Hello all,

I'm having this issue where my Fly-back power supply is not outputting clean 24VDC.

I have attached images of what the scoped waveforms looks like

1. Gate voltage at FET: Scope CH1 Yellow
2. FET drain voltage: Scope CH2 Blue
3. Gate voltage at pin 19 of LT8316: Scope CH3 Purple
4. Output voltage ripple across cap C86: Scope CH4 Green

I have tried adding a 28V zener (SMBJ5362B) and a 2200pF cap (VY1222M47Y5UQ63V0). None of this helped.

I've followed the recommend circuit analog devices has provided and the issue is still there.

Does anyone have any recommendations?

Would a zener at 26V help?

 

All Switching-Power-Supplies produce Noise, especially Boost and Flyback Circuits.

To get rid of enough Noise to suit your application,
it will all come down to how You handle the Output-Filtering.

A couple of Capacitors,
( a large Electrolytic, and several different sizes of Ceramic-Capacitors )
"may" do the job.

If they don't, then You will have to add an Inductor to the mix, or maybe even 2 stages,
( L-C-L-C ),
and if that's still not "perfect" enough for your requirements,
You may have to add a Linear-Regulator after all that.

The values and types of Components used in the L-C Filters
are heavily dependent on the Frequency that the Power-Supply uses.
.
.
.

 

I'm am little concerned that your isolated secondary is returned to mains neutral and not to PE That is very unusual - Neutral is NOT Ground. And I wonder if you're getting noise injected into your DC from the dirty mains. I'd disconnect that to leave your secondary isolated from the mains and then test it. If your 'scope ground is connected to mains neutral you stand a good chance of frying it - you might be lucky because your 'scope has an isolated supply but you can't be sure that anything you connect your supply to will be similarly isolated.

If you look at this reference design (different chip,same concept) you'll see that the mains input is completely floating as it should be.

 

All Switching-Power-Supplies produce Noise, especially Boost and Flyback Circuits.

To get rid of enough Noise to suit your application,
it will all come down to how You handle the Output-Filtering.

A couple of Capacitors,
( a large Electrolytic, and several different sizes of Ceramic-Capacitors )
"may" do the job.

If they don't, then You will have to add an Inductor to the mix, or maybe even 2 stages,
( L-C-L-C ),
and if that's still not "perfect" enough for your requirements,
You may have to add a Linear-Regulator after all that.

The values and types of Components used in the L-C Filters
are heavily dependent on the Frequency that the Power-Supply uses.
.
.
.

I've tried adding a bigger Electrolytic that didn't work.

I do have a switching regulator that brings it down to 5VDC

I also do have an inductor before the regulator, but for right now I have R103 and R104 removed. I have a wire just across the pads.

 

I'm am little concerned that your isolated secondary is returned to mains neutral and not to PE That is very unusual - Neutral is NOT Ground. And I wonder if you're getting noise injected into your DC from the dirty mains. I'd disconnect that to leave your secondary isolated from the mains and then test it. If your 'scope ground is connected to mains neutral you stand a good chance of frying it - you might be lucky because your 'scope has an isolated supply but you can't be sure that anything you connect your supply to will be similarly isolated.

If you look at this reference design (different chip,same concept) you'll see that the mains input is completely floating as it should be.

View attachment 347244

For Main GND due to the spec I have to have it connected to neutral. I've scoped it and it is clean. Right now the 120VAC is coming from a VARIAC. I mean it's isolated at the moment since the MAINS-P is not directly connected to MAINS

I see for their snubber circuit they also use a Cap, Res and a diode, I saw some others use that combination too. I may try that next. I don't understand what C11 is doing in that schematic?

 

For Main GND due to the spec I have to have it connected to neutral.

It is dangerous to have the Main's neutral connected to the output of your circuit, as that can lead to the main's voltage appearing on the output, if the main's hot and neutral are accidentally reversed.
There is no need to have the Main's neutral to be connected there, so please remove it for safety.

 

I've never seen a VARIAC that's isolating since they are almost always an autotransformer with a tapped winding.

I don't understand what C11 is doing in that schematic?

C11: Capacitor added across the drain and source of the MOSFET Q1
There is a method of adding a capacitor across the drain and source of a MOSFET in order to reduce surges at turn-off due to high-speed switching. This is also one kind of snubber. However, greater losses result, and so attention must be paid to temperature increases. Here, a 10 to 100 pF capacitor with a withstand voltage of 1 kV is used.

It is dangerous to have the Main's neutral connected to the output of your circuit, as that can lead to the main's voltage appearing on the output, if the main's hot and neutral are accidentally reversed.
There is no need to have the Main's neutral to be connected there, so please remove it for safety.

So I said earlier, but it seems to have fallen on deaf ears....

 

electro's by themselves are not good at filtering spikes due to ESL / ESR, try 1uF MLCC and 10uF MLCC

flybacks are inherently noisy due to the large fast leading edge triangular current spike on the output,

also your probing can affect what you see on the scope, try connecting the shorted scope probe to Vout - if you still see noise - it is all CM - subtract this from what you see when probing from 0v to Vout

 

Often it works much better to CLAMP a spike instead of filtering it.

 

Hello all,

I'm having this issue where my Fly-back power supply is not outputting clean 24VDC.

<...>

The first thing to do is to ensure that what your scope is seeing is real!

This requires very good probing technique; Don't use the ground clip with your normal probe. The spike may be picked up via magnetic coupling to the probe and not actually be present in the circuit.

I find the best way is to hard wire a coax directly to the PCB minimizing the the area of the loop between the signal and ground. Connect that coax directly to the scope. In some cases I have soldered a suitable connector (SMA, SMB or other small coax connector) to the PCB then used an SMA to BNC cable to connect to scope.

Only by doing this or other similar means will you have confidence that the spike you see is really there.

Kevin

 

Considering that the supply has the output common tied to the mains common, it is not suitable for an application that requires a low-noise output. It would be OK for motors,solenoids, and lighting, but not for any low-noise application.
AND, if the noise is at TP#1 as shown in the one circuit, that is at the filter input, not the output.