Vbe of 2n3904 in led driver circuit

[Kenneth Tan]

I've build this circuit to drive my powerleds. Works very well.
The question is about the Vbe voltage on the feedback transistor.
The schematic states that it is 0.6V
The datasheet states that it is 0.65V min and 0.85V max
How can I best calculate the resistor value for getting a fixed current?

Should I assume 0.6V or do I need to measure it in the active circuit.
And what determines the 0.2V min/max difference?

The leds I use use 125ma at the brightest that I need them (light vs heat) and a pwm is going to be attached to dim them from that setting.
There are 5 leds in parallel (3W COB leds).
So the total current will be 625mA. With a 12V PSU connected the voltage over the mosfet is low and keeps it cool @max current.

 

[Colin Mitchell]

How many LEDs in each string?

 

[AnalogKid]

Vbe increases with base current, decreases with temperature, and increases with collector current. Also, it varies from one part to another. Whatever the textbook says is the correct value for a small signal transistor, know that these things *always* come with a range of values. Figure out which end of the range is the more conservative for your design, start there, then adjust for the real world.

ak

 

[Colin Mitchell]

Don't frighten him with all that gobbleldy-gook.
All you have to do is put a 10R resistor for R3 and drive it with PWM.
Now reduce the resistance until you get the brightness you want by adding resistors across the 10R.
Do not below one ohm.
The transistor is taking so little current that things will not change at all.
The circuit is actually a CURRENT-LIMITING circuit or MAXIMUM CURRENT circuit because it is being turned on and off during dimming.
The MOSFET will not be cool because it is acting in place of the current limiting resistor that should be in the circuit.
I really don't know what current you are drawing because your request is all over the place.

 

[AnalogKid]

Actually, his request was clear, concise, included both context and a schematic, and asked a specific question about a specific issue. I didn't see your schematic.

ak

 

[Colin Mitchell]

Nobody answered the question or even got close.
"Figure out which end of the range is the more conservative for your design," ????????
The fact is the transistor will draw almost no base current because the collector current will be less than 0.5mA and the base voltage only rises above about 0.65v as the transistor is required to deliver more current - towards its maximum.

 

[(*steve*)]

Knowing that there are some variables, I would calculate the resistor value based on a 0.65V Vbe since, as Colin suggests, Ic will be low and thus the required Vbe and consequent Ib will be low.

Vbe increases as Ic increases, and that describes the reason why there is a range of Vbe

The exact Vbe required is also variable between transistors.

After doing the calculation above, you may also find that the exact resistor value is not available, and even if it is, the resistor week have it's own tolerance.

So, do the calculation based on the minimum Vbe in the datasheet then choose a resistor and measure the actual current. If you need a very tight tolerance you might need to adjust the resistor individually and/or use a better circuit.

 

[Kenneth Tan]

Thank you all for the comments.
Some of the comments were very helpful.
I think now that I'm going to use the circuit with it's pro's amd con's, by limiting the led current, based on the highest Vbe and the selecting the current limiting resistor with that voltage and desired max current. The pwm will do the rest.
The aluminium substrate the cob leds are mounted on is screwed into wood shelves so I don't want to heat them up to much.
The initial reason for my question was that the curve in the voltage/current diagram for these leds get goes up exponentially as the voltage reaches it's maximum and the difference in the Vbe has a big influence in the current as we get close to the max voltage.

I use 5 cob leds in parallel. They are 10.5 volt/260mA Max.
No need to go into a in parallel vs series discussion. These Cobs are parallel chains of series leds in a "gel".
I use them @ 150mA each. Total @750mA
I made a spreadsheet showing me the resistor values @750mA ranging from 0.6V upto 0.85V Vbe.
0.8 Ohm to 1.13 Ohm.

Then I reversed the calculation using the min and max resistor and looked at the current it would generate if I do 't know the exact Vbe.
The best result was when I took the highest resistor value at the highest Vbe.
With that it would give me a best case current of 750mA @ 0,85 Vbe.
Worst case it would be 530mA @ 0.6Vbe
The difference in voltage over the leds is 0.5V (10.0 - 10.5V ) with lumens @10V at an acceptable brightness ( worst case maximum brightness)

Love to see comments if I made a stupid error in my reasoning. Learning as I go along...
Thanks again!

 

[Kenneth Tan]

Don't frighten him with all that gobbleldy-gook.
All you have to do is put a 10R resistor for R3 and drive it with PWM.
Now reduce the resistance until you get the brightness you want by adding resistors across the 10R.
Do not below one ohm.
The transistor is taking so little current that things will not change at all.
The circuit is actually a CURRENT-LIMITING circuit or MAXIMUM CURRENT circuit because it is being turned on and off during dimming.
The MOSFET will not be cool because it is acting in place of the current limiting resistor that should be in the circuit.
I really don't know what current you are drawing because your request is all over the place.

I understand your approach, but I would like to build a circuit first, that has it's own maximum current, then dim this by using PWM. If something would go wrong in the programming the Leds would burn out if the input voltage is variable and the controlling port into the Gate of the mosfet would stay high.
With this circuit the input voltage has a bit bigger variance, with only the power-dissipation based on the Rds-on and Vin-Vleds-Vbe of the mosfet as a limiting factor.

 

[Kenneth Tan]

How many LEDs in each string?

5 COB leds in parallel, with individual strings in the COB. @10,5V I believe 3 leds in series.

 

[Kenneth Tan]

Knowing that there are some variables, I would calculate the resistor value based on a 0.65V Vbe since, as Colin suggests, Ic will be low and thus the required Vbe and consequent Ib will be low.

Vbe increases as Ic increases, and that describes the reason why there is a range of Vbe

The exact Vbe required is also variable between transistors.

After doing the calculation above, you may also find that the exact resistor value is not available, and even if it is, the resistor week have it's own tolerance.

So, do the calculation based on the minimum Vbe in the datasheet then choose a resistor and measure the actual current. If you need a very tight tolerance you might need to adjust the resistor individually and/or use a better circuit.

I just did just that. Only bases on the highest Vbe. Wrong?!?

 

[(*steve*)]

I just did just that. Only bases on the highest Vbe. Wrong?!?

No. You're just being conservative. That approach will pretty much guarantee a current slightly lower than desired.

in another post you worry about the exponential v/i curve of the LEDs -- don't. This circuit controls the current, so a slight variation will result in a small change in current, not a small change in voltage. In fact this circuit is designed to eliminate problems caused by the way the LEDs v/i curve changes with temperature.

 

[Kenneth Tan]

No. You're just being conservative. That approach will pretty much guarantee a current slightly lower than desired.

in another post you worry about the exponential v/i curve of the LEDs -- don't. This circuit controls the current, so a slight variation will result in a small change in current, not a small change in voltage. In fact this circuit is designed to eliminate problems caused by the way the LEDs v/i curve changes with temperature.

I know the leds are current controlled, but a 0.2V difference in Vbe makes a great difference in output current.
If I would choose the lowest Vbe of 0.6V and would choose the resistor based on that, the current difference if the transistor has a Vbe of 0.85V would be 1006-750=256 mA (or 1/3rd) of the chosen current extra.

That's the core of my question. How to handle the theoretical and practical Vbe of this cheap transistor as this influences the output current greatly. (I got 100 pcs of 2n3904 for $.90 on ebay)\

So for now, let's be conservative and build it with that in mind. maybe lower the resistor if field testing allows.

 

[(*steve*)]

Post a link to the datasheet you're using. The higher value of Vbe is almost certainly specified for a high Ic, and the low one for a low Ic. You have a low Ic in this circuit.

 

[Kenneth Tan]

https://www.sparkfun.com/datasheets/Components/2N3904.pdf

 

[(*steve*)]

Ok that datasheet suggests a wide range of values. However it is also giving Vbe(sat) figures and in your case the device will not be saturated. You can also note that in the examples the base current is much larger than Ic/hfe. In your case, since you are not forcing saturation the base current will be much smaller leading to a Vbe that is somewhat lower.

Some datasheets have graphs that show more detail that can be used to estimate typical values. This is not one of them.

 

[Kenneth Tan]

Ok that datasheet suggests a wide range of values. However it is also giving Vbe(sat) figures and in your case the device will not be saturated. You can also note that in the examples the base current is much larger than Ic/hfe. In your case, since you are not forcing saturation the base current will be much smaller leading to a Vbe that is somewhat lower.

Some datasheets have graphs that show more detail that can be used to estimate typical values. This is not one of them.

Thanks Steve for the assuring advice. going the conservative route and then I'll lower the current limiting resistor on the fly and test it with fully heated leds.

 

[(*steve*)]

Your first test should be done with something other than the LEDs. Just in case you've wired things incorrectly or made a gross calculation error...

Pick a resistor that will drop about the same voltage at the expected current (e.g. if you're looking to power a LED that drops about 12V at 700mA, use an 18 ohm 10W resistor).

Once you're confident it's not grossly inaccurate, drop in the LEDs.

 

[Kenneth Tan]

Your first test should be done with something other than the LEDs. Just in case you've wired things incorrectly or made a gross calculation error...

Pick a resistor that will drop about the same voltage at the expected current (e.g. if you're looking to power a LED that drops about 12V at 700mA, use an 18 ohm 10W resistor).

Once you're confident it's not grossly inaccurate, drop in the LEDs.

I don't understand.
The circuit works well, I was just trying to figure out the 0.2 Vbe difference of the NPN transistor and it's influence on the circuit/currrent through the leds.

 

[(*steve*)]

ok. Perhaps just file that away as best practice for next time.

Another alternative is to test the circuit using a power supply capable of current limiting the output. Setting the current limit to (say) 20% more than the expected current will also provide you with a safety net.