Powering bridged 3886s?

[unitgain]

Hi everyone! I have a question, please forgive me if it's stupid.

I have four LM3886s and I want to make a little amp to power a subwofer (one channel). The speaker is the Shiva, it has two voice coils 8 Ohm each, so I can run with the load impedance of 4, 8, or 16 Ohm.

However, the only power transformer I currently have is 2x50V DC after rectifying and filtering. This doesn't go very well together with 3886s! Unfortunately, power toroids are few and far between where I live. There's a little chance I'll find something more convenient.

I can do one of the following:

1. Regulate the excessive 8 to 10 volt away, run into 16 Ohm load (8 Ohm per side).
2. Parallel the secondaries to get 1x50V, run into 8 Ohm load (4 Ohm per side). Since it's a bridged configuration, I don't need the centre tap (I think). I can get the ground reference with two resistors.
3. Ditch the idea to use the chips.
4. Try to get a different trafo anyway
5. Anything else?

Also, will two chips be enough or do I need to use all four (in parallel)?

Any idea or advice will be greatly appreciated!

 

[analog_sa]

I like 2. Seems like a good idea and at least i can't see any reason to shoot it down.

 

[leadbelly]

I can do one of the following:

1. Regulate the excessive 8 to 10 volt away, run into 16 Ohm load (8 Ohm per side).
2. Parallel the secondaries to get 1x50V, run into 8 Ohm load (4 Ohm per side). Since it's a bridged configuration, I don't need the centre tap (I think). I can get the ground reference with two resistors.
3. Ditch the idea to use the chips.
4. Try to get a different trafo anyway
5. Anything else?

Also, will two chips be enough or do I need to use all four (in parallel)?

Well, FWIW, my opinion, in order of preference:

4
2
5: parallel the secondaries, build the single-supply circuit version in the National datasheet times 2, use a DRV134 to drive the inputs

As for whether 2 or 4 chips, only you can answer whether your transformer can power them and you are willing to source 0.1% resistors, etc. It is much more work, but the driver can handle the power.

 

[ve2lui]

A 3886 has a better performance at higher impedance.
I made much experimentation and comparision with
a tube amplifier and I learned much from this.
I would use a 16 ohms load, inverted amplifier.
You may get less than 0.00001% THD!
More, I would try a small value resistor in series at amp output, like one or two ohms.
How it sounds? It does not sound! It is a "perfect amplifier".
I am pretty sure that any difference, at such a perfect performance level is from
interaction of speakers with what is offered to them by the amplifier.
I have built a very precise tube amplifier and after tweaking to have the same "presentation" to the speakers, it sounded exactly
the same, using a switch to A / B compare. No one could see a difference. Without the "tweaking" there was one for sure!

 

[unitgain]

leadbelly,

I'm trying to source some toroids but without much luck.

The one that I have is 500VA which I think is enough to power four chips, and then some. As for 0.1% resistors I don't think I need them, because if I decide to use four chips I can run two independent channels, each to its own voice coil. So there's no need to parallel them.

 

[leadbelly]

The one that I have is 500VA which I think is enough to power four chips, and then some. As for 0.1% resistors I don't think I need them, because if I decide to use four chips I can run two independent channels, each to its own voice coil. So there's no need to parallel them.

Actually, 500VA is only marginal for 4 chips since the general guideline is 150VA per chip, but you should still be OK.

Yes, when you first said parallel, I assumed you meant paralleling chips, and not paralleling amps.

You may just want to try yet another option: just build 2 of the single-supply version circuits, and run 1 to each voice coil. (1) low risk, sure to work (2) if you're not satisfied with the power, you can build 2 more of the same circuit, and put 2 DRV134's in front of these 4 chips and run them bridged, without having to undo the original 2 circuits

 

[daatkins]

Another alternative is to un-wind some of the turns from the secondary of the transformer - that would lower the voltage.

Nice one,
David.

 

[unitgain]

Actually, 500VA is only marginal for 4 chips since the general guideline is 150VA per chip, but you should still be OK.

Hm... This App Note says one 385VA transformer per channel in the BPA200 amp. It is my understanding that they mean a stereo amp and each trafo powers four chips, am I wrong?

 

[leadbelly]

Hm... This App Note says one 385VA transformer per channel in the BPA200 amp. It is my understanding that they mean a stereo amp and each trafo powers four chips, am I wrong?

National's app notes are written for the audience of a designer developing a commercial product with typical economic constraints, in this case they are speaking of the example of meeting the spec of 200W into 8ohms for a commercial product, so they spec a 385VA xfrmr and 40000uF per rail. The GC experience on this forum is to go bigger on xfrmr and smaller on caps to get best performance.

 

[unitgain]

Another alternative is to un-wind some of the turns from the secondary of the

David, have you done such a thing? I'm a bit afraid, the winding looks uniform and symmetric, what will happen if I break the symmetry? And what will happen with the rated power? I'll need to remove about 20% of the winding!