I think you mean discrete vs. integrated circuit...
And also for amplifier you mean that huge ,spooky metallic thing that sucks lots of current from the mains ?
From another point of view , I think ICs for their convenience,size etc. represent the nearly totality of BF amplifying devices in the world : tvs, radios, cellphones,pcs ...
BF=LF
And also for amplifier you mean that huge ,spooky metallic thing that sucks lots of current from the mains ?
From another point of view , I think ICs for their convenience,size etc. represent the nearly totality of BF amplifying devices in the world : tvs, radios, cellphones,pcs ...
BF=LF
Last edited:
Since using chip amps would be more cost effective,
the only reason that remain is that discrete designs
are simply better.
Hello
I have TDA7294 and LM3886 chip amps, they sound good but very inferior to a very good discrete amp.
Chip amps need more compensations because they oscillated a lot without it, so this heavy compensations kill the sound.
The circuit of a chip amp are all on a small silicon die, the input and Vas section are too much close to the output section, that's no good and do accentuate the oscillations.
With a discrete amp you can chose a topology who will be easyer to be stable. In the Vas (voltage amplification stage) you have a capacitor between the base and collector of the transistor to stabilise the amp, if you use to big of capacitor value there, you will have a dull sounding amp, but you can reduce a lot the size of this capacitor by using a small value phase lead capacitor between the collector of the Vas transistor and the base of the negative side of the LTP input transistors. Those thing can't be done with a chip amp, you can play arround it but not inside of a chip amp.
Bye
Gaetan
I have TDA7294 and LM3886 chip amps, they sound good but very inferior to a very good discrete amp.
Chip amps need more compensations because they oscillated a lot without it, so this heavy compensations kill the sound.
The circuit of a chip amp are all on a small silicon die, the input and Vas section are too much close to the output section, that's no good and do accentuate the oscillations.
With a discrete amp you can chose a topology who will be easyer to be stable. In the Vas (voltage amplification stage) you have a capacitor between the base and collector of the transistor to stabilise the amp, if you use to big of capacitor value there, you will have a dull sounding amp, but you can reduce a lot the size of this capacitor by using a small value phase lead capacitor between the collector of the Vas transistor and the base of the negative side of the LTP input transistors. Those thing can't be done with a chip amp, you can play arround it but not inside of a chip amp.
Bye
Gaetan
Last edited:
On an IC, the transistors are formed on a single silicon wafer, apparently. It is very difficult (if it is not impossible at all) to optimize all the transistors for where they are used in the circuit. Input stage, VAS, output stage, etc, all have different requirements. With discrete design, you can pick transistors suitable for your design.
Also, all the resistors and capacitors within the IC are formed on silicon, obviously. They pose another challenge for an IC designer. As you guess, the properties of resistors and capacitors are difficult to control on IC, while it's just a design consideration/choice in discrete design. Not to mention the difference in sonics for those components (please, no flame war).
The another issue is the thermal coupling of the all the components on the IC. You don't want your input stage, or VAS thermally coupled to the output stage, for example. But on IC, you have no choice.
Here and there as these factors accumulate, there is no question why a well designed discrete amplifier could be better than the best IC amplifier. Well, there are a lot of ill-designed discrete amplifiers out there (counting mine, too, hehehe), so be careful when you pick a circuit for building your own, though.
Nonetheless, designing/building your own amplifier is always fun and rewarding experience, I believe.
Best regards,
Satoru
Also, all the resistors and capacitors within the IC are formed on silicon, obviously. They pose another challenge for an IC designer. As you guess, the properties of resistors and capacitors are difficult to control on IC, while it's just a design consideration/choice in discrete design. Not to mention the difference in sonics for those components (please, no flame war).
The another issue is the thermal coupling of the all the components on the IC. You don't want your input stage, or VAS thermally coupled to the output stage, for example. But on IC, you have no choice.
Here and there as these factors accumulate, there is no question why a well designed discrete amplifier could be better than the best IC amplifier. Well, there are a lot of ill-designed discrete amplifiers out there (counting mine, too, hehehe), so be careful when you pick a circuit for building your own, though.
Nonetheless, designing/building your own amplifier is always fun and rewarding experience, I believe.
Best regards,
Satoru
Last edited:
You might want to look at this thread:
http://www.diyaudio.com/forums/chip-amps/163385-so-just-how-good-can-chip-amp-9.html#post2149038
http://www.diyaudio.com/forums/chip-amps/163385-so-just-how-good-can-chip-amp-9.html#post2149038
To add a little salt...
ICs have several advantages: they offer excellently matched transistors (now that do a lot of people here so it must be a good thing ;-)), transistors can be exactly optimized for their task, compensation can be made external or internal and highly optimized (i.e. done correctly). Practically no parasitic inductances/capacitances, modern processes allow extremely fast transistors (high bandwidth). Last, the chips can be produced pretty cheaply.
No reason why it should not be possible to build a high quality discrete circuit as IC.
Problem though, that commercial ICs have a power budget, means power consumption has to stay extremely low which asks for compromises. But that's not a technological limitation in principle.
Still I'm building only discretes ;-)
Have fun, Hannes
ICs have several advantages: they offer excellently matched transistors (now that do a lot of people here so it must be a good thing ;-)), transistors can be exactly optimized for their task, compensation can be made external or internal and highly optimized (i.e. done correctly). Practically no parasitic inductances/capacitances, modern processes allow extremely fast transistors (high bandwidth). Last, the chips can be produced pretty cheaply.
No reason why it should not be possible to build a high quality discrete circuit as IC.
Problem though, that commercial ICs have a power budget, means power consumption has to stay extremely low which asks for compromises. But that's not a technological limitation in principle.
Still I'm building only discretes ;-)
Have fun, Hannes
You cannot easily change the doping on IC for different transistors... You only can change the size and shape of them.
I found a lot of statements in forums here for having better matched transistors in IC. I thought the matching is usually in range of 5% and better matching is achieved by trimming here and there (please collect me if my understanding is wrong, this is a good chance for me to correct my mistakes, as usual). Of course, thermally coupling discrete transistors (if not they are a single-chip dual type) is a headache. But isn't the thermal gradient formed within IC (from output stage toward input stage) bigger headache??
I read the thread about sumaudioguy's impressive chip-amp (225dB open loop DC gain?? no way). I think this is an extreme example of how far people can push the poor little chip...
In essence, I think the general idea is that: if a chip amp is designed right, they sound good: if a discrete amp is designed/constructed poorly, they don't sound good. Chip amps are more forgiving, while discretes have more freedom in design.
Personally, I find more fun in designing/building discrete amplifiers.
Best regards and have fun!
Satoru
I found a lot of statements in forums here for having better matched transistors in IC. I thought the matching is usually in range of 5% and better matching is achieved by trimming here and there (please collect me if my understanding is wrong, this is a good chance for me to correct my mistakes, as usual). Of course, thermally coupling discrete transistors (if not they are a single-chip dual type) is a headache. But isn't the thermal gradient formed within IC (from output stage toward input stage) bigger headache??
I read the thread about sumaudioguy's impressive chip-amp (225dB open loop DC gain?? no way). I think this is an extreme example of how far people can push the poor little chip...
In essence, I think the general idea is that: if a chip amp is designed right, they sound good: if a discrete amp is designed/constructed poorly, they don't sound good. Chip amps are more forgiving, while discretes have more freedom in design.
Personally, I find more fun in designing/building discrete amplifiers.
Best regards and have fun!
Satoru
Is there a benefit in matched transistors since the output
stage is usually a quasi complementary one ?...
DC gain of 225dB
Realizing an amplifier is a simply a filter with gain then the sketch below is correct and the two filter elements exactly do simply add up. Each has been characterized individually. Do you not agree with this practice? Accepting that is the case then the numbers I give are conservative for open loop gain at DC. The sketch does not show the global feedback loop as that was the discussion.
It was said by a great audio designer "that if more gain is good then most is best." With that philosophy this IC amp was developed. Please, I did it, it works as stated, and has huge gain. Please accept this. I know it is difficult for many to believe because they are not yet of the understanding reached for amplifier design to allow them to make such a high gain amp.
Can say reading these forums has reminded me of a few things forgotten in passage of time and learn a few new ones. I hope this is a chance for others to learn also.
-Sum
Realizing an amplifier is a simply a filter with gain then the sketch below is correct and the two filter elements exactly do simply add up. Each has been characterized individually. Do you not agree with this practice? Accepting that is the case then the numbers I give are conservative for open loop gain at DC. The sketch does not show the global feedback loop as that was the discussion.
It was said by a great audio designer "that if more gain is good then most is best." With that philosophy this IC amp was developed. Please, I did it, it works as stated, and has huge gain. Please accept this. I know it is difficult for many to believe because they are not yet of the understanding reached for amplifier design to allow them to make such a high gain amp.
Can say reading these forums has reminded me of a few things forgotten in passage of time and learn a few new ones. I hope this is a chance for others to learn also.
-Sum
Attachments
Oh, I wasn't clear enough. The matching matters on the input stage. In the case of differential VAS, matched pair is a good thing, too.
Best regards,
Satoru
If two amplifier units (or filters) are cascaded, then there is no "open loop" gain for the total, as both of them are configured in their own closed loop already.
For an example of an implementation of 180dB open loop gain by multiple opamps, take a look at pp 26 of 1994 February issue of LT Journal (by Linear Technologies). The 180dB gain is not at DC but 10KHz, though.
http://www.linear.com/ltmagazine/LTMag_V04N1_Feb94.pdf
Isn't your topology like the above article??
Best regards,
Satoru
I tend to build high power amplifiers for discos and bands so tend to use big power MOSFETs.
Basic amplifier IC's just dont provide enough power for me.
No, not at all exactly.
With those filter characteristics in mind then what I say still holds true and the amp has 225dB gain at DC. There is no local DC feedback. Why do you think it needs a servo? Can you imagine trying to trim that out with a pot???? That would be a good chuckle!
I do not mean to be rude to anyone but I am also tired of people trying to figure out how I made this cheap amp into one that others say competes at the $3000 level as far as sound goes. I am not going to answer any more questions about this specific design, thank you. Everything I have posted (except for an obvious typo in the distortion matrix) is correct for the "as built" one. Will say one thing, I used parts on hand and not the optimum values. Factor of two was close enough. Think filter with gain is far more useful than thinking "amplifier."
Please, forget everything you believe you know about amplifiers and go to the lab and play!
Here is one many will not read:
No. AN9420.1
-Sum
P.S. Of course for $$$$$ then I will
Last edited:
Thanks for the hint.
To me, these stones seem to be a step into the right direction, making a good trade off between design-simplicity and -freedom, between discrete and integrated and use of space.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.