Switching Regulator for Audio Amplifier

[Vladimir Vassilevsky]

It is not very obvious.
In the area of interest, the amp PSRR is iversely proportional to the
frequency. The filtering depends on the affordable L and C, asumming the
tougher requrements at the higher frequencies. Plus the increased
coupling over the air. EMC is the issue, also.

The 30...40kHz is where the switching losses are about equal to the
active losses, asumming FETs, copper and magnetics are reasonably
utilized by current and voltage.

The LM5116 has a reference design for 5V @ 5A that runs at 250kHz and
over 90% efficiency at 5A.

This is a trifling low power toy with the vast underutilization of the
components.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

 

[Bob Eld]

I'd like to know why Nat Semi quote 94V (no signal) and 84V with signal. Since
signal level isn't specified, are they saying 1mV of input @ 94V will kill it ?

Someone at Nat Semi needs to be shot.

Anyway, I suggested a better option, also from National.

Graham

Probably second break down. As you know, bipolars can only take a reduced
amount of current for a reduced amount of time as the collector to emitter
voltage is increased. The highest voltage is specified at a low collector
current and high collector currents are specified at reduced collector
voltage, power dissipation not withstanding.

Second breakdown occurs in the region of both high current and high voltage
on the device for a specified amount of time. Power amplifiers, power supply
pass elements and similar applications require close attention to second
breakdown. I suspect the 94 volt number is with quiescent current only. An
unspecified signal could send the device into the second break down region
when driving a load.

It does seem like a nutty way to specify max voltage though. Max voltage
should ALWAYS include worst case max current under load, or actually under
short circuit, current limiting conditions if you want the device to
survive.

 

[Phil Allison]

"Bob Too Elderly"


** Piss off you stupid -

OVER SNIPPING WANKER !!!

 

[Phil Allison]

"Bob Elderly Fool "

Probably second break down.

** BOLLOCKS.

Much more likely to be simple Vce breakdown if one of the output devices is
ever subjected to 90 volts or more.

The LM3886 chip *includes* fast SOA limiting to prevent "second
breakdown".




...... Phil

 

[Fred]

No, this is a real circuit.
I am also assumimng you will run both left and right channel from the same supply,
needing 2x value of caps and current.


Ok, use a lower voltage then, but the issue was about the type of rectifier configuration.

I get the strange feeling you have ground floating in your diagram, this is not true now is it?
that 50V has a middle tap I hope?

There are actually a standard range of torioid transformers for amplifiers in the shops,
I dunno if 50V (or 2 x 25 what it should be) is a standard value.

My bad, They would have to be center tapped windings. The
transformer could have 2x50 center tapped windings, but I think that
is not standard. One winding would power the left channel, the other
winding the right. I have been looking and it appear standard values
are 2x45 and 2x50 in the 500VA-600VA range. In that case I have to
get a custom one wound. And if I did that I could order a 2x48ct
500VA, which would give me a larger voltage safety margin.

 

[Fred]

No, this is a real circuit.
I am also assumimng you will run both left and right channel from the same supply,
needing 2x value of caps and current.


Ok, use a lower voltage then, but the issue was about the type of rectifier configuration.

I get the strange feeling you have ground floating in your diagram, this is not true now is it?
that 50V has a middle tap I hope?

There are actually a standard range of torioid transformers for amplifiers in the shops,
I dunno if 50V (or 2 x 25 what it should be) is a standard value.

The data sheet for that TDA7294 has an interesting booster
arraignment. Did you try that? I think it would work for the device
I used too. I've seen bridged and bridged-parallel arraignments for
the chip I used, but nothing like those booster darlingtons. I wonder
if any cross-over distorion shows up.

Fred

 

[MooseFET]

I see what you're getting at. I don't recall having seen that technique used which
is interesting.

The other side of the equation is that audio power amps don't need rock solid
regulated rails, in fact some people claim they sound more 'dynamic' when the
regulation is quite loose.

Others claim if you put a line of blue marker ink on the heat sink 53%
of the way down it, you improve the dynamic clarity.

 

[neon]

There is a lot of bad information and very good advice on these replys. The rejection from power supply to output on most amp is -90db that is a lot. Look at the max voltage the amp can sustain and apply raw power is silly otherwise to regulate POWER for an amp. and a switcher as opposed to regulator for this kind of application is faulty thinking. And it is never true that the higher the switching frequency the lower the noise.A T HI FREQ. capacitors behave like inductors because how they are made. solution geta better power amplifier. lm3886 is 68w 20 to 94volts I Don't think that is going to give you problem.

 

[Eeyore]

Others claim if you put a line of blue marker ink on the heat sink 53%
of the way down it, you improve the dynamic clarity.

But not green ? That's only for CDs !

The dynamic thing does actually have some merit amongst similarly amps of continuous
power rating since the one with sloppy regulation will have higher off-load rail
voltages and will reproduce brief transients without clipping more easily. The design
principle goes back decades in fact.

Graham

 

[Jan Panteltje]

My bad, They would have to be center tapped windings. The
transformer could have 2x50 center tapped windings, but I think that
is not standard.
Yes


One winding would power the left channel, the other
winding the right.

I see no reason why not use the same supply for left and right?
After all supply rejection is good.

I have been looking and it appear standard values
are 2x45 and 2x50 in the 500VA-600VA range.

Usually, toroids for audio power come with 2 windings,
like 2 x 27 V for example.

In that case I have to
get a custom one wound. And if I did that I could order a 2x48ct
500VA, which would give me a larger voltage safety margin.

Holy golly, oh me oh my, ooops, 500VA, for 50 W out!

If indeed the chip is your problem, and you already killed a few,
then why not use a decent chip from some other manufacturer that is less critical,
or go discrete.
A am not saying National makes bad chips, I am saying others make sometimes
better ones.

Transformers are very expensive, heavy.
Can you get a TDA7294 over there?
I run it with 2 x 33V transformer, single phase rectification.
The advantage of single phase is less diodes, but also as the + is at maximum,
then the - is at minimum, keeping the total more constant.
Once killed one when I accidently shorted the mounting tab against ground,
for the rest it works at full power now for more then 8 years.
If I drive it full, then I start smelling speaker voice coil, that chip is 100% stable.
Yes it is +- 50V no signal, and they say +- 40V with signal (whatever that weird spec may mean),
so I run it at 46 volt, meet halfway ;-) and it works fine.
(That voltage will drop a bit at full signal, so maybe it is in spec anyways ).
It has overload protection, needs no RC network to be stable, it is a good chip.
That chip is less then 10 Euro here, or say about 15 $.
Then you should be able to use your existing transformer perhaps.

Anyways, just my view and experience.
Make sure you have a decent fuse too.

Going discrete is not difficult too, what I used to do before there were these chips.
And may be just as expensive, or even cheaper.

 

[Jan Panteltje]

The data sheet for that TDA7294 has an interesting booster
arraignment. Did you try that?

No, I would go discrete for higher power.

 

[Eeyore]

A am not saying National makes bad chips, I am saying others make sometimes
better ones.

There is already a higher power Nat Semi 'Overture' series power amp. I posted the link before.
http://www.national.com/mpf/LM/LM4780.html

Looks like you use it in bridge mode for 120W.

Graham

 

[Jan Panteltje]

There is already a higher power Nat Semi 'Overture' series power amp. I posted the link before.
http://www.national.com/mpf/LM/LM4780.html

Looks like you use it in bridge mode for 120W.

Graham

But it needs that nasty Boucherot RC network to be stable (example dataheet).
The TDA7294 does not.
Perhaps because it is DMOS.
And National does not dare give a diagram of what is in that chip.

I have driven weird things with that TDA7294 amp, including ultrasonics...
It always is stable.

 

[[email protected]]

The dynamic thing does actually have some merit amongst similarly amps of continuous

power rating since the one with sloppy regulation will have higher off-load rail
voltages and will reproduce brief transients without clipping more easily. The design
principle goes back decades in fact.

Anyway to intentionally accomplish a regulation that gives this effect ..?
(that makes the amplifier sound more "dynamic")

 

[Eeyore]

Eeyore wrote

But it needs that nasty Boucherot RC network to be stable (example dataheet).
The TDA7294 does not.
Perhaps because it is DMOS.
And National does not dare give a diagram of what is in that chip.

I have driven weird things with that TDA7294 amp, including ultrasonics...
It always is stable.

Unless I've confused it, STM is about to obsolete its DMOS IC amps though.

Graham

 

[Eeyore]

Anyway to intentionally accomplish a regulation that gives this effect ..?
(that makes the amplifier sound more "dynamic")

Sure. Skimpy transformer and small reservoir caps.

Graham

 

[Fred]

I see no reason why not use the same supply for left and right?
After all supply rejection is good.


Usually, toroids for audio power come with 2 windings,
like 2 x 27 V for example.


Holy golly, oh me oh my, ooops, 500VA, for 50 W out!

If indeed the chip is your problem, and you already killed a few,
then why not use a decent chip from some other manufacturer that is less critical,
or go discrete.
A am not saying National makes bad chips, I am saying others make sometimes
better ones.

Transformers are very expensive, heavy.
Can you get a TDA7294 over there?
I run it with 2 x 33V transformer, single phase rectification.
The advantage of single phase is less diodes, but also as the + is at maximum,
then the - is at minimum, keeping the total more constant.
Once killed one when I accidently shorted the mounting tab against ground,
for the rest it works at full power now for more then 8 years.
If I drive it full, then I start smelling speaker voice coil, that chip is 100% stable.
Yes it is +- 50V no signal, and they say +- 40V with signal (whatever that weird spec may mean),
so I run it at 46 volt, meet halfway ;-) and it works fine.
(That voltage will drop a bit at full signal, so maybe it is in spec anyways ).
It has overload protection, needs no RC network to be stable, it is a good chip.
That chip is less then 10 Euro here, or say about 15 $.
Then you should be able to use your existing transformer perhaps.

Anyways, just my view and experience.
Make sure you have a decent fuse too.

Going discrete is not difficult too, what I used to do before there were these chips.
And may be just as expensive, or even cheaper.

The people I discussed this with seemed to think a separate
transformer for each channel was best. Failing that, at least using a
separate winding for each channel was recommended. The idea of using
the same winding for both channels seemed to be appalling to most.

Failing the separate winding for each channel, in order to avoid
having to buy two transformers or have a custom one wound, I suppose I
could use two sets of filter caps at the value we determined and add
some additional diodes to prevent a load on one channel from drawing
down the caps on the other.

Aside from that, I have acquired a design for a discrete MOSFET based
power amp with good specs in the 150W to 200w range. Although, less
then vastly, it is significantly more elaborate then this.

I believe it would be uh,,, prudent of me before delving into that, to
at least get this current relatively simple product working to my
satisfaction.

Fred

 

[Jan Panteltje]

The people I discussed this with seemed to think a separate
transformer for each channel was best.

Are those the same people who recommend oxygenated copper for the speaker leads?

Failing that, at least using a
separate winding for each channel was recommended.

You can do that with the single phase circuit as in my amp.

The idea of using
the same winding for both channels seemed to be appalling to most.

Supply rejection of that chip is ?dB?
If you know the max voltage change a sudden load variation in one channel
can cause, (say from silence to full power 1000Hz), then apply that supply rejection,
and that is what you get as transient in the other output.
As the channels likely carry related audio material, maybe you
will not be able to hear it.

Aside from that, I have acquired a design for a discrete MOSFET based
power amp with good specs in the 150W to 200w range. Although, less
then vastly, it is significantly more elaborate then this.

There are simple and complicated amps, build something simple, but especially
something you understand, run a simulation in LTspice, play with it,
LTspice is free, and easy to use, that saves buying transistors,
and shows you in a few seconds what would otherwise takes hours of measuring
and experimenting.

 

[Jan Panteltje]

Unless I've confused it, STM is about to obsolete its DMOS IC amps though.

Graham

Well, the future..
It is listed on the ST site,
http://www.st.com
and can be orderded, pressing 'Purchase' will show what is available around the world.
http://www.st.com/stonline/products/literature/ds/1057/tda7294.htm

A google search for 'end of life DMOS stm' fund no real links.
You have shares in national????

 

[Fred]

Are those the same people who recommend oxygenated copper for the speaker leads?


You can do that with the single phase circuit as in my amp.


Supply rejection of that chip is ?dB?
If you know the max voltage change a sudden load variation in one channel
can cause, (say from silence to full power 1000Hz), then apply that supply rejection,
and that is what you get as transient in the other output.
As the channels likely carry related audio material, maybe you
will not be able to hear it.


There are simple and complicated amps, build something simple, but especially
something you understand, run a simulation in LTspice, play with it,
LTspice is free, and easy to use, that saves buying transistors,
and shows you in a few seconds what would otherwise takes hours of measuring
and experimenting.

I've been using LTSpice for years now.

Its awesome!

Fred