7

I am designing a constant current source to drive a laser diode. I need it to be adjustable (from 0 to 100mA max) and stable enough as I will be using it as a light source in a Foucault mirror tester for my amateur telescope mirror making project.

After some reading I came up with the following general design, however I have some questions.

  1. Is this an adequate design overall (not overcomplicated)?
  2. The signal on the non-inverting input of U5 will be just a DC setpoint. Can I just choose the cheapest op-amp available (lowest bandwidth) or are there other considerations I need to make?
  3. Should I care for loop stability and compensate for it (C2-R3 or other solution)?
  4. The current source needs to be current limited to not damage LD1. Should I size the feedback resistor R4 (with VD) with this in mind, or should I trust the circuit and configure the max current indirectly with R2-RV1 and set R4 for low voltage drop and to get an adequate voltage feedback?

Thank you for your help.

Constant current source

nonius
  • 95
  • 4
  • 3
    it would be better to stick the diode between R4 and Q1. otherwise, temperature changes in beta will change the diode current although the current in R4 is constant. – tobalt Oct 02 '21 at 10:42
  • 1
    @tobalt that wouldn't work so well if the power rails are limited but, maybe I misunderstand your implied circuit. – Andy aka Oct 02 '21 at 10:50
  • 3
    @tobalt. That is a good point if accuracy is such an important concern. Maybe a fact that helps in this aspect is that the low current allows the use of a transistor with high beta. I've simulated an 80C delta with LTSpice here and found a couple hundred uA error with Vce below 1V. – devnull Oct 02 '21 at 11:49
  • 2
    @Andyaka is correct that the required VCC will be increased by the forward voltage of LD1 when the diode is stuck on the emitter side of Q1. If this is indeed not tolerable, you can use your circuit but replace Q1 with a higher beta one as suggested by devnull. I.e use a FET for virtually infinite beta. A JFET in particular will also provide natural current limiting. A MOSFET might be needed if you cant find a JFET with sufficient saturation current \$I_{DSS}\$. – tobalt Oct 02 '21 at 12:49
  • 2
    A similar answer which discusses a precision current source was discussed here: https://electronics.stackexchange.com/questions/563888/how-to-reach-50na-p-p-current-noise-in-200ma-current-source/563902#563902 TL;DR: the main noise sources are in descending order: a) the reference voltage, b) the current setting resistor, c) the VD supply voltage. A and B also contribute drift. – tobalt Oct 02 '21 at 12:55
  • 2
    How much accuracy you need, exactly? If you are setting anything with a potentiometer by hand, you can't really get any better than 5% or maybe 2% of the pot range just because you can't be that much accurate. This is why I think your design is over-over-complicated in this regard. On the other hand, you may consider other important features in your design - like, e.g. exponential regulation (better granularity at low current) or slow rising of the current (safer for eyes, as far as I can tell you are expected to look at the beam, allow yourself a time to blink if the intensity is way off ). – fraxinus Oct 02 '21 at 21:41
  • 1
    @fraxinus, thank you for your question. In practice, I would just adjust the potentiometer by hand until I'm satisfied with the light intensity resulting from the test I'm running. A testing session will have 1h at most, and the light will be captured by a camera, not by my naked eye. So in this sence, I don't need the circuit to be precise as I am not quantifying the laser diode light output as a function of its current, I just need it to be adjustable and stable. – nonius Oct 02 '21 at 22:07

2 Answers2

5

The current source needs to be current limited to not damage LD1. Should I size the feedback resistor R4 (with VD) with this in mind

If the BJT is saturated, the opamp will increase its output voltage until it clips, and output current regulation will not work. The BJT should have a base resistor to prevent both excessive opamp output current and excessive BJT base current when that happens. This resistor should be placed "after" C2, between the top of C2 and the base of the transistor, so that even when the BJT is saturated, this does not influence the high frequency feedback/compensation path through C2.

You can put a high value resistor between the wiper of the pot and ground, so that when the wiper fails open, the opamp gets 0V input, and not an undefined input that depends on the polarity of its input current.

Your choice of transistor is good, as the datasheet plot (top left page 5) shows it still has hFe of 100 with Vce below 0.1V at Ic=0.1A.

So you should size R4 to get the maximum desired current at the minimum supply voltage, with a margin of about 0.5V Vce on the transistor. It is important to select a low VceSat transistor as you did ; if it runs out of steam at low Vce, base current will increase, and that adds up to emitter current which is measured on the resistor, and current becomes inaccurate.

(I mean min/max supply voltage if you use batteries, but if Vcc is regulated, that doesn't apply)

Then, with this value of R4, at maximum supply voltage, and with the transistor fully ON (Vce=0) calculate the current through the laser diode: it should not damage the diode.

If current is too high, then increase the reference voltage to drop more volts across the resistor.

Then calculate max dissipation in the transistor, which will not occur at max current but below that, and make sure the SOT23 package can handle it.

bobflux
  • 70,433
  • 3
  • 83
  • 203
4

Can I just choose the cheapest op-amp available (lowest bandwidth) or are there other considerations I need to make?

The op-amp inputs need to work down to the ground (the most negative) power rail if you want full control down to near 0 mA. The op-amp output also needs to be able to go down towards the most negative rail (ground in your circuit).

If you are concerned about accuracy of the set-point the input offset voltage might be a concern if you decided to use an LM324 op-amp but, it depends on your spec and the value of resistor R4.

Should I care for loop stability and compensate for it (C2-R3 or other solution)?

Almost certainly when using an LM324 (for instance), the loop will be stable because the BJT is operating as an emitter follower and therefore adds no gain and very little relevant phase shift. In other words, when using a low frequency op-amp (like the LM324), you can probably not use C2. However, it's worth leaving R3 in place just in case.

The current source needs to be current limited to not damage LD1. Should I size the feedback resistor R4 (with VD) with this in mind

Yes, I'd do that (belt and braces).

Andy aka
  • 434,556
  • 28
  • 351
  • 777