Designing Adjustable Voltage SMPS

Status
Not open for further replies.
F

fethiyeli

Full Member level 4
Joined
Feb 9, 2013
Messages
207
Helped
10
Reputation
20
Reaction score
12
Trophy points
1,298
Visit site
Activity points
2,986
Hi, i am planning to make a adjustable voltage smps but there are a few questions i wanted to ask. SMPS has 0-50V and 0-5A output and i would like to use half bridge topology. PWM IC is UC3846. The first question is that how to find the proper output inductor value and according to textbook of "Switch-Mode Power Supplies Spice Simulations and Practical Designs" output inductor value becomes about 2mH for minimum output voltage which has been taken about 0.1V. I dont know whether this inductor value is ridiculous. Another question is feedback. Do i need to do feedback and other calculations according to minimum and maximum output voltages ? I am bit confused because output voltage is adjustable. How can i do the calculations and make the psu stable for all output voltages ? Thnx.
 

fethiyeli said:
SMPS has 0-50V and 0-5A output
.........
inductor value becomes about 2mH for minimum output voltage which has been taken about 0.1V. I dont know whether this inductor value is ridiculous.
Click to expand...

Do you wish to get as low as 0.1 V output? Then you need an extremely short duty cycle. The design tool calculated a value which is compatible with all the specs you gave it.

- - - Updated - - -

You may decide you do not need such a wide range of output.
 

Controlling very narrow duty cycle will be very difficult i guess. So how can i control very narrow duty cycles without changing the given output voltage range ?
 

One way is to greatly lower the switching frequency at the low voltage end.
One microsecond on and 100 microseconds off (1% at 10Khz) is a lot more practical than 200nS on and 20uS off (1% at 50Khz).
 
Do you plan to work with devices that are ruined by voltages greater than 15V? Then you might want to build in an option to set the smps at 15V, and obtain the range of 0-15V by plain resistive drop. That's how I made my power supply. A 12V transformer and 2N3055 transistor. It goes from 0 to 16V by changing bias. Up to 3A (if I can tolerate sagging output V). It has served me well for over 30 years. (The transistor sometimes gets hot enough to burn my skin, but the entire metal enclosure is a heat sink.)

Having it regulated would be a convenience, but I think that regulating it would inhibit the ease of getting the entire range of versatility.

And I have thought it would be a fun project to upgrade it to a variable smps (now that I have a notion how a buck converter works), but I want it to go down to 0V while being sure it will give smooth operation, etc.
 
I have done some research about variable smps but no one has stated about controlling very narrow duty cycles. Only the textbook of "Switchmode Power Supply Handbook 3rd Edition" said that the frequency is needed to get lower for very low duty cycles. Now, i am planning to adjust the frequency when the output is lower than say 25W.
 

Warpspeed has explained how to do it, which is by reducing the frequency such that the resulting very narrow pulses are significantly wider than the sum of Ton + Toff.
 

If you just run your PWM control chip, using a potentiometer to directly control the duty cycle, you will usually find that it can never go any narrower below some minimum, the pulse just suddenly disappears.

The data sheet sometimes says something like minimum pulse width x%.

There are all kinds of unique problems of designing switch mode power supplies that have to work over a very wide operating envelope of both load and output voltage, its a very complex project to tackle.
 
I know that the project is pretty complex but i really want to build it. This project is complex because the output voltage is adjustable so the whole calculations about feedback, output inductor value etc should be calculated according to whole output voltage range. Changing the frequency and fixing the minimum pulse width are easy i think. The question is how to determine and calculate the feedback, etc as i said.
 

All of my own commercial bench supplies use linear regulators, and they get over the dissipation problem by switching transformer secondary tappings, using a sort of auto ranging technique.
As I twiddle the voltage potentiometer up and down, I can hear the relays clicking.

If I was going to build myself a bench switching supply, I think I would use the switching supply part just as a front end to feed a linear post regulator. That has several advantages which include a very clean low noise output, better transient response, and faster current limit.

A 5 amp capable switcher will always need a significant output capacitor, and that is going to produce some nasty fault currents that there is nothing you can really do much about in the way of setting up a current limit.

But if you have a liner post regulator, you can set the output current limit to milliamps, and shut it down very fast. Also going right down to zero volts becomes much easier to do.
 
This setup could be useful to experiment with. The 555 timer IC has an easy method to change duty cycle. Change voltage at the 'ctl' pin to do this.

The 555 cannot tolerate 50V. Therefore the NPN & PNP arrangement is a means by which its low voltage output can control 50V.



The buck converter acts approximately as your specs state, up to 43V, up to 5A. Performance under 1V is uncertain.

There is no feedback loop. Voltage is unregulated. It should be possible to add regulation with a few more components.
 
Building variable linear power supply is pretty easier than smps one. I also would like to improve myself about smps and power electronics. Blocking the short circuit coming from output capacitor can be eliminated by putting a SCR between the output capacitors and load. Gate of the SCR can be triggered in case of short circuit. There also must be a dummy load between the output capacitor and SCR. This protection may be optional. I just wonder how to calculate feedback, etc.
 

An adjustable constant current supply can be pretty useful sometimes.

For instance, setting the output current at exactly one amp, and measuring the voltage drop across something, where millivolts = milliohms.

Your SCR crowbar will not allow you to adjust the current flow through something that is almost a dead short.

Another example is powering CMOS circuits where the current is practically nothing.
If you set the short circuit current very low, say at 20mA then anything really bad that might happen, is likely to just drop the supply voltage to zero rather than produce smoke and flames.
 


Adjustable constant current supply is pretty useful as you said but in case of short circuit, constant current protection may not be sufficient for protection so the whole charge from capacitor will go through the load. So using the SCR as short circuit protection is actually not a crowbar circuit. Crowbar as i know is shorting the output with thyristor in case of overvoltage. I will use the SCR like a diode between the capacitors and the load.
 

fethiyeli said:
Adjustable constant current supply is pretty useful as you said but in case of short circuit, constant current protection may not be sufficient for protection so the whole charge from capacitor will go through the load.
Click to expand...
If the linear post regulator is itself current limited, the output current cannot exceed that, no matter what happens. You can tweak it up to the full five amps, or right down to zero.

What you set with the output shorted, is what you will get.
 

Now i am thinking not to use linear post regulator. Even there is a constant current protection in smps, there nothing to do in case of short circuit at first. Since there is no protection between output capacitors and the load. After the short circuit and discharged capacitors, constant current is able to control the output current. Without protection between the capacitors and the load, there will be huge current spike. After that current spike, constant current protection can control the output current.
 

I will use the SCR like a diode between the capacitors and the load.
Click to expand...
Sounds erroneous. A SCR can't switch-off DC current, at least not without a second commutating switch.

The problem of instantaneous current limiting refers to inherent limitations of a switched mode power supply in achieving current and voltage dynamic. Most modern lab supplies are solving it by combing a switched mode regulator with a linear output stage.
 

Yes you are right. Using a SCR is erroneous. What about using a BJT or MOSFET between the output capacitor and the load.
 

Warpspeed said:
That is exactly what you do when you build a current limiting post regulator.
Click to expand...

Ok now i got There was a protection circuit in my mind and i was always comparing it with what you said. I will share it after i have finished. Thnx.
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…