SMPS - Buck converter - HOW !?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi people !

I want to design a HUGE power amp :devilr:, 4ch, 300W RMS each one, bridgeable... Buts as I need 500VA PSU for each channel, I'll need a 2000VA PSU :hot: :bigeyes:

But, there's no space enought to put a commom transformer inside the case, and even if it's outside, I think it's impratical... :whazzat:

So, I need a swithing power supply that takes 127 or 220V and convert it to + - 50V.

Look, this is how the supply should work, I think...

http://schmidt-walter.fbe.fh-darmstadt.de/smps_e/abw_hilfe_e.html#top


Any ideas?
 
Subwo & Eva, make very good points,

You should have line isolation if possible, by the time you purchase such a large transformer for isolation... why not simply derive your B+ from a standard bridge rectifier / filter?

It will be very difficult to find/build the ferrite transformers required for a good buck design in that power range.
 
Yes... Realy not easy, but I can build one supply for each channel, then it will be more easy...

I was looking PC supplys, that is of course, an offline buck converter, and the supply use 400v fast switching transistors, look:

http://www.ortodoxism.ro/datasheets/mospec/MJE13007.pdf

One ordinary PC supply can deliver about 200W, just with TWO of that transistors, ONE Mosfet (30v 0,022Ohms 40A) and a small transformer...

That's why I'm learning about switching buck converters, I think it's a real good solution, and the more important, it's REGULATED...

When we use the ordinary LARGE transformer, the voltage falls down very easily when we turn the bass and volume up, then the amplifier can't deliver the rated power...

For example, I made a stereo set for the computer of one friend, using two TDA2050, that each one deliver 30W @ 4Ohms @ 22.5v THD < 0,5%... Then I bought 18+18 4A transformer, after retification deliverier 27+27V... When the music was playing loud, the voltage falled down to 19+19... OK, but when stand by returned to 27+27 that BURNED the TDA after a while... Then I bought the 15+15 transformer that deliveried 22.5+22.5 in STAND BY mode, and when loud, 14+14v :xeye: :whazzat:

Then, I rated the amplifier 19+19 W RMS.... Because:

Amplifier efficience is

22.5v + 22.5v = 45v
(45v / 2*1.41)²/ 8 (load) = 32W RMS (100% efficience)

Ok, as the Datasheet says, the amplifier in these conditions above, can deliver 25W, so:

32 --- 100
25 --- X

X = 78,125 % of efficience.

Then

14+14v = 28v
(28v / 2*1.41)²/ R (load) - 22% (78% eficience) = 18,5W RMS...

If I used the swithing buck converter:

22.5 + 22.5 = 45v
(45v / 2*1.41)²/ 4 (load) = 64W - 22% (78% efficience) = 50W RMS :bigeyes:

That power they call "music" power, that is the power if the supply can deliver continuous voltage, without any slow down...

And that's the difference between normal PSU (LARGE transformer with common retifier) and the SMPS Buck converter...

:D


Waiting for opinions ! ;)
 
Yes... Realy not easy, but I can build one supply for each channel, then it will be more easy...

I was looking PC supplys, that is of course, an offline buck converter, and the supply use 400v fast switching transistors, look:

http://www.ortodoxism.ro/datasheets/mospec/MJE13007.pdf

One ordinary PC supply can deliver about 200W, just with TWO of that transistors, ONE Mosfet (30v 0,022Ohms 40A) and a small transformer...

That's why I'm learning about switching buck converters, I think it's a real good solution, and the more important, it's REGULATED...

When we use the ordinary LARGE transformer, the voltage falls down very easily when we turn the bass and volume up, then the amplifier can't deliver the rated power...

For example, I made a stereo set for the computer of one friend, using two TDA2050, that each one deliver 30W @ 4Ohms @ 22.5v THD < 0,5%... Then I bought 18+18 4A transformer, after retification deliverier 27+27V... When the music was playing loud, the voltage falled down to 19+19... OK, but when stand by returned to 27+27 that BURNED the TDA after a while... Then I bought the 15+15 transformer that deliveried 22.5+22.5 in STAND BY mode, and when loud, 14+14v :xeye: :whazzat:

Then, I rated the amplifier 19+19 W RMS.... Because:

Amplifier efficience is

22.5v + 22.5v = 45v
(45v / 2*1.41)²/ 8 (load) = 32W RMS (100% efficience)

Ok, as the Datasheet says, the amplifier in these conditions above, can deliver 25W, so:

32 --- 100
25 --- X

X = 78,125 % of efficience.

Then

14+14v = 28v
(28v / 2*1.41)²/ R (load) - 22% (78% eficience) = 18,5W RMS...

If I used the swithing buck converter:

22.5 + 22.5 = 45v
(45v / 2*1.41)²/ 4 (load) = 64W - 22% (78% efficience) = 50W RMS :bigeyes:

That power they call "music" power, that is the power if the supply can deliver continuous voltage, without any slow down...

And that's the difference between normal PSU (LARGE transformer with common retifier) and the SMPS Buck converter...

:D


Waiting for opinions ! ;)
 
Yes... Realy not easy, but I can build one supply for each channel, then it will be more easy...

I was looking PC supplys, that is of course, an offline buck converter, and the supply use 400v fast switching transistors, look:

http://www.ortodoxism.ro/datasheets/mospec/MJE13007.pdf

One ordinary PC supply can deliver about 200W, just with TWO of that transistors, ONE Mosfet (30v 0,022Ohms 40A) and a small transformer...

That's why I'm learning about switching buck converters, I think it's a real good solution, and the more important, it's REGULATED...

When we use the ordinary LARGE transformer, the voltage falls down very easily when we turn the bass and volume up, then the amplifier can't deliver the rated power...

For example, I made a stereo set for the computer of one friend, using two TDA2050, that each one deliver 30W @ 4Ohms @ 22.5v THD < 0,5%... Then I bought 18+18 4A transformer, after retification deliverier 27+27V... When the music was playing loud, the voltage falled down to 19+19... OK, but when stand by returned to 27+27 that BURNED the TDA after a while... Then I bought the 15+15 transformer that deliveried 22.5+22.5 in STAND BY mode, and when loud, 14+14v :xeye: :whazzat:

Then, I rated the amplifier 19+19 W RMS.... Because:

Amplifier efficience is

22.5v + 22.5v = 45v
(45v / 2*1.41)²/ 8 (load) = 32W RMS (100% efficience)

Ok, as the Datasheet says, the amplifier in these conditions above, can deliver 25W, so:

32 --- 100
25 --- X

X = 78,125 % of efficience.

Then

14+14v = 28v
(28v / 2*1.41)²/ R (load) - 22% (78% eficience) = 18,5W RMS...

If I used the swithing buck converter:

22.5 + 22.5 = 45v
(45v / 2*1.41)²/ 4 (load) = 64W - 22% (78% efficience) = 50W RMS :bigeyes:

That power they call "music" power, that is the power if the supply can deliver continuous voltage, without any slow down...

And that's the difference between normal PSU (LARGE transformer with common retifier) and the SMPS Buck converter...

:D


Waiting for opinions ! ;)
 
Yes... Realy not easy, but I can build one supply for each channel, then it will be more easy...

I was looking PC supplys, that is of course, an offline buck converter, and the supply use 400v fast switching transistors, look:

http://www.ortodoxism.ro/datasheets/mospec/MJE13007.pdf

One ordinary PC supply can deliver about 200W, just with TWO of that transistors, ONE Mosfet (30v 0,022Ohms 40A) and a small transformer...

That's why I'm learning about switching buck converters, I think it's a real good solution, and the more important, it's REGULATED...

When we use the ordinary LARGE transformer, the voltage falls down very easily when we turn the bass and volume up, then the amplifier can't deliver the rated power...

For example, I made a stereo set for the computer of one friend, using two TDA2050, that each one deliver 30W @ 4Ohms @ 22.5v THD < 0,5%... Then I bought 18+18 4A transformer, after retification deliverier 27+27V... When the music was playing loud, the voltage falled down to 19+19... OK, but when stand by returned to 27+27 that BURNED the TDA after a while... Then I bought the 15+15 transformer that deliveried 22.5+22.5 in STAND BY mode, and when loud, 14+14v :xeye: :whazzat:

Then, I rated the amplifier 19+19 W RMS.... Because:

Amplifier efficience is

22.5v + 22.5v = 45v
(45v / 2*1.41)²/ 8 (load) = 32W RMS (100% efficience)

Ok, as the Datasheet says, the amplifier in these conditions above, can deliver 25W, so:

32 --- 100
25 --- X

X = 78,125 % of efficience.

Then

14+14v = 28v
(28v / 2*1.41)²/ R (load) - 22% (78% eficience) = 18,5W RMS...

If I used the swithing buck converter:

22.5 + 22.5 = 45v
(45v / 2*1.41)²/ 4 (load) = 64W - 22% (78% efficience) = 50W RMS :bigeyes:

That power they call "music" power, that is the power if the supply can deliver continuous voltage, without any slow down...

And that's the difference between normal PSU (LARGE transformer with common retifier) and the SMPS Buck converter...

:D


Waiting for opinions ! ;)
 
OK, consider this

Use simple bridge rectifier and large caps connected directly to 120 volt line... makes 140 - 170 Volts depending on load. The ripple can be very high.

For isolation use a "FORWARD CONVERTER"... basically a buck converter with a tranformer. build 4 of the forward convertors, one for each channel.

This will bring the size down, so you can find/obtain ferrites for transformers and inductors. If you run at 50 - 100 Khz, the size will

A forward converter is in the AN19 application note.
 
Sounds like you need to build one of these:

A&T labs K6 1KW Switching power supply

It's unusual to find complete plans for very powerful SMPS's floating around. This design is simple and very robust.

The PCB artwork needs a little cleanup in a graphics program, since some traces are too thin, some too close together, some pads too small, etc. I've got a cleaned up version of the PCB graphics, if anyone is interested, since I've built this thing.

IMHO, if you've never built a switching supply, you should not start with something so big (and dangerous). You absolutely need some pertinent tools to undertake such a project, like an oscilloscope, function generator, and some powerful, isolated, variable AC and DC sources. You really need to develop a "feel" for how an SMPS works (not just theoretical -- practical ).
You need to understand what every component does in there. I mean, if it doesn't work the first time (and it won't), then what? You could lose an eye, cause damage to property, and be charged with criminal negligence, all in the process of tinkering with 1-2 KW SMPS's.

For a hobbyist, building an SMPS is often done in stages: you build one section, test it, build another, test under reduced voltage, get it working without feedback, then apply feedback, etc, etc..

And, of course, you can forget the buck converters; the only topology indicated for the powers you seek is the full bridge.

The buck is not isolated. This thing needs to be isolated.
I'm not sure what computer power supply you looked at, but they are all ISOLATED supplies, with the flyback and forward topologies being very common. And you cannot compare SMPS's providing 5V and 3V at high current, with a +/-80V amplifier supply. Design goals and approaches are different. Your voltage requirements present a different set of problems.

Andy
 
"http://schmidt-walter.fbe.fh-darmstadt.de/smps_e/smps_e.html

This reference is indeed a comprehensive topology workshop. However you need to look past the first entry, down towards the multiple-switch topologies, for the powers you need.

The author correctly qualifies the buck:
"The Buck Converter converts an input voltage to a lower output voltage. The buck converter often replaces the traditional analogue voltage regulator.

That is , it doesn't replace the power transformer; it replaces what comes after that-- the voltage regulator.

And as for forward converters, ....nando needs 2000W...

Building 4x 500W FWD converters is probably not going to be a cakewalk. The switching semiconductors are under much greater stresses than in the full-bridge topology.
The transformers won't be small ('cause the forward is not very efficient at squeezing the most power from the magnetics). You're looking at ETD49 cores maybe, or 59's.
A lot of very high voltage electrolytics.... And the need to keep them all cool, since they only have a lifetime of about 2000-3000 hours, and that lifetime is halved with every 10 degree Celsius rise in ambient temperature. So the SMPS needs to be very efficient, and not generate a lot of heat...

A 2000W PS that's 90% efficient releases 200W of heat. If it's 80% efficient, it puts out 400W of heat. For comparison, when a resistor dissipates about 2 W of heat, it becomes hot enough that you cannot touch it with your hand any more.
Think about that.

Andy
 
The full bridge would definately be the way to go... but the floating gate supplies and all that... tuning and snubbing for big bridge etc... the coupled-bucks with current-mode control would be bit easier on the brain.

I don't know if nanno is a hobbyist or an engineer... you know what I mean. Power sup this size is a tough project for a seasoned pro. You could add DC current scope probes to list of goodies too.

Seems like 8, 50 volt switchers at 250 watts ($1500) might be the least anguish. Geeze, it's easy enough to roast $500 in FETS & drivers just getting one running.
 
On that note, you could get this puppy:

http://www.allelectronics.com/cgi-bin/category.cgi?category=480&item=PS-337&type=store

It actually gives you +/- 45V at 2 A+ (I've tested it).
For five bucks, you can string 8 or 10 of these in whatever series/parallel combination makes your amplifier happy, and you're away in no time.
They're a steal, considering you cannot get useful voltages like dual 48V in surplus SMPS's these days.
And because it's Lambda, it's overengineered and underspecified.

BTW, the K6 full bridge has no floating gate supplies; all gates driven through a gate drive Xformer (turns given in schematic). The controller (and any full-bridge controller for that matter) takes care of gate timing to prevent shoot-thru. My prototype gives clean square-waves at the switching node with no snubbers whatsoever. Honest to god, I've had less headaches building this thing than I did other single-switch supplies, with all their acane spiking and ringing that blew FET after FET until you get the snubber values right.
It's an iterative process you know... yeah, right... iterate the semiconductors in the wastebasket.

Andy
 
Nando,

If your project doesn´t have problems with weigth and/or size, I strongly suggest to go to traditional unregulated 60Hz power supply.

The voltage drop showed in your measurements seems to me being a very bad transformer or problems with rectification/filtering.

Can you give us more details of that pwr supply?

Best regards,
 
I think nando said

"...But, there's no space enought to put a commom transformer inside the case, and even if it's outside, I think it's impratical... "

I you bought a 500W 60Hz transformer from anywhere, the shipping charges alone would make you cry.
I have one, and it weighs about 15 kg. And can you cat the exact voltages you need? ( and not pay retail ).

And then multiply that by 4.
 
blmn said:
Nando,

If your project doesn´t have problems with weigth and/or size, I strongly suggest to go to traditional unregulated 60Hz power supply.

The voltage drop showed in your measurements seems to me being a very bad transformer or problems with rectification/filtering.

Can you give us more details of that pwr supply?

Best regards,

Yeah... Maybe a very bad transformer, that thing smells bad, and cost pennies...

The retification is pretty good for this set... 4 caps 4,700 uF, on a center tapped transformer... Diodes are 4 A each one, complete retification (four diodes, bridge).


vectorplane said:
I think nando said

"...But, there's no space enought to put a commom transformer inside the case, and even if it's outside, I think it's impratical... "

I you bought a 500W 60Hz transformer from anywhere, the shipping charges alone would make you cry.
I have one, and it weighs about 15 kg. And can you cat the exact voltages you need? ( and not pay retail ).

And then multiply that by 4.

YES, see, 60Kg of transformer, that's INSANE auhiahauiahiaa...


blmn said:
Actually I missed that Nando´s period. :ashamed:

But I think the physical size question must be correct adressed.

See this link http://www.partsexpress.com/pdf/avelspecs.pdf for a 7kg/6.4" diamm/2.6" height 1KVA transformer.

Here in Brazil (Nando´s land) we have a manufacturer charging similiar pieces for less than Us$ 10 mail shipping costs.

Best regards

WHAT !? ??? WHERE ????

:bigeyes:

Thanks to all ! :D
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.