UC3843 BUCk Buck Circuit Implementation

nos001

UC3843 BUCk Buck Circuit Implementation [Copy link]

 

The UC3843 BUCK step-down circuit works normally at no-load and low current. From the current sampling using 1R, it can be seen that the design itself is used in low current situations.

Since the input voltage is up to 60VDC, LM2576HV was used before, but it was easy to burn out because it was not a genuine product and the genuine product was too expensive. This is the origin of the simple and complex.

UC3843 itself has become the hottest device. Is the current driving the MOSFET too high? I don't know if there is any good way to solve this problem.

Before implementation, I simulated many times to see if the bootstrap power supply could work properly. It actually worked very well.

In addition, the pinout of TL431 is inconsistent with the TI manual. The reference pin REF and the adjustment pin K of the SOT23 package are reversed, which really screwed me.

maychang

In your electrical schematic, what is the relationship between SGND (pin 5 of 3843) and the ground wire symbol (negative terminal of the rectifier bridge output)? Are they the same point?

nos001

maychang posted on 2019-7-21 19:49 In your electrical schematic, what is the relationship between SGND (pin 5 of 3843) and the ground wire symbol (negative terminal of rectifier bridge output)? Are they the same point?

The characters were lost when the screenshot was taken. It is the input power ground PGND, and SGND is the floating ground of the UC3843 power supply.

伟林电源

1. Is the 3842 too hot? It is probably because your VCC voltage is too high. You can add a current limiting resistor and a voltage stabilizing Zener diode.

2. SOT-23's 431 has two kinds of pin positions, so please pay attention when you buy it.

nos001

Weilin Power Supply published on 2019-7-21 21:21 1. Is 3842 too hot? It is probably because your VCC voltage is too high. You can add a current limiting resistor and a voltage stabilizing Zener tube; 2. There are two types of pins for SOT-23's 431...

When I put my hand on 3843, it feels a little hot. If the ambient temperature is high, it will probably reach 60℃. I don’t know if it can withstand it.

In fact, the operating voltage of UC3843 is only 12V, which has nothing to do with the input voltage. The input power supply only starts 3843 through a 20k resistor. After startup, it relies on bootstrap power supply, which is only 12V.

伟林电源

nos001 posted on 2019-7-21 21:43 I feel a little hot when I press my hand on 3843. If the ambient temperature is high, it will probably reach 60℃. I don't know if it can withstand it. In fact, the working voltage of UC3842 is only...

Of course 12v is fine for 3842, 9v is fine for 3843. If you don't believe it, you can try it and see if the surface temperature of the chip will drop.

伟林电源

maychang

　I have a different opinion on the Buck circuit controlled by 3843.
　Chips like 3843 were originally designed to control flyback converters. Two feedback loops were designed for this purpose: current loop and voltage loop, and a high-voltage self-starting circuit was also designed.
　If this non-isolated converter with an output of 12V is changed to a flyback converter circuit, and the 5th pin of 3843 is connected to the negative output terminal of the rectifier bridge, there are several advantages:
1. The optocoupler can be omitted, and the TL431 and the attached resistors and capacitors can be omitted. It is originally a non-isolated converter, so why use an optocoupler? If the 5th pin of 3843 is changed to ground (which is also the power input ground and output ground), the two resistors at the output end of the power supply can be directly sent to 3843. There is an operational amplifier and a voltage reference inside 3843, but these two things are not used in the original figure (the output of the optocoupler is connected to the COMP terminal of 3843, that is, the output terminal of the internal op amp of 3843)
2. Q4 and R16 can be omitted. These two components are designed to generate oscillation when 3843 drives a converter working in forward mode. These two components are not required when 3843 drives a flyback converter.
　A transformer is needed to change to a flyback circuit. However, the Buck circuit already has an inductor. Now the inductor is just replaced by a transformer, and the cost and volume increase are not much. The flyback circuit outputs 12V. Unlike ordinary flyback circuits, the transformer does not need a third winding to power 3843. It can directly power 3843 from 12V. If the output voltage is incompatible with the power supply voltage of 3843, a tap can be added to the secondary winding of the transformer. With a transformer, it is easy to output multiple groups of different voltages or output a power supply suspended above the ground.
　The disadvantage of changing to a flyback circuit is that the drain of the MOS tube must be added with an RCD absorption circuit to avoid excessive peak voltage caused by the primary leakage inductance of the transformer.
　But in general, changing to a flyback circuit has more advantages than disadvantages.
　I wonder what your opinions are.

@PowerAnts @Wei Lin Power

伟林电源

The 38xx series is very classic and is widely used. The PSR structure with primary side feedback is one of the options, and the buck can also be implemented without optocoupler feedback.

PowerAnts

maychang posted on 2019-7-22 10:57 I have some different opinions on the Buck circuit controlled by 3843. Chips like 3843 were originally designed to control flyback converters. For this reason...

The original intention of non-isolation is to pursue simplicity and cheapness while achieving higher efficiency than isolated conversion. However, when the transformation ratio exceeds 3-4, the duty cycle is too small, the switching loss surges, and the effective value of the ripple current of the input filter capacitor is also relatively large, and the efficiency and cost of non-isolation are no longer advantageous.

PowerAnts

Weilin Power published on 2019-7-22 15:31 The 38xx series is very classic and is widely used. The psr structure with primary side feedback is one of the options. Buck can also use non-optical coupler feedback...

38XX does PSR?

伟林电源

PowerAnts posted on 2019-7-22 21:20 38XX does PSR?

Is it not possible? My last post was a reply to maychang.

伟林电源

nos001

Weilin Power Published on 2019-7-25 02:34

In fact, the problem can be solved by an MP4569, which occupies a much smaller PCB area. It is easier to buy a genuine product.

I checked Mouser, Huaqiang Chip City, and Digi-Key, and they all cost $8 to purchase 500+, so I'd better buy it myself.

Taobao is much cheaper, why is the price difference so big? How big is the difference? I can't figure out the routine.

伟林电源

nos001 posted on 2019-7-25 20:36 Weilin Power posted on 2019-7-25 02:34 In fact, an MP4569 can solve the problem, and it takes up a lot less PCB area. If the genuine product is easy to buy. ...

Of course, there are too many choices nowadays. Although 384x is a bit old, for novices, once they have mastered it, it will be easy to deal with other fool chips.

nos001

Weilin Power Supply published on 2019-7-25 20:57 Of course, there are too many choices nowadays. Although 384x is a bit old, for novices, once they have mastered it, they can easily deal with other fool chips...

Why are Taobao products so cheap? Are they fake? Are they disassembled? Mouser is American, so they are a rip-off. Some Taobao products have blurry typing, so fake. In short, there are a lot of good things, but it is hard to buy them.

nos001

maychang posted on 2019-7-22 10:57 I have some different opinions on the Buck circuit controlled by 3843. Chips like 3843 were originally designed to control flyback converters. For this reason...

The transformer of EE13 also takes up more space than the inductor. Besides, the inductor is still needed for rectification. There is also the RCD you mentioned. When the PCB has area restrictions, it feels easier to layout the PCB if there are two less components.

The reason why the adjustment MOSFET is placed at the high end is determined by demand. The 48V power supply is used to drive the solenoid valve, which is controlled by an N-type MOSFET. This requires that the power supply of the control chip and the 48V power supply share a common ground, otherwise logic conversion is required. If there are many control paths, it will be troublesome.

As for the feedback method, if the adjustment tube is placed on the high end, voltage feedback will be used. As the load changes, the simulation will not converge. It seems that the use of optocouplers is better. In short, I am a novice in power supply, so I chose a method that I think is slightly safer.

伟林电源

nos001 posted on 2019-7-25 21:42 The EE13 transformer also takes up more space than the inductor. Besides, the inductor is still needed for rectification. And the RCD you mentioned, when the PCB has area restrictions, 2 yuan less...

That being said, as a learning and research project, studying 384x is better than studying those stupid chips.

nos001

Weilin Power published on 2019-7-25 22:02 That being said, as a learning and research, studying 384x is better than studying those stupid chips.

This circuit is only suitable for outputting 9-15V, otherwise bootstrap power supply will not be feasible.

伟林电源

It's not impossible, but it's just a bit more complicated.