Lm3886 with electrostatic speakers

[beanbag]

Hello folks,

I built an LM3886 amp about a decade ago which I have been using to drive a Martin Logan electrostat speaker for a few years now. It seems to be working fine. I may upgrade to a larger speaker, which has a lower impedance. If the amp doesn't work well (can't output enough current), then is the solution as simple as lowering the supply voltage a little bit and paralleling in another unit?

If I recall, my amp has an inductor paralleled with a small resistor to protect it a little from capacitive loads, plus a zobel network.

One of my concerns is if I use .1 Ohm paralleling resistors, doesn't that increase the output impedance by .05 ohms?

The current speaker I have claims 6 Ohms dropping to 2 ohms at 20 kHz, and the (potential) new one would be 4 ohms dropping to 1.2 ohms. Martin Logan keeps claiming that their speakers should work fine with amps that are rated for 8 ohms.

 

[AndrewT]

The output stage dissipates a lot more heat when the load is severely reactive.
Protections in 3886 may trigger at quite low levels of output when attempting to drive an electrostatic.

Adding series output resistor helps with stability.
A paralleled 3886 needs these resistors for balancing of output, so they achieve a double purpose.

But a many paralleled 3886 may be required and a build may be the only way of determining what can be achieved.

 

[geraldfryjr]

I had a similar discussion a while back with a fellow DIYer and he said that ha had a PA100 driving his panels just fine.
I don't recall the size of his panels or what step-up transformers he is using.

Personally, I would consider a 3 or 4 of them in parallel on a +/-30V to +/-35v supply to be good for reliability and still get a good voltage swing and have enough current for such low impedance reactive loads down into the 1 to 2 ohm range.

I now finally have enough chips to build such a beast and is on my things to do list for this summer.

The amp I have in mind will employ a DCservo for each LM3886's and a nested feedback loop ( compounded opamp design) setup like the My_Reference design.

I had found a design that does this with two LM3886's in parallel and all you have to do is add one or two more paralleled stages.

It used a LME49720 (I think may be it was LME49740's)for the first stages as any good High quality opamp will work.

I have some LME49860's and LME49740's that I may be using.
I haven't decided yet as there are so many to choose from.

It all depends on how well their Vos matches up for the DC servo sections.

FWIW

jer

 

[beanbag]

The amp I have in mind will employ a DCservo for each LM3886's and a nested feedback loop ( compounded opamp design) setup like the My_Reference design.

I had found a design that does this with two LM3886's in parallel and all you have to do is add one or two more paralleled stages.

It used a LME49720 (I think may be it was LME49740's)for the first stages as any good High quality opamp will work.

link plz

 

[geraldfryjr]

I don't have the URL hand for the circuit that I had mentioned but I will search for it.
I did print off the schematic and when I find that sheet maybe the URL will be on it.

Meanwhile I have been also thinking about a Paralleled TDA7293 version as well were one chip acts as a master and the rest are slaves.

http://www.diyaudio.com/forums/chip-amps/197398-tda7293-single-bridge-parallel.html#post2729693

Here is a forum that shows a 3 and 4 chip configuration of a TDA7293

VNAV • Xem ch? ?? - TDA7293 parallel

Here is another great thread of DC servo PA100 designs,

http://www.diyaudio.com/forums/chip-amps/103308-pa100-dc-servo.html#post1229583

There are many very good threads on this type of design in this forum, but when I find the one that I am looking for I will post the link for you.

This thread is the one that really sparked my interest,

http://www.diyaudio.com/forums/chip-amps/166722-problem-parallel-lm3886-pa-100-a.html#post2183632

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

There is another thread that shows a very large DCservo paralleled amplifier of the design in the link above and I believe that it was stated that as many as 8 or 16 devices had been tested!


jer

 

[geraldfryjr]

Here is an example of a Composite input stage.

Sorry, I called it compounded design and for some reason that name always seems to stick in my mind.

But the My _Ref designs are based on this concept,

http://www.diyaudio.com/forums/chip...ntation-amp-error-correction.html#post3506072

http://www.diyaudio.com/forums/chip-amps/138762-lm1875-nested-gainclone.html#post1816178

Here is the original one that was published in Radio Electronics back in the early 90's

http://www.diyaudio.com/forums/chip-amps/197384-composite-amplifiers.html#post2725392

They also had a BPA version in the article that was a 70watt amp using 4 x LM1875's, it is very hard to find that schematic.
I still have the original magazine somewhere.

Here Is a BPA200 based on similar techniques,

http://www.diyaudio.com/forums/chip-amps/131216-wendy200-gainclone-based-bpa200.html#post1630270

jer

P.S. Here is the start of the sort of thing I am looking for,

http://www.diyaudio.com/forums/chip-amps/140966-gain-clown-composite-amp.html#post2175950

and,

http://www.diyaudio.com/forums/chip-amps/138762-lm1875-nested-gainclone.html#post1747778