Uc3823n PWM Controller

Edwin,

Can you post the rest of your schematic?

Also what happens if to remove the NAND and IR21844 and drive the main switching FET directly with the UC3823? i.e non synchronous mode for the moment.

Regards,

Richard.

Hi richard,

Thanks for your reply. This is the complete schematic. And a correction here... I used a 4uF bootsrap instead of the 0.01uF shown here.

In effect, as you suggested, I tried non-synchronous firing once. I immediately stopped because the rms current being drawn from the DC power supply was in excess of 1.2A which is not normal at off-load. That drawn in sync mode is 0.085A.

Thanks.

Edwin

Edwin,

Here are some comments on what I can see of your schematic

1. Location of R11, R12 and the boost feedback resistor shoal be close to the UC3823 on your PCB. The traces from pins 1 and 2 are high impedance and can be susceptible to noise.
2. This is a closed loop system with no apparent feedback compensation. You shoal add a compensation network around the error amplifier, between pins 1 and 3 of the UC3823.
3. There doesn't appear to be a voltage feeding the floating gate drive of the IR21844, i.e. anode of D1 appears to just connect to two capacitors. I suspect "+15V_BOOST_1" should also connect to this node.
4. Dead time on IR21844 is  programmed with a 200k resistor. This gives a dead time of 5us while switching period is 10.6us. I think the dead-time should be much smaller, in the region of 500ns.
5. The flying capacitor is only 0.01uF, seems low, I would expect this to be 0.1uF to 1.0uF.
6. Also I don't see the decoupling caps for VCC of the IR21844, again something between 0.22uF and 1.0uF would be suitable.

Regards,

Richard.

Hello Richard, I am grateful for your pertinent remarks. 1. I have in effect located R11 and R12 close to the UC3823 2. You got me thinking here... I need to learn how to apply compensation here. Seems a bit complex. Reading about it. Meanwhile, any rapid tips as to how I can easily go about it? 3. D1 is in effect connected to the supply of the gate driver. Due to the low visual quality of the diagram, it is not easy to put into perspective. 4. Dead time: I am aware of this situation. I used a 10kOhm resistor instead. 5. The bootstrap (flying) capacitor is a 4uF. This should be very sufficient. 6. I used several decoupling capacitors. I did this set up on breadboard. I was forced to use several cables. Maybe this could be a source of EMI. However, I now strongly suspect compensation is the main issue... Thanks Regards. Edwin

Edwin,

Switch mode power supplies on a breadboard are always difficult due to the high di/dt 's and dv/dt's it their circuits.

Regarding compensation for a boost take a look at topic 3 of our seminar series, see link: http://focus.ti.com/docs/training/catalog/events/event.jhtml?sku=SEM408003

Richard.

Hello Richard,

Thx for the documents. The first thing I noticed with topic three is that equation 15 (on page 7) does not seem complete. Gvd(s) does not contain a term in s. Don't know if this is just a slip or a miscomprehension on my part. Then, what is the difference between control-to-output and duty cycle-to-output trnsfr functions?

Anyways, I have referred myself extensively to "Fundamentals of Pwr Electronics" by RW Erickson. Actually, the parameters for my ideal boost are: Vin = 42V; Vout = 189V; Power = 1875W (3750 W initially desired, but a lower power prototype for now); L = 1mH; C = 1000uF. The bode plot of the T(s) transfer function for the uncompensated system ( T(s) = Gvd * H / Vm ; where H = 0.014) shows that the the cross-over frequency is 96 Hz! This is really low! But I have rechecked calculations and it seems correct. With the help of MATLAB, I designed a type II compensator that moves the cross-over frequency of the system to 490Hz and a phase margin of 89.2 degrees. According to topic 3 (of TI), if R6 = 144 kOhm, this will correspond to R8 = 1kOhm; R5 = 4.45 MOhm; C4 = 2.181nF; C5 = 13.67pF. Left for me now to test these in the lab...

Just two questions:

(1) given that I am switching at 100kHz, will this cross-over frequency of 490Hz be high enough for the control system to work well?

(2) I made a similar analysis of the buck converter. The uncompensated cross-over frequency is about 1kHz. Is this high enough NOT to need type II compensation? Maybe so, because the buck gave the required output voltage of 52V (actually it gave 51.8V).

Just thought I should let you know that my ultimate aim is to build a UPS with a 1.875kVA power rating and a less than 5% THD.

Thanks alot for your help and understanding. Regards. Edwin

Hi Richard,

Compensation didn't work. Here is a block diagram of how I modelled the system. The parameters I used are

Vin = 42V

Vout = 189V

L = 1mH; C = 1000uF

(Neglected circuit non-idealities).

Switching at 50kHz. I know the power electronics part is ok because I replaced the Pwm controller with a function generator and it worked. Please help. Thanks

Regards

 

 

Edwin,

Try placing a 10nF cap between pins 1 and 3 of the UC3823. This will add a pole to help compensate the circuit.

In the referenced paper, the Z parameters are in the s domain, as they refer to the complex impedance of the output cap and inductor.

Richard.