Noise and osci in audio amp

[Dominik Werder]

Power supply has a torodial transformer and big capacitors, there is no

I bet there is that 66 kHz there if you look closely actually. Crank up the
oscilloscope gain.

ok, I have to check this on monday, I have no own oscilloscope..

The impedance of the supply needs to be low at high frequencies too. This almost
invariably requires 'local decoupling' of the power supply with film caps
typically close to the active circuitry.

Very good to know. I tried it with 0.01uF caps, but unfortunately it
didn't changed anything.

You don't need to do that. You *do* however need experience. It's quite unlilely
to be a component and very likely to be layout or a design oversight. You should
check your 'earth' paths too ( which have finite resistance ) to make sure
you're not coupling the output back to the input.

Yes, that's exactly why I'm building this amp, for the experience and
the fun

Thanks!
Dominik

 

[Dominik Werder]

I don't know if you mentioned it and I missed it, or not- Is this amp

your design, or something you've purchased?

No, I looked at many schematics and then I modified one a bit and now
I'm trying to build it, so no warranty

Dominik

 

[Dominik Werder]

When a power amplifier is oscillating, it is running near, or at the

full output level. This causes heating in the outputs, and the
ultrasonics can burn out the speaker(s). When you add a lower frequency
signal, you are modulating the oscillation which may or may not sound
OK, but it is not the proper way to run any amplifier.

Uh good point.. Because the 66kHz signal I have on the output is not at
full output level at all, only 6V..
But on the other hand it is also there if I don't apply any input.
...that's gonna be a long debugging process.

Regards,
Dominik

 

[Dominik Werder]

Hello!

While playing around to find the problem, I doubled the capacitor
between collector and base of the BJTs that drive the output stage BJTs
and this helps a lot, sound is much better, but I have to wait until I
can look at the output with an oscilloscope.
Besides that I still have no clue where this problem originates...

Regards,
Dominik

 

[Eeyore]

ok, I have to check this on monday, I have no own oscilloscope..


Very good to know. I tried it with 0.01uF caps, but unfortunately it
didn't changed anything.

I'd recommend 100-330n plastic film caps for good measure. 10n is near useless for
power circuits at audio frequencies.

Yes, that's exactly why I'm building this amp, for the experience and
the fun

Be very aware of common current paths. This is how the some proportion of the output
can easily couple back to the input.

Are you familiar with 'star-point' grounding ? If so, use this technique to establish
a separate ground with no significant current flowing through it to the input annd
feedback side of the circuitry.

Graahm

 

[Eeyore]

Uh good point.. Because the 66kHz signal I have on the output is not at
full output level at all, only 6V..

Quite typical for this kind of oscillation.

But on the other hand it is also there if I don't apply any input.
..that's gonna be a long debugging process.

See what happens with the input shorted. You may be getting some capacitive
coupling to it.

Do you have a schematic you could post somewhere suitable ?

Graham

 

[Eeyore]

How much high voltage surface mount work do you see?

I try and avoid it actually but I can tell you I've seen plenty of *leaded* 0.1uF
100V MLC caps expire in flames when used across 33V power rails.

Graham

 

[Eeyore]

Hello!

While playing around to find the problem, I doubled the capacitor
between collector and base of the BJTs that drive the output stage BJTs
and this helps a lot, sound is much better, but I have to wait until I
can look at the output with an oscilloscope.
Besides that I still have no clue where this problem originates...

I didn't realise it was a discrete design. That indicates a problem with the
feedback stability - actually the entire transfer function of the gain block .

Whose design is this ?

Graham

 

[Michael A. Terrell]

I try and avoid it actually but I can tell you I've seen plenty of *leaded* 0.1uF
100V MLC caps expire in flames when used across 33V power rails.

It sounds like its time for you to change vendors.


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Dominik Werder]

I'd recommend 100-330n plastic film caps for good measure. 10n is near useless for

power circuits at audio frequencies.

uh ok, I admit that I didn't calculate it, so I have to buy some of
these first (I'm not a professional as you know)..

Be very aware of common current paths. This is how the some proportion of the output
can easily couple back to the input.

Are you familiar with 'star-point' grounding ? If so, use this technique to establish
a separate ground with no significant current flowing through it to the input annd
feedback side of the circuitry.

No, didn't know of star grounding yet, but I think I've found some good
material to read on the web grounding seems even more important than
I thought, but fortunately it seems that most of my wireing already
follows this star grounding philosophy..

bye!
Dominik

 

[Dominik Werder]

See what happens with the input shorted. You may be getting some capacitive

coupling to it.

Do you have a schematic you could post somewhere suitable ?

I decided to try this:
http://www.ptt.yu/korisnici/s/r/srmarkovic/poweramp.htm
http://www.ptt.yu/korisnici/s/r/srmarkovic/P400Tips.txt

I liked it because it is all built from discrete components, and I'm
doing it for fun and learning

bye!
Dominik

 

[Eeyore]

It sounds like its time for you to change vendors.

They were perfectly good ones. I recall some were AVX for example ! I forget what the
others were but 'nice blue ones' -lol. Not cheapies at all.

It caused us considerable puzzlement. It seems to be a switch-on event caused by high
dV/dt.

Graham

 

[Eeyore]

uh ok, I admit that I didn't calculate it, so I have to buy some of
these first (I'm not a professional as you know)..


No, didn't know of star grounding yet, but I think I've found some good
material to read on the web grounding seems even more important than
I thought, but fortunately it seems that most of my wireing already
follows this star grounding philosophy..

Good. That's very important. I can barely exaggerate just how important in fact.

Graham

 

[Eeyore]

I decided to try this:
http://www.ptt.yu/korisnici/s/r/srmarkovic/poweramp.htm
http://www.ptt.yu/korisnici/s/r/srmarkovic/P400Tips.txt

I liked it because it is all built from discrete components, and I'm
doing it for fun and learning

Are you using the exact same components ( esp semiconductors ? )

I *would not* use TIP41&42 for drivers myself btw ( very slow devices exactly
prone to this kind of trouble ).

I could be mildly critical of the design generally btw.

Graham

 

[Alan B]

No, didn't know of star grounding yet, but I think I've found some good
material to read on the web grounding seems even more important than
I thought, but fortunately it seems that most of my wireing already
follows this star grounding philosophy..

Most people don't know proper grounding, and I'm not excluding many of
those actually working in the industry. It's poor technical education, I
think, maybe because it's too easy to just toss the concept of "ground is
zero volts" and forget about teaching return currents.

Just remember that all the current that goes in has to go back to where it
came from, and it will go back any way it can find. If you take care to
think about each bit of the circuit's current return, you'll be doing good
design. It sounds like maybe intuitively you've already done that.

 

[Michael A. Terrell]

They were perfectly good ones. I recall some were AVX for example ! I forget what the
others were but 'nice blue ones' -lol. Not cheapies at all.

It caused us considerable puzzlement. It seems to be a switch-on event caused by high
dV/dt.

So, were you "induced" to change the design?


--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Dominik Werder]

Are you using the exact same components ( esp semiconductors ? )

Yes.

I *would not* use TIP41&42 for drivers myself btw ( very slow devices exactly
prone to this kind of trouble ).

Is there any other device that comes to your mind for that usage?

I've looked a bit around the web by myself, but.. is there any website
where transistors are listed by there characteristics or do I have to
check every datasheet to find semiconductor that fits my needs?

I could be mildly critical of the design generally btw.

I'm very interested in any critique!

btw, no news with the amp because I have so few spare time to spend on
the project the last days..

thanks for all your posting!!

Dominik

 

[Eeyore]

So, were you "induced" to change the design?

We substitued polyester film ( mylar ) types in those positions. Very rarely even a wound
mylar type would burn up too ! Haven't seen the problem with 'box polyester' types though.

Graham

 

[phaeton]

I too am interested in power transistor pairs for power amping. ;-)

-phaeton

 

[Eeyore]

I too am interested in power transistor pairs for power amping. ;-)

Here's my advice then...... !

Typically, the larger devices used as the main output transistors are relatively
slow. The driver devices are typically somewhat faster.

This is sightly less true of modern devices and especially Japanese devices where
they've had durable fast power devices since the 70s but I'll continue.....

There is typically a tradeoff between device speed and durabilty ( notably second
breakdown characterisitcs - an interesting subject in its own right ) although once
again the Japanese parts seem to obey different trules ( presumably due to smarter
device design ).

By slow I mean an fT of say 800kHz of the 2N3055 of old to the 2-4MHz of more recent
Motorola/On-Semi parts like the MJ150xx series.

By fast I mean an fT of > 10 MHz and today more like 20MHz.

Slow devices cause a lag in the transfer function which can give rise to problems
with loop stability which may exhibit itself as low or high level oscillation which
is typically above 20kHz. ( i.e inaudible ).

Such oscillation will cause possible failure due to overheating of the output devices
due to charge ( carrier ) storage issues. Note that mosfets don't suffer in the same
way.

So, the trick is to ensure that the output stage ( normally emitter follower ) is
preferably as fast as possible. But you want the ruggedness of what may be slower
devices too.

This is acheived by using fast drivers coupled to slower main output devices but
ensuring that that the drivers carry a significant proportion of the quiescent bias
current. Indeed in my own designs and also many others in pro gear especially - the
main output devices are in fact non-conducting in the quiescent condition. This also
help thermal stability !

I'm quite fond of the MJs for output devices ( esp metal can TO-3 ) and Toshiba for
drivers. ( select according to your voltage, current and dissipation needs ). Toshiba
also make some good fast devices for the main output with respectable dissipation too
- but not in metal can.

Graham