BPA600?

JaredC79

Banned
2015-09-02 4:32 am
NJ
Can 2 bpa300 or pa300 boards be linked by a DRV for a bpa600? Also does anyone have a good negative for the opamp biased 150 or other solid designs for this chip? My printer isn't large enough for a 300 as it takes 2 pages and I'm not so good with computers to where I can print it on an angle to fit. I have a bunch of these chips and power needs of all sorts of designs that are sonically pleasing would be great. How is JIms audio boards?
 

Arty

Member
2011-02-24 5:04 pm
it is not impossible to paralell them, but it would take a lot of effort, and would be quite useless, as to utilise its power you would need some pretty low impedance speakers.
simply not worth it.
if you need more power /i'd actually say for more than 50 watt into 8 ohm/ you better lean towards class D amps.
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
pa is the parallel chipamp configuration.

National have used two of these to create the BPA (bridged parallel amplifier).
You cannot bridge an already bridged amplifier.

Have a look at the app notes. National show all the details there.

Yea I read them in and out. How does bridging and paralleling give u 4x gain on 2 but not 6x on 3 or 8x on 4? Btw is that really the best way to run that chip? I read Modulous's (sorry met a lot of people) page and he really knows that piece in and out. I have to imagine other and better ways to run it.
 
It could be done but paralleling to BPA300's is not the proper way.

First start with two PA300's and then bridge them.
Do use some DC servo amps on each section as well.

I started on such a monster design a few years ago starting with a BPA400 using LM4780's, but other pressing projects got in the way.

It was when DIY'er Tomchr started his very informative threads, my initial goal was to do a BPA800 to drive my ESL's.

FWIW

jer :)


P,S. There are two post here with some links on all of the info that I could find on the subject at the time,

http://www.diyaudio.com/forums/chip-amps/249484-lm3886-electrostatic-speakers.html#post3785881

And this one mainly,

http://www.diyaudio.com/forums/chip-amps/166722-problem-parallel-lm3886-pa-100-a.html#post2188439
 
Last edited:

Arty

Member
2011-02-24 5:04 pm
Alway gotta have more cowbell man! What if it was for a sub or I load my speakers to 16 ohms? Sometimes people's needs and wants don't always align with what u need or want

you can parallel even 100 of them, with a 16 ohm load it will mkae no differece in the positive way, just degrade sound.
you could paralell bpa300 amps, witch would be dreafull to start with, but you won't get even a fleefart more if the load is not adjusted to suit it.
and btw, it would need a load with less impedance.

one you attempt paralelling 3,or4,or more chipamps, and then bridge them its a better solution, but still would be dreadfull as to utilise the extra current capacity you would still need a low impedance load.
witch makes kindof less sense than you think.
automotive subs can be configured to be 2 ohms, or even less, 1 ohms impedance, that could use this beef of current, but it will most likely sound horribly in a living room.

the problem is, ability to produce more output current does not alone yield any power. you need higher voltage, or a more demanding /low impedance/ load.
bpa amps achieve more output voltage by running in bridged configuration, hence have larger swing at the output voltage wise. now, higher voltage into the same load requires more current. that is why its paralelled, to have that extra current demand met.
any further paralelling is only reasonable if your load requires more current.
you can nt increase the voltage swing any more, as the chip can not handle more. once bridged, it can not be bridged again.

and there is your dinner.
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
you can parallel even 100 of them, with a 16 ohm load it will mkae no differece in the positive way, just degrade sound.
you could paralell bpa300 amps, witch would be dreafull to start with, but you won't get even a fleefart more if the load is not adjusted to suit it.
and btw, it would need a load with less impedance.

one you attempt paralelling 3,or4,or more chipamps, and then bridge them its a better solution, but still would be dreadfull as to utilise the extra current capacity you would still need a low impedance load.
witch makes kindof less sense than you think.
automotive subs can be configured to be 2 ohms, or even less, 1 ohms impedance, that could use this beef of current, but it will most likely sound horribly in a living room.

the problem is, ability to produce more output current does not alone yield any power. you need higher voltage, or a more demanding /low impedance/ load.
bpa amps achieve more output voltage by running in bridged configuration, hence have larger swing at the output voltage wise. now, higher voltage into the same load requires more current. that is why its paralelled, to have that extra current demand met.
any further paralelling is only reasonable if your load requires more current.
you can nt increase the voltage swing any more, as the chip can not handle more. once bridged, it can not be bridged again.

and there is your dinner.

So really the best way to run the chip is by itself? i do not know much about this device and that is why i asked. i need to make a ton of amps (power needs as small as 20w to as much as 500, class D isn't an option) and i have a bunch of these chips as well as the tdas and a ton of solid state stuff so i figured i could make a high wattage mono block out of the chips for the subs and call it day. it seems as if your saying regardless of current and number of chips, gain is not increasing and the sonics are consistently degrading?

one particular tower is a 3 way, active crossed all 5 drivers are 8 ohms and the sensitivity is very high for each (all in excess of 98dbspl). Looking at amp specs (all amps pretty much not just selected ones) thd+n seems to get worse the heavier the load and a difference in performance at 4 ohms vs 8. i would imagine the trend carrying over and increase the performance at 16 ohms and actually what many feel is the best full range driver in the world (Frestrex) is a 16 ohm speaker.
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
Use one 50W chipamp to drive each speaker driver.
98dB/W @ 1m with a 50W (+17dBW) into 8ohms amplifier gives you 115dB @ 1m capability (98dB +17dB)

I was just using that as an example - those drivers will be done with a solid state design.

I did come up with an array of 4 drivers that's perfectly flat from 68 to 20k without a crossover. Each driver is 91 db 8 ohms making the cab over all 100 db at 1 watt - interesting design and I don't think a full range design without a CO has been done before. I can swap them out for 4 or 16 ohms If needs be. As there is a total of 20 Of them I may take the last 4 for smaller stereo zone and seperate the box and will probably make this design in a number of different ways. As a system it forms sort of a quad line array and I'll have to play with the wires so that sound cancelation is splitting the room in half so anything hard right or left stays hard right or left. Considering the power needs (35/70w @ 8ohms) 3886 may be a perfect fit to bi amp or quad amp them.

Since you seem to know this chip better then anyone and my only experience is with a 6 x 6 mono which can certainly be upgraded and changed as component selection was less then stellar, only has Rcas, tolerances can be tighter w some resistors, and no real solid high grade caps were used in the ps or the design, I used a bridge rectifier instead of the fast acting slow recover schottsky diodes and it has Ir, clip indication discrete outs as well as bridging to 3 8 ohm channels (4 will give off magic smoke) and 12v triggers for each as well as mute. It is however 1 rack unit which is nice due to a large od and flat Plitron toroid. I have to check what the toroid is rated at again and I don't have any solid equipment to test thd, SNR, THD+n, etc however once my friend is done with a large project I can test it out. I plan to make as many speakers active as possible to cut down on speaker wire lengths it does however sound good, not superb. With a lack of knowledge in chips would the best you can hope for is speed and low distortion values like your designs as I don't see too many ways to capture a "sound". I did recently pick up a number of the lme4930s of whatever that chip was that can take 200 v as well as the tdas w the dmos so you can run additional transistors for more cowbell and gain. Long story short with the 3886 is the modulous as good as it gets and while it may not have a signature sound it seems like you are getting as close to the originally recorded content which is desireable in a lot of applications. How would you define the sound? Is it fat sounding or sterile? I am very open minded about new things and ideas - why does it seem that solid state guys even ones who are very talented dislike and almost skoff at silicon? I made a reference of possibly using the modulous on my tweeters due to the power meshing well with a smaller transformer and the clean tone and he was awefully adamant that chips couldn't give as good of a sound as a bipolar and he was very adamant about bipolars vs lfets which I tend to use for simplicity purposes as well as they seem to be optimized for audio while they do have limitations. I guess there is always a trade off but with my current skill set they seemed like the best fit and I have an lfet amp I purchased that i feel sounds stunning. Is it common for people to be so opinionated or do chips have their short comings even in a solid circuit?
 
"an array of 4 drivers that's perfectly flat from 68 to 20k without a crossover. Each driver is 91 db 8 ohms making the cab over all 100 db at 1 watt"

mmm i would stay with a more reasonable number, like 97 dBL.
Only if the four drivers all couple together over the whole frequency range and only if all four drivers have the same output as the others over the whole frequency range.
Otherwise he is just telling us lies.
As it is, stating 100dB was a lie.
 

Arty

Member
2011-02-24 5:04 pm
well.. sure in real life he can be happy to reach something like 95 ish dBL combined, but i did not want to go that drastic.
not to mention most possibly since it would not be a single source echo and radiation patterns will make sure that with more sources, you gona get a quite bumpy freqvency responce right to begin with. in an extreme case it may even produce.. less at some ranges than a single driver.
 

tomchr

Member
Paid Member
2009-02-11 12:58 am
Calgary
www.neurochrome.com
ok makes sense now. For high power application i would assume you are not a fan of this chip?

What would you consider "high power"? 1 kW? I'm not a fan of any chip amp at those power levels. I'd use a Class D amp.

The LM3886 in a composite amp works well for the 120 W (4 Ω) power level if you run two of them in parallel on ±35 V rails.
Getting two LM3886es to play nicely is a challenge and designing a composite amp is yet another challenge. I've solved both of these.

You can in theory get more power by bridging two parallel LM3886 amps, but in practice, you'll need eight LM3886es in a BPA configuration to be able to drive a 4 Ω*load to the rails. It may be possible to build an amp like that and have it perform well, but there are better ways to build 400 W amplifiers.

Tom
 

JaredC79

Banned
2015-09-02 4:32 am
NJ
"an array of 4 drivers that's perfectly flat from 68 to 20k without a crossover. Each driver is 91 db 8 ohms making the cab over all 100 db at 1 watt"

mmm i would stay with a more reasonable number, like 97 dBL.

i know it isn't precise but isn't the math roughly 3db additional sensitivity for every driver in an array? starting at 91, 2 would make it 94 and 2 more would be 6 additional DB. Is there something personal i should know about? I am asking questions to learn which seems to be the reason of the vast majority while other who really know this seem be doing everyone a favor by educating. If you aren't a fan of the bpa300 we get it and you made it clear - additional chips sound terrible. Is there an issue or is this just standard operating procedure? i would prefer to get it out in the open so no additional time is wasting with pettiness and we can all progress rather then bicker.
 
i know it isn't precise but isn't the math roughly 3db additional sensitivity for every driver in an array? starting at 91, 2 would make it 94 and 2 more would be 6 additional DB.
That's not quite right...
You get +3dB for each DOUBLING of the number of drivers, so it's only going from one driver to two drivers that you get +3dB output. This assumes that the same amount of power is being delivered to each driver no matter how many drivers you have.

So:
1 driver --> 2 drivers = +3dB
2 drivers --> 4 drivers = + another 3dB for a total of +6dB compared to a single driver
4 drivers --> 8 drivers = + another 3dB for a total of +9dB compared to a single driver