Can LM3886 BPA300 beats Plinius quality?

[rhythmdiy]

Hi,
I was thinking to build a BPA300 amp for my 3 way system but I would like to know how does the sound compare to Plinius? as I heard that Plinius is top notch performer.. can LM3886 BPA300 beats Plinius quality?

[danielwritesbac]

Its worth a shot.
See also the little transistor buffer at decibel dungeon.

[sangram]

Build, and then compare. Is LM3886 better than xxx (insert choice of brandname here) is not good enough. Which specific amp, how old, which speakers, what kind of music? Too many variables, you see?

I find that the LM3886 starts facing obstacles at high levels or difficult loads, and is almost unusable when it runs into both simultaneously, no matter what the capacitance or heatsinking.

It's a very sturdy and miraculous little chip (survived reversed power supply for 35 seconds, mine did!) that is capable of excellent sound. Quality components help.

I finished another one last night with 80,000uF for a single stereo pair, and it's burning in yet. Holcos in feedback and Ri positions, and a BG MX 47uF for Ci. No snubbers, but 100nF X7R 1206 caps soldered directly on cap pins. As usual, done fully P2P. Very deep bass, and very little loss in midrange resolution. All in all much better than the LM4780 BPA I built from Peter's kit, if a little too glassy at the top end. Very different every time I build them, like cooking I think. Small changes teach a lot.

But it's no world-beater, and has its own limitations. the short answer is if you can afford to, get both. If you can't and have the money, buy the brand. Then you have someone to blame. If you don't have the money, build it. I have the money, but I'd rather build.

[rhythmdiy]

I have a 10 inch vifa woofers PL26w ones their extension is very very good but as you know they require 125RMS into 8 ohms min to work properly... Ofcourse I bridge it to get good voltage output but all the probs comes with the current output... Or else I might have just took this amp and might have built a massive gainclone..with 10 chips in parallel in BPA 500 just to drive my speakers..

I have three way speaker setup with 10 inch Vifa PL woofer and 61/4 wavecor Neo midbass with 1 inch neo tweeter from wavecor... Now the mids and highs are just incredible but the PL26 woofer needs more current than the mids and tweets..

Ive contacted Jeff Rowland and he said that no speaker is able to take up beyond 8 amps and the chips are able to deliver when the concentra 2 or Model 112 have those to feed... but but the major question comes only after reading lots of reviews is that the bass is not tight enough with these amps... so now tell me whats the way to get tighter bass like Brystons?

I ve auditioned plinius but some say that Jeffs rowlands are better in some ways when compared to plinius so now the point is that how do i solve this issue?

Ive simulated using the overture design guide.xls where I punched in certain values which gives enough output

[sangram]

I would suggest biamping your speakers, and using a separate amp for the woofers.

If you can drive the top with a single chip, optimised for quality and the bottom with your huge BPA, I guess you'll get what you need.

I'm not understanding where the JR and Plinius amps are entering the equation. If you're referencing those kind of amps, there are issues well beyond the circuitry and components used. If you keep thinking, you'll neither build nor buy anything.

[rhetoric]You can beat a Veyron with a car built in your backyard but the thing is, you need to have a big enough backyard, the right tools, and the right experience. Not that the amps you mentioned are equivalent to a Veyron, but you get the idea. [/rhetoric]

To bass: Higher capacitance helps stiffen the supplies, but I find diminishing returns after about 20,000uF per chip, and the best 'compromise' is about 4700uF per chip but with a very heavily oversized transformer. That is for my ears and my speakers. Larger capacitance generally hurts the midrange, but up to about 10,000uF per chip the degradation is not too much. And I use locally available caps bypassed with smaller caps, and recently SMD caps directly on the power pins.

As for your speakers, be careful how hard you drive them. From a voltage perspective it's difficult to get more than 40 volts at the output of any kind of chipamp, even in a bridged amp. That's a lot of power though, and if your speakers are 8 ohm then any number of parallel chips beyond 2 is not going to help increase the output further, or enhance the current output unless your speakers are difficult loads.

I'm driving a pair of Dynaudio BM6 monitors with just one pair of 3886 per side (PA100) and 40,000uF of capacitance. It helps me do everything that a commercial monitoring setup would, with no shortage of anything. If your speaker is indeed current hungry, you may want to start here.

Edit: I looked up the 26W specs and they are 87dB efficient. With losses in crossover, about 85? The first thing to do is lose the passive crossover. You should be able to get back up to the 87dB figure, 2dB of output is not something to sneeze at and you obviously like it loud. Then use a single pair of LM3886 in bridged configuration with a 30V supply (each chip sees a 4 ohm load/they'll be fine), as your first starting point, and then go from there.

[danielwritesbac]

Originally posted by rhythmdiy
I have a 10 inch vifa woofers PL26w ones their extension is very very good but as you know they require 125RMS into 8 ohms min to work properly... Ofcourse I bridge it to get good voltage output but all the probs comes with the current output... Or else I might have just took this amp and might have built a massive gainclone..with 10 chips in parallel in BPA 500 just to drive my speakers..

I have three way speaker setup with 10 inch Vifa PL woofer and 61/4 wavecor Neo midbass with 1 inch neo tweeter from wavecor... Now the mids and highs are just incredible but the PL26 woofer needs more current than the mids and tweets..

Ive contacted Jeff Rowland and he said that no speaker is able to take up beyond 8 amps and the chips are able to deliver when the concentra 2 or Model 112 have those to feed... but but the major question comes only after reading lots of reviews is that the bass is not tight enough with these amps... so now tell me whats the way to get tighter bass like Brystons?

I ve auditioned plinius but some say that Jeffs rowlands are better in some ways when compared to plinius so now the point is that how do i solve this issue?

Ive simulated using the overture design guide.xls where I punched in certain values which gives enough output

Tighter bass?

Okay! That's an easy one. You tune your power circuit for it.
The CarlosFM 2005 supply is one such example. My LM1875 for beginner's thread contains another such example and it doesn't use large slow caps. The footnotes on both designs are reasonably interchangable, especially the Rail2Rail (sink the mids) capacitor, (2.2uF to 4.7uF, 250v capacity) located at the power terminals of the amplifier board. That one little part does wonders for clearing up the midrange.
See also Snubbed Rectifier, especially the customary "ring around the rectifier" layout, as this allows using more voltage, getting more power, and having less heat.

See also a buffer/preamp, because the customary spot for a lot of audio tinkering/fine-tuning is a buffer or preamp. A smaller scale amplifier is generally an easier fight to pick than any attempt of voicing a larger scale amplifier. So, tinker the pre.

Fast caps aren't that rare after all. Nichicon's ES and FW are purpose made for high-resolution use. Whatever you think of that statement, I think that purpose-made components can be a lot more accurate than a random selection. I was especially impressed with these because they perform like the modeling software predicts. That's rare.

The usual 10" speaker will come to the limits of fidelity at approximately 70 watts. A pair of these with a pair of LM3886 is capable of damaging the interior of a house. You can, in fact, bridge the LM3886; however, it should be run on reduced voltage. Realistically, a single LM3886 is capable of most of the output of either its bridged (voltage limit) or paralleled (heat limit) version if you choose the right voltage for the job. This means that the next step UP from a single chip is probably bridge+parallel.

At the estimated efficiency of your speakers, they could burn out before making a concert presentation. But that isn't the point. The point is that the speaker will reach its fidelity limit at approximately half of its power handling capacity. Nobody really wants to hear the voice coil slamming into the magnet when music is expected instead. The 125w rating is probably when you'll hear noise.
Since it is difficult (not impossible) to get 70w from a single LM3886 without setting off the spike, you can use a paralleled LM3886 instead, and as this will spot half the load, then the Spike won't pipe up. I think there's your clean 70w. A pair of AudioSector 4780 boards is an option for a good place to start. You can add a more elaborate power circuit and a buffer/pre if you wish.

EDIT: Weird question though. The LM3886 is known for tight bass. Otherwise, I think that nobody would put up with the presentation it gives for midrange. Ah, but that's just my opinion. I also think its nicer paralleled.

[pacificblue]

Originally posted by danielwritesbac
it doesn't use large slow caps.

Originally posted by danielwritesbac
See also Snubbed Rectifier, especially the customary "ring around the rectifier" layout, as this allows using more voltage, getting more power, and having less heat.

Originally posted by danielwritesbac
The usual 10" speaker will come to the limits of fidelity at approximately 70 watts.

Originally posted by danielwritesbac
The point is that the speaker will reach its fidelity limit at approximately half of its power handling capacity. Nobody really wants to hear the voice coil slamming into the magnet

Originally posted by danielwritesbac
The 125w rating is probably when you'll hear noise.

Where do you read and hear such stuff? Or are you making it all up by yourself?

Originally posted by danielwritesbac
Fast caps aren't that rare after all.

Last time you wrote about fast capacitors, you could not explain, what they are. Can you now?

Tip: [F] = [As]/[V]

[danielwritesbac]

Originally posted by pacificblue
Where do you read and hear such stuff? Or are you making it all up by yourself?

That's ridiculous; however, actually, I would love to take credit for making it up. lol!

There is perhaps the error of omission about the snubbed rectifier. Not only can you use more voltage, the actual concern is that you'll need more voltage (just a bit). With 1 little cap per diode, the most weak (least useful) efficiency of the diode is hindered. That's a slight voltage drop. I think thats nice because the voltage will fluctuate less during the amplifier's operation.
We're supposed to run that LM3886 chip on DC after all, aren't we? Fluctuations are AC. Stop that and you get a cooler amp.

Okay, so I guess all of it was obvious. What were you after anyway?

Perhaps a clarification? When I said "large slow caps" I actually meant that the design doesn't use midbass noise to level its frequency response. Therefore there are no 4700uF caps on the power board and there are no caps larger than 600uF on the amplifier board.

I'm NOT against using 10,000uF caps on power supply boards, although it is also not my preference. I AM against using ONLY 4700uF caps on power supply boards because of the likelihood of causing ringing,

Originally posted by pacificblue
Last time you wrote about fast capacitors, you could not explain, what they are. Can you now?
Tip: [F] = [As]/[V]

No. I can't explain fast caps. My math skills are not up to translating advertisments and ratings into action.
However, I was able to get some information from Peter and Carlos about preferred caps. One in common was Nichicon FW. I think its most interesting to use them at the amplifier board, because any reasonably designed power supply board doesn't need speciality caps.

[Juergen Knoop]

[tubesguy]

Originally posted by Juergen Knoop

I'll second that, and add a