Boost SMPS is suffering from power source instability?

[cupoftea]

Hi,
We are doing 90W boost SMPS
5VIN, 24VOUT, 3.75Aout, 350kHz, Synchronous but initially we have disabled the synch fets and just using them as diodes.

All the components are as per LTpowerCAD and it indicated phase margin of 61 degrees and gain margin of -24dB. Xover is some 2kHz…..well below the RHPZ (according to LTpowerCAD)
Top and bottom FETs are each a pair of parallel BSC027N04LS’s (directly in parallel, gates and sources directly shorted together).
The board isn’t so well layed out, and there is no ceramic input capacitance…just three 47uF SMD Electrolytics. Switching node not directly over sensitive tracks though.

The odd thing is that it runs much better on the eLoad than the resistive load...ie much less instability when on the eLoad.

With resistive load, it starts going unstable as we get up to the full load……we are supplying it from a CPX400D power supply, (20A max) and I suspect that when CPX400D gets near its max current, we are seeing an instability brought on by our Boost SMPS, and the output filter of the CPX400D. (As you know, high step up boost is a "wild horse" with its low freq RHPZ)

Cpx400D bench PSU
https://www.aimtti.com/go/cpx/index.php?p=index-cpx400d


When the instability starts, we actually see the vin oscillating from about 2v5 to 5v5 at lowish frequency. (ie not anywhere near 350khz)

Do you agree, that we should run it at 80% load, to be sure and avoid the current limit of the CPX400D, and also add a massive output capacitor bank to the CPX400D, so that even if our Boost does go unstable, the input voltage remains “solid” at 5V. I was thinking we could do this with a massive bank of super-capacitors?

..The CPX400D manual says nothing of a maximum output capacitance for it…

CPX400D manual

https://resources.aimtti.com/manuals/CPX400D+DP_Instruction_Manual-Iss13.pdf

..the alternative of course is that we just open up the CPX400D and remove its output filter inductor, assuming it has one…which is surely must?

Do you also agree that when you are bringing up an SMPS...you should never see the input voltage going unstable...you must always supply it with such a huge capacitor bank that the input voltage should remain "flat", and stable, even if the SMPS goes unstable?

 

[FvM]

It's not necessarily the output filter of your source supply. It can be the constant voltage controller interacting with your DUT, or a combination of both. Switch mode converters expose a negative input impedance when driving a constant load.

Do you see the oscillation starting before the source supply current limit is exceeded?

Even better than a big parallel capacitor may be a "lossy" capacitance, capacitor with series resistor.

 

[silango]

Check the stability of the input supply.Simply incresing the capacitor value is of no use.infact it will increses the instabilty under some conditions.

 

[cupoftea]

Do you see the oscillation starting before the source supply current limit is exceeded?

..Good point......when the vout is rising up toward 24v at start up.....just for that short time, when the error amp of the boost is obviosuly saturated, we do not see the vin oscillations...so i guess FvM you are right, we are seeing the feedback loop of the CPX400D and that of the boost SMPS, interacting (fighting) with each other?

 

[Akanimo]

Unless your peak ripple current is only about 10%, you have exceeded the current limit of your power source. My first concern is that the boost is simply getting starved. At 5 V, you can get up to 40 A if you operate the CPX400D in parallel mode. I think you should start from there.