Solution Finder Get Tech Support

300W Offline SMPS but only 80W average

Posted by: treez on

Hi,

Our load can be 300W, but at most for 5 seconds. The Load duty cycle is 0.26, so the average power is only 80W

Vin = 100-265VAC, Vout = 37V

Do you believe the HyperPFS PFC followed by the HyperTFS 2 tran forward is the way to go?

We dont want to do the Duty cycle>0.5 thing, so can we avoid that high side power rail which allows >Vin across LPRI and hence D>0.5?

What about the bias flyback in the hiperTFS, does it cause noise issues at all? (having two smps controlled by the same controller, with the possible crowding of feedback signals and switching nodes.?

Comments

Submitted by PI-Wrench on 03/01/2023

The HiperPFS and Hiper TFS2 looks like a good combination for your purposes. You could also consider using the combination of a Hiper-PFS and Hiper-LCS, though the LLC transformer design will be more complex.   

The flyback and forward sections of the Hiper-TFS2 have separate controllers - the flyback section is very much like a TinySwitch-4.  I am inquiring about the feasibility of disabling the extended duty cycle of the TFS2. The extended duty cycle does offer some advantages - use of a lower PIV output rectifier, for one, better brownout capability, and higher efficiency.

FYI, the >B+ rail you mention feeds the high side switch. 

The design will change a bit depending on whether you select 66kHz or 132kHz operating frequency for the TFS-2. At 132 kHz, you probably will need to insert a minimal gap in the main transformer (we used 0.0005 inch thick polyester tape), so that the main XFMR primary can ring down far enough to allow the diode that goes from the primary bias to the high side supply to charge up the high side bias - otherwise, you can encounter missed pulses at light load. At 66kHz operation, this is not a problem.

Submitted by treez on 03/01/2023

Thanks, how would the Hiper-LCS be with fast load transients....sudden no_load_to_full_load and vice versa?

Submitted by PI-Wrench on 03/01/2023

OK - let me get this straight, what you are really asking for here is a power supply that idles at zero load, and peaks to 300W at 26% duty cycle?

An idle load would ease the transition from minimum to full power.

Submitted by treez on 03/01/2023

Thanks, just a few Watts you think?

Submitted by PI-Wrench on 03/01/2023

It will depend on how fast you can tune the control loop. The  Hiper-TFS2 may be better in that regard than the Hiper-LCS. I just finished looking at a 180W, 12V supply using Hiper-TFS2, and the output voltage excursion for a 3% to 100% load transient was about 700mV, with recovery in about 500us. At 37V output, you should be able to do better, especially if you don't choose too large of an output choke value.

Submitted by treez on 03/01/2023

Thanks, with opto feedback, do you think we can get crossover >2kHz?

Perhaps if we use one of those fast optos where you have to give them a 5v VCC?

I must admit i see your point, i never associate the LLC with a good transient response, more a slow, quiet, lowloss, stately kind of thing.

Submitted by PI-Wrench on 03/01/2023

Gain crossover of >2kHz should be achievable, even with a fairly standard optocoupler. The only caveat I have for your situation is that you will need to play some tricks with the TL431, as it is only rated to ~35V.

Submitted by treez on 03/01/2023

Thnks, i read Basso, and he says for good feedback loop bandwidth..use sec side bias supply and opamp error amp....so we may do that?

Specially since HiperTFS is voltage mode........which i dont associate with conduciveness to fast feedback loop(?)...but with Hiper-TFS its got some tricks up its sleeve like feedforward etc?

Submitted by PI-Wrench on 03/01/2023

You can try opamp feedback, but you might want to use something a little better than a n LM358.

I took a look at the datasheet for TFS-2, and we have features like duty cycle limiting and XFMR saturation protection, but other than that - voltage mode.  I also took a look at our DER-360 12V/190W reference board using the TFS7703H, and its loop crosses over at 5.4kHz, so decent loop response is achievable with a standard TL431/optocoupler setup. The output voltage sag for the DER-360 reference board is only 200mV for a 3% - 100% load step, so I suspect the loop is faster than the previous example I cited. The amplitude of the initial transient when the 300W load is applied will be determined by the ESR of the output caps and the amplitude of the load step. This is assuming a stable loop that does not overshoot. The absolute maximum rating of the TL431 is actually 37V.  Standard practice is to use ~80% voltage derating, so placing an 8.2V zener diode in series with the TL431 cathode will drop sufficient voltage to get the desired derating.  You may need to use a slightly higher voltage zener, as I doubt that there will be sufficient TL431 cathode current to push the diode current past the knee - so, a slightly "starved" zener.