07:53:17 pm »
(1)  Is there anything in this circuit that strikes you as faulty or terribly incorrect?  Are any of my assumptions flawed or incorrect?

When the output is off, the external capacitor used for frequency compensation is charging up.  This is called integrator windup and it results in the slow recovery.

Quote
(2) Is there a better way to go about compensating this OPA for a better transient response and reduce the ringing on the rising edge?  Can I do this empirically as before?  I do not have access to a VNA to do fancy stuff like the loop gain/phase measurement ...

The configuration of Q1 is a major problem; it is adding uncontrolled voltage gain within the feedback loop.  Reduce the value of the series base resistor so that it only suppresses parasitic oscillation and add a low value resistor in series with the emitter to stabilize Q1's transconductance.

With Q1 fixed, the external feedback capacitor should not even be required removing the slow ramp up.

Quote
(3) Why does the OPA seem to quickly slew to 500mV, but then exhibit that slow ramp-up to threshold?

I suspect this comes from the operational amplifier recovering from saturation of its internal circuits.

Quote
(4) Can this integrated OPA do what I am asking of it, given its GBWP and slew rate?

The topology is the problem and not the speed of the operational amplifier.  A faster amplifier will not recover from integrator windup any more quickly.

Quote
(5) Any other ways to improve this circuit by adding or deleted external elements?

I might use a different circuit configuration.  Replacing the operation amplifier with an operational transconductance amplifier would be one option but they are not common.  (1) This provides a direct current output so Q1 is no longer needed and frequency compensation is considerably simplified.

There is a trick I would at least evaluate though.  Ground the operational amplifier's output through a low value resistor to a low impedance source, and then use the positive supply pin to drive the laser diode.  This has the effect of replacing Q1 with the high side transistor of the operational amplifier's output stage and the output resistance now controls the transconductance.

U111 in the first example shown below gives some idea of what this looks like.  The power supply pins become current outputs, the normal inputs stay high impedance inputs, and the output becomes a low impedance input.

The second example below shows the same idea but the output of the operational amplifier is being used for local feedback.  Essentially this circuit configuration is an operational amplifier with a current feedback amplifier added to the output.

Quote
(6) Would an NMOS work better here, if I were somehow able to get the circuit to not oscillate?

A bipolar junction transistor is actually a better choice here because of low and predictable threshold voltage.

Quote
(7) Eventually I would like to get this circuit to work up to 10kHz modulation rates with 1% min duty cycles.  Would that be possible given this OPA?

Absolutely, the operational amplifier is more than fast enough.

(1) You could use a 723 voltage regulator; it has one.

I will implement some of your suggestions here, David.  Already have had some success with the base resistor swap out ... this and some other changes from replies above have been helpful.  Once I figure this out, I will report back here with an updated status. 

I've been keeping scope shots along the entire process for posterity and will post back some of my findings.   

Thanks everyone!  Couldn't have gotten this far without all ya'll.
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2850
  • Country: se
  • Country: Broken Britain
Re: Laser Diode driver - kHz modulation & 1% duty cycles
« Reply #28 on: April 26, 2019, 09:43:31 pm »
(7) Eventually I would like to get this circuit to work up to 10kHz modulation rates with 1% min duty cycles.  Would that be possible given this OPA?

With a 12pF photo diode and a 3.5MHz op amp 1us wide pulses will just be a little rounded hump. I use just a PD and 1 or 2 fast transistors to view square-ish 1us pulses.
.  That took much longer than I thought it would.
 

Offline LaserSteve

  • Super Contributor
  • ***
  • Posts: 1354
  • Country: us
Re: Laser Diode driver - kHz modulation & 1% duty cycles
« Reply #29 on: April 29, 2019, 11:54:02 am »
I received your PM, fast drive is not my cup of tea, unfortunately.

Having had my share of blown diodes due to poorly designed drivers in the laser show industry, I can suggested the following general additions.  You haven't lived until 16 single mode red laser diodes die in a brief flash and go LED. This happened  due to defects in a brand new commercial driver board, and at 35$ a diode...  Some of my diodes cost 200$ each, so I tend to carefully obtain commercial drivers. When I'm not in the lab, my laser diodes are used in arrays for combination into one beam, so I tend to drive 4 diodes in series at a time.  I only need 30 Khz at up to 1.5 amps,  and no light feedback.  I buy the drivers  from a vendor known to not have issues.

So I suggest the following, as my concerns are more practical, and not about the driver bandwidth:

Add a  Lasorb,  make sure you have a fast, low Vf,  reverse protection diode in series with the device,  An independent upper limit current clamp circuit is a must, and insure you are  ramping the power to the output stage after you know the op-amps in your circuit have stabilized.   Thus some form of soft start. 

A warning about working with LDs on the bench,  the little 10 to 100  uf capacitors across the current limited output stages of bench power supplies tend to store charge and blow up diodes, as does the surge that many  PSUs have during  startup.  In other words, commercial constant current bench supplies often have only the steady state current well defined.   :-\

If I could, I'd have a normally on FET (Depletion?)  of some kind across the diode, until the circuit is stable.  Normally closed mechanical shunt relays have traditionally been used for this, shorting the diode till the laser head  cable is connected, or the PSU is stable, and they just are not fast enough.

The Lasorb is a static discharge  protection device that has a structure like a Mosfet/SCR hybrid and is triggered by a fast  DV/DT.   I know the inventor personally, they do work, and I strongly suggest having one within 6 cm of any valuable  LD. 

www.lasorb.com

I would have also sent you to  Libbrecht and Hall as a start. There is a paper that was written in response to L&H that is worth a read:

An Ultrahigh Stability, Low-noise Laser Current Driver with digital control Christopher J. Erickson, Marshall Van Zijll, Greg Doermann, and Dallin S. Durfee Department of Physics and Astronomy, Brigham Young University.
REVIEW OF SCIENTIFIC INSTRUMENTS 79,  2008

Good, it is on line:

https://www.physics.byu.edu/faculty/durfee/Publications/Erickson08au.pdf

Which was a follow on to this:

https://tf.nist.gov/general/pdf/739.pdf

If you don't have academic access, emailing the professor will generally result in your receiving a preprint, if they are allowed.

If I had to design a very low current diode driver, I'd drive the diode off a high side current source and shunt it with a FET or Bipolar for the modulation, being careful to ensure there is a always a below threshold "leak" of current through the LD to protect it from surges.

Where you really should just go is here:

I'd suggest taking a look at  APC style LD driver chips with PD feedback  by  ic_Haus:

https://www.ichaus.de/keyword/Laser%20Diode%20and%20LED%20Drivers

Take a look here for addressing some additional concerns, his writeup is pretty good..

http://hololaser.kwaoo.me/laser/red_diodelasers.html#LDdrv
http://hololaser.kwaoo.me/electronics/myLDdriver.html

One last note, the traditional current source used for testing LDs by the laser show industry and laser hobbyists is a LM317 configured as a current source with just a non-inductive  current sensing resistor.  If the leads are kept short and the input power is filtered, this combination tends to lead to long diode life when used by amateurs.  Evidently National's  LM317 has the right startup characteristics, but others work as well.  Hence my note on shunting the diode for regulation. 

Steve



« Last Edit: April 29, 2019, 01:41:01 pm by LaserSteve »
"When in doubt, check the Byte order of the Communications Protocol, By Hand, On an Oscilloscope"

Quote from a co-inventor of the PLC, whom i had the honor of working with recently.
 

Offline LaserSteve

  • Super Contributor
  • ***
  • Posts: 1354
  • Country: us
Re: Laser Diode driver - kHz modulation & 1% duty cycles
« Reply #30 on: April 29, 2019, 02:17:40 pm »
I finally found something EE worthy

Design and Stability Analysis of a CMOS Feedback Laser Driver
P. Zivojinovic, M. Lescure, and H. Tap-Béteille
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 53, NO. 1, FEBRUARY 2004

IEEE is anal on uplinks, or I'd post it for you..  However you may be amazed about what shows up in Google Images for a preview..
While the pictures show an array of FETs, the circuit is modeled on traditional components.

I'm pretty sure from reading it that they modeled it on the bench before making the IC...  >> then 1 Mhz from my reading.

Steve
« Last Edit: April 29, 2019, 03:18:35 pm by LaserSteve »
"When in doubt, check the Byte order of the Communications Protocol, By Hand, On an Oscilloscope"

Quote from a co-inventor of the PLC, whom i had the honor of working with recently.
 

Offline smoothVTerTopic starter

  • Regular Contributor
  • *
  • Posts: 151
  • Country: us
Re: Laser Diode driver - kHz modulation & 1% duty cycles
« Reply #31 on: June 03, 2019, 02:58:07 pm »
Sorry its been a long time.  I mostly figured out this laser driver, and this post is to update my findings and trial-and-error scope shots.
The topology I settled on came about after taking suggestions from prior posters:



At first pass, the circuit didn't have the schottky on the OPA output;  there was no Re;  there was no Rpe;   Rbase was a few kohms;
 and Cf was about 1nF.   Theory of operation is during the laser "on" time, Q2 is off and all feedback current is generated by the monitor photodiode.  OPA1 pushes enough base current to bring the inverting and noninverting inputs to the same potential.  During the laser "off" period, Q2 injects a small current into the inverting node, raising the voltage there, and then OPA1 turns current off to the base of Q1.  Consequently, this puts OPA1 into saturation because the + and - inputs are rather far apart in potential.



The yellow trace is the base voltage;  the green trace is the voltage at the inverting node;  the red voltage is a pseudo-differential voltage across the laser diode by taking 2 scope probes across the diode relative to GND and doing some math in the scope.  As you can see, (1)  The output oscillates  (2) There's a large negative going excursion of the OPA output as soon as Q2 turns off; I think this has to do with the Cf capacitor's stored charge suddenly being reversed, as well as the "integrator windup" effect.   (3) There is a long slow ramp of of the base voltage, not desirable for generating a sharp, fast laser pulse.

The first thing I changed was to put a schottky on the OPA1 output, to prevent that large negative going excursion:



This seemed to help the recovery time from saturation.

Next, I empirically determined a Cf value in my circuit that would just quell the oscillations on the rising edge of the base voltage.  This improved my waveforms considerably:



Still wasn't happy about that long, rising edge.  Another poster suggested to (a) lower the base resistor just enough to prevent parasitic oscillation and (b) add an emitter resistor to stabilize Q1's gain.  Well, all this sort of reminded me of BJT design class and made a lot of sense. 




That rising edge was a little strange looking so I had to go back and forth and trying out different values for Rb/Re. 
After playing around with values for Rb and Re, and lowering Cf even further, I finally got something approaching a good waveform with good turn on/turn off characteristics:



The laser diode voltage falling slope is determined by a large part to Rpe ... Rpe is basically there for that purpose, plus a larger purpose of always allowing a tiny bit of current through the laser diode so it is faster to turn on from the off state, with less of a transient excursion.  Of course, power consumption was an issue in my design so it had to be a balance between response and power consumption.

I am sure there are many, many things I can improve on this circuit.  For example:   I used an AD8029 for OPA1, for the only reason that (a) it was available in my lab and (b) it is a tiny package size and space was a consideration. But perhaps there is another OPA better suited for this role?   

Also:  Q1 I found the NPN's that work best with this circuit are high-beta, low Vce NPN's. Why is this so?  This one in particular is a 2SD1979GSL NPN.  I basically soldered and de-soldered NPN's here until I found something that seemed to work best.   What combination of characteristics are best for Q1 in this application?    Some NPN's I tried oscillated no matter what ... some had such a slow turn-on that the 2us pulses above never reached steady-state. 

Also:  Re adds to the power dissipation of the entire driver; it stabilizes Q1's Vbe and thus its gain, but also is in series with the laser drive current, raising the voltage dropped by it, and this necessitates a higher laser supply voltage.   I settled on a value of 15 ohms.   I would like to lower this resistor as much as possible but it seems the lower value I choose, the worse that rising edge of the base voltage appears.    How can I choose Q1 so that I minimize Re as much as possible?

Thanks everyone for contributing, I hope these screenshots sort of show what was happening during the design/tuning process of this circuit.


« Last Edit: June 03, 2019, 03:03:23 pm by smoothVTer »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf