Hello guys,
I am confused with a question for a while, i have refered several web tutorials, but cant find exactly what i am looking for, so i thought i might ask here. coz i know many of you have golden ears.
so my question is;
what is the advantage of a class AB transistor/MOSFET based amplifier over similar powered class AB ic amplifier? I mean, suppose there is a class AB circuit based on Transistors/MOSFET which can deliver 55-60w @ .9Thd upto 40 watts, and similar circuit based upon class AB op amp chips like 3886 ( just assume that the 3886 circuit can also deliver 55-60w @ .9Thd upto 40 watts) .
My doubt arised from the observation that, despite considering the complications in transistor/MOSFET based designs, or even
the large number of passive components needed to design for such circuits, most professional av recievers ( i specify multi channel av recievers,because its more complicated and requires more components than stereo receiver) opt for transistor/MOSFET based circuits than simply using gainclone chips. Is there exist a special reason for that??
If its about power output, well most of these receivers are around70-150w / channel, and that can be achieved by parallel gainclone chips , right?? ( well this can make more sense if it was above 500 watts / channel, for adopting transistors/MOSFET designs)
Or
is it because of heat dissipation of these chips ( considering that these chips put out lot of heat than individual Transistors).
Or
is it that golden ear things? dose these sounds better than chipamps? ( wowww , how can anything be better than these lm gainclones????)
Or
is it about current/voltage handling capacity ( as Transistor circuits can be designed to accpet more volatge than pre configured IC designs ) and hence can be fed with higher volatges to get rated power with lesser current which in turn will resuce the cost and size of high current transformers .
please provide me with some comparison and reasons..
thank you.
I am confused with a question for a while, i have refered several web tutorials, but cant find exactly what i am looking for, so i thought i might ask here. coz i know many of you have golden ears.
so my question is;
what is the advantage of a class AB transistor/MOSFET based amplifier over similar powered class AB ic amplifier? I mean, suppose there is a class AB circuit based on Transistors/MOSFET which can deliver 55-60w @ .9Thd upto 40 watts, and similar circuit based upon class AB op amp chips like 3886 ( just assume that the 3886 circuit can also deliver 55-60w @ .9Thd upto 40 watts) .
My doubt arised from the observation that, despite considering the complications in transistor/MOSFET based designs, or even
the large number of passive components needed to design for such circuits, most professional av recievers ( i specify multi channel av recievers,because its more complicated and requires more components than stereo receiver) opt for transistor/MOSFET based circuits than simply using gainclone chips. Is there exist a special reason for that??
If its about power output, well most of these receivers are around70-150w / channel, and that can be achieved by parallel gainclone chips , right?? ( well this can make more sense if it was above 500 watts / channel, for adopting transistors/MOSFET designs)
Or
is it because of heat dissipation of these chips ( considering that these chips put out lot of heat than individual Transistors).
Or
is it that golden ear things? dose these sounds better than chipamps? ( wowww , how can anything be better than these lm gainclones????)
Or
is it about current/voltage handling capacity ( as Transistor circuits can be designed to accpet more volatge than pre configured IC designs ) and hence can be fed with higher volatges to get rated power with lesser current which in turn will resuce the cost and size of high current transformers .
please provide me with some comparison and reasons..
thank you.
The main tradeoff is thermal. The amount of power that's dissipated in a Class AB output stage depends strongly on the rail voltage. High power design -> high rail voltage -> high dissipated power. In an IC design that's all dissipated into one relatively small package. In a discrete design, the power is dissipated in multiple devices. That's much easier to manage.
I suggest reading the Thermal Design section of my Taming the LM3886 Pages. Play with the rail voltage. Use ±35 V for a high-power chip amp; ±45 V for a 100 W and ±65 V for a 200 W design. Watch what happens with the device temperature as function of dissipated power.
Tom
I suggest reading the Thermal Design section of my Taming the LM3886 Pages. Play with the rail voltage. Use ±35 V for a high-power chip amp; ±45 V for a 100 W and ±65 V for a 200 W design. Watch what happens with the device temperature as function of dissipated power.
Tom
One factor in favor of the IC amplifier (within its package dissipation limits) is the extremely tight thermal coupling between the output devices and the bias control circuitry, and the thermal overload protection circuitry.
This allows temperature compensation, and thermal protection to be less conservative, yet more fast and precise. FWIW.
One must also consider engineering decisions based on NIH (Not Invented Here) Syndrome.
-Peter
This allows temperature compensation, and thermal protection to be less conservative, yet more fast and precise. FWIW.
One must also consider engineering decisions based on NIH (Not Invented Here) Syndrome.
-Peter
Last edited:
so this is literally all about termal management,right? ya.. it is true. we all have seen how much amount of heatsink is need to cool down a single 3886.surely that will count on bulkiness and cost, i mean, suppose for a 120w /channel @ 7.1 will require 16 of 3886 ic and it will require a huge heatsink (may be almost double or triple times larger than those used in transistor designs) .
but any other reasons? what about sound quality ( the golden ear thing!)
but any other reasons? what about sound quality ( the golden ear thing!)
Hi Showly,
Forget golden ears! You can easily measure the advantages of a good discrete design - as long as it is built well. You also have thermal modulation problems when the front end is that close to the power sections. That's also a fact.
So what is good about chip amps? They are cheap, and probably the fastest way to realize an "okay sounding" amplifier. In over 35 years, I've had everything cross my bench, and you do learn how to listen for problems and quality. So in a way, I got spoiled by good sound. I would rather that then never know what great sound is.
So if you're learning how to build stuff, a chip amp may be the way to go, but by no means is that the end of your journey. Also, if all you'll feed the amp are MP3 files, even a crappy chip amp will sound fine - go that way so you don't waste money. But if you really want quality, that is a journey of steps. You don't get there in one, and your choice in amplifier may change as you go through life.
Don't worry about "the best" right now. Define what you listen to, and the quality of speaker you are using and define your quality from that. Besides, the "best" amplifier is a moving target for many folks.
-Chris
Forget golden ears! You can easily measure the advantages of a good discrete design - as long as it is built well. You also have thermal modulation problems when the front end is that close to the power sections. That's also a fact.
So what is good about chip amps? They are cheap, and probably the fastest way to realize an "okay sounding" amplifier. In over 35 years, I've had everything cross my bench, and you do learn how to listen for problems and quality. So in a way, I got spoiled by good sound. I would rather that then never know what great sound is.
So if you're learning how to build stuff, a chip amp may be the way to go, but by no means is that the end of your journey. Also, if all you'll feed the amp are MP3 files, even a crappy chip amp will sound fine - go that way so you don't waste money. But if you really want quality, that is a journey of steps. You don't get there in one, and your choice in amplifier may change as you go through life.
Don't worry about "the best" right now. Define what you listen to, and the quality of speaker you are using and define your quality from that. Besides, the "best" amplifier is a moving target for many folks.
-Chris
Thermal modulation... These factors need to be considered in light of the listening environment.
In an intense monitoring environment, really pushing the amps a bit, sure.
In a domestic listening environment, not so sure. Maybe I'm spoiled by 95dB/W/m speakers in a domestic environment.
Horses for courses.
Cheers
In an intense monitoring environment, really pushing the amps a bit, sure.
In a domestic listening environment, not so sure. Maybe I'm spoiled by 95dB/W/m speakers in a domestic environment.
Horses for courses.
Cheers
Last edited:
ya.. i have built many.. from LA4440, TDA2003,2005,2030,2050,7560 ,LM1875, few stks, and some 3055 ,2sc3858/2sa1494 , and now some class D chips. from all of these i love 1857 most. its seems bright.. but the sad fact is, i always use some kind of eq (hardware/software) while listening to music, ( for now i always use viper in my S7 and asus laptop) . so may be i am not hearing something ( btw.. phips says that i have passed few levels of golden ear challenge, lol) . I am into multichanel audio now.. so alwsys built 5.1 systems now, and my preference is 1857 always, with 2*2050 for sub. but now i want to built something more powerful, i have those golden 3886 chips,30+ 1875, some 7294 , and about 5 MOSFET boards (32-0-32v @ 100w/channel). For now i am in confusion.. should i go for MOSFET or the IC's. btw.. i never listen music at ear shaking volumes . (though i have built few huge 3way boxes which can handle 200w rms /channel, along with 500w/12 inch sub . powerd by 150w/channel MOSFET amps for channel's and 300W MOSFET board for sub. it can rattle my window panes even at 20-30% volumes. And my mom hates that.. she even threatens me to burn my speakers. So insted of sitting in between ashes, i decided to keep it down now a days.. using my yamaha NSP 40 , which gives 30*5 + 70*1 5.1 setup, ( ya , still i keep it below a total of 100 w or something.
so i dont want to drive louder, l like to drive crisp. 1875 keep it nice, but it seems lil less . so i was thinking about building something with 50+ watts /channel. and about 100 for sub. options are pretty straightforward ... 3886 /7294 or the MOSFET again. but what really concerns me is.. MOSFETs are Monsters, they have burned some of my pretty speakers in the past, ( Few 2*8ohms and Few 4ohms). Now my yamaha is 6Ohms@30w rms per channel, and the sub is powered. i dont want to loose them. So what you guys suggest, MOSFETS or Lm? And i will keep it down , about 20-30% to protect speakers.
btw,i listen to 320kbps mostly, with some 256/128/high resolution 24bit ( its so hard to get high resolution tracks in indian music, if there is some one who knows. pls tell me ) . i can differentiate between 128 and 320, but cant differentiate between 24 bit 192khz and a 320kbps!!!
may be my golden limit is 320kbps lol!
so i dont want to drive louder, l like to drive crisp. 1875 keep it nice, but it seems lil less . so i was thinking about building something with 50+ watts /channel. and about 100 for sub. options are pretty straightforward ... 3886 /7294 or the MOSFET again. but what really concerns me is.. MOSFETs are Monsters, they have burned some of my pretty speakers in the past, ( Few 2*8ohms and Few 4ohms). Now my yamaha is 6Ohms@30w rms per channel, and the sub is powered. i dont want to loose them. So what you guys suggest, MOSFETS or Lm? And i will keep it down , about 20-30% to protect speakers.
btw,i listen to 320kbps mostly, with some 256/128/high resolution 24bit ( its so hard to get high resolution tracks in indian music, if there is some one who knows. pls tell me ) . i can differentiate between 128 and 320, but cant differentiate between 24 bit 192khz and a 320kbps!!!
may be my golden limit is 320kbps lol!
I propose another 4 factors in favor of discrete over IC.
1. cost
LM3886 is quoted at $3.26 for 1k pieces. That's more expensive than a pair of output BJT or MOSFET bought at large quantity. And AVRs need 2 ICs for each channel which makes things even worse.
2. 3886 is rated at 34W for 8ohm. This is not enough for most home use amplifier. Bridge 2 ICs will boost power but each IC sees half the load. So if this bridged amplifier is driving a 4 ohm load, each IC see only 2 ohm at its output. This might not be stable for the IC.
3. Marketing might be another issue here. I think it's really hard to come up with good marketing material to explain why simple IC is used. There's still lots of audiophiles overthere who don't like ICs.
4. Scalability is another issue. With a discrete design, it's easy to scale up or down. While for ICs, higher/lower power normally means another IC. This has 3 consequences. First, experience with certain IC is useless if another IC is chosen. This should not be overlooked since all kinds of small issues will surface if enough units are build and used in all kinds of situations. Second, if this particular IC is obsoleted by the manufacture, a new design has to be created to replace it. Third, using multiple ICs for various products means for each IC, the purchase quantity is low which leads to higher cost.
So overall, I think amplifier builders have enough reason to favor discrete circuits over ICs. I've seen reviews on active monitors where ICs are more common, since they market speakers instead of amplifiers.
1. cost
LM3886 is quoted at $3.26 for 1k pieces. That's more expensive than a pair of output BJT or MOSFET bought at large quantity. And AVRs need 2 ICs for each channel which makes things even worse.
2. 3886 is rated at 34W for 8ohm. This is not enough for most home use amplifier. Bridge 2 ICs will boost power but each IC sees half the load. So if this bridged amplifier is driving a 4 ohm load, each IC see only 2 ohm at its output. This might not be stable for the IC.
3. Marketing might be another issue here. I think it's really hard to come up with good marketing material to explain why simple IC is used. There's still lots of audiophiles overthere who don't like ICs.
4. Scalability is another issue. With a discrete design, it's easy to scale up or down. While for ICs, higher/lower power normally means another IC. This has 3 consequences. First, experience with certain IC is useless if another IC is chosen. This should not be overlooked since all kinds of small issues will surface if enough units are build and used in all kinds of situations. Second, if this particular IC is obsoleted by the manufacture, a new design has to be created to replace it. Third, using multiple ICs for various products means for each IC, the purchase quantity is low which leads to higher cost.
So overall, I think amplifier builders have enough reason to favor discrete circuits over ICs. I've seen reviews on active monitors where ICs are more common, since they market speakers instead of amplifiers.
@yuan 201,thats a pretty good explanation bro. so actually thre is nothing that stops an audio lover, to build an AV reciver using ICs like LM3886 or 7294 , besides the thermal management. and 30-40 watts /channel seems to be enough for normal home use ( from my experience, ya it is less disturbing to parents.. lol) . and one of my 60w/channel using stk ic + 100w sub amp using another stk (60*5+100*1 @ 8 ohms) which was a sony circuit , seems to be lil over kill in terms of power. btw , its seems to lack some crispness ( brightness + clarity) than my 1875 build.. so thats why i am trying to build new amp... ( lol my room is now full of these stuffs , 7.1 Transistor + ic, tda 2030 based 5.1, tda 2050 based 5.1 ,7264 based 5.1( currently using it),An17831a based 5.1 ( its also beautiful, but little bit high on the trebles, and some class d boards including 4*stereo tripaths..) and now considering a 3886/7294 based 5.1..!
hahah.. if you ask why this much multichannels... i will have only one answer... this is what i do.. and its my aged hobby..
hahah.. if you ask why this much multichannels... i will have only one answer... this is what i do.. and its my aged hobby..
actually i would say the IC chips have advantage over transistors in many ways.
matching of components for example.
and the price is really decent too, you should consider more than just the output pairs. most of the chips have everything covered so to say, can't really beat that.
power is not an issue, 20 ish watts is far more than enough, it really depends on the speakers. if you need high power however, thermal problems ma arise. or not, if you need really a lot of power, then go for class D chips. lots of power, tiny heat.
ease of use, consistency, all favour chips.
power scaling allso, never has it been so easy and cheap to make 300+ watt amps than nowdays.
quality wise...
well composit amplifiers are just amazing. precision opamp with a decent poweramp.
matching of components for example.
and the price is really decent too, you should consider more than just the output pairs. most of the chips have everything covered so to say, can't really beat that.
power is not an issue, 20 ish watts is far more than enough, it really depends on the speakers. if you need high power however, thermal problems ma arise. or not, if you need really a lot of power, then go for class D chips. lots of power, tiny heat.
ease of use, consistency, all favour chips.
power scaling allso, never has it been so easy and cheap to make 300+ watt amps than nowdays.
quality wise...
well composit amplifiers are just amazing. precision opamp with a decent poweramp.
I think that a discrete transistor class AB has bass authority that a chip amp can't touch. Having built about half a dozen discrete transistor amps now - they just seem to have easy effortless bass. I am finding that I particularly like big fat vertical MOSFETs (like Vishay IRFP240/9240) for the output stages. They have an inherent ability to flow huge amounts of current and have a very low on state impedance so can drive difficult loads very easily. Well designed discrete amps also have lower distortion than basic chip amps (souped up ones like tomchr's mod86 being the exception).
Well designed discrete amps beat pretty much all chip amps. However, this doesn't mean chip amps have anything inferior per se. This rather shows the IC industry doesn't have much incentive to design such ICs. The LM3886 was around for 15 years and is still one of the best amp IC available while CPU 15 years ago is much less powerful than the chip in my mobile phone. If TI or ADI is serious in developing the best performacing amp IC, I think they'll be able to deliver something.
As stated, the thermal limitations of the small IC package becomes an issue at higher power and ICs like the LM3886 deliver better than enough distortion performance for mass market consumer goods. I don't see chips like this used much outside mid range guitar amps, small bass practice amps, small subwoofers and few compact stereos (most of these stereos use STK modules). Most modern receivers use an IC driver and discrete output. This seems to be the common arrangement of most $500 and less home theater receivers over the last 20 years or so.
Discrete amplifiers are only as good as their designer. There is a lot of knowledge and good components available now for building great discrete amps but it wasn't always this way. Look at the schematics for earlier designs that were used in consumer goods. After repairing a lot of vintage equipment, there was a lot of laughable marginal stuff out there, though my hindsight is 20/20.
Discrete amplifiers are only as good as their designer. There is a lot of knowledge and good components available now for building great discrete amps but it wasn't always this way. Look at the schematics for earlier designs that were used in consumer goods. After repairing a lot of vintage equipment, there was a lot of laughable marginal stuff out there, though my hindsight is 20/20.
The main issue is that chip amps are designed for economy. Nothing past that, and that is what TI has designed. We can improve the external circuits around them but we are stuck with the built in performance.
Hi Arty,
-Chris
Hi Arty,
Well, closer than a random pick off tape maybe. They have not been optimized any further from the mask set, so you aren't going to see the high performance that matching might bring to the table. That's just wishful thinking. Remember, chips are there for economy and nothing more.
-Chris
Guys what I dislike of I.C.'s is that to get a replacement if the smoke came out if the sets are not new anymore. Sound? discrete and the circuits do not really differ. In 1971 I had an "ic" amp and the output stage was dead. I opened the i.c.and fitted transistors with pieces of wire to the drivers. It worked. The internal wiring was difficult to solder as rosin core did not take easy. I was challenged and won the waver and the stakes a laugh and smile on my dial
IC chipamps are used to save big manufacturers some assembly money.
The single chip replaces at least ten discrete components and could replace 100 discrete components.
They are quick to assemble and due to very few components relatively few mistakes can be made so the manufacturer has fewer warranty returns.
That is the be all and end all of chipamps. Quick and cheap to assemble by robots.
The single chip replaces at least ten discrete components and could replace 100 discrete components.
They are quick to assemble and due to very few components relatively few mistakes can be made so the manufacturer has fewer warranty returns.
That is the be all and end all of chipamps. Quick and cheap to assemble by robots.
surely you could hand match transistors closer, but the ic is made out of the same material. it gives a pretty good edge.
somethng you will never have with discrete components.
The LM3886 isn't too shoddy out the box with a proper layout and if you look at some of the composite designs they are close to state of the art within their power limitations, which are a lot lower than discrete. If you need 200W or 30A peak current discrete wins hands down, no contest. But for 30-40W into sensible loads they can be as good as it gets, and a great introduction for the DIY'er who is just starting out. For the money can't be beat. Certainly nothing to be ashamed of.
- Status
- Not open for further replies.