Frequency and gain matches are not the problem you may imagine. Naim used slugs like BD743 and BDY58 before using Sanken LAPTs to very little advantage, IMHO.Can somebody enlighten me why an amp with two 2SC5200 cannot beat an amp with a complementary pair of 2SC5200/2SA1943?.......Certainly an amp has to amplify the positive and negative signal portion symetrically, but from the datasheet of all the NPN/PNP pair, people rarely see there is a perferct beta/ft match
It is the transfer of current from two similar devices that are connected asymmetrically with different design driver circuits and consequent bias errors that is the problem. One transitor is an Emitter follower and the other is part of a CFP pair. This is not going to be neat and tidy. See Douglas Self's "Audio Power Amplifier Design Handbook" - The Output Stage chapter - any edition.
The illustrative wing-spread diagrams of Quasi output stages show a horrid kink in the transfer of current when crossing over. This can be greatly reduced or least made symmetrical by using complementary devices - ostensibly mirror image transistors for minimal disturbance at crossover. That is the reason so much time and effort over so many years went into complementary types for audio power devices.
However, as ilimzn points out, there are now techniques that get around the problem. Chip amps, with their close thermal coupling, allow some great results with quasicomp. topology but the basic Quasi design is technically a poor cousin, unless as I said, you are actually looking for distortion as Naim and some other manufacturers clearly were.
Hi ilimzn, thanks for your information. What are the totem-pole output approaches? Could you give an example of mosfet outputs with totem-pole?
Hi Ian, thank you very much. Your post is very informative and does help me to understand the issue with quasi-complementary output.
Since the basic quasi design has its inherent problems, eg. distortion incurred at cross-over, why amps using quasi-complementary output still achieve very low distortion such < 0.02% at 10KHz? Does that really matter to make the THD to lower levels? Or, is there some other drawbacks with quasi-complementary that are not acceptable by current Hi-Fi standard?
By the way, what would you comment Quad design by using same polarity output transistors and current dumping. The very same quad concept has been implemented from Quad405 to Quad909. The later has been manufactured and has a not-bad-sales in the 21st centry, I believe.
There is a paper by JOHN VANDERKOOY AND STANLEY P. LlPSHlTZ which explains in detail how the current dumping works and what advantages it has.
Jazz
Last edited:
Where you have a large crossover incongruity, you could expect the arising distortion to dominate there, whatever bias you choose. See NAD L50 and 310 models where this is forced and noticeable. Don't forget that much of Naim's distortion was also due to the forced current imbalance of the input LTP collector resistors. That won't alter with output stage bias. A complementary OPS could well reduce that .02% to <.002% THD in an otherwise conventional amplifier with say, 28 dB feedback. Here is the reference to Self's diagrams plus coverage of the Baxandall diode for improving Quasi crossover: Audio Power Amplifier Design - Douglas Self - Google Books
I couldn't fully describe how I judge amplifiers other than by listing the differences or irritations I can identify when using a test amplifier in my own system. I usually focus on stereo image quality with familiar recordings but I make lots of mistakes too. Don't even get me started on how mood, listening fatigue, predisposition to brand, differing systems, room effects, programme and unreliable memory make a complete mess of my listening tests, not that I'm a great fan of formal ABX testing.
For distortion measurement, I use a standard 8R low inductance dummy load, inbuilt tone generator a 24 bit reference tone generator of the usual 100,1k, 10kHz standards. I find 20kHz testing is more for academic interest or bragging rights. I test at 1, 5W and sometimes at 10W intervals up to brief full power testing because it can be destructive and inappropriate for amplifiers intended for music. I measure with the FFT feature of my USB Picoscope (16 bit) for simplicity or for more serious measurements, it's an ESI Juli@ with ARTA software or the recent USB Quantasylum QA400 soundcards.
This is mostly O/T, so running a Q & A on Audio electronics would be better in another thread. The best you could do for yourself though, jazzclassics, is buy a book from one of our respected authors and study it gradually and properly. Most of your technical concerns will have been treated systematically and the myths spread about many products and designs can be seen in fair comparison, without the spin or anecdotal guff spread all over the various forums and websites by their fans
for German explanation go to the first pdf file (page 4) by post #39 about
http://www.diyaudio.com/forums/solid-state/154493-your-opinion-about-schematics-4.html
an english description in detail you will find there
http://www.sg-acoustics.ch/analogue_audio/power_amplifiers/pdf/audio_power_amp_design_comments.pdf
also of interest in this case are this URLs:
http://www.diyaudio.com/forums/solid-state/72729-emitter-resistors.html
http://www.diyaudio.com/forums/solid-state/233861-driver-emitter-resistors.html
http://www.diyaudio.com/forums/solid-state/233595-power-resistors-power-amps.html
http://www.diyaudio.com/forums/soli...paralleled-place-one-output-transistor-4.html
http://www.diyaudio.com/forums/solid-state/125650-mjr-7-mosfet-amplifier-m-renardson.htmlhttp://www.diyaudio.com/forums/soli...opologies-100ma-idle-sound-closest-class.html
http://www.diyaudio.com/forums/soli...io-amplifier-books-overview-google-books.html
Last edited:
Where you have a large crossover incongruity, you could expect the arising distortion to dominate there, whatever bias you choose. See NAD L50 and 310 models where this is forced and noticeable. Don't forget that much of Naim's distortion was also due to the forced current imbalance of the input LTP collector resistors. That won't alter with output stage bias. A complementary OPS could well reduce that .02% to <.002% THD in an otherwise conventional amplifier with say, 28 dB feedback. Here is the reference to Self's diagrams plus coverage of the Baxandall diode for improving Quasi crossover: Audio Power Amplifier Design - Douglas Self - Google Books
I couldn't fully describe how I judge amplifiers other than by listing the differences or irritations I can identify when using a test amplifier in my own system. I usually focus on stereo image quality with familiar recordings but I make lots of mistakes too. Don't even get me started on how mood, listening fatigue, predisposition to brand, differing systems, room effects, programme and unreliable memory make a complete mess of my listening tests, not that I'm a great fan of formal ABX testing.
For distortion measurement, I use a standard 8R low inductance dummy load, inbuilt tone generator a 24 bit reference tone generator of the usual 100,1k, 10kHz standards. I find 20kHz testing is more for academic interest or bragging rights. I test at 1, 5W and sometimes at 10W intervals up to brief full power testing because it can be destructive and inappropriate for amplifiers intended for music. I measure with the FFT feature of my USB Picoscope (16 bit) for simplicity or for more serious measurements, it's an ESI Juli@ with ARTA software or the recent USB Quantasylum QA400 soundcards.
This is mostly O/T, so running a Q & A on Audio electronics would be better in another thread. The best you could do for yourself though, jazzclassics, is buy a book from one of our respected authors and study it gradually and properly. Most of your technical concerns will have been treated systematically and the myths spread about many products and designs can be seen in fair comparison, without the spin or anecdotal guff spread all over the various forums and websites by their fans
Like many amateur audio diyer, I rarely had any formal education in electronics and know very little about either audio engineering or circuit design. I have been reading D. Self's Audio Power Amplifier Design Handbook since 3 years ago. The book is full of information and wisdom of the author and other distinguished audio amp practitionals, which I for the time being have difficulty to comprehend. However, I remember that Self also stated in this book that: 'Fully complementary output devices have been available for many years now, and you may be wondering why it is worth examining configurations that are obsolete. The answer is that in the world of hi-fi , no circuit concept seems to ever quite die...' I would like to add that you may dislike or unappreciate the quasi-complementary design, but forcing others to accept your opinion is another thing. I raised the questions because John has made a statement without reasons to backup.
For the Naim, it could be explained that the THD is dominated by its imbalance at the input stage, so quescent current from mAs to Amps does not matter much. However not every design allow you vary the quescent current so much and maintain THD<0.02%, right? And since you have reduced THD to that level, how much benefit can you gain for your audio system by reducing it futher? Just like the THD of Naim input stage masked the THD of the output stage, the distortion of a conventional speaker system will mask the THD of a well designed amp such a Naim.
IMHO, the topic in HiFi is interesting in that it is not only technical but also philosophical. A listening system must incoporate human beings as an integral part of the system. So subjective judgement does play a role. That, I guess, is why your stated mistakes happen. Since an audio amp is designed for human ears, I think discussions on philosophy is inevitable.
By the way, I think this forum is a place where people have freedom to express or exchange as long as it does not infringe the forum rules or offend others.
Look for the QUASI amplifier threads on this forum, there are MOSFET and BJT versions described. These designs show how to make a completely symmetrical output stage using the same polarity devices in a totem-pole configuration.
For some other ideas, you can look at OTL designs using tubes.
Yet another approach is the 'Circlotron' configuration whichis not totem-pole but completely symmetrical using only one type of output device.
I would like to ask what actual resistors that I can buy can be used as 0.1 Ohm emitter resistors in the attached circuit. There are 4 output transistors per rail.
My calculation:
4x5W = 20W @ 0.22 Ohm per resistor
Current per resistor = sqrt(P/r) = sqrt(5/0.22) = 4.77A correct to 3 s.f.
Current in 0.1 Ohm resistor = 4x4.767A = 19.1A correct to 3 s.f.
Power requirement for 0.1 Ohm resistor = I^2xR = 19.07^2x0.1 = 36.4W correct to 3 s.f.
What can I use to get a 36W, 0.1 Ohm resistor?
Please refer to the attached circuit.
My calculation:
4x5W = 20W @ 0.22 Ohm per resistor
Current per resistor = sqrt(P/r) = sqrt(5/0.22) = 4.77A correct to 3 s.f.
Current in 0.1 Ohm resistor = 4x4.767A = 19.1A correct to 3 s.f.
Power requirement for 0.1 Ohm resistor = I^2xR = 19.07^2x0.1 = 36.4W correct to 3 s.f.
What can I use to get a 36W, 0.1 Ohm resistor?
Please refer to the attached circuit.
I refuse to be manipulated by your arrogant reply. Not writing why you arrived at your pseudo-conclusion is a clear indication.
You are being most arrogant by posting bad Math and then insulting those who correct your mistakes.
Nice way to get answers in a Forum
I challenge you to write what is wrong in my calculation. My calculation is based on logic, that is, four power transistors sinking current into the same 0.1 Ohm resistor. This means, I^2 grows by a factor of 16, which in turn implies, the power also grows by a factor of 16 for the same resistance value. Since 16x5W=80W, and the resistance is 0.1 Ohms about half of 0.22 Ohm, the answer should be half of 80W, id est, 40W.You are being most arrogant by posting bad Math and then insulting those who correct your mistakes.
Nice way to get answers in a Forum
Two different derivations lead to the same result. This is Physics, not electronics.
Your calculation based wrong starting data. You have 85V rails. The maximum peak output voltage is about 80V, which means 20A peak, if the load is 4ohms. As this is only half cycle, the calculated DC current is 6,4A, which means 4W dissipation on Your 0,1Ohm resistor.
Sajti
The 6.4A requires a peak current divisor of 3.125. For a sinusoidal signal peak of 20A, the root mean square is 20/sqrt(2) = 14.1A.
From where did you get the 3.125 divisor?
A 5W, 0.1 Ohm resistor is more readily available to me, rather than a giant resistor rated at 36W, and uses far less PCB space compared to a resistor array.
From where did you get the 3.125 divisor?
A 5W, 0.1 Ohm resistor is more readily available to me, rather than a giant resistor rated at 36W, and uses far less PCB space compared to a resistor array.
Let's try this from another perspective. Start with the assumed 80W out into 4 ohms. For half the cycle there will be one 0.1 ohm emitter resistor in series. For the other half, the other one will be in series. The power dissipation will scale so I see 1/2 times (0.1)/4 times 80W per resistor. That's 1W. To be safe use double that.
My calculation is based on making the four emitter resistors and the 0.1 Ohm resistor reach maximum power together. Your calculations show that the four emitter resistors are overrated at 0.22 Ohm, 5W.
So, using two 0.22 Ohm, 5W resistors in parallel will be more than sufficient.
Thanks to all for replying.
So, using two 0.22 Ohm, 5W resistors in parallel will be more than sufficient.
Thanks to all for replying.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.