I have been playing with the attached constant current driver. I am pleased with the way it performs.
The "problem" is the MOSFET. I used it because I have it. Even though it is designed to work as a switch it seems to perform well in this circuit. In searching for a potential alternative I have come to the conclusion that I have no idea what I am doing.
Is it worth finding a replacement?
If it is, what are the important parameters?
The MOSFET is an IRLML6344TRPBF. The maximum possible current is 30mA. At this point, the target current is 12mA. Vcc will be 1.8V to 5.0V (the LED is unlikely to light below Vf; the current will fall below 12mA as Vcc approaches Vf).
And thank you for the inspiration. I started with the double BJT current limiter you posted a few times. It also works well. I used 2N3904s. The only reason I ended up with a MOSFET is it works a bit better at lower voltages. I think it also uses a bit less current overall.
I highly recommend LTSpice. It did a very good job modeling the various circuits I wanted to try.
So what characteristics would make a FET (or indeed a BJT) specifically appropriate as a "switch" vis a vis a linear power amplifier?
High gain so it switches smartly from "off" to "on"? Well, not really critical, as you want linear amplifiers to have high gain also.
Linearity? Presumably less important for a "switch" but most devices are not particularly linear in any case, this is compensated for by employing negative feedback.
Low saturation voltage or resistance? Yes, but that is also useful for linear amplifiers.
Low gate charge retention (capacitance)? Yes, but that is also useful for linear amplifiers.
The only real difference between a pure on-off switch and a linear amplifier, is that the switch is expected to spend most of its time either fully saturated or completely off, which are the states in which it dissipates minimal power, so the power dissipation required for a switch is less than as a linear amplifier (which is of course, why we use switchmode power converters).
...which are the states in which it dissipates minimal power...
So the packaging and mounting may be different? A "MOSFET designed to switch" may be in a smaller less thermally conductive package?
IR does brag about how itty-bitty their power MOSFETs are (and how low the Rds(on) is). The datasheet for all of the other "power" integrated circuits I have found include information about heat sinking into the PCB (including thermal pads in the copper layer). The datasheet for this MOSFET does not. Just three tiny pads. I assume that means they expect the MOSFET to run "cool".
The low value of the Ron ( on resistance of the FET ), is important in a switching application as it reduces the power dissipated but it is not as important in an analogue application because there will be natural heat dissipated due to the mode of operation.
So you could say that a package with a low thermal resistance between junction and case would be better for linear operation, but you could equally say it was important for a switching application.
It is often the dissipation limits of the package that limits the "true" maximum current in a FET. Those headline figures on maximum current are as about reliable as any tabloid headline. That is the spin often obfuscates the truth.
Thank you both for the replies. They have been extremely helpful.
Is there an advantage to having a higher gain with the other transistor (Q1)? I assume a higher gain would reduce the overall power consumption. Less current would have to cross from the right to the left through Q1's base?
Edit #1: As far as LTSpice is concerned, it makes no difference.
Edit #2: Spoke too soon. 38.701274nA versus 28.208926nA. So it does make a very miniscule difference.
Edit #3: At 4.42mA that is 0.000237% "waste". I believe it is safe to call that negligible.
Edit #4: Yup. According to LTSpice, there is a very miniscule improvement with the 2N5089 vs the 2N3094.
