UCC28780: Question about EXCEL design Calculator of the UCC28780

Hello Mohamed,

Thank you for your interest in the UCC28780 ACF controller.

The Brown-in and Brown-out voltage levels for the UCC28780 are tied together by a factor of 0.83.  In other words, the Brown-out shut-down voltage threshold is fixed at 83% of the Brown-in start-up voltage level.  Usually, the start-up voltage is targeted to be a little below the minimum level of the input range (in this case Vin (min) = 97Vrms).  To allow for tolerances, I suggest to allow start-up about 5% below the normal range.  However you may also choose to start up above your minimum input, provide that the resulting Brown-out level does not exceed your minimum input target (97Vrms).

For an AC input of 97~165Vrms:
set C16 to “AC”
leave C17 alone.  C17 is a calculated result and should not be used as a user input. (This is a mistake in the tool.)
set C18 to “92” Vrms   (for example)
set C19 to “165” Vrms
set C20 to “165”

With 400Hz AC input, your bulk ripple voltage will be small.  The peak of 97Vrms = 137Vpk.  Your bulk capacitor value will be more likely chosen based on rms ripple current capability than for minimum hold-up voltage.  For an example, I will guess that your instantaneous Vbulk _min will end up to be around 100Vdc.
set C21 to “100” (this is Vdc, not Vrms)
set C22 to “97” Vrms
set C23 to “400” Hz

The choice of minimum switching frequency (fSW) depends first on whether you intend to use Si-based or GaN-based MOSFETs for the power stage.  GaN Fets have considerably lower Coss and can switch much faster than Si, but currently are more expensive.  The nature of ACF has it that fSW will vary roughly 2:1 over the line range and maybe more over load.  The lowest frequency is at low-line full load, and goes up as Vin goes up and load goes down.   Reasonable fSW ranges for GaN are from 400kHz to 700kHz and higher, while Si Fets operate from 200Khz to 400kHz. Even lower to 150kHz for very low input voltages.  The choice is often an economic one.
set C24 to “200” kHz (for example, for a Si-based design)

After the turns ratios are selected, the actual number of primary turns (Np) depends on a number of factors relating to transformer design.  That is beyond the scope of the datasheet and the calculator tool.  Np depends on the cores’ area factor, core loss parameters, allowable temperature rise, and choice of operating frequency.  This is a very iterative process, and you may need to adjust the target turns-ratios slightly to accommodate a whole number of turns (fractional ratios are okay, no fractional turns).  Calculation results of core loss and copper loss may influence a modification of your target switching frequency.  At these frequencies, make sure you use Litz wire.  It can be a very iterative process to compromise several opposing considerations to optimize the overall design, compared to optimizing only one single item at the expense of several others.

Regards,
Ulrich

Hello Mohamed,

Two tables of fault protection functions are listed on pages 30 and 35 of the UCC28780 datasheet.  Table 2 lists the controller's reactions to system level faults, while Table 3 lists reactions to pin-level faults.  Section 7.4.10.6 provides details for the output short-circuit protection.  After detection and shutdown, the controller will automatically attempt to restart after about 1.5 seconds.  If the output short persists, it will continue to cycle through detection, shutdown and 1.5s wait period to keep the stress level down.

I don't understand the phrases "increase the transparency time" and "high current call", but it may be a matter of translation.  I guess you may be referring to transient response time (which generally should be reduced, not increased) and high current load? 
In any case, the main issues with an output capacitance of 20mF are:
a.  the time to charge it up during start up, and
b.  peak discharge current during an output short-circuit event.

During start-up in almost all power supplies, the output appears to be a short-circuit when the output cap (whether large or small) is at 0V.  The UCC28780 has a current limit, based on Vcst(max) = 0.8V, and this full current is applied to the output cap to charge it up to the regulation voltage.  The controller allows 160ms to operate at the current limit, then shuts down due to an OPP fault.  In most designs, the output reaches regulation far sooner than 160ms and the current falls to normal levels, so OPP shutdown is avoided. 

In your case the extra high output capacitance may take longer than usual to charge up, depending on the available current and any load that is applied during start up.  For example, for 35W rated output at 30V, you may have a ~45-W OPP limit which allows 1.5A charging current.  Charge-up time = 20mF*30V/1.5A = 400ms. Any loading during start-up will make it take even longer. This exceeds the 160ms OPP-fault window, so your system will not be able to start up with a reasonable OPP limit.  Please reconsider the need for such a large output capacitance.  Based on the controller fault timing, you'll need to reduce Cout to 10~20% of 20mF to reliably start-up, assuming a small to no load. 

The alternative is to design with lower Cout, and include an active circuit between the low Cout and the 20mF which delays charge up of the 20mF until Vout is regulated.  Further, it would limit the charge current to less than rated Iout (basically <1A) and prevent loading of the 20mF until it becomes fully charged over 600ms later. 

The other issue is the peak surge current during the short-circuit, which will flow through the connector and output wiring limited only by series resistance and stray inductance.  Everything in the short-circuit loop path must be capable of withstanding the peak current as well as any peak voltage from L*di/dt ringing that may result from the stray inductance.  This is true for any-size cap, but the peak magnitude is much higher for 20mF than for smaller caps, of course.

Regards,
Ulrich

Hello Mohamed,

Thanks for explaining "transparency time".  This parameter is also often referred to as "hold-up" time or "ride-through" time. 

In any case, I believe that your output capacitance target is greatly underestimated due to an equation error.  Your equation assumes that the total energy content of Cout will be drawn down to 0V to provide the 35W output power for 200ms.  However, I will assume that the output voltage must stay relatively regulated during the power outage and will only be allowed to sag by 1 or 2 volts to keep your load powered up.  The following equation then applies:  
Cout = Tr * P / (0.5*tol*(Vout1² - Vout2²)).  So, Cout >/= (200ms*35W)/(0.5*0.7*(30V² - 28V²)) = 172mF.   If your hold-up regulation can be looser than -2V, the cap size can go down, but if tighter, the cap size must increase more!

In any case, storing energy at low voltage is volumetrically very inefficient and costly.  Since you have at least 97Vrms input ahead of the ACF converter, it would be advantageous to store energy at a much higher voltage and allow the ACF converter to be "normal sized" and regulate from the high voltage hold-up.

Using my previous "guess" of 100V minimum bulk voltage (for example) and based on 97Vrms input (137Vpk) we obtain Cbulk >/= (39W*200ms)/(0.5*0.7*(137V² - 100V²)) = 2540uF.  (39W liberally accounts for ACF conversion losses; it's probably better than this.)  A lower minimum voltage will reduce this cap size further, but will also require the ACF converter to operate over a wider input range.  Still, it may be smaller and cheaper to pursue this option than using a massive low-voltage output cap.

That would also eliminate the issue of start-up OPP timing limitation or special current-limit circuits.  Anyway, the UCC28780 does automatically attempt to restart after each OPP-fault shutdown. It will continue to cycle through restart attempts whenever the 160ms OPP limit is exceeded, but will start-up and stay powered if the power is decreased below the limit within 160ms.

Regards,
Ulrich

Hello Ulrich,

It does make since for me, thank you so much for your answers !

Have a nice day.

Regards,

Mohamed