How do you probe signals on a switch mode supply board

[snarkysparky]

My home project is a switch mode SMAW welding current controller. I have it  built out now but need to do some testing.  Well I found that my particular circuit just laughs at any attempt to connect a scope to board level signals.  I have a closed loop hall effect current sensor for the main welding current and a differential op amp circuit for the voltage across the leads.  But I know enough now to know that I need to test this creation significantly before I can expect to run nice welds.  Well I can't connect the scope directly to the control circuitry as this is nothing but noise.  So I sent a SPI port out from the processor to another board ( off the main board ) where I drove all the SPI lines into optocouplers and which had local battery supply on the receiving end.  No DC connection to the main board.  This isolated SPI signal drove a two channel D/A converter for scope probing.  It works fine for testing at low power levels but when any welding commences it looks like a forest of noise.  It even destroyed the D/A converter on the optoisolated side that was only powered by a local battery.  The signal from the D/A during welding showed 5V peak to peak centered around ground even though the D/A cannot possible provide that.  Well I went nuclear option.  I ordered three fiber optic drivers,  three fiber optic receivers and three fiber cables to provide a debug channel.  I will send the two channel D/A converter date over the SPI so the scope can be 2 Meters away.  If that fails I give up.
How do professionals debug circuits that have grounds with insane noise on them like I would suspect most switchmode supplies to

[David Hess]

How do professionals debug circuits that have grounds with insane noise on them like I would suspect most switchmode supplies to

They fix the ground loop problems and use differential probes or isolated oscilloscope inputs.

You might also have problems with inductive coupling and flux leakage with high current switching.

[snarkysparky]

How do professionals debug circuits that have grounds with insane noise on them like I would suspect most switchmode supplies to

They fix the ground loop problems and use differential probes or isolated oscilloscope inputs.

You might also have problems with inductive coupling and flux leakage with high current switching.

Yes it does have insane flux leakage problems. Just clip ground to tip and wave it around and get big signals.

 I though of sending the analog out from a fully differential on board source and receiving with a fully differential receiver.  But thats a lot of op-amp stuff that is tricky to get right.

Mostly though i think it's the circuit ground is floating after full wave bridge rectifiers with respect to the scope ground. 

Thanks

[donmr]

See Dave's video on a 1kV isolated scope.

[David Hess]

You might also have problems with inductive coupling and flux leakage with high current switching.

Yes it does have insane flux leakage problems. Just clip ground to tip and wave it around and get big signals.

The way I have handed this in the past when the flux leakage could not be reduced is to use thin coaxial cable or twisted pair to make the probe connections.  This way the magnetic flux cancels out as much as possible.  Using an oscilloscope probe with ground lead is practically worthless in these situations.

If using a differential probe, twist the probe leads together tightly.

I though of sending the analog out from a fully differential on board source and receiving with a fully differential receiver.  But thats a lot of op-amp stuff that is tricky to get right.

I have done it and while possible, if you do not have experience with the types of circuits needed it will be a learning experience.  Maintaining good common mode rejection at high frequencies leads into video amplifier type designs.

Mostly though i think it's the circuit ground is floating after full wave bridge rectifiers with respect to the scope ground.

Probably one of the best solutions is an oscilloscope with isolated inputs.  They often work better than even differential probes.

[ocset]

Also, did you lay out the pcb and make sure
...switching currents  arent allowed to travel along long lengths of control ground
...power switching current loops are as low in area as possible...
tha attached gives a guide on pcb layout for smps if you wish.

Also, page 68 of the following shows a piccy of What Davi Hess  kind of said about not having a scope dangling ground lead.

https://ac-dc.power.com/sites/default/files/PDFFiles/rdr-641_40watt_constant_power_variable_output_power_supply_using_innoswitch3-pro.pdf?download=1

And yes coaxial probes are good....have a  BNC connector  one end for the scope and the other end just twizzle the outer ground sheath into a "wire" at the end and connect it with the inner conductor to the rails you want to probe. solder it to the rails for probing.

[snarkysparky]

Also, did you lay out the pcb and make sure
...switching currents  arent allowed to travel along long lengths of control ground
...power switching current loops are as low in area as possible...
tha attached gives q guide on pcb layout for smps if you wish.

Also, page 68 of the following shows a piccy of What Davi Hess  kind of said about not having a scope dangling ground lead.

https://ac-dc.power.com/sites/default/files/PDFFiles/rdr-641_40watt_constant_power_variable_output_power_supply_using_innoswitch3-pro.pdf?download=1

And yes coaxial probes are good....have a  BNC connector  one end for the scope and the other end just twizzle the outer ground sheath into a "wire" at the end and connect it with the inner conductor to the rails you want to probe. solder it to the rails for probing.

Thats part of my problem.  A good portion isn't on a pcb.  The power switches,  freewheeling diode and massive inductor are screwed to a board.  It's a learning experience.

I will try some of the suggestions for probing a noisy board.  Thanks

[ocset]

Thats part of my problem.  A good portion isn't on a pcb.  The power switches,  freewheeling diode and massive inductor are screwed to a board.  It's a learning experience.

Right OK.....at low switching frequencies where the switching is well damped (fets switch from on to off [and vice versa] relatively  slowly), that can possibly be workable...but really, the way to get proper layout with smps is with a pcb.....admittedly if youve got big inductors etc then its awkward to mount them on a pcb without bending it and cracking the surface mount control components etc...but then you  just have to have little daughter  boards.

In fact one reason why high power circuits are often made with lots of lower power ones in parallel is because its easier to mount lower power inductors etc on a pcb because they are smaller.

[drussell]

I'm no high power SMPS design expert, but if I were snarkysparky, I'm not sure I would put much faith in circuit design or layout advice from treez. 

No offense to treez, but if you look at his typical questions in his various threads on power supply designs you'll see why you should take what he says about circuit design with a grain of salt and double check with other, more definitive sources before taking as gospel truth.

[tautech]

See Dave's video on a 1kV isolated scope.

@snarkysparky, this ^ !
Particularly the bit where HV MOSFET switching is discussed. Lots of good stuff in that video.

[Mechatrommer]

See Dave's video on a 1kV isolated scope.

@snarkysparky, this ^ !
Particularly the bit where HV MOSFET switching is discussed. Lots of good stuff in that video.

right! $9600... (watched that yesterday)

[tautech]

See Dave's video on a 1kV isolated scope.

@snarkysparky, this ^ !
Particularly the bit where HV MOSFET switching is discussed. Lots of good stuff in that video.

right! $9600... (watched that yesterday)

NO !
Not the HW but the issues WRT to HV MOSFETS and how capacitance loadings could affect switching characteristics and circuit stability.