diyAudio (
-   Class D (
-   -   Line-powered SMPS for Power Amp? (

marcus66 11th February 2004 05:06 PM

Line-powered SMPS for Power Amp?
Hi all,

I'm curious to see if anyone on this forum has tried to build (from scratch) a SMPS power supply for use in a power amp, that would be powered from the line.
There are a few threads on SMPS supplies for use in cars, where it is an unavoidable solution, but I have not seen anything line-powered being discussed. Of course, such an amp built on a DIY basis would really be something out of the ordinary.

Needless to say, this would not be a beginner's project, and I am personally not about to start on something like that right now, but it's interesting to think about what components that would be needed, how to build something that would be safe to use whilst delivering high power, and so on. The goal would be a supply of, say, -60-0-60V at high current.

The U.S company QSC and the swedish LABGruppen both successfully market high power amps with switchmode power supplies, and QSC are nice enough to have schematics on their web page, also for current designs.
The QSC switchmode supply (for their PLX series) is surprisingly simple. They use a SG3525 pulsewitdh modulator to make two very rugged IGBTs drive a transformer in a push-pull configuration. The secondary side use one fast diode bridge per rail, and HF filtering is done using ferrite beads. No details on the transformer are included in the schematic.

I have a source for an iron powder toroid, Amidon T-400A-26, which is rated for 0-1MHz operation. This is 101 mm across and 33 mm high, with a 57mm hole in the middle. The cross-section is 743 mm2.
Maybe this could be used to wind a transformer? If you know that this is a case of 'forget it', then please reply and say so.
Otherwise, does anyone have a good hint on where to find info on calculating windings? What power could one hope for with the above toroid (which is quite big for a switchmode application). What frequency should one aim for? (This appears to be limited by the IGBTs).



maylar 11th February 2004 05:33 PM

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.

That said, it's done all the time in PC power supplies. They're a good place to steal high voltage electrolytics, transistors, bridge rectifiers, and even a core.

I can't comment on the core you have.

ChocoHolic 11th February 2004 07:02 PM

Hi Marcus!
Checking the limits what is allowed in this forum ;-) ? discussion about circuits operated at mains...!

You can contact me directly through mail
But I did not find the schematic which you have mentioned.
So for discussion I would need this or the direct link.
And some days patience as I am running low on time these
days. :xeye:

But please note. For such a design you do not only need the
design of the magnetic behaviour itself, but the transformer also must comply with the relevant safety standards. They typically define min required creepages, thickness of isolation, number of required isolation layers, test voltages,... fire retarding materials (UL-listing??)..... and more... !!!!

Also the equipment for working with SMPS at mains may not be
with you... differential probes, isolated current probes (?)....
And then the parasitic effects of the layout..... :hot:
If you only have the schematic, you will need a good portion of
SMPS-Design know how to get that thing running in proper way.
And in fact I would not encourage you for such experiments.

Last but not least: Do you plan to watch TV, or listen to the radio
when this SMPS is operating? ...have fun with some Voodoo-like EMI phenominas...


marcus66 11th February 2004 09:06 PM

Hi again,

I'm sorry if my post above was against the rules! I did check the forum rules and moderator's statement, and assumed the subject to be OK, but if discussing a SMPS is not allowed, I apologize!

I would not think of trying to design something like that without a plethora of safety precautions, such as an isolation transformer, for starters, and for now I just brought it up as a purely theoretical idea.

Anyway, thanks for the responses. I do not know if QSC still have the schematics online, as I downloaded them some time ago. Anyway, I did not suggest making a direct copy of their design, but rather that their schematics could be an inspiration.


Nelson Pass 11th February 2004 09:45 PM

I understand and appreciate the prohibition about circuits
on the AC line, but realistically, just about all these projects
have to deal with the AC line at some point. A DIYer can
fry himself wiring up a transformer, switch, fuse, filter, and
AC inlet.

It is also easily possible for power amps isolated from the AC
line to deliver lethal voltages, inside and at the output terminals.

I don't have an answer to this, but a blanket prohibition may not
be the best it.

Petter 11th February 2004 10:07 PM

Go to and search for SEM800, then search for 100's above and below that. You will find a lot of information. Also, application notes at are quite informative. There is also an extremely interesting/useful app note from

You can also purchase eval boards to get going quickly.


Eva 11th February 2004 10:32 PM

2 Attachment(s)
Iron powder cores are not suitable for magnetic coupling transformers due to its low permeability [mu=50 or so] that would require a huge number of turns and therefore huge leakage inductances

For your purpose, you need low permeability power ferrites [mu=2500 or so], higher permeability ferrites [mu=10.000] are more suitable for signal transformers

Low loses in the core up to high frequencies [>1Mhz] are not required except when you use high switching frequencies [>200Khz], for 30-100Khz most power ferrite materials are suitable [to get an idea, look for 3C85 and 'equivalents']

As an example, in the attached picture you can see a transformer with ferrite core that I use in one of my designs [It's something like an EE 42, In my country I can only get this kind of 'no-name' stuff at reasonable prices]

Windings are designed to operate at 31Khz [pretty slow for today standards but allows to use cheap bipolar transistors for switching and still get low losses, 4x MJE18008] and to mantain 1KW output [14,4V 72A regulated] with little air cooling [less than 8 watts losses in the copper]

Secondary consists of to layers, each with a 4-turn winding made witn 5x 1mm diameter wire and is sandwitched by two layers of primary with 24-turns made witn 1mm wire [sandwitching reduces leakage inductance to 8uH as seen from the primary side with shorted secondaries]

Insulation between primary and secondary in this prototype consists of 3 layers of common insulating tape [I can't find the usual mylar tape in my country] and wires are protected with PVC tubing when they enter and leave the bobbin

To saturate this transformer you need to apply more than +-360V square wave to the primary at 30Khz and 25șC, this is the main criteria to select its number of turns. I use to avoid calculations and do it empirically, measuring current versus volts*time/turns curves and selecting a good compromise [50 to 70% of the value needed to saturate the core at 25șC since at higher temperatures the core saturates faster]

Knowing a valid number of primary turns for your application, then you can select secondary turns for your desired turn ratio [more primary turns means more leakage inductance, ie: more time [lost] needed to reach the desired current with given voltage applied to primary and load to the secondary, about 200nS for this design]

PD: Anybody could suggest me a cool forum where switching knowledge could be freely shared? I feel this board is not the right place for serious switching discussion and power conversion has evolved a *lot* since 50-60Hz 'passive' switching

JonMarsh 12th February 2004 12:29 AM

As an applications engineer that works with SMPS all the time at power levels from 5w to 3kW, just a few points to suggest:

1. ALWAYS use an isolation setup for testing- at the least, a 120VAC:120VAC isolation transformer with fusing or circuit breaker, but if you can find something like an active inverter wtih protection functions like a Cal Instruments supply, that is even better, as you can test at varoius voltages and frequencies (Europe vs. USA, for example).

2. There are many standard references by authors like Abraham Pressman, Kit Sum, etc., on SMPS design; do a google search, take a look, and pick one that seems at your level to start. Even Borders and Barnes and Noble carry such books in their technology section.

3. As Petter suggested, Unitrode/TI references are good, and if you download their app notes, you'll have an interesting starting point.

4. Magnetics Inc. has a wide range of core products for SMPS, including Ferrite cores for transformers and MPP/Kook Mu cores for inductors. They have design calculation software also to help streamline the process. I did a 1 kW resonant reset forward convereter with a new HV SiC switch over the Xmas holidays (some holiday, huh?), and their site made it much easier to select the right parts and get the work done than most.

5. Simulation with a good simulator is a wise idea before you buy your first resistor or core. I've used most of the available ones in the business. (PSPICE, HSPICE, SABER, SPECTRE, SSPICE, IS-SPICE). One of the best performing ones is SIMetrix from Catena Software, UK; their Intro version of 4.5 will simulate a complete forward converter quite well; very stable and free from convergence issues, and MUCH, MUCH easier to use than current or past versions of PSPICE. They have reasonably good built in vertical MOSFET model, and good nonlinear magnetics models. The "paid for" versions are reasonably priced, though out of the range of casual DIY.... ;)


That said, SMPS design is not for the faint of heart, or someone who isn't really an engineer at heart, if not on paper. Just the EMC issues make people crazy....

Be safe, and have fun. If you can't get a SIM to work, don't worry about building the hardware.... :eek:

Best regards,


millwood 12th February 2004 01:15 AM


Originally posted by Eva
PD: Anybody could suggest me a cool forum where switching knowledge could be freely shared?

I would also love to know as well, tho. my primary interest is in either adapting ATX PS for audio applications or expansion around LM simple switcher chips.

subwo1 12th February 2004 01:42 AM

I have been experimenting with them for years and think it is possible to be careful. If you want to use less voltage, you can run a totem pole (half bridge) off of 160 volts. That keeps the primary voltages down to half what they would be otherwise. You just sacrifice some efficiency in certain cases, but that can be compensated for by designing-in a higher current primary side.

All times are GMT. The time now is 10:30 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2