# Low cost buck SMPS with IC AOZ1284 - 5V x 4A - values for COMP pin

I want to use a buck converter based on IC AOZ1284PI from Alpha & Omega Semiconductor. AOZ1284PI is a low cost buck controller. For example, on Digikey it costs USD 0.51 @ 500 whereas the TPS54340 from Texas Instruments costs USD 2.53 @ 500. The IC support up to 4A of continuous output current, so I want to use it to reduce a 12V rail to a 5V rail capable of delivering 4A. Below are the parameters that I used to make the calculations using the equations shown in the datasheet. I have calculated RC to be 180K ohms e CC to be 1.5nF. I would like to ask if somebody can do a revision on these calculations, and check if these values are really suitable, the value of CC and RC on schematic, that goes connected to COMP pin.

Parameters:

• Vfb = 0.8V
• Vout = 0.8 + (0.8 * 1800/330) = 5.15V
• Iout = 4A
• CO = 141uF
• COesr = 5 mOhm
• inductor = 4.7uH 8x8x4mm (SRN8040TA-4R7M). Current 5.8A. Saturation 6.7A
• FSW = around 1MHz (47K 1% resistor)
• fc = 30KHz (crossover frequency) or maybe 5khz if better
• Vin = 12V
• RL = 5.15V / 4A = 1.28 Ohm (datasheets states RL)

Calculations:

• fp1 = 1 / (2pi * CO * RL)
• fp1 = 1 / (6.28 * 141 uF * 1.28 Ohm)
• fp1 = 1 / (6.28 * (141*10^-6) * 1.28)
• fp1 = 882.2

• fz1 = 1 / (2pi * CO * esrCO)

• fz1 = 1 / (2pi * 141 uF * 5 mOhm)
• fz1 = 1 / (6.28 * (141*10^-6) * (5*10^-3))
• fz1 = 225866.2

• RC = fc * (VO/VFB) * ( (2pi * CO) / (GEA*GCS) )

• RC = 30000 * (5.15/0.8) * ( (6.28*(141*10^-6)) / ((200*10^-6)*(4.5)) )
• RC = 190009.25
• RC = 180K...

• CC = 1.5 / (2pi * RC * fp1)

• CC = 1.5 / (6.28 * 180000 * 882.2)
• CC = 1.5*10^-9
• CC = 1.5nF...

Then, finally

• fp2 = GEA / (2pi * CC * GVEA)
• fp2 = (200*10^-6) / (6.28 * (1.5*10^-9) * 500)
• fp2 = 42.4

• fz2 = 1 / (2pi * CC * RC)

• fz2 = 1 / (6.28 * (1.5*10^-9) * (180000))
• fz2 = 589

Schamatic:

Below I show my current layout, from top layer:

On the bottom layer, I have a good ground poor.

Regards.

EDIT1 Schematic and layout after user Verbal Kint answered to the question. 12V input, 5V output @ 4A

• A 30-kHz crossover goal is rather aggressive and it is not possible to comment without having the control-to-output response of your buck converter featuring the caps ESRs. Beside, unless there is an internal pole in this chip (which I doubt), you must compensate the loop with a type 2 compensator, not a type 2a as implied by your configuration. It also implies that the op-amp has enough OL gain at 30 kHz and considering the poor data-sheet content, I don't know how you can check this. Commented Feb 4, 2020 at 12:10
• Which range of crossover frequency do you think would be more suitable? There is no need to be 30Khz in fact. Commented Feb 4, 2020 at 14:10
• According to the datasheet drawing of the typical application circuit, there is only one cap and one resistor in series on COMP pin. There is no cap between COMP pin and GND. Commented Feb 4, 2020 at 14:24
• researchgate.net/figure/… Commented Feb 4, 2020 at 14:28
• Well, I have seen the data-sheet and it is difficult to give less information : ) A second cap. across the $RC$ network is necessary to roll-off the gain at high frequency. A more reasonable crossover frequency would probably reside between 1 kHz and 10 kHz, depending on your experience with this exercise. Commented Feb 4, 2020 at 14:49

As I said in the comments, before attempting to stabilize a converter of any type, you need its control-to-output transfer function. The problem here is that the data-sheet of this Alpha & Omega chip is eloquently empty so difficult to figure out what its internals are. Anyway, I have captured a schematic using Elements, the SIMPLIS free demo version in which I included the output caps. ESRs and the inductor resistance:

The chip can switch up to 6 A with a 0.22-$$\\Omega\$$ internal sense resistance and there is probably some internal slope compensation but there are no details. To compensate this guy, I have automated calculations as shown in the book I wrote a while back:

From the simulation, you first verify the operating point is ok, meaning the converter regulates and delivers 5 V from the 12-V input source:

The circuit switches at 1 MHz and delivers 4 A to the load. The feedback voltage is around 900 mV and we can extract the control-to-output transfer function now:

If we pick a 10-kHz crossover frequency, we extract the following data from the graph: the magnitude at $$\f_c\$$ is -5.8 dB while the phase is -82°. Enter these data in the automated sheet and run the simulation again. Look at the compensated look gain and check it is ok:

Oui ! Perfect, a 10-kHz 0-dB crossover frequency with the wanted 70° phase margin. A rock-stable design for this operating point. You must now explore various situations (load changes, input voltage changes, output cap. value and ESR spread etc.) to make sure the stability is not at a stake in any of these situations but for a simple project it should do well. The calculated components values for the COMP pin are: $$\R_2=76.4\;k\Omega\$$, $$\C_2=55\;pF\$$ and $$\C_1=835\;pF\$$. You can round these values to the closest normalized values of course. Good luck with this design!

• Thanks very much for your work, I really appreciate your help. To be clear, on my schematic I will do RC = 82K (1%), CC = 820pF (NP0 5%) and CP = 56pF (NP0 5%), thats ok? I have added updated schematic and PCB layout on the bottom of the topic, please check them now. I have added the CP cap to the layout. Another question: If I want a range of input from 10 to 32V instead of 12V fixed, the values of these components would be different right? Or the circuit should work fine also with the values you give? Thanks again. Commented Feb 5, 2020 at 15:09
• No problem, I'm glad if I could help. Because it is current-mode control, it should be rather insensitive to input voltage variations. Even if the operating mode changes (CCM to DCM in high line), I don't expect any issue. Good luck. Commented Feb 5, 2020 at 15:12
• CP and CC values are correct on the schematic of the bottom? Just to have sure, maybe I inverted them Commented Feb 5, 2020 at 15:14
• They look ok to me. Commented Feb 5, 2020 at 15:17
• Ok. Thanks very much, again. Extreme good work. Regards. Commented Feb 5, 2020 at 15:19