LED constant current driver

[Mosaic]

Hi all:
Does anyone have a circuit for a running a 7 segment display with a transistorized constant current driver that permits a single resistor to set the current for all the segments?

 

[MrAl]

Hi,

Do you mean a single constant current supply or do you really mean a single resistor to set multiple constant current supplies?

You do the latter but not the former because the segments are in parallel.

Also, do you mean just one single digit or multiple digits? There are driver IC's out there that do this.

 

[audioguru]

Why do you need a brightness adjustment?
My TV, clock radio and DVM use LED 7 segment displays and do not have a brightness adjustment. Their brightness is fine anytime.

 

[Mosaic]

MrAl , Yes Mr. Al , I but I don't know any sourcing constant current drivers that can handle the higher voltages of series LEDs. Eg 12- 15V
Multiple digits.

 

[MrAl]

MrAl , Yes Mr. Al , I but I don't know any sourcing constant current drivers that can handle the higher voltages of series LEDs. Eg 12- 15V
Multiple digits.

Hi,

What are you saying yes to? Is it that you want multiple current sources controlled by one resistor?

 

[Mosaic]

If u look at this circuit , it requires a sense resistor to current limit for each segment that is made up of 3LEDs for an average 9.6Vf.


So I will need 7 of these circuits (no decimal). To adjust the current limit settings I'd have to vary all the sense resistors. I was wondering how I could make the current limit for all the segments reference a single resistor.

This is a current sink circuit. I can do a source version with PNPs.

 

[KeepItSimpleStupid]

Could you not use PWM and say create a blanking circuit or PWM the digit power source? Just a thought?

 

[Mosaic]

Here's a variation that works. But I don't think it subscribes to the KISS corollary of circuit design. Does anyone see an alternative?

 

[ronsimpson]

Here is a idea. Four channels but you can see how to make it a seven channel. All the transistors have the bases tied together. There is a pot that makes a voltage common to all the bases. D1 is used to temperature compensate for the B-E voltage of the transistors. The "buffers" could be any TTL type parts or a u-processor.
The voltage across R1---R4 sets the current. The current should match pretty close.

 

[Mosaic]

Hi Ron:
I don't see how that regulates current equally across the LEDS. There isn't any feedback for each LED.

Edit:
Rather than the 225mW SOT23, bipolar, mmbt3904s I am considering using this NFET (logic) as it has much higher dissipation ratings (10x the bipolar) and would allow a wider range of supply voltage & LED currents.
**broken link removed**
Its only 34 cents at Newark.
https://www.newark.com/vishay-siliconix/sia936edj-t1-ge3/mosfet-n-ch-20v-4-5a-powerpack/dp/47X1986

 

[ronsimpson]

Hi Ron:
I don't see how that regulates current equally across the LEDS. There isn't any feedback for each LED.

Look at T1. There is feedback.
If the logic out is at or near 5V then there will be no current in the collector of the transistor. No LED current.
If the logic out is at 0 volts then the voltage at the wiper of the POT will be across the emitter resistor(s).
>At 1V and 1k ohms =1mA, 3V=3mA The supply voltage does not matter. If the supply is 5V or 10V or 20V the LED current will be the same.
>All the LEDs will have the same current with in the accuracy of the resistors.
>

 

[MrAl]

Hello,

In Ron's circuit it also helps to know that a constant voltage on the base keeps a pseudo constant voltage across each emitter (sense) resistor. A constant voltage across a resistor results in a constant current through that resistor and that's the goal.
Since this current can be varied with either the emitter resistor value or the base voltage, we elect to change the base voltage so that we can control all N stages.

The current through the collector is the current through the emitter minus the current in the base emitter, and the current in the base emitter comes from the voltage setting and the emitter resistor. The Beta of the transistor used must be high enough to ensure the current in the collector can always be produced by the small base current. Roughly a gain of 40 should do it which i think is what most small transistors can do these days. It can vary up or down quite a bit too as the following should illustrate.

The collector current (our goal) can be shown to be approximately equal to:
Ic=((Vb-Vd)*B)/(Re*(B+1))

where Vb is the base voltage to ground, Vd is the base emitter voltage drop, Re is the emitter resistor, and B is the DC Beta.
For small variations Vd does not change much, and Re is a constant resistance, and holding Vb constant (our set point) we can reduce this to:
Ic=K*B/(B+1)

where K is a constant, so we can see that Ic is proportional to B/(B+1).

Computing this for B of several values, we get:
B=40, B/(B+1)=40/41=0.9756 (nominal value)
B=20, B/(B+1)=20/21=0.9524 (half value)
B=80, B/(B+1)=80/81=0.9877 (double value)

Now comparing the two extremes for half and double the Betas, we see that the collector current will only change by about 4 percent for these fairly wide extremes. So we end up with a relatively good current regulator.

 

[Mosaic]

I simulated it and it works! The feedback needed some tweaking though.

That approach cancels the opamp requirement in the other circuit shown.

I learned something new & useful. This seems to be an extension of the emitter follower.

BTW, would replacing the bipolars with low Rds logic level NFETS be an improvement?

 

[MrAl]

Hi,

N Mosfets would require a different biasing technique which wont help the efficiency anyway. Probably just not worth the effort and the turn on voltage is temperature sensitive.

 

[ronsimpson]

BTW, would replacing the bipolars with low Rds logic level NFETS be an improvement?

On the hobby level you should use transistors not FETs.
I can make it work with FETs but.....long story.
This counts on the "turn on voltage" B-E or G-S should be the same on all transistors (npn or N-FETs).
Inside a IC this works very well with any type of transistor/FET. All the transistors are made at the same time on the same piece of silicon and will have the same temperature.
For a hobby you should use transistors of the same type (2N2222A or 2N3904 or...) and from the same batch (date code).
The NPN will have a B-E voltage of about 0.65V while the MOSFET will have a G-S voltage of about 3V to 5V depending on the type of part.
I used a diode to compensate for the B-E voltage while I should have used another transistor.
On the MOSFET design it is important to use another MOSFET to compensate for the G-S voltage.

Because you can withstand a +/-10% error with out any effect you could use transistors from different date codes but you should at least use from the same maker. (IT or Sanyo or Motorola.......) I can get the current to match to 1% with some effort. You LEDs are not matched so why get the transistors very close?

 

[EinarA]

Connecting R5 to the LED supply will cause the bias voltage to change with supply voltage. Replacing D1 with a PNP as an emitter follower will reduce this, but often this circuits works if you just drive the bases of the NPNs from the wiper of the pot; using a low value for it ( 1K or less.) There will be some modulation of the current as the number of segments on varies, how much will depend on Ic and transistor gain.

 

[Mosaic]

The emitter follower is to buffer the base ref voltage?

 

[MrAl]

Hi,

There will be some change in set point with Vcc yes, but it depend on the ratio of R5 to the pot value, assuming the +5v source to the pot is regulated.

In these kinds of circuits it's not unusual these days to see the use of a voltage regulator diode to provide the bias voltage. The voltage is still adjusted with a pot, but it may be set up a little differently.

 

[EinarA]

I interpreted an earlier post to mean that the Vled will be varying. The PNP is connected base to wiper, collector ground and emitter to R5 and all the NPN bases. R5 should be sized to provide more current than the NPNs will require at the lowest Vled. The current in the pot will now be reduced by the gain of the PNP and will provide a more constant bias voltage.

 

[ronsimpson]

Here are some options:

D1 & R5 can be removed. D1 compensates for the B-E voltage drop in the transistors.

This version reduces the current in the wiper arm of the pot by 1/100 to help a problem.
You can see the voltage on the wiper arm is raised one B-E and then brought back down by one B-E drop.

In this version; If the pot is set to zero ohms there is no voltage on the emitters and no LED current.
If the pot is set to 1k then both the pot and R6 have 2.2 volts. All the emitters have the same 2.2V and the LEDs have current.
When I do this at work I try to make the current in T5 match the current in T1...T4. But I think this is good enough for a hobby.