data/text/DLP2/883046

UCC28600: TINA example model setup, changing the transformer to a coupled inductor

Part Number: UCC28600 Hi, I would like to design a multi-ouput flyback converter based on UCC28600 controller IC. I would also like to simulate the circuit, therefore I was trying to change the transformer to a coupled inductor (because later I will need more than one output) in the given TINA example model to see if it is even possible to simulate the same system but with a coupled inductor with the same properties used with the transformer. Unfortunately, if I use a coupled inductor, the simulation result is completely different than before. Is there any way to change the transformer to component with which I would be able to simulate successfully a multi-output flyback topology? Thank you in advance! Bálint

I suggest you simulate just with single output. Then you use cross regulation to understand other output. It is mainly a transformer simulation which you need to build your own model. Basically just to find out the leakages and coupling factors

Hey, First, I really appreciate your answer, thank you very much! I took your advice and I was trying to simulate the circuit only with one output. Right now, all of the component values have been calculated but the circuit either works in a really weird way, or I get the infamous convergence problem. For example, the weird thing is that voltage top of the current sense resistor has a waveform which is far from ideal, it doesn't have the ramp increase, instead the voltage changes instantly. If the current source is around 4 A, than the duty cycle of the OUT pin looks strange as well. It has a kind of a normal cycle, after that some really thin pulses and it keeps repeating. I attached the current version. Do you have an idea why is it working like this? Edit.: I left the given transformer model in the simulation, I just changed its values based on my calculations. mine.TSC

It looks some setup not good. Can you start with the TINA design (attached here, "slum608.TSC") fromcompany web and change one thing at a time to see which change causing the convergence problem? I just ran "slum608.TSC" and it worked ok. slum608.TSC

Hi, Since then I did what you suggested, and the main reason why the circuit was not working is the incorrect secondary diode I've chosen. My problem now is the following: The attached simulation works as expected until 1.4A of load. The problem is, if I go higher than this value, the controller is unable to hold the output voltage on the desired 12V, even though all the values (OVP, current sense etc) have been calculated for 197W output power with 85% efficiency, therefore if the output current would be around 16A, it still should be able to work. Now I'm a bit stucked. I think the chosen OVP resistor values and current sense protection (Rcs and Rpl) have been calculated correctly, I don't really see why the controller stops the operation. Unfortunately, I don't know TINA that well, but my first thought was maybe the diode on the output causes some problems, as the current one is able to handle 1A maximum forward current, but changing that value doesn't help. Maybe somehow the state machine starts to work incorrectly? May I ask your help in this regard? Thank you in advance!step_by_step.tsc

Can you probe each node to see if any hint? Also, when 12V is not held, does it drop or shutdown? If it drops, which means some setup is likely saturated. If shutdown, it means some fault triggered. You need to check one at time.

12V slowly drops, it is quite clear that the controller doesn't put energy from the primary side, so the capacitor on the secondary side just loses its charges and the voltage drops in the same ratio. This is how it looks the drop after the load reaches the problematic level. And there are weird spikes in the current signal after the 2ms mark if I zoom in. This happens at 2A of load. This thing goes even more weird, if I change the load to 16A. In that case the controller keeps the primary side transistor open, and although the voltage on the output drops, the waveforms look really strange after the load goes on.

Can you compare your simulation circuit and the original? My review of your circuit looks you removed the feedback loop compensation components on TL431 so your circuit cannot make correct loop compensation.

I added the loop compensation as you suggested by following the steps in the "Compensation Design With TL431 for UCC28600" (SLUA671) document. The circuit unfortunately still doesn't work as expected. I can go a bit higher with the current up to around 4A (you can see a picture about a simulation with 1.6A load below), but it is still far from the final 16A. A was also trying to add the Zener network later before Rled, but it didn't help either. Based on the transfer characteristics, the compensation network should do the work, I changed the crossover frequency to 3kHz of Vo/Vcomp, it also meets with all the stability conditions. The other really weird thing for me (although maybe I just misunderstood something) is the switching frequency. Shouldn't it work with at least 40kHz bursts? Right now, depending on the load, the switching frequency varies between ~3kHz - 7kHz. And the pictures: Unfortunately it is quite hard to figure out if it reaches 12V or not, because the simulation stops with convergence problem a lot of times. Sometimes pressing 'retry' works, but I don't know how accurate the result going to be after this error. Also if I don't change anything in the example simulation, but I change the simulation time to 30ms, it doesn't run. It quits with convergence error. So back to my circuit: It reaches the target output voltage @1.6A, but during the simulation I had to press 'retry' on the convergence error once. At 3.5A at the start it looks like the system operates in CCM, but the controller is a DCM one, so it shouldn't happen right? The Bode of Vo/Vcomp I also attached the updated simulation circuit. 5824.step_by_step.tsc

How did you decide the loop compensation? How did you design your converter parameter values?

For the loop compensation as I said I followed the method presented in SLUA671. The following MATLAB code shows my implementation s = tf('s'); Rd = 8; ESR = 50e-3; Rcs = 151e-3; Vtl431min = 2.5; Vdd = 5; Vcesat = 0.3; Vf = 0.7; Ibias = 1e-3; CTRmin = 0.4; Rpullup = 20e3; Rledmax = (Vout-Vf-Vtl431min)/(Vdd-Vcesat+Ibias*CTRmin*Rpullup)*CTRmin*Rpullup; R1 = 10e3; Vsample = 2.5; Rlower = R1*Vsample/(Vout-Vsample); Vin = 325; Voutmax = 13; Nps = 15.27; D = Nps*Voutmax/(Vin+Nps*Voutmax); H = Nps*D/(5*Rcs)*Rd*(ESR+1/(s*Co))/(Rd+ESR+1/(s*Co)); %transfer function of Vo/Vcomp bode(H); Gmid = 10^(12.65/20); %to have 3kHz as cross-over frequency Rled = 499; R2 = Gmid*R1*Rled/(Rpullup*CTRmin); [z,p,k] = zpkdata(H); p = cell2mat(p); fz = abs(p(2,1)); fp1 = 40e3; %suggested pole fp2 = fp1; C2 = 1/(2*pi*R2*fz); C1 = 1/(2*pi*R2*fp1); C3 = 1/(2*pi*Rpullup*fp2); Copto = 2e-9; F = (s*R2*C2+1)/(s*R1*C2*(s*R2*C1+1))*(1/(1+s*Rpullup*(C3*Copto/(C3+Copto))))*Rpullup/Rled*CTRmin; %transfer function of compensation network bode(F); bode(H*F); Are you curious about all the converter parameters (transformer parameters etc), or only the ones which needed for the controller (OVP, current sense, Css etc)?