symmetrical SMPS

[franksterb92]

I'm trying to make a compact audio power amp with integrated smps and doing research on the smps as that's what i know the least about I'm needing a +- 28ish v 4a output and just wondered on what topology I should go with I'm thinking half bridge but not quite sure and another thing is the transformer if there is any thing out there about making or selecting a proper transformer thanks everyone

[TimNJ]

Before you get too far down the rabbit hole with a switch-mode solution, I would at least consider the traditional approach of a toroidal mains frequency transformer with a bridge rectifier and some reservoir capacitance. A 250VA toroidal transformer is about 4.5 (OD) x 2.0” (H) after a quick google search. Reservoir capacitance can take up a good chunk of space, but usually people go overkill on this so maybe it won’t take as much space as you think. Anyway, just consider it!

As a frame of reference, the latest/most efficient switch-mode topologies, like LLC, can get about 120-150W out of a 2” x 4” x 1.5” volume (no fan).

[Terry Bites]

Make sure what that ever you do, buy from a reputable source. Theres no reason not to use an smps if its properly designed and applied.There won't be many dual 24V smps off the shelf. So you'd most likely end up using two. A DIY psu will be a serious struggle to design from scratch. Loads of fab deisgns, but finding/ making the magnetics will be a major pain.

The ubiquitous 24V transformer bridge diode combo will generate about 30V you might want to get a trafo with mains adjustment taps.
One with a screen winding is deisrable.
Audiofools will sell yo all kinds of exotica that you just don't need.

[franksterb92]

really the main reason i want to do a smps is that i have never done a mains smps so id love the learning experience plus the space a linear would take

[Faringdon]

The easiest SMPS for you woudl be 250W BCM Boost PFC, followed by two flybacks, albeit with big-ish ferrite cores....one of the high number ETD or PQ or similar.
One flyback for +28V, 4A, and the other for -28v, 4A.

Power integrations may even  have a design spreadsheeet for you to do a 120W flyback.

Or else a PFC followed by Half Bridge as you say...or even a full bridge...you can easily do full bridge in current mode, which is easier.......as you suggest you coudl do a split coil secondary and have both outputs off the one tformer. You coudl use bootstrap hi side gate drivers so you  can have an easy hi side gate drive for the full bridge.

But two big Flybacks in BCM mode woudl be easy, if a little bigger than a single bridge.

[Terry Bites]

Most power supply chip companies offer free design software.  TIs WEBENCH® Power Designer. LTpowercad and others. https://www.onsemi.com/design-tools/inputs Powersim is very useful but you have to pay for it.
NXP and ST WTF. That's to be expected I guess.

You can wade through https://www.ti.com/lit/pdf/tidubk6?keyMatch=25-W%2088%20EFFICIENCY
To get the design equations

[franksterb92]

this is probably a dumb question to you guys but anyway does the current flow through the transformer so would the transformer have to be rated for say 5 amps to handle 3 amps cause from schematics im looking at and just from the little i know about this stuff i would say yea it would have to other wise it would burn up but idk since this isnt like a linear transformer this is why i want to do this ive worked a ton with linear stuff but i know so little about smps 

[jonpaul]

HypeX from Netherlands specializes in SMPS for audio PA

https://www.hypex.nl/

j

[JohanH]

Connex have a range of SMPS for audio, but doesn't seem to have any specifically with +-28V https://connexelectronic.com/product-category/power-supplies/switched-mode-power-supply/

[Faringdon]

If you download the free LTspice simulator you can run the attached very basic half bridge which provides your spec with a half bridge.

If you do choose a half bridge, then probz best to use rail split caps , and fets, rated for the full rail voltage, and use plenty of slope compensation in current mode.
Nobody has yet (publicly) defined the minimum amount of slope comp that you can use in current mode half bridge to get balanced operation of it.
Some people couple the output inductors to improve cross reg of the outputs even more.

[analityk]

If you want try something in SMPS I suggest my way. Let's start with something useful but relatively easy. I have been built transformer powered by atx 12V that's give me +-28V on secondary output for my old amplifier. It can draw 14 A from power supply so it was something.

Transformer winding is painful, but not is so painful like 300V/100uF touches by your finger, even long time after switched of power supply.

Another problems are: leakage inductance, flux walking, core saturation, overcurrent protection, output voltage controls, asymmetric load, emc and you may be not afraid about power factor correction.

[Faringdon]

Yes, i mean you could get eg an OTS 300W isolated output PSU....say with a 24V output, or 48V...then just do two flybaCKs from that to get you the 28v's.
...Or...one  boost or buck to get the +28v...Than also a cuk or flyback to get the -28v.

If you want to see detail of transformer design, then input a design of a two tran forward on power integrations site...it gives you the full transformer design bit by bit.
You can carry It over to half bridge or whatever.
I mean , you could do two HyperTFS two transistor forwards if you wanted, one for +28 and one for -28
....if you did that, power integrations (power.com) site will give you all the design detail.
Just fill in their software app.
They probably actually have say a 30V, 4A design on their app notes....just copy that if you want....and use their software aswell.

Be careful when you have big pcbs with heavy-ish heetsinks and transformers on them...you cant then put SMD stuff eg 1206 caps on the board as they will suffer bend stress and/or pop open cct.
So use daughter boards on headers for your surface mount....have them on headers.
You can even make a platform pcb on a header...then  lay a daughter pcb on that and solder it on like an smd component.

Heres a PI design...they may have others nearer your spec
https://www.power.com/design-support/design-examples/der-483-184-w-non-pfc-stage-forward-power-supply
2 like that and youre laughing.

They also do PFC's...and take you through all the design.

So a Power.com PFC followed by a power.com two transistors forward (x2)...would do it for you, and give you all the design detail needed.

[franksterb92]

In my research I've come across this video now things are so much clearer in the world of smps

https://youtu.be/9bFVTa8FEM4

[Faringdon]

Also, please use this free smps course...full of demo ltspice sims of the principles

SMPS course_Big folders
https://drive.google.com/file/d/17ij_pQLkAzzYeL_TRNci86sXHMirKHmD/view?usp=sharing
SMPS course_little folders
https://drive.google.com/file/d/1WvElhDNfHI0EBei4rouSNEdWWTO1oCtJ/view?usp=sharing

Since you are a candidate for half bridge....may i ask, and this is to your benefit, do you know the difference between flyback smps and halfbridge smps, in that in the half bridge, the primary power current creates (in ideal case) no flux in the core, since the field of it is cancelled out by the flux from the secondary current......and such is why its only the magnetising current in a half bridge that can  saturate the core.
Also, do you comprehend, that with most ferrites you shoudlnt have magnetising flux in half bridge going from +0.3Tesla to -0.3T at frequencies above about 70khz (at least)..as it will overheat  the core with core losses.

And do you thus see why a half bridge core doesnt need a gap....and that a flyback core needs a gap.

And are you aware that you shoudl keep your windings away from the gap.....so if its a centre leg gap...then pack out your bobbin so your turns are at least 3  wire diameters away from the gap...due to frIngIng flux .
heating.
And are you aware that in a flyback  (all SMPS) you need a ultra faSt output diode.
Also, are you aware of the scourge of reverse recovery current in boost PFCs in continuous mode....and that you can mitigate it by switching on slowly the fet...or using sic fet.

[franksterb92]

ok ben playing around with ti workbench and this is what i got so far. please criticize but just know im just a hobbyist not a "real Engineer"   and ive changed to a flyback not halfbridge

[franksterb92]

i do kind of know the difference between half and flyback just from looking at schematics it seems a flyback your pulsing the primary to ground via a ic and half your driving gates to do the pulsing at the high side this may be way off but im just speculating from schematics. im still learning about smps so some of the other lingo is sort of over my head at this point but thanks for pointing out those courses. ive ben watching a YouTube channel called Haseeb Electronics he has alot of the in deep videos at least for me at this point in my smps learning

[Faringdon]

Please confirm you dont need PFC?....the analysis will differ if you do.
AYK, you dont have PFC at the moment.
PFC is a regulatory thing....i think its  EN61000-3-2 (?) "Harmonics on the mains" (?)
If you are doing high volume, then your country may object to no PFC SMPS being connected up by all your customers...since it causes high apparent power draw. (lots of reactive current sloshing about in the elec grid wires, and overheating the transformers etc)

The flyback is slightly easier to design if you do use PFC....because PFC gives you a constant 390VDC input rail, no matter what is the mains voltage.

Also, you can have  feedback resistors off both rails so as to balance out the rails better. (Going off the subject of PFC now)
Ie, split the feedback current, half from each rail.
Also, please confirm your VAC input, eg 100-265VAC, or 100-130VAC etc.

Also, is the load  a class D amp?...or a Class AB which would be much higher average power?

Be sure to sandwich wind the txformer.....best probably 2 primary layers sandwiching the two secondaries....but you coudl do PSPSP for even better coupling but probably not needed.
Make each coil a full layer on the bobbin, for better coupling.
Remember skin effect so use Litz or >1 strand in each coil.
Make sure inner coil is not too near the centre leg gap....so "pack out" the bobbin if necessary, if you have a gap >0.5mm, approx....or use Litz.
You can "pack out" a bobbin just a bit with fishing line , and then tape wound over it.....it makes a nice flat "bed" for the next coil to lie on top of....and gets the inner coil away from the gap

If you use Triple insulated wire for pri or sec, then you wont need margin tape, which is fiddley.
Eg Furukawa TEX-ELZ wire.

Once you decide whether you are having PFC or not...then  , possibly  could start talking pri/sec turns ratios, and pri and sec peak and rms currents, at lowest vac input....and looking at the ripple current in the input capacitor.

[Faringdon]

Attached is your flyback, needs a bit of touching up

This is for no PFC
Mains 100-265vac

Transformer is ETD44 (TDK) with 1mm centre leg gap
AL = 262

Primary is two  sandwiching coils in pllel, each is 14 turns of 2 strands of 7/0.22 TEX-ELZ
Secondaries are each 9 turns of 3 strands of 7/0.22 TEX-ELZ
They are sandwiched by the two pri's

If you make a prototype, then it'd be best to just put pads near the bobbin, and terminate the strands to those pads...rather than fiddling with terminating the TIW to the txformer pins.

Quite a lot of dissiaption in the sense resistor with UCC38C44

Also, i gave it a few sic diodes in parallel for the secondary....since this is in CCM throughout to keep the peak current down.
Be sure and pick a FET with enough voltage for the reflected voltage and the leakage inductance spike

Attached is LTspice sim of it, and excel  calcs doc

[franksterb92]

awesome thank you for that and the prototype advice thats a really good idea

[franksterb92]

this is just a personal thing i really dont plan on selling it but i would like to add pfc at some point but i just have to keep the psu as small as possible oh and my mains from different times of playing with linear transformers is anywhere from 110 to 121 vac

[Faringdon]

for 110-121vac, i'd either just leave it, or change NP/NS to 1.059  with LP=50uH.
Gives slightly lower  peak currents.
I assume you know about using  the right hand corkscrew rule for assessing how to get the polaritys right in the flyback txformer.
When fet is on, or when diode is conducting, the flux "arrows" in the core should go the same way. (since its a flyback).

[franksterb92]

ok so should i connect earth to all grounds pre transformer i was wondering about this today while making little changes to the kicad project

[Faringdon]

Its up to  you if you ground the isolated secondary or not.
In the primary side, earth only connects to the Y caps....which connect to line, and to neutral.
C1  and c2 on page 9
https://www.power.com/sites/default/files/documents/rdr-648_150watt_power_factor_corrected_llc_power_supply_using_hiperpfs-4_and_hiperlcs.pdf

ok so should i connect earth to all grounds pre transformer i was wondering about this today while making little changes to the kicad project

...sorry to say this, but i hope you understand what bits of the primary side you shouldnt touch....all of it......there is nowhere in the primary side that you should connect earth, other than  the Y caps connection.....if you earth anywhere primary side, post rectifier, you will blow it up....and touching there can kill.

May i ask do you understand about isolation ratings of windings in the transformer, and that a certain level of isolation is needed for these windings, otherwise you could electrocute yourself when touching the secondary aswell....the furukawa triple insulated wire i told of is safe if used.  (TEX-ELZ), or indeed TEX-E

There is some safety stuff in here
https://emcfastpass.com/wp-content/uploads/2017/04/an15.pdf
...related to offline flybacks.

Also, remember to add discharge resistors into your schems and boards......Mains AC shocks are bad enough, but  HVDC shocks are worse, and when you switch off, that 180uF cap at your front end will store a lot of high voltage dc, which can give you a nasty shock......its easy to forget that when you unplug your SMPS, the caps will still be charged up.

[franksterb92]

yea i understand the isolation stuff its just little things like the ground i was curious cause the ti workbench had different symbols for the primary and secondary and again i know these are separated just for the reason you mentioned that this stuff can kill you as if the grounds are all connected to the same one a high emf event could make its way to the secondary part of the circuit so yea the y caps to earth but the rest of the pre primary side just tie to the bridge rectifier negative then... i know it seems like i don't know what I'm doing but I do understand the severity of the project and i do have a mains isolated transformer on my bench with a variac as well but i do appreciate the concern and all the tips couldn't ask for a better forum 

[Faringdon]

AYK, "Bridge negative", is alternately Line  for 8.33ms, then Neutral for 8.33ms...etc etc + etc.... (because of the way the rectifier bridge works) But yes, you tend to refer to it is "primary ground".