dslreports logo
 
    All Forums Hot Topics Gallery
spc
Search similar:


uniqs
14293
TheMG
Premium Member
join:2007-09-04
Canada
MikroTik RB450G
Cisco DPC3008
Cisco SPA112

TheMG

Premium Member

SMPS transformer core and winding considerations?

I want to build a SMPS to deliver regulated +/- 30VDC rails from input voltage of 11V to 15V DC. Output power need at least 150W (approx).

I'm thinking of using the classic TL494 as the SMPS controller. Unless someone has something else to suggest. This is my first experiment with SMPS technology, so I'd like to keep it simple for now.

Seems fairly straight forward, but the part I'm unsure of is the transformer. I want to do this with mostly parts that I have available. I was thinking of re-purposing ferrite cores from old defunct PC power supplies I have kicking around.

How do I determine the optimal frequency and number of turns in the primary, as well as the maximum allowable power so I do not saturate the core?
public
join:2002-01-19
Santa Clara, CA

public

Member

said by TheMG:

I want to build a SMPS to deliver regulated +/- 30VDC rails from input voltage of 11V to 15V DC. Output power need at least 150W (approx).

I was thinking of re-purposing ferrite cores from old defunct PC power supplies I have kicking around.

Cores are often epoxied, and shatter when disassembled. Also unknown material not easy to design in.

How do I determine the optimal frequency and number of turns in the primary, as well as the maximum allowable power so I do not saturate the core?

E=4B Ac Nf x 10-8

E=applied voltage (rms)
B=flux density in gauss
Ac=core area in cm2
N=number of turns
f=frequency in Hz

42625UG PQ26/25 core P material
PUSH-PULL SQUARE WAVE
100kHz 100W 40mW/cm3 loss @1000G
250kHz 200W 100mW/cm3 loss @1000G

Start at 200kHz 1000Gaus

I have pq2625 in stock.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

1 edit

lutful to TheMG

Premium Member

to TheMG
Click for full size
Typical boost converter designs use an inductor.

a) You can calculate L for a DIY toroidal inductor using many online calculators. I show typical power inductors - there are not that many turns but the wire is thicker gauge than typical transformer wire.

b) You may also have some old telecom-style DC-DC converters which have excellent planar snap-on ferrite cores. You make the winding on the PCB itself. See diagram above.

*** TI doc on DC-DC boost converter calculations:»www.ti.com/lit/an/slva06 ··· a061.pdf

You may also look at Linear Technologies, specially the amazing LT8705 high power buck-boost IC which can convert any DC voltage to any other DC voltage at very high efficiency. »www.linear.com/product/LT8705
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

Typical boost converter designs use an inductor.

a) You can calculate L for a DIY toroidal inductor using many online calculators. I show typical power inductors - there are not that many turns but the wire is thicker gauge than typical transformer wire.

Even when isolation is not required, for a 150W flyback you will find the inductor impractical.

b) You may also have some old telecom-style DC-DC converters which have excellent planar snap-on ferrite cores. You make the winding on the PCB itself. See diagram above.

That is an expensive way for a novice to learn.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

3 edits

lutful

Premium Member

Click for full size
said by public:

Even when isolation is not required, for a 150W flyback you will find the inductor impractical.

I have to use high power DC-DC boost converters to make 12V from battery bank into 48V for PoE at hundreds of watts. I also follow boost converters for electric/hybrid cars like the 10KW shown above. Usually they are not isolated.

Also, I think planar inductor design from 48V/200W class telecom power supply will be OK for 30V/5A application. Even if measured L value is off by 50% , we can adjust switching frequency to compensate.

*** TheMG See Profile I just noticed you also want -30V from +12V ... curious what is your application. How much current from +30V and how much from -30V. Will the GND need to be floating or should it be bonded to GND of 12V side?
TheMG
Premium Member
join:2007-09-04
Canada
MikroTik RB450G
Cisco DPC3008
Cisco SPA112

TheMG

Premium Member

said by lutful:

*** TheMG See Profile I just noticed you also want -30V from +12V ... curious what is your application. How much current from +30V and how much from -30V. Will the GND need to be floating or should it be bonded to GND of 12V side?

Audio. Essentially the goal is to power an existing class AB audio power amplifier (that currently operates from mains power using a 60Hz transformer) from 12VDC nominal for use in car.

I know I could just buy a car amp and be done with it, but that's not the point. This is also meant to be somewhat of a learning experience into SMPS design.

I had another thought... would it be possible to re-purpose the ferrite toroids from the output filter choke from PC power supplies to make the transformer (of course, would have to re-wind them)? If one toroid core can't handle the power, is it possible to stack multiple toroids? I think I've seen this done before.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful

Premium Member

Click for full size
Since it will be for audio, isolation from the car battery/alternator is good. Obviously you should NOT use the audio amp as a test load for your experiments. Use matching high wattage resistors for now.

I have seen quite small toroids with 3 windings like the one shown above in similar applications. The primary inductor does most of the voltage boosting so turn ratio does not need to be like transformers.

Maybe start with equal turns on all 3 windings and then lower the 12V side one turn at a time.
TheMG
Premium Member
join:2007-09-04
Canada
MikroTik RB450G
Cisco DPC3008
Cisco SPA112

TheMG

Premium Member

said by lutful:

Since it will be for audio, isolation from the car battery/alternator is good. Obviously you should NOT use the audio amp as a test load for your experiments. Use matching high wattage resistors for now.

I've got no shortage of dummy loads and high power resistors, so that's not a problem.

Yes, isolation is desirable. Voltage and load regulation is also desirable, so there will need to be a feedback loop. This will most likely involve optoisolators.
said by lutful:

I have seen quite small toroids with 3 windings like the one shown above in similar applications. The primary inductor does most of the voltage boosting so turn ratio does not need to be like transformers.

Maybe start with equal turns on all 3 windings and then lower the 12V side one turn at a time.

Time to start experimenting! Hopefully I don't pop too many transistors.

shdesigns
Powered By Infinite Improbabilty Drive
Premium Member
join:2000-12-01
Stone Mountain, GA
(Software) pfSense
ARRIS SB6121

shdesigns

Premium Member

said by TheMG:

Yes, isolation is desirable. Voltage and load regulation is also desirable, so there will need to be a feedback loop. This will most likely involve optoisolators.

You can do decent regulation by placing a diode on the primary and measuring the flyback voltage (would be negative in the 3rd diagram.)
public
join:2002-01-19
Santa Clara, CA

public to TheMG

Member

to TheMG
said by TheMG:

I had another thought... would it be possible to re-purpose the ferrite toroids from the output filter choke from PC power supplies to make the transformer (of course, would have to re-wind them)? If one toroid core can't handle the power, is it possible to stack multiple toroids? I think I've seen this done before.

Those are not ferrite, and are not suitable for transformers.
public

public to lutful

Member

to lutful
said by lutful:

I have seen quite small toroids with 3 windings like the one shown above in similar applications. The primary inductor does most of the voltage boosting so turn ratio does not need to be like transformers.

Instead of posting silly block diagrams from sales articles, try to actually design something manufacturable with all of the pesky details.
Like unpaid interns in old companies you will be amazed it is not so simple.
public

public to TheMG

Member

to TheMG
said by TheMG:

Time to start experimenting! Hopefully I don't pop too many transistors.

First design a working power train. Only then worry about the controller, feedback and isolation.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

1 edit

lutful to public

Premium Member

to public
said by public:

Instead of posting silly block diagrams from sales articles, try to actually design something manufacturable with all of the pesky details.

Way back in March 2006, you provided quite helpful feedback when I posted photo of a 48V PoE power supply we used to manufacture for our own radios (»Re: Applianced based M0n0wall ) and you may have also come across various radio boards designed by my old company Radialink in the wisp forum. I often have people designing stuff for me.

*** Those silly block diagrams are not actually from sales articles although I have no qualms about using them to illustrate an idea.
You will find them in EE texts and course materials.

The first diagram shows the inductor as an essential component, the second one shows how a transformer in this setup is really still the same boost inductor on the primary side. I made up the third composite diagram to show OP winding and diode polarity for +/-VOUT.
lutful

lutful to TheMG

Premium Member

to TheMG
said by TheMG:

there will need to be a feedback loop. This will most likely involve optoisolators.

You could also consider CMOS isolators which are simpler to use and actually provide a bit more isolation.
»www.silabs.com/products/ ··· ult.aspx

Although you mentioned TI PWM controller, any microcontroller or even the LT6994 Timerblox IC could handle the feedback(s) and drive the MOSFET. I said feedback(s) because you have to keep both -30V and +30V stable.
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

I said feedback(s) because you have to keep both -30V and +30V stable.

Unless the accuracy is critical, coupled inductors, and a single control loop are adequate for multiple outputs.
Again the power train is the real design work.
TheMG
Premium Member
join:2007-09-04
Canada

TheMG

Premium Member

Since the loading will be approximately equal on both outputs, feedback on only one output will do fine.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

3 edits

lutful

Premium Member

said by TheMG:

Since the loading will be approximately equal on both outputs, feedback on only one output will do fine.

I am seeing that audiophiles go to great lengths to make dual rail power supplies with single point ground. Probably they will buy the best 300W pure sine wave inverter and stick with the original power supply.

You could also do something like that by converting 12V DC to +/-22V AC with a center tap transformer which will give approx +/-30V DC after rectifier. It seems the exact DC voltage is not as important as very high capacitance and very low ripple.

»sound.westhost.com/project15.htm
TheMG
Premium Member
join:2007-09-04
Canada
MikroTik RB450G
Cisco DPC3008
Cisco SPA112

TheMG

Premium Member

said by lutful:

I am seeing that audiophiles go to great lengths to make dual rail power supplies with single point ground. Probably they will buy the best 300W pure sine wave inverter and stick with the original power supply.

Meh... this is by far not supposed to be an audiophile project in any way. This is primarily a learning project, as I've stayed away from and ignored SMPS technology up to this point and I thought I should try doing something with SMPS as it has so many possible applications to continue ignoring.

If it ends up working, it will go in my car to power a cheap (so cheap it's not worth stealing) subwoofer to give my factory car audio a little more "oomph".

If it wasn't for the desire to learn and experiment a little, I'd just go to the store and pick up a cheap $30 car amplifier and be done with it.

SparkChaser
Premium Member
join:2000-06-06
Downingtown, PA

SparkChaser to TheMG

Premium Member

to TheMG
I realize you way want to get in to SMPS more deeply but have you looked at the TI webench. I haven't tried since the National buy out but it worked pretty well for me on a couple of designs.

»www.ti.com/ww/en/analog/ ··· er.shtml
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

1 edit

lutful to TheMG

Premium Member

to TheMG
said by TheMG:

If it wasn't for the desire to learn and experiment a little, I'd just go to the store and pick up a cheap $30 car amplifier and be done with it.

I am inspired to do some experiments with DIY toroid transformers ... hopefully not just vicariously.

I will personally use a microcontroller for the PWM and above block diagram. The feedback will be from 3.3V to power the controller. That will make approx +/-33V on the other 2 coils if we use 10X number of turns than the 3.3V coil.

I put in LDOs for the audio supplies, but I have to look if 35V/5A capable ones exist and if they can be hooked up for negative voltage. If they don't exist, it is relatively easy to implement a DIY version.

If the toroid is smaller, switching frequency needs to be higher to get same power. This site has complete info with color codes and L value for specific number of turns: »toroids.info
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

If the toroid is smaller, switching frequency needs to be higher to get same power. This site has complete info with color codes and L value for specific number of turns: »toroids.info

Another less than brilliant idea.
RF toroid materials are not suitable for power application. When you do not have a clue, you are not helping.
Just use a buck derived push pull with coupled inductors, and pq2625 based main transformer as described in the first reply, and stop creating misinformation.
public

public to SparkChaser

Member

to SparkChaser
said by SparkChaser:

I realize you way want to get in to SMPS more deeply but have you looked at the TI webench. I haven't tried since the National buy out but it worked pretty well for me on a couple of designs.

»www.ti.com/ww/en/analog/ ··· er.shtml

Webench or ltspice do not have non company parts, although it is possible to add models. Both are still valuable tools.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

2 edits

lutful to public

Premium Member

to public
Toroid specs at a specific switching frequency does not change based on the application or personal opinion. The higher the switching frequency, the smaller the toroid. That is why we have even multi-Mhz PWM controllers today.
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

Toroid inductance at a specific switching frequency does not change based on the application or personal opinion. The higher the switching frequency, the smaller the toroid. That is why we have even multi-Mhz PWM controllers today.

Try to read application notes from one of the magnetic core vendors before you make complete fool of yourself in a public forum.
You are thus far following the steps that clueless interns in old companies take to create fiascoes. If you wish to continue, consider selling tickets to your power supply demonstration.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

1 edit

lutful

Premium Member

said by public:

said by lutful:

Toroid inductance at a specific switching frequency does not change based on the application or personal opinion. The higher the switching frequency, the smaller the toroid. That is why we have even multi-Mhz PWM controllers today.

Try to read application notes from one of the magnetic core vendors

TheMG See Profile has expressed an interest in reusing ferrite cores. He may have some powder cores, but we don't know yet. He said already this is just for learning ... it is possible to stay within max temperature/current limit for a specific ferrite core with a smart controller. This is not that different from max-power-point tracking solar controller design.

*** "gapped ferrite toroid for power inductors" has lots of tables and charts.
»www.ferroxcube.com/appl/ ··· oids.pdf
lutful

lutful to TheMG

Premium Member

to TheMG
Click for full size
*** TheMG See Profile ... just noticed that Project 89 from Elliott Sound Products happens to be an automotive 150W switchmode power supply for +/-30V audio.

Two different versions of complete schematics and lots of guidance: »sound.westhost.com/project89.htm

BTW they used ferrite toroidal core (from power inductors) to make the DIY transformer. The larger diameter reflects lower switching frequency of just 50Khz using an older design PWM IC.

Their DC-AC inverter stage could be simplified using an H-bridge IC (used for DC motors) and the switching frequency can be made much faster (at least 150Khz) to allow a smaller toroid and smaller capacitors. Anyway another option for your experiments.
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

Two different versions of complete schematics and lots of guidance: »sound.westhost.com/project89.htm

Avoid the obvious errors when recreating this.
Unregulated converter has many advantages. It is compact, low mass, very efficient and easily radiation hard. Good choice for ion thrusters.
Poor guy does not understand that without inductor it must not have dead time. Due to his errors he gets 75%.
For low step up target should be 94% efficiency.
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

lutful

Premium Member

said by public:

said by lutful:

Two different versions of complete schematics and lots of guidance: »sound.westhost.com/project89.htm

Avoid the obvious errors when recreating this ... Poor guy does not understand ... Due to his errors he gets
Because you wrote earlier that TheMG See Profile should ignore everything else and just follow recommendation in your first post, I went back and read it carefully. You provided some specs of a transformer core and then mentioned that you have it in stock. Perhaps that kind of conflict of interest is driving you to post insults again and again.

I am familiar with your posts for 7 years now while I just came across Elliott Sound Products last week. Probably their design capabilities are respected a bit more by audio enthusiasts.

»sound.westhost.com/appno ··· dex.html
»sound.westhost.com/projects.htm

When I posted Project #15 as an example of the quest towards maximum capacitance and minimum ripple among audiophiles, you had no criticism. But obviously that power supply is even less efficient ... probably because power consumption is less of a concern for them.

Anyway, I mentioned project #89 only because it matches what OP is trying to build and also happens to use a DIY transformer wrapped on toroidal core just like he wanted to try out.
public
join:2002-01-19
Santa Clara, CA

public

Member

said by lutful:

Perhaps that kind of conflict of interest is driving you to post insults again and again.

You consider pointing out your graphic demonstrations of ignorance insults.
You have obviously never designed anything manufacturable, as your posts demonstrate. Sad...
lutful
... of ideas
Premium Member
join:2005-06-16
Ottawa, ON

2 edits

lutful

Premium Member

said by public:

said by lutful:

Perhaps that kind of conflict of interest is driving you to post insults again and again.

You consider pointing out your graphic demonstrations of ignorance insults.

Let's recap some of the "ignorance" you have pointed out in this thread:

a) A boost "inductor" is not suitable for high power switch mode boost converter design ... even if isolation is not desired.

b) High frequency core materials are not suitable for high frequency power supply designs.

c) Toroids from common mode chokes can't be used to make inductors or transformers.

If you go back and read my responses, you will see that each time I tried to respond to your mistaken beliefs without making it a personal attack. Including this time.
said by public:

You have obviously never designed anything manufacturable, as your posts demonstrate. Sad...

What about Elliott Sound Products? They have obviously manufactured many different items and yet you insulted their designs.

I used to be an IC designer for many years before building WISP products for some years. No, I am not a DIY PCB designer but many professional PCBs were designed under my direct supervision - that is how I make a living since 2009.

Ironically, my most recent design was a 99% efficient 12V to 48V boost converter which uses FM-modulated RF switch and tiny air core inductors. I just hinted at that earlier because I did not want you to argue such designs won't work.

I am able to do that kind of cutting edge "block diagram" designs mainly because I am not limited by the knowledge of so called "experts" in some field.

*** some more complex PCB layouts
»Re: $100.00 CPE, If you build it will they come?
»Re: Too many cards in one box?