I'm having a problem with a two channel mosfet dimmer that I'm constructing. When I had only one channel (T1_signal and Load_1_to_AC-) populated on the PCB in the schematic below the dimmer works fine. But when using two channels they interfere with one another. If channel 1 is fully on and channel 2 fully off then channel 2 turns on during the top half of the AC sinewave. The reverse is also true (if you swap places with 1 and 2 in the previous sentence).
Below the schematic is another example where channel 1 on the dimmer is set to about 20% intensity (mosfets on for about 2ms). In the first scope image channel 2 on the dimmer is set to 100% intensity (mosfets always on). Magenta (3) on the scope is measured at the drain of Q1 and cyan (2) is the T1_signal. As seen the top half of the sinewave does not get dimmed/chopped to 20%/2ms. The second scope image shows the same scenario except with the intensity of channel 2 on the dimmer set to less than 20% (mosfet on time for channel 2 less than the 2ms on time for channel 1) and now the upper half of the sinewave gets correctly dimmed.

I tried putting a diode between R9 and GNDA, thinking the problem had something to do with current moving from the gate of the mosfets that are on to the gate of the mosfets that should be off thus turning them on, but that didn't change anything.
Can anyone figure out what is going on here?

Thanks in advance,
Lars E

 

Right now Q2 and Q4 partially in parallel.

So when the circuit shall be two independent "light dimmers", so there must be a second Gate Supply.

 

Right now Q2 and Q4 partially in parallel.

So when the circuit shall be two independent "light dimmers", so there must be a second Gate Supply.

Would it be possible to isolate the mosfet pairs from eachother with a diode between the source and GNDA? (One on each pair, anode to GNDA)

If not, what parts of the gate supply could be shared (not duplicaded)? Can I share D1 and R2 between them?

 

No, as I see it you need to make an extra Gate Supply circuit, so there will be a GND_B and GATE_DRIVE_B for U3, Q3 and Q4.

 

No, as I see it you need to make an extra Gate Supply circuit, so there will be a GND_B and GATE_DRIVE_B for U3, Q3 and Q4.

I agree.
The two "GND" voltages are not at the same voltage but will vary with the state of their respective MOSFETs' gates.
That's giving the odd results you are seeing.
You thus need two separate GATE_DRIVE supplies and two separate "GND" connections.
(They can share D1 but not anything else).


Sorry to hear you already have a PCB.
A simulation of the circuit before you built it would have revealed the problem.
Below is an LTspice simulation of your essential circuit.
Note how the two GND voltages are different, and change as their respective MOSFETs switch on and off, thus they can't be tied together.

 

I see, thanks @Kjeldgaard for the swift reply, and thanks @crutschow for the very detailed explanation! (any chance you could post the LTspice-file?)

 

Attached is the LTspice file.

 

Attached is the LTspice file.

Thanks alot! This might be a bit off topic but I've been trying to add the 1N4007 you use in the simulation without success.. I've added the line below to my standard.dio (located in C:\Program Files\LTC\LTspiceXVII\lib\cmp) and restarted LTspice multiple times and even the whole computer but still the diode does not show up in LTspice. I've tried removing the standard.dio file and then LTspice kindof hangs at startup so I know it's reading the file.. what am I doing wrong?

.MODEL 1N4007 D(IS=2.55E-9 RS=0.042 N=1.75 TT=5.76E-6 CJO=1.85E-11 VJ=0.75 M=0.333 BV=1000 IBV=9.86E-5 Iave=1000m Vpk=1000 mfg=GP1000V1A type=silicon)

Screenshot http://pasteboard.co/3IaA283Td.png