Help me with comparator and multiple PSUs

Hi,

I've been working on a lm3409 circuit for analog and PWM dimming of LEDs. Since the chip doesn't support turning off the LEDs with analog dimming, I added a comparator that would turn off the EN pin on the chip when the voltage is close to 0V.

I tested everything on a protoboard and ran a simulation in LTSpice and everything seemed to work. The only thing I didn't add on the protoboard was the LED power supply.

The board takes a 0-5V dimming signal, 5V for the comparator and a much higher voltage PSU for the LEDs (48V in my case). When I connect the LED PSU, the comparator stops working. Here is the schematic:

**broken link removed**

I tried a different PSU (thinking that my original one was causing ripple?) but that didn't make a difference. With the LED psu on, the voltage VIN+ on the opamp is much higher.

I'm guessing I'm missing something very simple, but given that this is my first circuit and that my EE knowledge comes from the internet on the evenings, I don't know what it is.

Any help would be greatly appreciated.
Thanks

What is the unidentified circuit and where are you applying +48V ? If it appears on VIN+ and R3 isn't properly grounded the potential divider formed by R3 and the ground resistance will apply a fraction of that 48V to the VIN- pin of the comparator and give spurious results.
Presumably the unidentified **broken link removed**circuit is rated to handle 48V ?

Hi Alec,

I assume you mean the 10 pin component? That is the LM3409. it can handle up to 75V. For the purpose of this issue, the LM3409 can be considered an open circuit because the problem is still there even without it being populated.

Here is the PCB:
**broken link removed**

Thanks.

Can you draw the whole circuit with part numbers?

Here are some measured voltages using 5V control + 12V LED power supplies. The only voltage that changes is the dimming voltage, which messes up everything. RED is without the LED PSU, blue is with the LED psu.
**broken link removed**

ronv, sure, while I'm doing that, the BOM is.
**broken link removed**

EDIT: Labeled components:
**broken link removed**

resistors are Vishay CRCW 1/10W 1% series except for R1 : WSL2512R2200FEA

C1 and C2 are C1210C225K1RACTU
C4 is C0603C104J5RACAUTO
C5 is CGA3E2C0G1H471J
C3 is C1608X7R1C105KT

Am I correct? PREADJ is you dimming signal and Vout from the op amp goes to EN on the pwm IC?. I'm thinking with the 48 volt supply you cant get the comparator to switch?

ronv, Spot on.

According to the data sheet, there is a 1.24V zener from IADJ to GND inside the LM3409. For 0 < PREADJ < ~ 1.25V (0-312mV at IADJ without the LED psu) , when the LED psu is turned on, the voltage at IADJ is constant ~10mV. When PREADJ goes above 1.25V with the LED psu on, the comparator turns on the EN (VOUT) pin and enables the IC. At that point there is a little bit more voltage at the IADJ pin with the LED psu than without. But then gets interesting. Once the EN pin is enabled (preadj goes over 1.25V), on the way down to 0V, it never switches off. with preadj at 0V, the voltage ad IADJ is about 130mV!


I took a few data points with the multimeter. This is what they look like:
**broken link removed**

x axis is the preadj voltage. 1V on the EN pin is considered ON (easier to plot).
NPSU - led psu OFF
PSU - led psu ON
VIN+ is the voltage on the non-inverting pin of the opamp and the IADJ pin of the LM3409.

Ideally the yellow and brown lines should match (when the IC is on and off), as well as the red and green.

Also, the voltage on the LED PSU doesn't seem to affect this plot. I tried with 15V and 12V and the result was the same.

This is only a gut feeling.... I think the input impedance for PREADJ is to high. I don't know what you are driving it with but I would try to reduce it to say 910 ohms and 300 ohms from 91k and 30k. This will keep the 5 ua current source in the IC from changing the voltage as much. If you can't drive the 5 ma for the low values even 9.1k and 3k might help a lot.

ronv, you should listen to your gut That was it. Tried with 121Ω and 49.9Ω (not exactly the same divider but had this around) and it worked like a charm. Thanks very much!