Go Back   Home > Forums > Amplifiers > Class D

Class D Switching Power Amplifiers and Power D/A conversion

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 12th February 2004, 04:22 AM   #11
Symx is offline Symx  United States
diyAudio Member
Join Date: Aug 2003
Default Line-powered SMPS for Power Amp?

[QUOTE] (maylar) Offline converters are not for the weak of heart. You're dealing with 170 VDC or so on the primary side, and you can't connect a 'scope anywhere because the mains are not at common negative.

This is true of course unless you use an Isolation tranformer (1:1) for testing purposes only. If you do not have one You can use Two back to back Step-down Transformers in series IE: Two 120v - 12v secondaries tied together so as to have 120v - 120v Isolated from ground.
  Reply With Quote
Old 12th February 2004, 04:33 AM   #12
AJT is offline AJT  Philippines
diyAudio Moderator
AJT's Avatar
Join Date: May 2003
Location: Palatiw, Pasig City
hi subwo1,

i make a living out of repairing pc psu's, i have done HEC's And Enermax and ANTEC pc psu's, lots og generics too...the generics i do modding by replacing parts with higher capacity ratings and the impact i would say is dramatic specially for the cheaper psu's.

i would be happy if there is a permanent thread regarding smps on this website....

i have a question for you, i dismantel some of the traffos and i find some with gaps in the center leg, while others do not have, can you differentiate between the two? i understand the gap is there to avoid core saturation, but i would like to know the effect of the gap on power delivery if any...

thanks a lot in advance...

i hope to be able to rig up oone of my own pc psu someday....i am not in a hurry as i realize i have lots to learn, repairing is easy, but designing ang building one from the ground up is another .... i have a collectiion of 3c85 type cores which i understand to be good up to 200khz...
  Reply With Quote
Old 12th February 2004, 06:15 AM   #13
subwo1 is offline subwo1  United States
diyAudio Member
subwo1's Avatar
Join Date: Jan 2002
Location: North American Continent
Hi joan2,

Those cores probably are gapped because the topology is push-pull. That approach tends to place a DC bias on the transformer which causes heating and a tendency to saturate. I have noticed the heating problem to be somewhat severe on ungapped cores I tried to use for push-pull.
  Reply With Quote
Old 12th February 2004, 07:42 AM   #14
Eva is offline Eva  Spain
diyAudio Member
Eva's Avatar
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
Ferrite cores are gapped to reduce its high permeability and allow some energy to be stored in the air and not just all the energy in the core

I think that the most easy way to understand this is to apply constant volts/turns to the core, while measuring the current waveform in the coil, and when current gets too high let the coil freewheel and return to the supply all the energy it has stored [>10Mhz dual oscilloscope is mandatory and digital storage is a great thing to analyze curves without the circuit running]

Provided the vots / turns is keep constant, the sloppe of the current / time curve is nothing but the inverse of the inductance shown by the core for that current [all non-air-cored inductors have variable inductance depending on current]

As current increases, inductance suddenly changes from the high value given by the core permeability to the low value given by the coil itself as there were no core [leakage inductance].

This is called 'saturation'. It ocurrs when the core is 'full' of energy and its permeability suddenly decreases and can be easily seen as a knee in the current waveform [past this point, the core must be switched off quickly to prevent the current to rise without control]

The curves for ungapped cores show very high [magnetizing] inductances and very low saturation currents, so its main use are magnetic couplings where you expect the transformer to act as an ideal transformer capable of directly transferring long pulses of energy between windings, storing very little energy in the core itself

Examples of magnetic couplings are push-pull magnetically-coupled buck converter topologies used in AT & ATX PC power supplies and other medium or high applications [>200W]

As you progressively increase the gap, you can see how the [magnetizing] inductance decreases and the saturation current increases

If you try to calculate total stored energy for different gap lenghts [E = .5 * L * I^2] you can see how the value increases as gap increases, but the relationship appears to be very complex [not linear nor quadratic]

This means that cores are gapped to use them as inductors, whose energy storage capability, inductances and saturation currents can be adjusted with gap lenght and number of turns

Since the ferrite is still showing much higher permeability than if there were no core, gapped cores require much less turns to achieve the same inductance as an air cored inductor, and therefore, they provide lower leakage inductance between coils, allowing them to be used as thight coupled inductors

The most ussual application for coupled inductors are flyback converters, the standard topology for low power [<100W] supplies [VCR, DVD, TV, monitors, PlayStation, battery chargers, low power or standby section of AT & ATX, etc...]

The flyback topology is very demanding on leakage inductance since the energy is stored in the core through the primary and then it's released through the secondaries, so gapped coupled-inductors designed to have good coupling between winding are mandatory to reduce the energy dumped back to the primary while the leakage inductance of the secondaries is 'energized'

PC power supplies include :

- An ungapped ferrite core from common mode EMI filter [should have low losses up to several Mhz, not present in very crappy units]
- An iron powder core from differential mode EMI filter [dual inductor, not present in very crappy units]
- A gapped ferrite core for the standby flyback converter [not present in AT]
- An ungapped ferrite core for the magnetic coupling of the main buck converter
- An iron powder core as a coupled inductor for the main buck converter [some ATX use another core for 3.3V]
- An ungapped small core from the current transformer that drives switching transistors
- A saturable reactor from some ATX that employ pulse-delay techniques to get thight regulation on 3.3V output
- Ferrite bead inductors from pi filters on the output [crappy supplies doesn't include them]

In buck converters, the energy is stored and removed from the inductor through the same winding so they are forgiving on leakage inductance. The reason to use a coupled buck inductor in PC supplies is to provide some cross-regulation between 12V and 5V outputs since only 5V output has regulation

The reason to use iron-powder cores instead of gapped cores in the buck inductor of AT & ATX supplies is that low permeability iron-powder has really very progressive saturation, so you can reliably use a core smaller than required on its non-linear inductance zone to reduce size and costs at the expense of having a ripple current that increases with the load, instead of being constant like with an ideal inductor

PC power supplies are filled of cost-cutting tricks

Millwood : Don't expect much more than 400W output from a modified 200/250W AT/ATX PS [AT ones are easier to modify since there is no standby section to deal with]
  Reply With Quote
Old 13th February 2004, 03:27 AM   #15
AJT is offline AJT  Philippines
diyAudio Moderator
AJT's Avatar
Join Date: May 2003
Location: Palatiw, Pasig City
hi EVA,
thanks for the information....i relize i am still at the bottom of the learning curve and have a long way to go, right now i am learning to dismantle choppers without breaking the core, unwinding the coils to find out their tricks, later on i will use them....

i do dual psu mods, by combining atx generic 300watter with a broken 5v standby circuit, with another 300watt atx with good 5v standby to power on both boards, one board is dedicated for mobo, while the other one is for peripherals, i can mount them together in one psu casing, results seem to be very satisfactory, a lot of cost savings can be realised out of these el-chepo psu's

i wish there could be a forum dedicated to smps for diy'ers..
  Reply With Quote
Old 13th February 2004, 06:06 AM   #16
Eva is offline Eva  Spain
diyAudio Member
Eva's Avatar
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
I've been taking a look to QSC PLX series power supply schematics

What I've seen is a simple very low cost SMPS :

- Simple half bridge topology [low efficiency], same topology as AT power supplies

- No output regulation at all

- Rudimentary peak current limiting

Most of the complexity of the circuit is dedicated to inteligent shutdown and restart in case of any fault

Anyway, this kind of simple half bridge unregulated power supplies may be useful as a first project for people interested on SMPS for audio
  Reply With Quote
Old 14th February 2004, 12:12 AM   #17
diyAudio Member
Join Date: Mar 2003
Location: Martorelles (Barcelona)
Default SMPS for Power Amp

Hi Eva,
A three years ago I listened the Link Audio system, the price was aprox 13000€ and the use the SMPS system its incluide SMPS system
power suply.
I ´dont remenbert now onther brithis company that use that use the same technology, it is not imortant.

You Say:
What I've seen is a simple very low cost SMPS :

- Simple half bridge topology [low efficiency], same topology as AT power supplies

- No output regulation at all

- Rudimentary peak current limiting

Most of the complexity of the circuit is dedicated to inteligent shutdown and restart in case of any fault.

For example I Listened In london the trivista amplifiers at the sound was excepciional and it don´t use SMPS.

The SMPS was invent for all perfomance, Ok the problem in the old system is the red ( in spanish red electrical).
Yes when you rectifier the red you have the 125Hz, this is a bad , ok
only you need a carefully with the ground loops.
In the SMPS system you can converter the line system at the High frecuency this is Ok, but the only problerm is when you have that traslate the same energy in the all the audio espectrum.

Eva.Donde trabajo en electromedecina utilizamas la SMPS solo
por lo que has explicado, pero en audio es otro campo

Eva Check.
They don´t use SMPS

Un Saludo.

Jesús Puerto
  Reply With Quote
Old 14th February 2004, 02:57 AM   #18
Account Disabled
Join Date: Apr 2003
Location: US
I thought I would post something that I just started to work on for your comments. please note that this is not meant to be correct or be actually built. It is at this stage just for discussion.

this is a 555-based smps. the 555 acts like a pulse width modulator (shamelessly stolen from National's datasheet on LM555, figure 8/9), control'd by a voltage signal on its pin 5. its output, pin 3, is attached to a mosfet Q1's gate. so Q1 will turn on and off based on pin 3. when Q1 is turned on, C2 gets charged via L1 (D1 is the free wheeling diode). R2 acts like a constant load, and variable current source, I1, acts like a variable load (in this case, at 6Khz). R3 samples the voltage across C2, and feeds it back to the 555's pin 5.

so by adjusting the wiper on R3, you get a desired voltage on C2.

R1/C1 is usual charge/discharge network for a 555 astable.

at this particular configuration, the output is rock solid, after the 50ms or so, at 7.738v. So it appears to work on paper.


1) will it work in reality?
2) what can be done to make it better topology-wise? and why?
3) how to optimize parameter values?
Attached Images
File Type: jpg 555 smps.jpg (81.4 KB, 1259 views)
  Reply With Quote
Old 15th February 2004, 05:41 AM   #19
Eva is offline Eva  Spain
diyAudio Member
Eva's Avatar
Join Date: Oct 2003
Location: Near the sea
Send a message via MSN to Eva
Congratulations, you have figured out by yourself the basic 'buck converter' topology with voltage contol [actually some kind of histeresis peak voltage control]

It may work in reality with some changes

Output filter is not realistic, you can´t use 47mH inductors with 0 ohm DC resistance nor 470uF capacitors with 0 ohms ESR [not to talk about the size and weight of a 47mH 2A inductor]

The output filter is nothing but a simple 2th order filter feed by a PWM signal that has AC and DC components. The filter should give good attenuation to AC components [whose minimum component is the switching frequency] and output the DC component 'clean'

So if you operate at 100Khz then your output filter should provide good attenuation [ie:60dB] at 100Khz and higher frequencies [you may add an aditional LC filter to smooth things further]

The output filter is placed in the 'transfer function' of the converter, so try to simulate 47mH and 470uF as a 2th filter with little or no load and you will see the huge resonance peak it causes

To be able to apply feedback in a reliable way to the whole thing, you have to design the filter to have little peaking and apply some degree of frequency compensation to the 'error amplifier' in order to prevent positive feedback [like in power amplifiers] [usually a zero near one of the output poles of the filter]

The way in wich the MOSFET is driven is also unrealistic, a consistent square wave between gate and source must be used

The control circuit is also unrealistic since it appears to work at a variable frequency and this makes it hard to filter the output, try to use a PWM whose duty cycle is controlled by an operational amplifier whose output is proportional to the difference between output voltage and desired output voltage [gain should be rolled off to avoid feedback at the switching frequency and above]

You should try to bread-board something with a SG3525 o TL494 [typical cheap control ICs], an iron powder inductor, proper mosfet drive

For that kind of converters, schottky diodes are recommended since they have negligible reverse recovery time, so you can swicth a MOSFET over the diode when it's conducting without causing very high current transients due to the diode apearing as a short-circuit for 100ns or so until it stops conducting [look at diode datasheets for reverse recovery waveforms and timing]
  Reply With Quote
Old 15th February 2004, 03:37 PM   #20
Account Disabled
Join Date: Apr 2003
Location: US
thanks eva, you raised a lot of good points and I don' thave all the answers.

the frequency of the 555 is determined by r1/c1.

Actually the circuitry has a foudamental flaw: it didn't have a trigger signal, .

it did work in my simulations but seems to suffer from low reaction time (with the trigger at 100khz). not sure why.

to have a consistant Vgs switching signal, I would have to use a p-channel mosfet. but then the phase of the 555 modulator is 180 degree out. so I wouldn't know how to make it work that way.

what I am really interested is to have a similar circuitry but works off line voltage.

I guess we aren't there yet.
  Reply With Quote


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

Similar Threads
Thread Thread Starter Forum Replies Last Post
High powered SMPS design tstitans Power Supplies 14 24th July 2009 04:32 PM
Simple 12V SMPS for small powered devices EWorkshop1708 Power Supplies 16 13th April 2008 10:46 PM
Batteries powered line preamps. beppe61 Solid State 12 6th March 2006 11:32 AM
Line array for HT with built in powered woofs elambert Multi-Way 0 27th April 2005 01:44 PM
Battery powered Balance Line Stage icceman Pass Labs 10 26th December 2003 01:35 AM

New To Site? Need Help?

All times are GMT. The time now is 03:50 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2