dual pwm interference issue

Hi guys and gals, I'm trying to work out a solution to a little circuit I've created. Let me start by saying that I'm no electronics guru. I can follow logic and flow but never learned the intricacies of circuit electronics. I'm hoping someone here can help me understand where my problem lies and how to overcome it. OK - enough of that, here's my issue:

I have two LED DRL modules and want to run them with dual PWM controlled dimmer modules. I wanted each light to have two independent dimming modes (hence the twin PWMs). The PWM modules I'm using are commercially bought (555 chip based) dimmers. They regulate the pulse width on the negative output side leading to the lights and can be varied through a potentiometer (1K).

To give each PWM two dimming states, I've taken out the potentiometer and ran those lines through a DPDT relay and then to two 1K trimpots. So depending on the trigger state of the relay, the circuit runs from either one or the other potentiometer and this gives me my dual mode dimming. All works great - but here's the issue:

When I introduce the second PWM (same principle as above), the output lights all start to pulse (i.e. slightly vary in brightness). I'm guessing that the PWM pulse frequency waves are somehow cancelling each other out or clashing because they are not in sync. Since the PWM signal is on the negative side I'm not sure how to isolate or block the interference between the two modulators.

I've attached the drawing of the circuit board I made. Apologies if I've used incorrect symbology so please ask if anything seems unclear.

Thanks for any help.

Hmmm.. image didn't attach. I'll try again with the circuit drawing:

Are you "beating" (mixing) the 2 signals - do you have sufficient track separation and decoupling (caps across each 555 etc)?

I assume by dual you do not mean 2 PWMs feeding the same LEDs...

With a circuit diagram I might be able to suggest more.

Thanks for the quick reply. No, the dual PWMs are meant to be independent of each other. Each bank of LEDs has its own PWM.

I'm not sure what you mean about the decoupling. The 555 based PWMs are unmodified apart from the changes mentioned above.

With regards to the diagram.. is the 2nd post not showing it? Seems to come up alright when I'm viewing it here.

Thanks for your help.

That's a wiring or layout diagram - ie, "physical" - not circuit. I usually only reverse-engineer physicals when I am cracking something for my own benefit, though I do allow the "relay pin physicals" of the typical DIN (Bosch etc) relays as often used on this site. [My circuit diagrams show the coils & contacts so that the circuit (behavior) can be understood.]


In short, you may have to separate the tracks & wires carrying the PWM pulses, or make them heavier (especially the +V & GND supplies; maybe by adding solder to the tracks), or add decoupling caps.
The latter might include a biggish cap across the power input and rails (eg, 10uF to 10,000uF electrolytic caps) and small fast caps (0.1uF or 0.01uF) across signal (PWM) lines as well as power rails.

Maybe discarding the PCB is the solution - ie, wire direct to relay sockets and keep the 2 "channels" separated.   


I've given some more (optional) misc info below, but circuit noise is an art in itself. PCB layouts can be critical. In some cases, PCB tracks require shielding tracks (GND) on both sides. That's where ground planes and multisided-PCB layer order require their expertise.


Decoupling caps are those used across power supplies to filter out spikes. EG - 555s are notorious for switching transients and often have a 0.01uF or similar as close as possible to its supply pins - ie pins 1 & 8 - but most digital chips have the same. Otherwise spikes thru the power rails can cause undesirable triggering.

But such transients can also be coupled across tracks and wires that are parallel and too close, hence why power and signal or audio wiring requires separation, else decoupling.
This (inductive or capacitive) coupling increases with current and steepness of the waveform. And PWM like digital signals are "sharp" edged square or rectangular waves.

Digital signals can be decoupled (filtered) with caps, but the cap needs to be small so as to filter out the high frequency noise but not smooth out the digital signal. EG - you do not want a cap across a PWM to turn it into a "continuous" sinusoidal wave (despite what some Swedish "expert" BMW owners seemed to think - dig dig! ) as that defeats the purpose of PWM.

oldspark wrote:

---

I've given some more (optional) misc info below, but circuit noise is an art in itself. PCB layouts can be critical. In some cases, PCB tracks require shielding tracks (GND) on both sides. That's where ground planes and multisided-PCB layer order require their expertise.

---

Dang.. I almost knew there wouldn't be a simple answer :)
I did try the non-PCB way as well thinking I may have screwed up somewhere in the circuit design. Unfortunately it's the same outcome. I can balance the waves if the trimpots are set absolutely the same between the PWMs. As soon as one of them switches to a longer pulse width the problems re-emerge.

Thanks anyway for your help. I may have to discard this concept and just go back to running a single PWM.

Again, if you can provide the circuit diagram.

It does sound like you are combining both PWMs onto the same LEDs, but if you say they are electrically separate... (excluding crosstalk etc).

OK.. I've tried to draw up the circuit of what I've done. Unfortunately I don't have the circuits for the PWM as they have been bought commercially. All I've done to them is to take out the potentiometer and used the three pins to splice it into my circuit (i.e. going to the two potentiometers; one or the other depending on the relay). The LEDs are wired directly into the pins on the PWM (no cross over there) but the PWMs share the main power feed and the ground.

What value potentiometer?
That pot in 555 PWM circuits is usually connected to a diode; a diode is an AM detector (receiver) and you have long leads to external pots.

If might be paralleling the kit's pot circuit and going below (say) the minimum 1k required in that circuit (to protect the 555??).    

But ensure the GND paths are good - ie, enough track width or solder, and the 555 GNDs and relay-coil's GNDs are reasonably independent (not through each other).

Thanks for your help Oldspark. The potentiometers are all 1K.
I've taken a snapshot of the PWM in default trim. The ribbon wire coming off the board top left is where I'm replacing the standard pot with my circuit.



Luckily I had a couple of these spare so I tried wiring them direct to my lab power supply; so no using my circuit or the twin trim pot idea. Same issue. As soon as you combine the two dimmers onto the same ground they interfere.

I'm guessing there's probably better ways to skin the cat for what I want to do but here I was hoping for a relatively simple and compact solution. At the moment I run the lights just by using a controlled power supply and then through a single PWM to get the dimming state. I might just stick with that and be done with it since my circuit design knowledge isn't quite up to the task of designing a complete solution.

I'd think AM would not be a problem with a 1k trimpot (it's when you have 50k - 2M etc).


But who's replaced D1 with a resistor?
And what value is that resistor (in case it is the inline supply for the 555 & 358.

Usually PWM circuits only involve the 555, it is rare I see OpAmps (LM358) added.