What are the most powerful chip amps?

[nigelwright7557]

If you want more power then you might consider the IRS2092 based systems.
They can go up to Kilowatts.
Quite a selection on ebay or other vendor websites.

 

[tomchr]

LM3886
42V -max supply -220.5W average power into 4 ohms @100% efficiency
?W in 4 ohms @10% THD 68W in 4 ohms @ 0.1% THD

You may want to re-read the LM3886 data sheet...

The max instantaneous output current of the LM3886 is 7 A, guaranteed. 11 A under "typical" conditions (25 ºC, average performing IC, etc).

Let's do the math at 8 Ω load:
Vout = I * R -> Vout = 7 * 8 = 56 V (peak).
This is higher than the ±42 V allowed for the LM3886, so with an 8 Ω load, the LM3886 is limited by its supply voltage. The max (theoretical) amount of power you can get into an 8 Ω load using an LM3886 is:
P = Vout_peak^2/(2*R) -> P = 42^2/(2*8) = 110 W.
This number assumes no voltage drop across the LM3886, which is obviously not the case in practice.

For the 4 Ω case:
Vout = I * R -> Vout = 7 * 4 = 28 V (peak).
This means that in the 4 Ω case, the LM3886 is limited by its output current. The max (theoretical) power you can get from the LM3886 without hitting the current limit is:
P = Iout_peak^2 * R/2 -> 7^2 * 4/2 = 98 W.
Again, this number assumes no voltage drop across the LM3886, which is obviously not the case in practice.

You can get the drop-out voltage from the data sheet figures and re-do the math if you feel like it. The drop-out voltage causes the actual output power to be approx. 65-70 W into 4 Ω (±28 V rails) and 50-60 W into 8 Ω (±35 V rails) for the LM3886.

Some would argue that you get more power by bridging. That is true if the amplifier is voltage limited. If it is current limited (as is the case with the LM3886 and 4 Ω loads) you actually don't gain anything at all by bridging. Recall that in a bridged amp, each amp half sees half the load impedance. Thus, running an LM3886 in a bridge with a 4 Ω load is equivalent to loading a single LM3886 with a 2 Ω load. That's not a recipe for success.
With the LM3886, you are much better off running two of them in parallel. That has its own challenges as the two amp halves will set up a standing current between them. This current will need to be minimized for good performance above a few watt of output power.

Unfortunately, the laws of physics are rather unforgiving. A Class AB output stage is only so efficient. This means you'll be dissipating a large amount of power into a small package if you are using a chip amp. This also means that if you want more output power, your best option is to use multiple devices in parallel as this spreads out the dissipated power among multiple device packages.

I'll soon have a replacement for the Parallel-86 ready. It should be good to go around the middle of October.

Tom

 

[shredhead]

Thanks for that breakdown Tom, I didn't think of going off of current and working back to rail voltage. Has anyone verified the datasheet on the clipping point of the chip in a resistive load?

 

[chrisb]

FWIW, I've heard a few chip amps over the past dozen or so years, and the most immediately relaxing and musical was the Modulus 86. Its "relatively modest" rated power when compared to something like a Sure TK2050 "100W" - air quotes - belies its ability to deliver dynamics and finely resolved inner detail on medium efficiency speakers. In my brief exposure to the Mod86 last month, it was running a 2-way MTM with Alpair drivers and parallel passive XO.

The new replacement for Parallel86 (LM4780) should be very interesting indeed

 

[AndrewT]

surely if in a livingroom you need anything more than 20 watts / channel you are using spaceheaters, lost your hearing, or you are attempting to ventillate your brain.

but the question may just be theoretical, you know.

I disagree.
If I am listening @ ~2.5m to a pair of speakers with a sensitivity of 88dB/W @ 1m and the average power delivered in the speakers passband is 1W then the average spl from those two speakers is ~83dB
That is not particularly loud, A person sitting near me could just about hold a normal level conversation without shouting while the music is playing at that average level. Consideration for that person would demand I turn the music/audio down or off to continue that conversation.
But let's just assume the two of us are listening to the 1W of average power level.
The peak transients can be anywhere from 6dB to 30dB louder than that average level. I allow for 20dB peaks above my average levels. That requires both the amplifier and the speakers to handle 100W peak transients. But even that only gets that listening session to peak transient SPL levels of 103dB. That is not exceptionally loud.
If you restrict yourself to 20W, then your peak SPL level is down at 94dB.
Most listeners would not be satisfied with that level for reproduction. Mostly because the levels required to avoid clipping of the peaks would require an average level of ~74dB

 

[Arty]

it be verry interesting to log current and voltage comming from your amplifier under regular use. just to see what is really going on.
continos power and instantenios/spontanius are not quite equal.

 

[shredhead]

I agree with Andrew 100%. Average power is fine to get an idea of how an amp will perform with steady state sine waves but it really doesn't translate well into dynamic source material at all. A couple of years back I was surprised to find how easy it was to clip a 300W/channel pro audio amp into a set of mains on a home theater using dynamic source. I did some power testing with a more powerful amp and clocked peak power of over 600W at reference level per channel.

There are a lot of good sounding yet not so efficient speakers out there and with power being cheaper than ever per watt, we aren't as limited as things once were. After figuring in the size of the room and any kind of EQ boost given to the amp, it is very easy to compress the output of low powered amplifiers. It's also doesn't hurt to design the system so that the amplifier has a few dB of headroom above the peaks for better performance.

 

[Arty]

300 and 600 watt / channel.
peaking out.
surely its better to have and notneed than not have but need,
but there is a definition for overkill.
but what can i say. been there too, but won't go back.

 

[shredhead]

surely its better to have and notneed than not have but need,
but there is a definition for overkill.
but what can i say. been there too, but won't go back.

Agreed, I don't really engage in SPL drag racing. What amp(s) do you use for your mains? Class A?

 

[Voegelchen]

Have a look at

Here are some links to an article I've written about chipamps, including a powerfull LM3886 design:

http://www.ac-vogel.de/Download/Dow...ith integrated Power stages (Chipamps)__1.pdf

http://www.ac-vogel.de/Download/Dow...tage regulators for Audio Power Amplifier.pdf

http://www.ac-vogel.de/Download/Dow...sults%20%96%20in%20numbers%20and%20graphs.pdf

http://www.ac-vogel.de/Download/Dow... and pseudo differential Audio signals__1.pdf

http://www.ac-vogel.de/Download/Dow...tage regulators for Audio Power Amplifier.pdf


or in general:
Audio main

have fun

 

[Arty]

Agreed, I don't really engage in SPL drag racing. What amp(s) do you use for your mains? Class A?

94 dBL sensitivity custom drivers (6.5" midbass papercone, and a dome tweeter) driven by sta540 amplifiers. passive line level XO.
i bet they never seen more than a few watts. but no one sofar managed to tell.
everyone thinks it must be at least like 200 watts / channel.
could not be any more wrong than that.
i use a toroid ment for halogen lamps, 10,8 V AC what comes out of it.
no more than say.. 14VDC after rectification.
50 VA rated traffo.

does not takes more than that. and i can safely say i have no problems with dynamic material. surely there is some compression, but i bet its not a lot.

 

[dannyjmf]

So far what i had in my hands TDA8950 (class D ready build module however) in BTL into 4ohms it will produce about 400Wrms easily ... currently using it for subwoofer duty ...

Below shows my TDA8950 BTL module ... it will get hot if pushed hard ... need a fan to cool it down ... i think the heatsink is too small from the design

 

[dannyjmf]

if all you want is to get the most power, than class D is your pal.
if we are allowed to go with unusal impedances as load, you can squeeeeeezeee out killowatts from class D amps and mosFets without issues.
the limitng factor is the output voltage of the chip, from then on mooar output devices will yield moooar powa.
in class AB it can be simular, but harder to do.
bridged and paralelled chips that have slave/master mode make things less problematic, however roughly 500 watt is the feasable design.

discerte designs while certainly harder, can offer anything you like.
one of the lunatic solutions of the end of the tube era used tubes for voltage gain, and the output devices where so silly that the unit featured NO trnformator. basicly it was able to throw whatever you main powerlines where able to provide.
s the output swing was 200 voltish, and if you lost your brain somewhere in theory it was able to spit 6 amp current from regular household lines, up to 16 amp of current if you happend to live around industrial sites, and if you where total nuts and used industrial main lines, then 25 amper was the maximum.
since the main's line could not offer any more.
keep in mind this build was never adwised, not many exsisted, and out of that not many most of them burned or exploded or done all kinds of funky stuff.
in exchange it was the cheapest way to get your dream amp of absolute power, rated maximum to single channel 5 Kw. fidelity was no question, who needs fidelity from an 5 Kw amp ?? you could weld with those without problems.

but keeping it not silly, probabaly the tda7293 BPA application circuit is what makes somewhat sense in class AB.
and for class D the trusty TDA8954 is your friend.
anything more.. i would adwise to not fiddle with it.
it can be verry lethal.

that is why TDA8950 class D ic chip was there ... and the good news it can be purchased ready made module ... i had one the power output is really overkill for me ... used for subwoofer duty and at 60% output the subwoofer starts to make everything in the house rattle like crazy

 

[shredhead]

So far what i had in my hands TDA8950 (class D ready build module however) in BTL into 4ohms it will produce about 400Wrms easily ... currently using it for subwoofer duty ...

Below shows my TDA8950 BTL module ... it will get hot if pushed hard ... need a fan to cool it down ... i think the heatsink is too small from the design

From the datasheets (at what THD is not specified):

TDA8950
SE 2×150 W, RL=4Ω(VP=±37 V)
SE 2×170 W, RL=4Ω(VP=±39 V)
SE 2×100 W, RL=6Ω(VP=±37 V)
BTL 1×300 W, RL=8Ω(VP=±37 V)

TDA8954
SE 2×210W, RL=4Ω(VDD=41V; VSS=−41V)
‹SE 2×235W, RL=3Ω(VDD=39V; VSS=−39V)
SE 2×150W, RL=6Ω(VDD=41V; VSS=−41V)
BTL 1×420W, RL=8Ω(VDD=41V; VSS=−41V)

How many volts are you powering the 8950 with to get 400W average power? Maybe it's compressing a bit which would explain how hot it runs?

I have also found another contender:
TAS5630B
I don't understand the datasheet's numbers on max output though, does anyone have experience with this chip?

 

[DUG]

From the datasheets (at what THD is not specified):
...

Usually 10%.

Some also have numbers for 1% or 0.5%.

I usually consider power ratings at 1%. This roughly corresponds to what I have seen on a scope trace for the onset of clipping.

IMHO

 

[dannyjmf]

From the datasheets (at what THD is not specified):

TDA8950
SE 2×150 W, RL=4Ω(VP=±37 V)
SE 2×170 W, RL=4Ω(VP=±39 V)
SE 2×100 W, RL=6Ω(VP=±37 V)
BTL 1×300 W, RL=8Ω(VP=±37 V)

TDA8954
SE 2×210W, RL=4Ω(VDD=41V; VSS=−41V)
‹SE 2×235W, RL=3Ω(VDD=39V; VSS=−39V)
SE 2×150W, RL=6Ω(VDD=41V; VSS=−41V)
BTL 1×420W, RL=8Ω(VDD=41V; VSS=−41V)

How many volts are you powering the 8950 with to get 400W average power? Maybe it's compressing a bit which would explain how hot it runs?

I have also found another contender:
TAS5630B
I don't understand the datasheet's numbers on max output though, does anyone have experience with this chip?

Well i powered it using 37V +/- DC ... i assuming it close to 400W into 4ohms due to at 8ohms it is already 300W (on its hand it already killed 2 of my subs)

 

[shredhead]

I usually consider power ratings at 1%. This roughly corresponds to what I have seen on a scope trace for the onset of clipping.

Same here.

 

[shredhead]

Well i powered it using 37V +/- DC ... i assuming it close to 400W into 4ohms due to at 8ohms it is already 300W (on its hand it already killed 2 of my subs)

Nice. Nothing wrong with some hot bass.

 

[NanoFarad]

Big problem is package thermal resistance.
Since it´s basically the same as 2 TO220 cases side by side, you can´t expect wonders.

Both TDA728x and LM3886 have same case dimensions, so same thermal resistance to heat sink.

LM3886 is known for its ruggedness, whiloe TSA has a long track of failures.

Not surprising: LM claims 50/60W output and obviously can handle that; TDA claims up to 100W per chip ... and fails regularly.

I use them to get 70W and then work fine.

As of chipamps, there are way larger and more powerful, they are physically larger of course, check the very popular Sanken STK series.
I know up to 200W RMS ICs, used in PA and Bass amplifiers, there might even be larger ones.
Of course, they are as large as the palm of your hand ... which is fine, in any case they must be bolted to an even larger heat sink.

Tda don't have long track of failures. Don't talk rubbish. I am using Tda7294 with success & satisfaction.
According to data sheet & IEC rules it gives 100watt of music power at 10% thd. Nothing wrong in this. But as far as HI-FI applications are concerned, it gives 82wRMS at 0.1% thd @4ohm; & don't forget it is almost a vintage integrated.
Much older than lm3886. Lm3886 is cool but 7294 is good & powerful.
I think your integrated(7294) was fake or maybe you did something wrong.
I like lm3886 specs & low reproduction but with little tweak & good speakers/enclouser tda7294/93 can beat lm3886 in sound reproduction.
Have a good day.

 

[nigelwright7557]

Tda don't have long track of failures. Don't talk rubbish. I am using Tda7294 with success & satisfaction.
According to data sheet & IEC rules it gives 100watt of music power at 10% thd. Nothing wrong in this. But as far as HI-FI applications are concerned, it gives 82wRMS at 0.1% thd @4ohm; & don't forget it is almost a vintage integrated.
Much older than lm3886. Lm3886 is cool but 7294 is good & powerful.
I think your integrated(7294) was fake or maybe you did something wrong.
I like lm3886 specs & low reproduction but with little tweak & good speakers/enclouser tda7294/93 can beat lm3886 in sound reproduction.
Have a good day.

If you dont mistreat the tda7294 it works well.

I designed a hybrid valve/tda794 amp and it oscillated badly.
I made the mistake of reducing the tda gain to about 7. If you read the datasheet it needs a gain of at least 22 to be stable. I managed to bodge it to stop oscillating by putting 1nf across + and - inputs.

Another mistake I made was making the feedback resistor path too long.
Apparently you get 1nH of inductance per mm ! So it needs to be as short as possible.

Another mistake I made was soldering the tda hard up against the pcb.
This caused the front pins to short out on the back pin pcb pads !
So a quick removal of the old tda and soldered in a new one 3mm above the pcb and it worked great.