Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Adding an LC post filter can make an SMPS unstable....use RC filter instead?

Status
Not open for further replies.
T

treez

Guest
As per page 32 of the attached….
“Switch Mode power converter compensation made easy “ by Robert Sheehan

….if you design an SMPS, and you compensate its feedback loop such that it is stable with an output capacitance of C1, say, then obviously it is stable.
Now add another capacitance to the output, call it C2…add C2 downstream of C1.

Now suppose that even with C2 added…the converter is still stable with good gain and phase margins..still stable without changing the compensation……

…Now add an inductor downstream of the output divider, and downstream of C1, but upstream of C2……. Now the SMPS could go unstable. (it will go unstable if the L_C2 resonant frequency is less than 3x the SMPS crossover frequency)

This is (to the simple minded like myself), not intuitively obvious…..
But page 32 of the attached confirms it.

..Now suppose we add a low value resistor instead of the inductor. Can you confirm that this will not go unstable?
 

Attachments

  • Output filter _pg32 _slup340.zip
    3.3 MB · Views: 126

Well just follow the guidelines in your app note. Also see this one. They both address the problem.
https://ridleyengineering.com/images/phocadownload/1 second stage filter design.pdf

Obviously the second LC needs to be 'far away' from the first LC cutoff but the important trick is to put most of the C at the output. So it's big L, small C, small L big C. Now you have little more sensitivity to external C than a single stage filter.

The 2 stage filter is a powerful tool, the small L and small C can make a big difference in ripple with little component size. I'm using it on TypeIII compensated buck converters with good results.
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
If you load a converter with a less than critically damped LC stage, where the Fo is close to or less than the crossover freq of the converter - then yes you are asking for problems - similarly if you power a large inductor from a DC/DC converter, this L is in parallel with the o/p C and can reduce the effective C seen by the converter - causing osc issues ... esp if you manage to resonate the parallel LC ckt near or just below the crossover freq ...

This is why building damping into a type 3 controller is a good idea ....
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks, i am thinking , the straight RC second-stage output filter may offer good enough filtration, and has no problem with any instability like the LC second stage filter.....the problem with LC type is that you never know if the load connected is going to have a big capacitance and take the second stage LC resonance below 3xcrossover.

Also, eg if the load is an inverter, it may actually comprise an L and a C at its input....and this again, as you know, would possibly bring about instability...so if one had added an output RC stage filter (downstream of the output feedback divider), then you at least get some damping of the load's possible LC input stage....and slightly reduce your chance of going unstable.
 

I highly doubt you can specify a useful RC filter without excessive losses or a gigantic C.

Second, understand that any phase shift risks stability: a single extra pole and its 90 degree phase shift will also destabilize your system if its gets 'too close' to the first LC.

See the attached where output C is swept by a factor of 5. You can see that the second LC doesn't budge meanwhile the first LC cutoff moves the same amount for both the single and double stage filters.


Post Filter.PNG
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Second, understand that any phase shift risks stability: a single extra pole and its 90 degree phase shift will also destabilize your system if its gets 'too close' to the first LC.
Thanks, I am speaking of cases where the output feedback is taken upstream of the RC filter. Though I believe we would say that the reduced phase shift of an RC post filter means that its not so bad as an LC post filter?
Also, we are currently doing a 12kW SMPS……the load will be eight brushless DC motors and their inverters……..we don’t know how much capacitance the customer has put at the input to their inverters…..we also don’t know what the resistance of the bus bars/wires will be going to the inverters from our power supply…..As such, we may well get a very significant output RC post filter….and go into instability.

…….I am wondering if there are any techniques for such “unknown loads”?

In fact, I wonder if we should use the Synqor NQ60W60HGX40 module instead of what we are now using. The NQ60W60HGX40 has an internal output filter of 40uF_330nF_13uF . Also, on page 12 of its datasheet it states that there is no upper limit to the amount of external output capacitance that can be added to it….this would appear to make a mockery of the App Note that I posted in the top post of this thread.?

Synqor NQ60W60HGX40 datasheet:
https://www.synqor.com/products/hvnq/nq60w60hgx40
 

any techniques for such “unknown loads”?

Yes, put damped C's on your power supply, and damping in the control loops, 3rdly specify max capacitance load, it is generally only a problem at light load / no load - as at full loads the R of the load damps any amount of C considerably ...
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks, and by damped C's i take thats C's with "added ESR" (series resistors added)?
 

The second paragraph from the bottom of page 7 of the below, basically implies that if an output inductor is placed in the DCDC module output, then the output capacitance can be above the datasheet_specified_maximum, “C(OUT_EXT)” . –Which begs the question, what in that case is the maximum output capacitance allowed?

https://www.vicorpower.com/documents/application_notes/an_Parallel_DCMs.pdf

This paragraph essentially implies that the C after an L in the DCDC module output, can be as big as one pleases……this goes against common theory?
Do you know what’s going on here?
The maximum output capacitance of the Vicor DCM3623T50M53TC200 DCDC module appears to be limitless?……as long as its pre-charged if above 2000uF. (otherwise the module want be able start up)
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top