Linear LED driver oscillates at 20MHz

linear led driver

Hello,

The following is a linear LED driver that runs off a 13.5V battery......


http://i26.tinypic.com/29o0kkp.jpg

Q8 and Q7 turn the whole lamp off if some of the LEDs fail open.

The LMV431 gives current regulation........

**broken link removed**

…bottom of page 14 on the above datasheet shows the LMV431 as a current regulator.

This linear LED driver draws a total of 0.518A.

It is all on a double sided PCB (2 ounce copper) of dimension 16cm by 8cm, LEDs one side, Electronics the other.

The other day we got the made-up PCBs back from the PCB place and tested them.

They were all drawing 10% less current than they should have been.

-So we scoped the collectors of the NPN’s …..
-and we saw a 2V pkpk sinusoidal oscillation at 20MHz (really 20MHz !!!!)

-The scope probe was just a ‘normal’ probe with a long-ish ground clip.

Anyway, to get rid of the 20MHz oscillation we put a few 10nF capacitors across the Base-Emitter junctions of a few of the NPN’s.

-After adding these 10nF capacitors, the 20MHz oscillation went and the linear LED drivers drew the correct amount of current.

Does ant reader know how this linear LED driver managed to oscillate with such magnitude at such a high (20MHz) frequency?

oscillating led driver circuits

Not surprizing when using a control loop without sufficient compensation. I suggest to connect a series resistor to the U1 sense
terminal and a C or series RC compensation network between snese terminal "cathode" (a kind of miller compensation). C1 should
be removed, I think.

linear led driver circuits

Thankyou FvM,

With an SMPS, oscillation due to control loop problems is usually at a few KHz.

Do you know why this linear LED driver goes at 20MHz?
It seems phenominally high frequency to me.?

linear led driver

I fear, my first explanation has been wrong. The TL431 gain at 20 MHz is too low to allow a >=1 loop gain near 20 MHz. So it's
most likely an oscillation of the output transistor stages. You should try individual base resistors of e.g. 50 or 100 ohms. Also less
parasitic design inductance should eliminate it.

circuit for 20mhz oscillator

FvM:- we managed to eliminate it (the 20MHz oscillation) with 10nF caps across a random 2 or 3 of the NPN's.

However, i am not sure why this oscillation occurred in the first place as parasitic inductance and capacitance surely cannot have been big enough to cause a 2V pkpk sinusoidal oscillation at 20MHz at the collector terminals?

These NPN's do appear to have acted like common emitter amlifiers and with a gain of about 8 (RC/RE)

I cannot imagine what caused the bae voltage to have an oscillation of 2/8 = 250mV.

As far as i knew, linear regulators are'nt supposed to oscillate.?

Added after 5 minutes:

Sorry, i've just re-examined the datasheet for the FMMT619 NPN's

**broken link removed**

...it says the fT parameter is 100MHz, and i wondered if we should have picked a transistor with a much lower fT value? -to avoid oscillation.

led +oscillation

Typically, you get oscillations in a common base configuration, in your circuit C1 achieves the necessary low impedance base conncetion. The transistor is operating as a colpitts oscillator and a kind of resonant circuit must exist at the collector node.