Switching Mode Power Supplies and Regulators

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
I've been putting some thought into power supplies, especially the high current variety. Modern tech has gone over to switchers for the most part. They are small, lightweight, and do the job.

Note: some of the following circuits I'll be discussing only pass muster because they do have a isolation transformer. The voltages and currents are NOT safe, and should be handled by experienced hobbyists.

One of the secrets for switchers is the use of torrid transformers instead of the bulky 50/60Hz units used in many wall warts. The latter are easy to use and pretty safe, but they are big chunks of iron and heavy. Torrid transformers can handle some respectable currents, but not at low frequencies.

So we need a circuit to make a high frequency medium current feed for a torrid. Not too difficult, fully rectified AC with nominal filtering will make 180VDC. Again, if this voltage were to be used directly this thread would be shut down in a heart beat, but the plan is to use it in a sealed circuit.

I am a math brick, to steal a phrase from a friend. I'm debating what the best frequency and windings for a home brewed torrid would be. A 180V P-P square wave at 1A would translate to 24V at 7.5A. A simple circuit such as this should suffice.



Regulation could also be accomplished by simply adding an optocoupler to turn the oscillator on and off with a simple comparator.

Which brings me to regulators, a different subject but related.

My first though probably will not work, though I am tempted to build it and see. This is because of one of the odd problems with MOSFETs, where they require a really substantial drive current to turn them and off quickly. The circuit to turn this on is adequate, but not so for turning it off. 10KΩ will probably not be enough.



To solve this problem I would probably use something like this design.



This is not the first thread I have done on switching power supply regulators, but I have a bit more experience under my belt. Some of this may wind up in the ebook, after experimentation. This is where this project I showed would come in handy.

Project : Electronic Load

Basically I'm putting this thread out for discussion and questions. I'm always willing to learn, though as I get older it does seem harder. I joke with my son, my Alzheimers is firmly under control. For some reason he doesn't find that funny.

Edit: People may wonder what the .zip attachment is about. I work from quite a few computers, so I'm using this first post as a stash point for my various drawings. Basically work in progress.
 

Attachments

Dodgydave

Joined Jun 22, 2012
11,285
If you need to isolate the output from the input supply,i use an opto coupler with a Tl431 regulator, classic design, otherwise you can use op amp comparator sensing. As for frequency most Atx, laptop psu's are in the 30-80Khz range,probably limited to the transformer size.

Another technique is to use a mirror winding that is used to self feed the osc chip, like this circuit using pin 7Vcc, and pin 2 Vfb sense, used alot in "Cheap " laptops and mobile phone chargers.

http://danyk.cz/uc3842_2.pdf
 
Last edited:

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
Actually, I was talking about torrids, and they look a lot different. I'm thinking of winding one to see how it works.
 

AnalogKid

Joined Aug 1, 2013
10,987
"Regulation could also be accomplished by simply adding an optocoupler to turn the oscillator on and off with a simple comparator."

The first switcher I had to maintain was like this, now called a hysteretic converter. Not super efficient, relatively poor output regulation, and relatively high ripple. But they need almost zero compensation in the control loop and are more intuitive than other topologies.

"10KΩ will probably not be enough."

Yup. The gate capacitance is on the datasheet, and you've gotta suck it out as fast as you pumped it in. Even the 7555 circuit probably won't give you the turn off time you want because its output stage p-ch pull up fet has much higher channel resistance than the n-ch pull down. If you don't want to use a commercial gate driver chip, think about a 3904/3906 pair as a driver stage.

I assume you know that Power Integrations has an entire product line of single-chip switcher parts, including some that cover your power range. Even if you want the fun of doing it the discrete wan (no sarcasm intended), their app notes and design kits are outstanding.

ak
 

ronv

Joined Nov 12, 2008
3,770
I'll throw in 2 cents worth even though I know nothing about transformers. :(
I think the main advantage is the high frequency allows use of smaller transformers and filters. So from that standpoint higher is better. I think the things that limit it is FET switching speed and core frequency losses.
I used a circuit like your second one for a little step down regulator for a gas gage in my electric golf cart. :D
You might need to add some hysteresis to slow it down and turn the FET around.
 

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
<snip>

Yup. The gate capacitance is on the datasheet, and you've gotta suck it out as fast as you pumped it in. Even the 7555 circuit probably won't give you the turn off time you want because its output stage p-ch pull up fet has much higher channel resistance than the n-ch pull down. If you don't want to use a commercial gate driver chip, think about a 3904/3906 pair as a driver stage.

I assume you know that Power Integrations has an entire product line of single-chip switcher parts, including some that cover your power range. Even if you want the fun of doing it the discrete wan (no sarcasm intended), their app notes and design kits are outstanding.

ak
Understand, I like to play around. By using simple chips I understand the mechanisms much better, and am likely to figure what is going wrong when it does (Murphy rules).

At 12V the 7555 has more than enough current capability. When you drop below 9V it becomes a problem, a major one at 3V.
 

Dodgydave

Joined Jun 22, 2012
11,285
All Switched mode PSU's use the same principles, pwm on the transformer, and monitor the output using an opto,or mirror windings, then shut down the pwm cycling so the voltage is regulated. The current sense pin usually monitors the Source pin or Emitter on the chopper transistor.
 

tcmtech

Joined Nov 4, 2013
2,867
As someone who has had to deal with the high capacity SMPS systems using in most modern welders and plasma cutter I am looking forward to your tutorial on what I consider small scale units. :)

(Unfortunately I think I have forgot more than I know now due to lack of regular use.) :oops:

BTW on most of the large multi KW and up stuff I worked on that had to have variable voltage and or variable current outputs the most common design on the primary side was a full H bridge with the windings in series with a capacitor between each half of the H bridge.

By varying the PWM duty cycle and or the switching frequency there was a wide range of stuff that could be done on the output side with surprisingly basic control circuitry for a feedback loop. :cool:

Now as far as the small roll your own designs I would highly recomend using purpose built drivers IC's and salvaged HF transformers or toroids for the primary components. Reason being salvaged components are fairly easy to reverse engineer their original circuits figure out what their base switching method and frequency are.
 

Thread Starter

Wendy

Joined Mar 24, 2008
23,415
At the moment this is basically my form of playing around.

I remember looking at the 50A and larger chargers, where they used SCRs to control the AC input to some massive power supplies.
 

AnalogKid

Joined Aug 1, 2013
10,987
I forgot to comment on best frequency. The vast majority of commercial supplies run between 50KHz and 200 KHz. It is a tradeoff between higher transformer core losses at low frequencies, and transistor switching losses becoming a larger percentage of the overall loss budget at higher frequencies. Again based on what I've seen in the market, 180W is the high end of the range for a flyback topology. A single transistor forward converter is more efficient and has lower noise, but there is all of that messy core resetting gunk.

ak
 

Evil Lurker

Joined Aug 25, 2011
116
With SMPS the higher the switching frequency the less inductance you need for your transformer. The lower the inductance the less number of turns of wire you need. The less wire you have the smaller core you can use and the less losses due to wire resistance. The other advantage is that ripple current requirements for your capacitors are reduced. However, at some point skin effect comes into play and to maintain efficiency you will need to use litz wire.
 

tcmtech

Joined Nov 4, 2013
2,867
I remember looking at the 50A and larger chargers, where they used SCRs to control the AC input to some massive power supplies.
Century and Lincoln welders used that design and variations on it a lot as a primary side control system on normal iron core transformer based models even up into the multi hundred amp ranges and to be honest it does work really well.

Being on the topic of this stuff I may have to put some time into digging though my old service files one of these days and see if I can find and schematics for various welders and plasma cutters that used HF power supplies and such.
 

cmartinez

Joined Jan 17, 2007
8,220
Century and Lincoln welders used that design and variations on it a lot as a primary side control system on normal iron core transformer based models even up into the multi hundred amp ranges and to be honest it does work really well.

Being on the topic of this stuff I may have to put some time into digging though my old service files one of these days and see if I can find and schematics for various welders and plasma cutters that used HF power supplies and such.
I'm interested... it would be real nice to have a peek at those plans...
 

Roderick Young

Joined Feb 22, 2015
408
Actually, I was talking about torrids, and they look a lot different. I'm thinking of winding one to see how it works.
One difficulty with toroids, in the likely event that you don't have a winding machine, is how you're going to wind it. I unwound a toroid intended for 230 VAC operation, and the primary had maybe 2200 turns. Even if you could slip your spool of wire through the donut hole, it would still be a real pain. Certainly, you can use less turns in your transformer, but will have to be very careful about duty cycle on the driver, to avoid excessive currents.

Here is a simple circuit that produces a regulated 5 volts at 1 amp from switched line voltage. The secondary is isolated, and so is the sense winding. The most intriguing thing about this circuit is the low parts count and extreme low cost. However, if you're writing a text, the analysis of the circuit may be too difficult for beginners. The idea of sensing directly on the output and using an optocoupler to control the primary in a bang-bang (that is, oscillator with fixed duty cycle, either on or off) fashion is a good one, easily explained. http://pididu.com/wordpress/blog/fix-a-cheap-supply-3/

I noticed also in your sense circuit that you seem to be using a diode drop as a voltage reference. I tried to do that in a regulator that I built, and found that it worked kind of well, except was not stable over temperature. I would get one voltage on cold mornings, and another on a warm day. Maybe that's ok in your application, but if not, consider whether using a low-voltage Zener is a possibility.
 

tcmtech

Joined Nov 4, 2013
2,867
I'm interested... it would be real nice to have a peek at those plans...
Put me on that list.
I will see if I can dig up the full service schematics for one of the Lincoln welders from the SP- xxx series. A few of the models had good schematics to read and were a pretty bulletproof control systems that used very common and basic parts making the ability to repurpose the regulation circuitry into a different applications on most anything that can run off a normal iron core 50/60 hz transformer dead easy.
 
Top