UCC28740: Based Constant current flyback converter development

Hello Aban,

Thank you for your interest in the UCC28740 flyback controller.  This controller is designed to regulate output voltage, but has a feature to regulate to constant current on an overload.  This is generally used for charging batteries, where the discharged battery pulls maximum current continuously until its voltage rises back to the charged level.  The constant current regulation is tailored to the battery rating to avoid overstress. 

The UCC28740 is normally a voltage converter and to reconfigure it into a constant-current (CC) converter may be a bit of a challenge.  I looked through the available reference designs but did not find any that specifically targeted CC operation as the "normal" state.  That is not to say that it cannot be done, but that we don't have a reference design which has already worked out the possible issues. 

As I see it, there are two ways to achieve this goal:
1.  Design it as an 80-V, 0.6A voltage regulator but run it with sufficient load such that it is always running at full current.  CC regulation will be about 5% in this mode. There is a limit to how low the output voltage can go.  This controller relies on reflected voltage on the AUX winding to bias the control, and VDD must remain above the UVLO threshold of ~8V.  Maximum recommended VDD is 35V, which corresponds to the 80V output level.  8/35 = 0.229, so rounding to Nas = 0.25 (for margin to UVLO), this suggests that Vout can drop from 80V down to ~32V before VDD is in danger of crossing the UVLO limit and shutting down the converter. 

The feedback loop is designed as usual, as a constant voltage output. To operate in CC, the load resistance must be initially low enough to draw 0.6A from 80V (RL = 133ohm), and should go lower from there.  Min RL (max load) = 32V/0.6A = 53ohm.  Lower RL than this will starve the VDD from sufficient reflected voltage and the device will shut down.  If your system can operate within these constraints, this is the simplest approach.
Note: while operating in CC mode, the voltage at TL431 REF input will be less than 2.5V.  It will be fixed at 2.5V only while in CV mode.

2.  The second way is to configure the secondary-side feedback to directly regulate output current.  To do so, you will need to sense the current through a current-sense resistor and compare it to a reference.  The error signal from that comparison is amplified and drives the opto-coupled feedback network. This method should have tighter control over the current and maybe achieve +/-1~2% regulation, compared to the ~5% in method 1, but requires extra circuitry. 
Also, it is still subject to the same reflected voltage limitation for VDD, unless you provide a separate bias supply on the primary side.    

In both cases, you will need an opto-coupler to maintain primary-to-secondary isolation, and you'll need a secondary-side reference and regulator.  A TL431 will work for both methods, and in nearly the same way.  In method 1, the TL431 sense the output through a resistor-divider from Vout and compares its REF input to an internal 2.5V reference. The difference in method 2 is that a separate current-sense op-amp will sense and amplify the current signal and use that as the feedback sense to the TL431 REF input. For example, a 1-ohm current sense resistor converts 0.6A into a 0.6V signal.  This signal is then amplified by 10 and a 6V signal is presented to the TL431 REF divider.  So the TL431 is essentially regulating to maintain a 6V "output", which is derived from a 0.6A current. In this example, sense loss is ~0.6W.  A smaller sense resistor can be used to lower the power, but the amplifier gain must be increased to get a decent signal level for the TL431 to regulate.    

In both cases, the TL431 (and other shunt regulators) has a max input voltage limit, in this case about 36V.  So it can't run directly from an 80-V output, but must use a linear regulator (or low-power switcher) to drop its bias voltage down to say ~12V.   

In the Excel Calculator Tool that you provided for review, some corrections are necessary on the "Start Here" sheet:
row 27 should be 30V, not 79V, to provide the CC range (from 80V to 32V)
row 28 should be 0.5V, not .0.5V, (which is just a typo).  But this is voltage transient parameter which may have no meaning in a CC design.
row 52 should be equal or slightly lower than the recommendation in row 51.
row 57 (Npa) should change to ~5.56 to maintain Nas = 0.25 when Nps = 1.39 (row 49).
row 86 should be 2.5V, not 80V.
row 87 should be 4uA, not 10000uA.
rows 89 and 91 should follow the new recommendations.
row 96 should be about 2kHz, not 80kHz.  (Optos are not that fast.)

I hope this information gets you on the right track.

Regards,
Ulrich

Hi Aban,

Thank you for the additional information.  It gives me a better idea of your application and possible options.
I believe that Method 1 can work well for you unless you need very tight current regulation.

24 LEDs at 3.4V each add up to 81.6V.  I recommend to set the nominal output voltage to 85V to give some headroom to get into CC mode.
81.6V/0.6A = 136R.  Estimating that 100R is equivalent to 18 LEDs, minimum voltage is 18*3.1V = 55.8V ( /0.6A = 93R).  This range fits well within the limits for keeping VDD alive.  In fact, you may be able to push down to 12LEDs minimum, where 12*3.1V = 37.2V. 
This voltage reflected to the AUX winding by a factor of 0.25 = 9.3V on VDD (neglecting diode drop).  It is cutting it close, though.
If it is desirable, you may adjust the AUX to SEC turns ratio Nas = Na/Ns from 0.25 (1/4) to something a little higher, like 0.275 depending on the granularity of the turns available. (For example, Nas = 0.25 = 8T/32T, while Nas = 0.281 = 9T/32T, or Nas = 0.276 = 8T/29T.)

Anyway, I checked the effect of row 52 and do see the difference.  Going back to the "Start Here" sheet, I changed:
row 24 changed to 85V  
row 26 changed to 0.6A to match row 25 (ignore the warning note)
row 31 to 90V for OVP
row 49 to 1.313 due to the higher Vout target
row 52 to 0.36R (row 51 recommends 0.337R, but 0.36R produces actual Iocc at 0.599A)
row 55 to 250uH
row 57 to 5.25 to reset Nas to 0.25 since Nps changed.

Even with these new values, you may need to do further refinement, especially with the transformer turns, since the ratios can fractional numbers but actual turns on each winding must be whole numbers.  You may need to adjust the ratios to accommodate whole number of turns.

Also, I suggest to use 2 or 3 current sense resistors in parallel to better tune the actual current limit in your application.
Use 2 or 3 parts with same value to get close, and one more parallel part of higher value to fine tune to the final Rcs value.

Regards,
Ulrich

Hello Aban,

To address your latest questions:
1.  Yes, that is correct, The TL431 REF will go to the divider from 80V (85V?) and the 12V regulator will power the otpo and Tl431 circuit.  It will not go directly to the TL431 cathode, however, but to the top of resistor Rtl as seen in the schematic diagram on the "SCHEMATIC AND BoM" sheet of the Excel Calculator.  

2.  The optocoupler really doesn't do anything during CC mode.  Basically the FB current is cut off.  However, the system will need something to regulate the voltage if the load becomes disconnected or drastically reduced. ( You touch on this point with your question #5. )  Without feedback current, operation at output current less than 0.6A will result in Vout increasing until it hits the OVP threshold detected on the VS pin.
OVP will cause the UCC28740 to shut down and restart.  If load is still less than 0.6A, OVP will happen again, and it will cycle off and on indefinitely, until input power is removed or the load is increased to 0.6A again.   

3.  VS pin does not set the output voltage but it does detect an output overvoltage from the output reflected to the AUX winding.  It also detects the Vaux zero-crossing to set up the timing for the next switching cycle turn-on.   

4.  The conversion ratios that you mention are true for a forward-mode converter, but this is a flyback.  The secondary winding of the flyback transformer has an inverted phasing with respect to the primary winding.  Therefore the secondary-side voltage that you calculated do appear on the SEC winding, but as a negative voltage applied to the output diode.  The maximum reverse voltage on this diode can exceed 400V in your design.  I think you'll probably need a 600-V 3-A ultra-fast diode for the output.

5.  A no-load condition (or light load) will operate in voltage-mode regulation and Vout will be regulated to 85V.  When the load is increased to 0.6A, it will still be 85V, but when the load is attempted to increase higher than 0.6A, the CC limit prevents higher current, so Vout will start to fall.  Iout will be regulated to 0.6A +/-5%.

Regards,
Ulrich