Several friends, along these last years, comes to me asking about chips...how to make them sound better.... well..... sadly i hate them, i hate chips.... and i have to answer things about them.
Will do no more...will point this thread, where i will say what i think about them...then people will come and read...and i will be free to be repeating my beliefs and result of experiences made along several years...there are some videos posted in youtube about.....type "destroyersoueu" there, in the youtube search bar, and you will find them.
Chip is something when one internal transistor burns, you lost the entire thing.
Chip is something misterious, you have transistors inside, and capacitors, and you cannot change them or replace them and tune them.. circuit is not correctly published, not complete.
Chip has an awfull thing, the case is small, and this was made this way to reduce costs and to make them better to be installed in micro systems.
Chip sounds very good if you know the tips and tricks to use them.... the tips and tricks are:
1) do not use full power
2) do not use high power
3) do not play loud
4) do not use 4 ohms
5) accept the thing has no power and adjust your volume very low
Why i dislike them?
That small case cannot transfer heat to heatsink..so... dissipation is a problem, the case is small, the contact area is too much small... to that case, 100 watts of dissipation is a hell optimistic...i think 80 watts may be the maximum.
Well...if you have 80 watts of dissipation (maybe if you're lucky, and if the environment is a miracle and temperature is 25 degrees celsius and heatsink does the miracle to be keept in 25 degrees celsius without warm up!)... in class AB amplifier, you cannot have more than 40 or 45 watts rms.... so...they make it this way.... 45 watts or something alike..... and if you install 4 ohms speaker..them it will give you 45 watts too...ahahahahahah!...because cannot give you more, or will overheat and melt the whole damn thing.
The chip has protections, and this kills the sound...it has over voltage protection...you install over voltage and it burns!
The chip has invertion protections....try to invert and see what happens
The chip has output short circuit protection...sometimes works.
The chip has overcurrent protection... this limits the audio dinamic... if you have a fortissimo, a peak of power, will be limited..so, will sound alike a battery portable radio from the seventies.
The chip has over voltage (output) protection...this clips the audio signal.
Well...cannot be perfect in sonics...it is not possible to be perfect this way.
If you can remove protection, them will burn...if you can do it will burn....it seems you cannot.
So.... if you can accept it was made, to operate around 5 to 10 watts to give you, unders this circunstances ,under this condition, a headroom to face some peaks..then they are fine, so good, and even better than a very good discrete.
But, will never give all that treble...and this because the distance inside the chip...the metal silicon block, are too small..so...output can be picked to input...to avoid this, the amplifier is heavily neutralized, this kills treble harmonics and sonic quality.
Also the gain condenser, is adjusted to avoid you to have deep bass..because deep bass triggers the protection circuits....so...you can increase the NFB gain condenser, that one is in series with a resistance..then you bass will be strong and you will have protection entering in action and your dinamics will be awfull.... unless you reduce your volume demand.
So.... the solution is to increase that condenser, to obtain deep bass and adjust the amplifier to low power...this way you will have headroom and nice sonics.
Accepting low power, then chips are fine, are great, lovely and have excelent sonics...depends on you.
I want distance from them.... big distance.
regards,
Carlos
Will do no more...will point this thread, where i will say what i think about them...then people will come and read...and i will be free to be repeating my beliefs and result of experiences made along several years...there are some videos posted in youtube about.....type "destroyersoueu" there, in the youtube search bar, and you will find them.
Chip is something when one internal transistor burns, you lost the entire thing.
Chip is something misterious, you have transistors inside, and capacitors, and you cannot change them or replace them and tune them.. circuit is not correctly published, not complete.
Chip has an awfull thing, the case is small, and this was made this way to reduce costs and to make them better to be installed in micro systems.
Chip sounds very good if you know the tips and tricks to use them.... the tips and tricks are:
1) do not use full power
2) do not use high power
3) do not play loud
4) do not use 4 ohms
5) accept the thing has no power and adjust your volume very low
Why i dislike them?
That small case cannot transfer heat to heatsink..so... dissipation is a problem, the case is small, the contact area is too much small... to that case, 100 watts of dissipation is a hell optimistic...i think 80 watts may be the maximum.
Well...if you have 80 watts of dissipation (maybe if you're lucky, and if the environment is a miracle and temperature is 25 degrees celsius and heatsink does the miracle to be keept in 25 degrees celsius without warm up!)... in class AB amplifier, you cannot have more than 40 or 45 watts rms.... so...they make it this way.... 45 watts or something alike..... and if you install 4 ohms speaker..them it will give you 45 watts too...ahahahahahah!...because cannot give you more, or will overheat and melt the whole damn thing.
The chip has protections, and this kills the sound...it has over voltage protection...you install over voltage and it burns!
The chip has invertion protections....try to invert and see what happens
The chip has output short circuit protection...sometimes works.
The chip has overcurrent protection... this limits the audio dinamic... if you have a fortissimo, a peak of power, will be limited..so, will sound alike a battery portable radio from the seventies.
The chip has over voltage (output) protection...this clips the audio signal.
Well...cannot be perfect in sonics...it is not possible to be perfect this way.
If you can remove protection, them will burn...if you can do it will burn....it seems you cannot.
So.... if you can accept it was made, to operate around 5 to 10 watts to give you, unders this circunstances ,under this condition, a headroom to face some peaks..then they are fine, so good, and even better than a very good discrete.
But, will never give all that treble...and this because the distance inside the chip...the metal silicon block, are too small..so...output can be picked to input...to avoid this, the amplifier is heavily neutralized, this kills treble harmonics and sonic quality.
Also the gain condenser, is adjusted to avoid you to have deep bass..because deep bass triggers the protection circuits....so...you can increase the NFB gain condenser, that one is in series with a resistance..then you bass will be strong and you will have protection entering in action and your dinamics will be awfull.... unless you reduce your volume demand.
So.... the solution is to increase that condenser, to obtain deep bass and adjust the amplifier to low power...this way you will have headroom and nice sonics.
Accepting low power, then chips are fine, are great, lovely and have excelent sonics...depends on you.
I want distance from them.... big distance.
regards,
Carlos
Last edited:
for the power dissipation, a normal LM3886 has a junction to case resistance of 1 C/W
add isolation + cooling paste and we got 1.5 C/W
with a fairly large heatsink of 0.5C/W we got 2 C/W
if we want to keep everything below 80 degrees C we can dissipate 80 - 30(temperature in my attic atm) = 50, 2C/W means 50/ 2 = max 25 watt.
i have to agree at this point with you.
that 1C/W is mean, you have to put a couple in parallel for lotsa power.
but my little brother of 13 can build a LM3886 amplifier and make it sound good.
thats the magic of them, IC 5 euro, transformer, 13 euro each (farnell sale), casing 21 euro each, rest of the parts around 10 euro each, used some old AMD CPU coolers.
So a 1 channel amplifier was about 50 euro's
cheap as can be, nice sound, and my little bro from 13 can make it.
for high end use, nope not my piece of cake, for a decent hifi amplifier, hell yeah
add isolation + cooling paste and we got 1.5 C/W
with a fairly large heatsink of 0.5C/W we got 2 C/W
if we want to keep everything below 80 degrees C we can dissipate 80 - 30(temperature in my attic atm) = 50, 2C/W means 50/ 2 = max 25 watt.
i have to agree at this point with you.
that 1C/W is mean, you have to put a couple in parallel for lotsa power.
but my little brother of 13 can build a LM3886 amplifier and make it sound good.
thats the magic of them, IC 5 euro, transformer, 13 euro each (farnell sale), casing 21 euro each, rest of the parts around 10 euro each, used some old AMD CPU coolers.
So a 1 channel amplifier was about 50 euro's
cheap as can be, nice sound, and my little bro from 13 can make it.
for high end use, nope not my piece of cake, for a decent hifi amplifier, hell yeah
Class D is a minefield for the uninitiated.
His 13 year old brother would never manage it.
Yes folks..enjoy to discuss the subject...as i told you, i am interested to
keep this first post saved in the forum files...then when people comes to me to ask things about chips, i will have this link to point them without having the need to talk about this subject over and over again.
Enjoy a good debate about.... i am old and already have my conclusions about..and they were all tested...so...i am not even curious to discuss the subject.
About this subject, i am not in the sittuation to say "I think so"...i do have already jumped to other stage about the problem os chips versus discrete.... i am in position to say to myself that i already know .and i should worry and spend my time with other things i really do not know...other kind of technologies.
I thank you by the cooperation and kind support, visiting the thread and i would be happy if you accept to manage the thread by yourselves, as i said..the thread, for me, is just a tool not to discuss the subject anymore.... if you want and like, or need to discuss..then have fun.
regards,
Carlos
keep this first post saved in the forum files...then when people comes to me to ask things about chips, i will have this link to point them without having the need to talk about this subject over and over again.
Enjoy a good debate about.... i am old and already have my conclusions about..and they were all tested...so...i am not even curious to discuss the subject.
About this subject, i am not in the sittuation to say "I think so"...i do have already jumped to other stage about the problem os chips versus discrete.... i am in position to say to myself that i already know .and i should worry and spend my time with other things i really do not know...other kind of technologies.
I thank you by the cooperation and kind support, visiting the thread and i would be happy if you accept to manage the thread by yourselves, as i said..the thread, for me, is just a tool not to discuss the subject anymore.... if you want and like, or need to discuss..then have fun.
regards,
Carlos
Last edited:
hi destroyer X, the most sensible approach is to combine the avantages of ICs and discretes, in an intelligent way, don't you think ? Kuroda did this long time ago, in September 1982. His design predates the (in)famous patented (!) "Alexander Current-Feedback Audio Power Amplifier" presented in the 88th AES convention in March 1990. It is a shame that Kuroda got eclipsed by Alexander. The Kuroda approach is simple, clean, subtle, also has curent feedback attributes along with high slew rates, and can cope with the most advanced output stages like soft non-switching class AB. Give him a chance. Have a try, maybe ? Here is a link : http://www.diyaudio.com/forums/solid-state/164476-kuroda-tl071-60w-fet-amp-1982-a.html
Last edited:
I just went to the LM3886 datasheet and was very surprised that the isolated one and the metal tab one both have theta-jc of 1 C/W. So if that's real (and I doubt it) you can save a bit by using LM3886TF. Interesting also to compare this figure with the 'competition' - the TDA7294 has 1.5 C/W so I guess its a smaller die.
Paralleing is the way to go with chips, then with paralleled chips we can also bridge. Once we do that though, there's no reason to spend $5 on an LM3886, may as well use a $0.3 TDA2030! They have 3 C/W so three of them is thermally equivalent to one LM3886 but they can be spaced apart making the heatsinking easier.
Can use an SMPSU which will be cheaper (what with raw materials like steel and copper prices going up, silicon keeps on coming down). With a bridged-paralleled amp, only need one 24V PSU (tweak its voltage up to 28V) to do 50W/4R comfortably. Price is under $10 for the supply.
Agreed, chips make it simple to get good enough sound that a kid can do it
Last edited:
paralleling and then bridging them doesnt make it easier, needs a PCB, needs 1% tolerence parts.
chip amps are lovely becouse there so simple, if they stop being simple why no build a discrete power amp?
24V PSU, i think you need a symmetrical powersuply so you need 2 of them anyways.
and i got some bad experiences with way to noisy SMPS units, if you know a afforable one thats good say so please (and shipping to the netherlands makes a transformer cheaper fast)
One of the big advantaged of Chip amps is, they don't require any bias and/or DC adjustment at the manufacture. Also they are more reliable then most discrete designs. From a manufacture point of view (for example active loudspeakers where you see them most) this are very strong and valid points to choose a chip over discrete.
Subjectively speaking, for small subwoofer or bass applications the BPA300 alike application of 6 chips sound superior to most discrete designs in my opinion.
The LM3886 (also in bridge parallel) turn out to be very very reliable if designed the right way.
With kind regards,
Bas
Subjectively speaking, for small subwoofer or bass applications the BPA300 alike application of 6 chips sound superior to most discrete designs in my opinion.
The LM3886 (also in bridge parallel) turn out to be very very reliable if designed the right way.
With kind regards,
Bas
This is good for short feedback paths. But bad for other parts. There will be always a coupling (magnetically) between input stage and output stage. Normally you wish to separate them from each other to avoid coupling.
With kind regards,
Bas
OK, point taken about the PCB, I've never tried to build a 'components in the air' chip amp though.
But 1% parts are death to paralleled chip amps - I only build paralleled with 0.1% resistors or alternatively hand selected 1% types matched to better than 0.1%.I've already designed and built several discrete power amps. I'm still a novice at the possibilities offered by chip amps, and chip amps are much much cheaper. A parallel-bridge chip amp is still easier to build than a bridged discrete design and absolutely no worries about bias which was always the bugbear with my discrete designs. I never did output protect either which is a bit of a headache - chip amps have that as standard.
Yeah, I originally thought that but I think it should be possible with just an active rail splitter. Going to try that anyway as no current needs to go to 0V.
When I started powering my chip amps with a pair of (very cheap, $22 for 200W) SMPSUs I also found the sound wasn't good - way too much sibilance on vocals. But I explored further and found I'd not done the power supply decoupling and regulator grounding correctly on the chip amp board. Fixing that fixed the poor sound. I do have mains input filters on the SMPSUs though. There are guys using 'Meanwell' (think its a Taiwanese brand) supplies into Tripath type amps getting good results, but my supplies are even cheaper (unbranded) than those.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.