So I am near finishing my build to create a sunrise/sunset high power LED lighting system for my aquarium.
As I started to build out the constant current component, I can across the following circuit. It is much simplier than mine. (Mine required an NPN N-Channel mosfet, a mosfet driver, an lm317 and various other resistors.) This circuit seems very simple and very efficient with limited overhead. Is it too good to be true? I have posted the picture, but the original posting is located here:Super-Simple Power MOSFET Linear Current Regulator - CandlePowerForums
I'd probably reduce the value of R1 to 100k to improve the transient response or even lower if the PWM frequency is really high.
R1 pulls Q1's gate up turning it on, when the current through R2 exceeds the turn on voltage of Q2 (about 0.6V), Q2 turns on which diverts the current to 0V, therefore reducing the qate voltage to Q1.
When Q3 is turned on, the gate of Q1 is shorted to 0V causing it to turn completely off.
The drop-out voltage will be equal to 0.6V plus the voltage drop of Q1, assuming the current is 0.6/0.82 = 730mA, the datasheet for Q1 says the resistance is 0.27R so the voltage drop-out will be about 0.8V
The current limit value depends upon the forward-bias base-emitter voltage of Q2. Since this voltage is temperature sensitive, the current limit will vary with temperature (≈-0.3% per degree C). So if that amount of change in the current limit with temperature is not a problem in your application, then the circuit should work ok for you. Assuming you are using the current limit to control the LED current, then that amount of variation will likely not be noticeable in the LED brightness level.
Great. To make this work for my situation, I changed the Q2 to a MPS8098_D26Z which can handle 60V. Here are the specs: **broken link removed** Any suggestions for Q3 so it can handle 60V? I saw a couple of options, but they operate at different frequencies than the 2N3906.
BTW Hero999 you have been a great help for me on this project.
If you need to power the circuit from voltages >20V, you need to add a zener diode across the gate of Q1 so it won't get damaged if a surge across the gate exceeds the maximum rating of 20V, a 10V zener will do.
The zener will also mean the voltage ratings of Q2 and Q3 don't need to change.
If you need to power the circuit from voltages >20V, you need to add a zener diode across the gate of Q1 so it won't get damaged if a surge across the gate exceeds the maximum rating of 20V, a 10V zener will do.
The zener will also mean the voltage ratings of Q2 and Q3 don't need to change.
while i do not think he cares about good temperature and accuracy, the best would be a dual transistor instead of his Q2 with the other one just using the BE junction as a diode to offset the tempco of the control transistor.
Okay how is the diagram below? Is there an advantage or disadvantage of the zener over the resistor? I am not mass producing these so cost between a resistor and zener is minimal?
while i do not think he cares about good temperature and accuracy, the best would be a dual transistor instead of his Q2 with the other one just using the BE junction as a diode to offset the tempco of the control transistor.
Is there an advantage or disadvantage of the zener over the resistor? I am not mass producing these so cost between a resistor and zener is minimal? View attachment 41652
A zener means it can be operated over a wider voltage range than a potential divider.
I suggest researching potential dividers and zener diodes and it should become obvious.
From what I could tell, the zener diodes switched faster and provided a more constant voltage than the resistor. One thing I did come across was a couple of posts showed a resistor in series with the zener. Is this overkill?
Another circuit using only BJT's View attachment 41619
A workaround if those transistors won't match properly is to add resistors to emitters for Q1 and Q2.
The disadvantage of a current-mirror, as shown in your attachment, is that you need to put as much current through the control mirror transistor as goes through the controlled transistor. Thus you double the power consumption (assuming the control mirror gets it's current from the same source as the LEDs).
Maybe an ECE background would help me better, but let's see if I am reading this correctly.
A resistor would provide a set amount based on it's resistance properties (i.e. ohms value) and only resist a certain amount of voltage. A zener is rated for a specific voltage and will maintain that voltage based on it's specs.
For my circuit because I have a fairly constant Vin and Vout, a resistor would work fine. However a zener would provide better flexibility if I wanted to change the PS or the number of LEDs in the circuit. Did I get that correct?
The disadvantage of a current-mirror, as shown in your attachment, is that you need to put as much current through the control mirror transistor as goes through the controlled transistor. Thus you double the power consumption (assuming the control mirror gets it's current from the same source as the LEDs).
Read it again and I agree. I will go with the one that started this thread. And BTW Hero999 you have been amazing with helping me. Thanks for eveything. I am learning alot.