Will NI 9265 solve my problem?

[gguo2]

I would like to drive 5 laser diode using some kind of current source in a sequential manner. Each laser diode need to be driven by a current of 2 mA to 12 mA in a 100 kHz saw-tooth waveform.  I want to switch the output between each diode sequentially. For example, after driving the first diode for one period of saw-tooth current, the current switch to drive the next diode(so the first one is off). After driving the fifth diode, the current switch to drive the first diode again.  

[crossrulz]

 The 9265 has a Max sample rate of 100kS/s.  So it will not be able to handle your waveform speed.  You need a sample rate more around 1MS/s.

[gguo2]

Thanks for the reply. Here I've got two more questions.

1. Is NI 9265 able to output a saw-tooth or sinusoidal current instead of constant current?

2. Is there any device that has a sampling rate above 1Ms/s?

[Henrik_Volkers]

I don't know a NI unit with current output and 1MSPS...

There are a lot of voltage out units with >= 1MPS  look for an AWG.

All you need is a voltage to current converter or voltage controlled current source with a bandwidth (for a sawtooth @100kHz) ~ 10 to 20  times higher .. say 2MHz.

EDN has published a design idea for a HF current source AFAIR.

And there's a small market for laser diode drivers with voltage input (current and/or photodiode feedback control)  .. not on the cheap end... but have other goodies to protect your diodes.. LD don't like current/voltage spikes ...easely turn dark or to a normal LED (BTDT :D)

switching the diodes can be done with a OptoFET .. if you use the PD feedback some more multiplexing is involved 😉

the 9265 can update the current output with 100kSPS with the wfrm you define.

If you light your LD... do you measure something? how fast can you sample?

[Henrik_Volkers]

If you want to MUX (multiplex) the LD driven by a current source, you have to make shure that the the current source doesn't run open (voltage run up) and is switched to the next LD -> voltage/current spike -> LD fried 😞

If possible, you can chain all LD serial to one current source and use OptoFETs to shortcut all but one LD.  OptoFETs should be closed by design ...

For LEDs I used this simple driver (OP needed that can drive the current (THS 4041?), buffer cap(s) and Rin (for 0V at open input) missing ) but I never validated it for save LD use....

But there are many circuits around ...

[gguo2]

Hi Henrik,

Thanks for your reply. May be we don't want to switch the current since it will cause a spike. What we really want is we can collect signal from each diode sequentially (even though the five diodes might be driven simultaneously). When we collect signal from one diode, we want the other channels off.

[Henrik_Volkers]

If I would need n (five) 10µs voltage ramps to control fixed defined current sources (no fancy, flexible fast change in these parameters) on low budged, but some time to build myself stuff, I would build 

n analog ramp generators (const. current into a cap, cap shortcuttet by a transistor controlled by a DO plus a bias + buffer with shut down (same DO signal) . Or one ramp generator and an analog mux ... but since LDs vary maybe individual ramps are better. Ten pots (five times current and bias)

(but I started burning my fingers with a solder iron >40 years ago 😉 )

More flexible would be 5 fast AO channels , however at 1MSPS update rate you would get a staircase with 10 current steps. (assuming you need a fast turnoff ramp, analog filters don't help much)