Simple Chip Amp for P to P wiring

[AndrewT]

Originally Posted by coldcathode

Hypothetical 32Watts Po would be 64/8 = 8 sqrt = 2.8 amps? So 1.4 from each side or is it peak of 2.8 from either side?

Yes, 2.8Aac into a 4r0 (resistive) load is equivalent to 32W average power.
This requires a maximum current of 4Apk into that same 4r0 load. i.e 4Apk from both supply rails alternately.

A 4ohm speaker can demand approximately three times the current of the nominal resistive load. So expect to design for 12Apk to drive your 4ohm speaker.
No National chipamp can meet this current demand. The highest current output chipamp (lm3886) is limited by specification to between 7A and something unspecified over 11.5A when the chipamp is cold.

I do not recommend a low sensitivity 4ohm speaker as the load for any chipamp. They cannot pass the transient current required for good quality sound at the louder SPLs.

Either, stick to medium sensitivity 8ohm speakers, or go discrete.

[coldcathode]

Andrew,

I am confused, the DataSheet for LM3886 specifically calls out a 40W/4Ohm load?

Now you are telling me it can't do this?

[AndrewT]

the National datasheet shows specifications for 4r0 and 8r0 loading, not real speaker loading.
They, like many other manufacturers, do not tell you what their equipment might not be good at. They emphasise what their equipment is good at.

[wintermute]

Hi Andrew, That would require an impedance of 2.3 Ohms.... Whilst I will concur than an impedance measurement of my MTM's when I put a zobel on them did have dips in the impedance curve almost that low, it is only at certain frequencies below and above the resonant freq of the box, and not across the entire freq spectrum...

The chances are the music source is not going to be exclusively in the freq range which just so happens to coincide with the minimum impedance of the speaker. Yes it is possible that there may be times when something might demand more current than the chip can deliver, but what are those chances..

ColdCathode, one thing is definite about chipamps.. When they are overdriven their clipping is not pleasant. If you have an impeadance curve of your speakers that may give you a better idea as to whether you would have a problem or not

Tony.

[danielwritesbac]

Any amount of watts along with a 4 ohm load and any amount of voltage will create either space heater or a lot of clipping. I have certainly done both with LM3886TF. The fortunate aspect of the 4 ohm load is less clipping.

There are several ways to cool it off, but there is only one with the extremely powerful dynamics necessary for a bass woofer amp to keep pace with a tube amp.

You need the PA100 design from National Semiconductor's AN1192.pdf and you can slightly alter the value of R-Out (page 8) to cool off the amp.

You also need an Antec transformer.

Here's some tolerances:
From 20 0 20 vac To 28 0 28 vac with a 20+ amper capacity bridge rectifier. -OR- From 20+20 vac To 28+28 vac with a pair of 10+ amper capacity bridge rectifiers.

From 3.5 ampers to 7 ampers transformer tolerance.

VA. . . the total transformer voltage multiplied by its amperage tolerance. Example 24+24vac * 4 ampers = 192VA, and that is a recommendable figure to use for a dual mono or monobloc. About 6a is good for stereo.

And, have a look at page 8. http://www.national.com/an/AN/AN-1192.pdf <click link

AndrewT may tell you that a larger size capacitance figure for the NFB cap will make for better quality bass. That is true. As a starting point, use a made-for-audio-power small canister Nichicon or Panasonic 220uF electrolytic // 22nf polyester.

For Pa100 bass amp use inexpensive 1&#37; metal film resistors throughout. If you happen to want a "resistor trick" then the resistor that is in series to the NFB cap may be 1/2w carbon 1% matched by ohmmeter.

Smaller caps for power at the amplifier board tend to make for less heat. For example 470uF//100nF or 220uF//220uF//100nF per each rail, per each chip. In this case, you'll need fairly short length power supply cables.

The sidenote in the CarlosFM power supply indicates a polyester capacitor (250v, sized from 2uF to 4.7uF) reaching from V+ to V- and whatever else that may do, it also makes the amp run cooler.

For power, look up the CarlosFM power supply as that can be helpful to a bass amp. See also Decibel Dungeon.

Use LM3886TF--the TF is conveniently pre-insulated. This is helpful to avoid voltage output upon the heatsink. For the "TF" Plastic to Metal thermal interface, it can be optimized with the filler properties of metallic thermal paste.

With the parallel amp, PA100, it is important that exactly the same signal goes into both chips. It is also important that DC offset for both is exactly the same (hopefully zero).

[coldcathode]

OK,

Now all of you have me more confused than ever. Let me try to explain EXACTLY what I am looking for and what I am NOT looking for also.

A: I do NOT need 100Watts, therefore I don't think that I need a parallel chip configuration.

B: ALL I need to understand is why the power supply for 40 Watts into an 8 ohm load needs to be so high? Using the math in the LM3886 datasheet indicates a supply voltage of 18V and 4.5Amps by my calc that is 81 WATTS so the amp is 50&#37; efficient?

C: Assuming all I want is 20 Watts into an 8 ohm load what do I do?

[danielwritesbac]

Originally Posted by coldcathode

OK,

Now all of you have me more confused than ever. Let me try to explain EXACTLY what I am looking for and what I am NOT looking for also.

A: I do NOT need 100Watts, therefore I don't think that I need a parallel chip configuration.

B: ALL I need to understand is why the power supply for 40 Watts into an 8 ohm load needs to be so high? Using the math in the LM3886 datasheet indicates a supply voltage of 18V and 4.5Amps by my calc that is 81 WATTS so the amp is 50&#37; efficient?
. . .

That all sounds quite logical if you were amplifying DC. However, you don't want to amplify DC.
You want to amplify AC, as in audio.
It is dynamic.

In order for a clean 20 watts of bass from a chipamp, you need to factor in dynamic headroom versus speaker efficiency. To keep up with your tube amp is beyond the capacity of the 200 watt amplifier listed in the AN1192 document.

You can cover up the ringing cries of anguish from a too small transformer if you use a capacitive multiplier circuit or a regulator in the power supply. Transformers ring like a bell, and the smaller the transformer, the higher the pitch. . . directly into the audio band. Its not all about amperage. Don't you want pretty bass?
Your power supply would be cheaper if your transformer is larger.

[Fenris]

B. 4 ohms is nominal, most drivers are lower at some point and also present a reactive load that is different from a 4 ohm resistor. Some posters just wanted to make sure you had anticipated the worst case scenarios. Also, don't forget the difference between DC and RMS power - that's where the other 50&#37; is hidden.

C. LM3886, preferably the TF insulated version. Very overkill, but we like overkill. 18-24vac, 50-100 VA transformer. 4-10,000uF or so of caps. Just my suggestion.

I've built dozens of chipamps, both bugsplat and PCB. I've moved on to tube amps, but still use chipamps where I think they're practical. My Home Theater uses 15 chipamps (5 channels, two woofers and one tweeter each) with an active crossover. Too expensive and power consuming to do practically with tubes and my relatively low effiency speakers.

[danielwritesbac]

Originally Posted by Fenris

. . . Some posters just wanted to make sure you had anticipated the worst case scenarios.. . . .

The worst case scenario, which was already illustrated, is this: Low resource power circuit makes retail subwoofer sound.

[Mark Allen]

I do not recommend a low sensitivity 4ohm speaker as the load for any chipamp. They cannot pass the transient current required for good quality sound at the louder SPLs.

Or since you are doing P to P wiring you could parallel two or three of the LM3886's for higher current capability and use a larger transformer.

Mark