Headphone Amp Redux

Well, they do have a certain elegance. Yin and Yang in harmony perhaps :-)...

Regards,

Chris

He likes extra parts in general... sort of the antonym of Mad Man Muntz ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

I was hoping Bitrex would answer - sort of like "write down the purpose of every component as if you were trying to sell it to ..."

Being just 2 transistors, it might have been less daunting then the whole design at once - and the answer might have convinced Bitrex that some simplifications were possible.

I tried, but he doesn't want to play :-(

I was more concerned with all the junk from the base of Q7 to the base of Q8.

And Q9... what's that all about? ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

Sorry, I have no qualifications in alternative realities :-D

I think Q7 may have been lifted from a design with a phase splitter - which he buggered up with his choice of q2/q3. I have to stop. Following this schematic is a sure path to madness (and I'm a little close already)

=A0 =A0 =A0 =A0...Jim Thompson

You need to use uniform sampling in the spice run. That is, use all four terms on the .tran line.

Sorry for the delay in my reply. Q11 and Q13 are part of a DC servo to ensure that the output sits near ground, since the amplifier is DC coupled.

You could do DC and AC in a single loop. That's the usual way ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

snipped-for-privacy@sushi.com expounded in news: snipped-for-privacy@q2g2000pre.googlegro ups.com:

Yes, and beyond that, you need a sufficiently small step time.

There's no point in interpreting FFT results beyond 250 Hz, if your step time is only (for example) 0.001 seconds (1000 sampling points per second).

Common sense should tell you that you need a minimum of four sampling points in the time domain in order to register a periodic wave (to be mapped to a frequency).

Warren

Yup, but it's hard to do without a differential pair where the inputs to the pair are sitting at 0 volts, and you're basically taking a current output into the VAS stage. To DC couple with single ended means the input is going to be sitting at some non zero potential, unless you start using Zener diodes as level shifters or something. That means the AC feedback is going to have to be capacatively coupled. I guess being able to use the combined feedback is one of the reasons that diff pairs are so popular as input stages.

I've seen some power amp designs that have the combined feedback and ALSO have a DC servo on the output. I guess some people are really paranoid about getting DC in their speakers.

Not hard at all. You can't take this...

and convert to split supplies ?:-)

Think on it for awhile. If you can't do it, ask, and I'll post it... it's a trivial set of changes. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

Sorry, I thought the "thread was dead" as there weren't any new responses for a while. Didn't we already cover some of this in an earlier post? Basically the "junk" is a modified diamond buffer, that both buffers the output stage from the input voltage gain stage and sets up the boostrapping to increase the gain of the first stage. As it stands with emitter degeneration I measured the gain of the voltage amplification stage at about 200. I think the amplifier is pretty linear on its own so not too much gain should be needed to give the negative feedback a chance to straighten out what nonlinearities remain.

Q9 is part of the output stage. R14 and the Vbe of Q9 set the quiescent current through Q8 and Q10. As the input to Q8 pulls down, current increases against R14, turning on Q9, pulling down the base of Q10 and shutting down Q10. The current from Q10 flowing through Q8 is then "replaced" by current from the load. On the positive swing Q8 shuts down, current through R14 starts to decrease and the base of Q10 swings up, sourcing current into the load and Q8. In practice the current through R14 is nearly constant. I've breadboarded the circuit and it does seem to work pretty well! Try it!

stable!

Too many parts :-(

Bitrex is now crowned the anti-Muntz ;-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

"Without a differential pair." That circuit's got one! :) Let's say that circuit is set up with spilt supplies. If we replaced the differential pair and its convenient inverting input sitting at 0 volts with a single ended stage, it seems like it would be much more difficult to unify the AC and DC feedback loops, without some sort of convoluted level shifting arrangement.

In any case, you're right in that changing the circuit to split supplies isn't too hard. Make ground Vee, and you can remove R10, C5, and C2. Connect the bottom of R3 to ground. You'd then want to connect R20 to ground, and change its value to be the same as R5 to equalize the bias current voltages generated at the diff pair inputs. Also, changing R18 to a current source would improve CMRR, and you could add some emitter degeneration to Q9 and Q10 to improve distortion. The value of C7 would have to be changed though to keep the gain crossover frequency the same, though.

Better to keep C5, adjusting its value for the change in R20. Who knows what DC might be on the input.

stable!

Something like this should really set your blood to boil, then:

It's an output stage that requires dual floating power supplies.

David Eather expounded in news:htKdnRNDrcC5ivrQnZ2dnUVZ snipped-for-privacy@giganews.com:

There used to be an annual Obfuscated C programming challenge (maybe there still is).

Do engineers here ever participate in a similar Obfuscated E. Design challenge?

Or is this only done by non e. engineers? :)

Warren

iirc, there's a circuit for a direct coupled tube amp in an old (1966 ?) edition of the Editors & Engineers Radio Handbook. fwir, it uses between

4 and 8 6080 double triode regulator tubes for the output stage and split supplies to allow direct coupling to the speaker. My copy of the book is in store at present, otherwise would scan and post so you could have a look. You might find it on the web though.

Did you try that amp circuit I posted a couple of days ago ?. Simulated fine at this end and only 6 transistors including the input differential pair...

Regards,

Chris