Advice for a choke for blocking parasitic switching spikes

  • Thread starter Thread starter artis
  • Start date Start date
AI Thread Summary
Minimizing switching interference in a switch-mode power supply (SMPS) for an audio amplifier involves addressing parasitic high-frequency oscillations and ripple on the DC supply rails. The user has implemented several strategies, including using ferrite chokes and snubbers, but still experiences significant noise. Suggestions include employing a Pi filter with low ESR capacitors to further smooth the output and potentially adding capacitors across the IGBT switches to suppress spikes. The discussion highlights the importance of layout and component selection in reducing noise, with a focus on achieving better performance without extensive redesign. Overall, the goal is to enhance the SMPS's effectiveness while learning from the experience for future projects.
  • #51
artis said:
I have done some experimentation in the past with the samps to find out the problems, I have also probed my secondary side and I am afraid that the capacitors are not the whole blame here, there is some small ringing present in the primary side of the switches so I think it all comes together to form this spike behavior in the secondary.
That is unlikely. The ringing you observe on the primary side will be a voltage waveform. The communication across the pulse transformer is primary current to the magnetic field, then to secondary current. The spike you are getting on the secondary is a voltage spike, almost certainly related to di/dt in the secondary wiring and capacitor inductance.
 
Engineering news on Phys.org
  • #52
Ok I understand your point , that even though the primary current in the transformer primary winding has some minor "tails" the secondary creates it's own spikes as the combined result of many factors such as not the best pcb trace layout, capacior self inductance etc right ? Would this be so because any minor but very sharp spike (almost vertical rise and fall ) wouldn't "get through" the transformer or atleast be attenuated enough so that the secondary filter caps can deal with it?Also , should the main chokes in the filter DC rails be rated for 10mH? because doing spice simulations changing them to lower inductance values like 5 or 3mH results in only a few dB less in the drop of the signal.
But that lesser inductance means a smaller inductor with less wire and less space.
 
  • #53
artis said:
But that lesser inductance means a smaller inductor with less wire and less space.
Engineering involves lowering the cost while still meeting all the specifications. You are looking for lower cost, but you have no specifications on amplifier immunity to supply noise, so you cannot yet know how quiet your supplies need to be. By all means lower the inductance of the chokes.

Switching amplifiers are significantly immune to power supply noise, indeed they should create as much noise on the supply as there was in the first place. Get the priorities right. It is usually easier to refine specifications than to over-engineer the hardware.
 
  • #54
In post #44 you show a bipolar supply with a common ground.

The supply and amplifiers do not need a power supply ground if you use bridged pairs of amplifiers. That eliminates the reservoir series capacitors and all the ground connections, which reduces ESR steps and inductance spikes. All ground conductor material can be usefully reassigned to the power rails.

Airbourne.png
 
  • #55
  • #56
Well my amps are just a stereo (2 channel ) version of an AB topology, rather powerful ones (16 output devices per channel) My psu is actually almost like you @Baluncore showed with a single thick ground wire going from the psu for both amp boards as well as for the speaker connectors, just with the addition that in the very psu secondary side itself the filter caps are connected such that each cap either from positive side or negayive one goes to ground in each of the two bridges, that is because my supply voltage is 85-0-85 or about 170vDC and my filter caps are rated 160v so I connected them like that, in order to rewire I would have to swap out my existing caps for ones that are rated for double the voltage , well at least 200+ VDC, then I could rewire like in your schematic.

The question then becomes , which is cheaper and would result in more noise reduction, the capacitor swap or simply adding the filters. Something tells me that simply rewiring the cap bank with new capacitors wouldn't fully get rid of the noise and a filter would still be needed the only benefit probably with smaller DC rail inductors.
 
  • #57
I doubled checked the schematic @Baluncore you last posted and yes seems the only difference in there is that instead of the caps in each of the two bridge rectifiers being added directly between the pos ad neg rails in my psu the caps are tied to the ground so between pos ad neg rails they are two in series with ground in the middle.
 
Last edited:
  • #58
artis said:
I doubled checked the schematic @Baluncore you last posted and yes seems the only difference in there is that instead of the caps in each of the two bridge rectifiers being added directly between the pos ad neg rails in my psu the caps are tied to the ground so between pos ad neg rails they are two in series with ground in the middle.
That is the fundamental point I was making. Series connection to the irrelevant ground gives half the capacitance, twice the ESR, and twice the inductance.
 
  • #59
Sadly I wasn't aware of this when I first ordered the caps and finished the smps, normally at least in linear psu it is considered an advantage that the bipolar rails allow the use capacitors with lower voltage ratings that have double the capacitance rating for the same price.

Well I guess it's parts search once again, have to check out how much would a 200v rated or more cap would cost. I wonder if my DC voltage is about 170/175, what would be the minimum safe margin of capacitor voltage to choose from. I guess 250v.
 

Similar threads

Back
Top