I have not yet come across speakers which are the limiting factor, it's always been the amplifier running out of puff that cripples the sound. At the recent show, ambitious JBL horns were slipping into PA sound mode because the amp wasn't up to it, and a highly ambitious panel speaker was only capable of a certain volume before starting to become ragged - to my ears a power delivery problem.
Frank, humor me, as a "design exercise'', and which chip-amp exactly would be able to accomplish such wonderful deed ?
You're welcome to BTL them, if you think it helps.
(something I did ~30 years ago, btw, parallel/bridge a bunch of TDA chips)
Yes, exactly that, with some, many(!) LM38xx chips. In the Rowland tradition, the chip amps are 'intelligent' power transistors - straightforward to get current delivery, peak to peak voltage is the more difficult to achieve. It becomes a nightmare of multiple, floating power supplies to ensure the peak voltage can be hit - so the complexity of feeding the chip rails correctly defeats the value of using this approach.
Kinda like "Chip foot" that big powerful chip amp only you seem to know .......
I rather suspect such comments are borne out of snobbery as I don't notice any technical arguments making sense here. The current limit is an issue for wanting more output power than for what its designed for, or for driving lower impedance loads than 4ohms. THD residuals aren't particularly significant in my experience (which goes back to when they first arrived on the scene, somewhere back in the mid 1990s if I recall correctly).
The biggest issue I've found with LM3886 is that its driver stage power supply can't be separated from the output stage - meaning the PSRR is ****-poor. However there are some ways to mitigate this - KSTR has presented a schematic which re-references the input signal to the -ve rail so that only the +ve rail PSRR matters.
LM3886 does eventually fall down against TDA8566 in terms of 'warmth' - the National part being somewhat less accurate in instrumental timbres (piano being an obvious one for me) but that's only based on driving the Philips chip into a fairly benign load (8R) so much more limited power output (around 12W vs 60W for the LM)
As for heat dissipation, like Frank says. Oh and avoid the isolated 'TF' version - its thermal impedance sucks as against the metal tabbed one.
Last edited:
Yes, in a very early knock up I did, the chip had minimal effective heatsinking - and this unit keeled over, stewed, with little provocation. For the working units I went with a heatsink shape that exploited the chimney effect, to get the heat away from the source as efficiently as possible ...
Tonality I haven't found to be a problem: to give you an idea, at one stage I had the tweaked Perreaux 2150B driving the right speaker, which was different from the left speaker, that was using my chip amp design to feed it, for quite a while. No-one picked that something was "wrong",
....
Mainly that the gains matched as is, and the sound was veering into and out of the invisible speaker mode, depending upon everything - at this point a listener is tuning into the sound 'picture', rather than what's coming from the speakers - the normal giveaways are not subjectively important in the listening.
No schematic, maybe there's one out there - its thing was voltage drive, 90V rails - poor energy reserves in standard form required about a 6 times boost in capacitance - this allowed it to keep up with the chip amp,
...
You're working hard on your belief system, a.wayne ... the only amplifier I've come across recently that impressed me as being more 'strong', and simultaneously clean, than what I normally am aiming for, is the Bryston 28B-SST2 1000W monoblocks ...
Last edited:
Haven't seen a chipamp with 90V rails - were you using TDA7293 in bridged mode to keep up with that? Or just not running flat-out?
My current hunch is that tonality is somehow connected with having an LTP input stage. LM3886 has some nice big (1k1) degen resistors in its LTP though which should make it more linear than a bog-standard opamp. However the LTP has EF buffers before it - this makes me suspicious of the HF performance of the on-chip CCSs.....
Stereophile says that 1kW Bryston has 160,000uF rail capacitance - sounds way too low for a kW amp. My last chipamp had that, for 20W.
My current hunch is that tonality is somehow connected with having an LTP input stage. LM3886 has some nice big (1k1) degen resistors in its LTP though which should make it more linear than a bog-standard opamp. However the LTP has EF buffers before it - this makes me suspicious of the HF performance of the on-chip CCSs.....
Stereophile says that 1kW Bryston has 160,000uF rail capacitance - sounds way too low for a kW amp. My last chipamp had that, for 20W.
Last edited:
Just not, running flat out. Chips were LM3875, I had the power supplies pushed right up to the spec'd limits, 42V I seem to recall - yes (looking at DS) that's right on the max. I could get away with this because it was fully regulated, with very large energy reserves.
The proof was in the eating, these chip amps could give me true 'invisible speakers' at solid volume levels, if all elements that mattered were in alignment.
The proof was in the eating, these chip amps could give me true 'invisible speakers' at solid volume levels, if all elements that mattered were in alignment.
No I'm not , i have owned those amps , not run mental Sims as you seem to enjoy , they are pretty pathetic at best ...
Brax,
Brystons are pretty pathetic too, much better than the Perreaux , but still not my cup of tea ...
Note: I have not owned a 28B-sst2 , but in typical Bryston fashion its a 4 ohm at best amp , like the 4B , et al i have owned ....
Stereophile:
Fig.3 plots the THD+N percentage in the 28B-SST's output against its output power into 8, 4, and 2 ohms. The amplifier comfortably exceeded its 1000W/30dBW specification into 8 ohms, delivering no less than 1300W (31.15dBW) at 1% THD+N. (The wall AC voltage dropped from 125.1V to 120V at this power level.) Into 4 ohms, the Bryston clipped at 1800W (29.5dBW, 117.5V wall voltage), though it couldn't maintain its high power into 2 ohms, delivering 1050W (24.2dBW, 116V wall voltage).
PS: Still waiting to gear that DAC you raving about , the one better than dCs ...
Last edited:
You should experience what is a available , a chip amp is barely listenable Frank, the top end is bad and the grunge hides low level detail. You can build a better amp than that frank ...
You'll be waiting a long time, don't hold ya breath as I'm not planning on sending you a sample
Perhaps you'll be able to cadge one from a beta tester though.- Status
- Not open for further replies.