John Curl's Blowtorch preamplifier part II

[gerhard]

Originally Posted by Juergen Knoop

I remember when Norsk Hydro introduced their satellite dishes. Afair they had 90cm diameter and where milled out of solid aluminium blocks.
I haven't read any complaints then and I find all the moaning about prices in the blowtorch thread astonishing immature.
The blowtorch preamp is a luxury product and it costs what it costs.
Regards

That's a myth that milling an enclosure from a solid block costs an arm
and a leg. We once built RF power amplifiers for an ultrasonic phased array
and had a technician build a shielding hood from sheet metal. Took him
half of a day for the first sample and it looked like ****.

Second try was to mill it from a solid brick of alu. It was really the size of a brick.
First sample took a day and a half, the next 31 took less than an hour each
on the CNC machine, unattended, including cooling fins, ribs, threads &
whatever. The swarf was returned to the aluminum smelter. That was more
cost effective than bending metal sheets.

We also did a chassis for an ultrasonics pipeline pig that housed some 100
circuit boards with conduction cooling. This was made from machined alu
parts and was soldered in nearby France in a molten salt bath. Although
some 3 meters long, it looked like machined from a single block, the joints
were visible only if you knew where to look.

Sometimes, these mechanics guys impress me.

Gerhard

[Wavebourn]

Originally Posted by syn08

It doesn't necessary do any good. Sinking (or sourcing) only is not a consistent or guaranteed solution. Not that there's anything else that you can do, other than using the opamp as designed.

When I have to use the gear that already uses opamps that according to my preferences were not designed for audio by definition at all I can at least minimize some of it's nastiness' turning a complementary emitter follower into at least a single ended one, or better into an open collector stage loaded on a plain resistor.

Your example about 0.25V shift of a 1.4V region was not fair. The point is, bias is needed to get as smooth as possible curve close to zero. Let it be sharply bent near any rail, or even at half of a rail voltage; I don't care about 6db of headroom if I have a pristine clean transition over zero as the result.

[syn08]

Originally Posted by Wavebourn

When I have to use the gear that already uses opamps that according to my preferences were not designed for audio by definition at all I can at least minimize some of it's nastiness' turning a complementary emitter follower into at least a single ended one, or better into an open collector stage loaded on a plain resistor.

...assuming (just as an example) the output stage is symmetrically layed on the chip, and the input stage is on an isotherm (not necessary true for a low power device). Otherwise, what you got in linearity by sinking/sourcing a current you lost (or worse) in thermally induced distortions.

[anatech]

Hi Steve,
Thank you sir.


Hi Joshua, Curly,
You fellas need to chill. You are finding things to be offended at that don't really exist. The result is nothing but chatter and OT posts.

I was about to remove an entire pile of posts from you two, and others that referred to those posts. So once again, the rules are pretty simple to follow.

You can argue (intelligently) about technical details or the main idea.
You may not bring up comments referring to personalities or the relative competence of another member. Those things may get you binned.
You may not inject off topic comments or arguments.

The last few pages have been depressing reading, seeing how much trash you guys dropped in here. We all suffer for it.

Joshua, your question on minimum transient load impedance is on topic and an excellent example of what you should be posting. But then you allowed personal pride, or prejudice to lead you to comment that you wouldn't look at one of the designs that Ovidiu did. Your loss, since Ovidiu is also an excellent designer. I can't understand the attitude. Why not examine these designs in secret? No one has to know. The added personal comments reflect only on you though. You only harm your own status in those posts.

Curly Woods, you need to stop the personal attacks on other members. You are oversensitive to what you see are attacks on John Curl for one. However, the way you go about responding to perceived threats is breaking the rules of behavior here. Again, you are capable of posting reasonable content, but you seem to be looking for a fight.

Subjective vs. objective arguments will persist and both sides will never back down

Well, you have to back down. You keep bringing this up time and time again. Your definitions do not agree with what is generally accepted for one, and it's been explained to you by several people. This is now a dead, rotting and stinking horse. Leave it alone, it's off topic for one and very tiring for two. Continuation may bring you bin time. You're the only one arguing this now.


-Chris

[anatech]

Hi syn08,

assuming (just as an example) the output stage is symmetrically layed on the chip, and the input stage is on an isotherm (not necessary true for a low power device).

That's been carefully designed in monolithic chips for some time now. Thermal gradients used to cause all kinds of trouble. I think it's safe to say that any chip designed for the ability to drive signal lines or low impedance loads (> 1K) have been laid out with thermal gradients in mind.

I think I saw a statement that pointed out that power output ICs were really the only problematic devices where thermal gradients may be a problem. This due to the higher temperatures created.

I'm not a fan of connecting a resistance from the output of an op amp to either supply. If you really wanted to do this, a current source would be the best answer. Of course you now have to limit the output voltage to a point where the current source was still stable. The newer crop of audio op amps are probably best used as they were designed to use.

Remember the uA739 and uA749? Two identical chips with the exception that the uA739 had a 5 K pull-down resistor and the uA749 did not. The reason was that the distortion was lower by using a constant current sink instead of a resistor in that position. It's true! You also get better isolation from the power supply by using a CCS in that position. Aside from price, I can't see any advantage in using a resistor for that.

-Chris

Edit: John, read over from about post #2234. That's what I'm concerned about. You should be too.

[Wavebourn]

Originally Posted by syn08

...assuming (just as an example) the output stage is symmetrically layed on the chip, and the input stage is on an isotherm (not necessary true for a low power device). Otherwise, what you got in linearity by sinking/sourcing a current you lost (or worse) in thermally induced distortions.

I don't care about VLF distortions on high swing that easily go under the rug swept by feedback loop: it is exactly what opamps can do nicely.

[janneman]

Originally Posted by anatech

Hi syn08,

That's been carefully designed in monolithic chips for some time now. Thermal gradients used to cause all kinds of trouble. I think it's safe to say that any chip designed for the ability to drive signal lines or low impedance loads (> 1K) have been laid out with thermal gradients in mind.

I think I saw a statement that pointed out that power output ICs were really the only problematic devices where thermal gradients may be a problem. This due to the higher temperatures created.
[snip].

Chris,

When I look at the data sheets for those high-current DSL drivers, that can drive 100mA or more into 50 ohms or less at -100dB distortion at 100kHz or so, I'm pretty sure that those guys licked the thermal transient problem.

If you look at audio opamps (designated audio opamps by the manufacturer that is) they are often crippled opamps compared to those wideband low distortion chips. It's dangerous to let the marketing guys decide what should be an audio opamp and what not. Wide bandwidth, high slew rate, highly linear, that's great for audio. Even when they are sold as shake table drivers or what have you.

Some of the best chips for audio can be found NOT by looking for 'audio opamp' but from searching high speed wide bandwidth catalogs.

jd

[Bonsai]

"When I look at the data sheets for those high-current DSL drivers, that can drive 100mA or more into 50 ohms or less at -100dB distortion at 100kHz or so, I'm pretty sure that those guys licked the thermal transient problem."

Absolutely correct. I think Scott commented on this fact about a year or so ago when someone else mentioned this. BTW, James Solomon covered this in c. '72 or '73 in one of the Natsemi (god bless them and keep them safe from Brian Haller and his kind) app notes.

[syn08]

Originally Posted by janneman

Chris,

When I look at the data sheets for those high-current DSL drivers, that can drive 100mA or more into 50 ohms or less at -100dB distortion at 100kHz or so, I'm pretty sure that those guys licked the thermal transient problem.

If you look at audio opamps (designated audio opamps by the manufacturer that is) they are often crippled opamps compared to those wideband low distortion chips. It's dangerous to let the marketing guys decide what should be an audio opamp and what not. Wide bandwidth, high slew rate, highly linear, that's great for audio. Even when they are sold as shake table drivers or what have you.

Some of the best chips for audio can be found NOT by looking for 'audio opamp' but from searching high speed wide bandwidth catalogs.

jd

Well said, and those wideband, high current output opamps don't need any extra load to linearize

A good example of marketing in action is OPA2604. This "audio amp" is as bad as you can get, from whatever perspective you are looking at (except of allowing +/-24V supplies, which is a nice feature). Though, this opamp has a well established reputation among audiophiles as "smooth and delicate sounding" and the rest of the usual crap. And of course, loading the output with a resistor is "improving the sound".

[janneman]

Originally Posted by syn08

Well said, and those wideband, high current output opamps don't need any extra load to linearize

A good example of marketing in action is OPA2604. This "audio amp" is as bad as you can get, from whatever perspective you are looking at (except of allowing +/-24V supplies, which is a nice feature). Though, this opamp has a well established reputation among audiophiles as "smooth and delicate sounding" and the rest of the usual crap. And of course, loading the output with a resistor is "improving the sound".

Sure, and without wanting to insult anyone, but as long as you refuse to accept that hearing and perception are two different things, anybody can sell anything to you

jd