Transformer flux imbalande solutions

Yes, this is how it is done usually. Unless the input power deeps badly, the PSRR of the amplifier makes the voltage regulation unnecessary.

Those things are intended to be very simple, cheap and reliable.

The symmetry is very important, the transformer and the PCB layout are made keeping that in mind. Also, the dead time of the push-pull is just sufficient to demagnetize the core.

Few hundred Watts is no problem, however it gets more complicated at the kWt level or higher.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Symmetry is easy, since they use TL494s. Any imbalance would be due to component variation only -- give or take 10ns basically.

Even if it does saturate, who cares? So one side of the waveform starts drooping and efficiency drops slightly. The MOSFETs are bolted to the extruded aluminum case, they won't care.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

The most critical is the symmetry of the transformer and the layout of the primary side. Since the current is ~ 100A, every nH does matter.

Customer service does care.

Small drop in efficiency -> big increase in the dissipated heat. The very important parameter for the boom-boom stuff is the time to thermal shutdown at max. power. If it overheats quickly, users will immediately start complaining.

The heatsink is one of the most expensive parts, so the trend is reduce its size as possible. Also, the magnetic components can make a lot of heat which is not very easy to dissipate.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

You wish! Half of all automotive amplifiers out-and-out lie. For instance, I have a "240W" car amp here which has a 5A fuse in the front. It uses a pair of TA7250's inside. One of the most humerous I've seen claimed "250W Class A MOSFET" output; it similarly had a bipolar chipamp inside. At least I haven't seen one with buzzwords like "billet" emblazoned on it.

Maybe the high end stuff has responsible design, I don't know. I do know I've seen lots of crap. Core saturation (if the designers even consider it.. or even *know* about the phenomenon!), is likely low on their priorities.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Typically such power ratings are 200-400% distortion. Probably fine for playing rap judging the sound quality of a typical blabering system.

Among the other things, I design the audio amplifiers. My clients are the respectable companies, and they are honest and picky about the performance parameters. Although it is more of the consumer grade stuff, it sustains the rated power with the rated distortion at the rated load for the reasonable period of time. There is certainly a lot of junk on the market, especially as the mass market is driven by the cost; fortunately there are still people who know the difference.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

Now wait, You may be mistaken on that, Years ago when I used to put that crap in cars, the name plate on the unit in the dash many times indicated the systems over all power. If you didn't do your home work correctly, you wouldn't know that there was an amp that you were suppose to install as a matching unit which would then make that statement true. I remember putting in EQ's that only had 5 watts of power with a name plate that indicated watts well over 100,200 etc.. This was because a amplifier was expected to be used and since most amps are concealed, people like to brag about their power, so it was stamped on the EQ or tuner for bragging purposes.

Of course, marketing knew there were tons of people like you out there and could sell loads of these units. Also, many dealers were ignorant to this and never stocked the matching amps.

"powerampfreak"

** The vast majority of SMPSs in pro audio power amps are also just like that - ie non regulated, square wave inverters. Only they run from rectified mains voltages like 340 volts DC.

Nowadays, most use hefty IGBTS for the job.

** That chance certainly exists but two things help.

Firstly there is a significant "dead time" between current pulses in each half of the primary AND secondly the core is never operated near magnetic saturation. In fact, the more current is drawn from the secondary side, the further from saturation the core becomes.

** Yep - low voltage, square wave inverters are pretty damn simple circuits.

But it took the development of high current power mosfets and IGBTs to make them viable in consumer gear.

.... Phil

ach

tic

he

ke

But if heading for high quality realiable stuff, why not go to current mode control to eleminate the risk of flux imbalance?

If the switching speed is a little slow, it does more to balance out the flux than the Rds(on). At the end of the on time the current will be higher when you near the saturation. A soft switching will cause the drain voltage to rise sooner on the side that has the higher current. Doing this heats up the MOSFETs but the slow switching helps on the EMI issues.

Really massive semiconductors are getting to be fairly low cost items. The complexity vs brute force trade off is moving towards the brute force side unless the complexity is inside an IC.

Though most of the designs use the TL494, SG3525 or similair push-pull PWM modulator running 50/50 (and a small dead time of course) I see the chance of achieving perfectly matched on-times and switching behaviour of both primary switches, next to impossible. There are different gate-charge requirements of all parallel FET's and the gate resistors have tolerances. It all ends up that one side will switch differently and thus perfect match is not possible. I realize there are several things to consider to balancing the flux symmetrically, and that is generally what I asked in the first post. I will summarize it as follows:

1. I realize the power transformer will be the component which affects flux symmetry most, keeping both primaries tightly coupled.
2. As direct and symmetrical pcb as possible.
3. Using a deadtime of 10% for a switching period. (I mean switch at

45/45 duty cycle)

1. Use no or low inductance secondary filter inductors, to help for the fact that there are no voltage regulation (really nothing to do with flux symmetry)
2. Use FET's for the switching devices, which makes use of Rds(on) to improve flux symmetry.
3. Since each switching side will contain several parallelled devices, the current handling capabilities will be very high and withstand a degree of flux imbalance without failure.

Is anything more worth to be added to the list?

Best regards

PS. Phil, where did you go? I'm a little curious why you tell me "piss off" as soon as I start to talk current mode control? Ehh...

So, they're supposed to be advertizing the complete system, which I suppose includes about five of these units? A convienient excuse, too bad it doesn't make sense. Still sounds like advertising sprayed on the outside with no relation to what's inside.

I wonder what the 1kW amps are marketed as, then? The ones with 12-16 output transistors inside. Last one I found was actually rated for 1kW, and I can believe it would actually do that. Was it actually for a 1250W system? Or (going by the same scale factor) a 4kW system?

You assume incorrectly. I didn't buy this thing.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

:1. I realize the power transformer will be the component which affects :flux symmetry most, keeping both primaries tightly coupled. :2. As direct and symmetrical pcb as possible. :3. Using a deadtime of 10% for a switching period. (I mean switch at :45/45 duty cycle) :4. Use no or low inductance secondary filter inductors, to help for :the fact that there are no voltage regulation (really nothing to do :with flux symmetry) :5. Use FET's for the switching devices, which makes use of Rds(on) to :improve flux symmetry. :6. Since each switching side will contain several parallelled devices, :the current handling capabilities will be very high and withstand a :degree of flux imbalance without failure. : :Is anything more worth to be added to the list? : :Best regards : :PS. Phil, where did you go? I'm a little curious why you tell me :"piss off" as soon as I start to talk current mode control? Ehh... :

A little extra primary resistance (smaller than ideal wiring) can also help as long as you don't risk overheating the transformer or burning a winding.

Look at the hysterysis loop of the core material. If it is a lowish per material it will gently go into saturation. If it is a square perm (like a magamp core) it will popcorm fets. Choose lower perm or a small gap.

lp

g.

r

ke a

Mook, I guess the use of iron powder cores will be the thing then, right? Or a gapped ETD49 or 59 core may be the thing aswell?

I've never seen a powdered iron core that's anywhere near suitable for transformer action. Even with an awful lot of turns, they draw as much magnetizing current as transformer current. I don't think any get over 1000 u_r, with most in the 100 range. Power ferrites are generally in the

2000-5000 range.

I don't have any problems using high permeability toroids in my induction heater. Same deal, I'm driving a square wave into a transformer at fairly high duty cycle (I happen to be using a TL494 to generate the signals). At lowest frequency, Bmax is around 200mT.

As an aside, it's funny that Mook mentioned magamps... I was experimenting with some just the other day. Got really low gain, needs a lot more turns!

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Believe what you want, but equipment like EQ's were intended to drive amplifiers. Not many were sold with the hefty amp in it. Labeling was every thing. It was common for an EQ to have 5 watts of power. Back then most amps had low Z inputs to support existing radios and thus, making the EQ's to appear the same was a logical approach.

Most respectable products indicated this clearly on the package so to not miss lead the customer. Many others, you had to read the fine print in the manual to know this.

You can't always blame the manufacturer of a product line if you are dealing with cheap ass dealers pulling the wool over your eyes. Most dealers that delt with that crap already knew this. So, who is really the blame for having customers getting miss lead ?

I still have an old EQ that is marked 350 Watts on the face plate, it only does 5 watts out how ever, the matching amp is long gone with the car, when I sold it.

Sure tape wound coils can be used at lowish frequencies (20-40Khz) for a SMPS. some of the tapewound cores have perms in the 10,000u area if you don't gap them. As long as you don't saturate them then the have very low magentizing current. The problem is if you hit the saturation level, POW! they make a very good short immediately.

2000-5000 is the low perm cores I was talking about. Stick with the 2000u range and you'll notice the B-H loop leans over like it has a gap in it. When you approach the limits it is a softer entry into hard saturation. It gradually (relatively speaking) draws more current and allows the Rds and primary resistance time to reduce the V.sec product for that side of the switching cycle.

The problem with implementing a current mode approach is the requirement of secondary magnetics. You you probaly have is a free running pushpull with only peak rectification on the secondary. This is cheap and effective. as long as you're careful with the design I don't think flux walk will be a problem.

If your really concerned use a IR2153 and drive it with a half bridge design. The transformer is AC coupled then. ;)

ch

er

n the

on

airly

ing

low

OW!

00u

=A0It

of

ith

=A0as

-

To AC couple you need one or more capacitors in series with the primary winding. These capacitors will take the full primary current... Also, I rather use fully isolated designs, instead of boot strap circuits. But for automotive use, I see the push-pull topology well suited and well worth a try. Also, the dual primary configuration will make the primary have double battery voltage and needs less step up ratio of the power transformer.

Heh, under automotive conditions, that seems daunting, but you only need 16V caps, too. They don't really even make those...

Yeah, I can't see using anything else at this voltage. Flyback for under

20W maybe, PP for everything else. Half bridge wastes half the supply voltage, full bridge burns double conduction losses. PP gets full voltage per half and single conduction losses. RMS current in the windings is slightly higher because it's half wave per side, but there's tons of space when you only need four turns or so.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms