SMPS design ?

[p55xp]

I need a power supply for my project - which is always power on and requires 12V, 1.5A. To make it small in size and light weight I have decided to use one switch mode power supply. Some of my friends told me that an SMPS cannot run conteniously, but I do not believe this.
But reality is I don't know how to design an SMPS of my need.
Please help me so that I can design an SMPS of my requirement.
Thanks.

 

[crutschow]

What is your primary power source?

 

[solis365]

can you just convert a PC power supply? its probably a little bulky but I leave my computers on for long periods of time. They are also often rated over 10A on the 12V rail.

 

[kripton2035]

it's really difficult to design a smps supply that operates from the mains (110 or 220v)
the main difficulty is the transformer, and the high voltages inside (300-400v) that makes the design dangerous if you're not familiar with it.
it's easy to design a linear power supply for 1.5A it would be ok
or you can buy a ready made smps supply on ebay, there are plenty and they are cheap.
like : **broken link removed**
or like : **broken link removed**
regards

 

[p55xp]

What is your primary power source?

Primary source is AC 220V.

 

[p55xp]

it's really difficult to design a smps supply that operates from the mains (110 or 220v)
the main difficulty is the transformer, and the high voltages inside (300-400v) that makes the design dangerous if you're not familiar with it.
it's easy to design a linear power supply for 1.5A it would be ok
or you can buy a ready made smps supply on ebay, there are plenty and they are cheap.
like : **broken link removed**
or like : **broken link removed**
regards

I know it's difficult but I need to cross the difficulty to design according to my spec.
Thanks

 

[q5101997]

Look inside a computer SMPSU and see whats involved. Now get 10 very tallented engineers involved and it should only take 12 months to desighn a safe, reliable and safe supply.

 

[dougy83]

Look inside a computer SMPSU and see whats involved. Now get 10 very tallented engineers involved and it should only take 12 months to desighn a safe, reliable and safe supply.

That's a little negative. I reckon I could design a "safe, reliable and safe" 12V 1.5A PSU using a 555 timer (or maybe some comparators if I felt like lashing out) and some other junk components in less time than that.

OP: If you think that you may have trouble designing the SMPS you could use a all-in-one chip to do the work for you. You'll notice they're very cheap even in unit quantities. You'll still have to select components that are correctly rated and follow appropriate layout guidelines to maintain a safe and reliable circuit. Some of the datasheets following may help you in that regard.

Have a look at the datasheets for the SMPS chips here:
**broken link removed**
A lot of the datasheets contain an example schematic or two, including the transformer winding info.

 

[Hero999]

Here's an idea based around a blocking oscillator, it's probably very noisy and not very efficient but it should work.

 

[p55xp]

Here's an idea based around a blocking oscillator, it's probably very noisy and not very efficient but it should work.

Thanks, HERO999 - this is a proper guide after so many ....
To be very frank I am inspired to design my own SMPS by looking inside a mobile charger, and again looking into CFL electronic ballast, and finally by the above post.
Now please tell me (1) how the chopping frequency is ascertained and control of this frequency to achieve output regulation in case of input fluctuation ?
(2) calculations involved in ferrite transformer design at that frequency.

Again - I have seen similar design in a DVD player, so I think there must be a way to effectively filter out the noise. Please detail the components and explain the circuit you post.
Thanks.

 

[Hero999]

1) The frequency will depend on the transformer and the load connected to it. As the load on the transformer is increased, the frequency will decrease.

2) Just use a ferrite with a more than a high enough power rating (always best to over-rate) for the circuit you're powering. The turns ratio, can be lower than you'd normally expect for a mains transformer because it's a flyback. Try 200 turns for the driver (D), 3 for the feedback (F) and 5 for the secondary (5). I'd recommend winding the driver, then the feed back and finally the secondary. Put a good layer of insulation tape between the feedback and the secondary. Note the phasing on the schematic.

How do you reduce the noise? You could add an inductor to the secondary side, I've edited the above schematic, see L1. The mains side should also have the a suppressor fitted.

For safety reasons C3 must be Y-rated and shouldn't be more than 5nF and if you can be certain of the polarity of the mains it's probably better to connect it to neutral. C3 helps to minimise EMI by diverting high frequency noise from the secondary side to the primary side. Its value is too low to pass enough 50/60Hz maind current to be harmful.

I've never build this circuit, it's just an idea. Always be careful when experimenting with mains voltage circuits. All creepage distances between the mains and DC side of this circuit should be >10mm. Test the circuit by connecting it in series with a 100W light bulb.

 

[Nigel Goodwin]

1)
For safety reasons C3 must be Y-rated and shouldn't be more than 5nF and if you can be certain of the polarity of the mains it's probably better to connect it to neutral. C3 helps to minimise EMI by diverting high frequency noise from the secondary side to the primary side. Its value is too low to pass enough 50/60Hz maind current to be harmful.

You should also have a suitably rated high value resistor in parallel with the capacitor, this prevents static buildup on the secondary which could give shocks, and even break down the insulation in the Y capacitor.

 

[Hero999]

That's a good idea Nigel, although I haven't seen one in any of the PSUs I've looked at.

According to the paper linked below the capacitor but be Y1 rated meaning it should be impulse tested to 8kV so the high voltage resistor should have a withstand rating.
**broken link removed**

 

[Nigel Goodwin]

That's a good idea Nigel, although I haven't seen one in any of the PSUs I've looked at.

It's standard in almost everything that uses a SMPSU - just checked a Pace analogue satellite receiver, that uses 10meg and 1nF.

 

[Hero999]

I couldn't find any Y-rated resistors using Google, perhaps they just use normal high voltage resistors?

 

[mneary]

C3 (the Y-rated capacitor) is often missing in designs I see where the 0V output is earthed to the green (or green/yellow) line.

 

[bountyhunter]

That's a little negative. I reckon I could design a "safe, reliable and safe" 12V 1.5A PSU using a 555 timer (or maybe some comparators if I felt like lashing out) and some other junk components in less time than that.

You can't use a junk circuit like that for an off line PS because it is too dangerous. You can not have a silicon path from the AC line to the output side, or you will get fried if the converter blows..... and they love to blow up, especially while you are fooling with them. You will need a half bridge converter so that the main power transformer provides AC-output isolation and you will need gate drive source crossing the isolation boundary using wither transformer drivers or opto drivers. You will need transformers properly designed for the tasks, heat sinks for the switching components and snubbers which will keep the switches in safe operating area under all load conditions. This is not trivial, I did it for about 25 years. It's not a beginner project.

 

[bountyhunter]

OP: If you think that you may have trouble designing the SMPS you could use a all-in-one chip to do the work for you.

There is no such thing. You still have to do all the hard work. Believe me, we would have bought a chip that did the work if there was one. BTW: you can learn a lot looking at a typical off-line PS board. There is usually a dotted line showing the HV/output side boundary line. The newer ones use dedicated power IC's with the switches inside but the transformer, filter, gate drivers, opto isolators etc all have to be designed and you also have to the loop compensation which is not trivial.

 

[Hero999]

C3 (the Y-rated capacitor) is often missing in designs I see where the 0V output is earthed to the green (or green/yellow) line.

That's certainly true.

And if the PSU is earthed but the DC side is floating, a capacitor (not always Y-rated) is often connected between the negative rail of the DC side and earth.

 

[p55xp]

Hello Hero999

(1) I have started gathering components for practical design - the transistor I'm going to use is an MJE13003 with a Ptot of 40W, a ferrite core from a salvaged CFL of 36W. But I am in great doubt about your following comment:

1) The frequency will depend on the transformer and the load connected to it. As the load on the transformer is increased, the frequency will decrease.

If frequency and load is inversely related then how the Vout regulation can be acheaved? I think frequency and voltage is directly proportional. May be I have misunderstood you - please clarify.
(2)My question was about the relation between frequency, voltage and turns.
According to you if I try with 5 turns as secondary to get an output of 12V, that means 200 turns at driver or primary have 480V. Is there really so high voltage ?! Please try to put some light on this.
Also please detail your schematic - it contains almost no component value. How should I proceed to construct?
I have really understood what Nigel has said because once I have experienced a severe shock from a detouched plug that was connected to a similar SMPSU