High-Quality Output Buffer for Battery-Powered Audio Circuit | Impedance: 50Ω

[axemaster]

Hi guys, I'm currently designing a battery powered audio circuit. I've already finished all the signal processing stuff, and now I'm looking for a suitable output buffer to drive the speaker, which is about 50Ω impedance.

Some of the requirements for the op amp buffer:

- Single supply operation with Vss ≥ 3V
- Rail to rail output
- Low distortion / noise
- Unity gain stable
- Output current of at least 100mA

I have already experimented with the BUF634, and it works, but at $8 it is just too expensive. The LME49726 might be better as it's only $1.50, but I can't find a PSpice or LTSpice model for it.

So I'm hoping that someone here will either 1) know how to get the PSpice model for the LME49726, or 2) know of a better op amp.

Thanks for helping me out!

 

[yungman]

Why do you need to simulate the circuit? Get one of them and hook it up! It said it's unity gain stable. Only thing missing is the stability when driving a speaker that is inductive and has a resonance peak. But that you cannot simulate anyway.

Obviously you are not looking for hifi audiophile quality, so as long as it is stable with the load that you use, it's all good. Just do it!

 

[jasonRF]

google is your friend - spice model at the ti web site:

http://www.ti.com/product/lme49726

EDIT: it is for the TI spice application, which works fine in my experience. You probably already found this! Sorry, I didn't pay enough attention to your initial post.

This looks like a nice op amp for your application. Note that it is surface mount; if you are looking for through-hole you will have to look elsewhere. One example (not as nice as the part you found) is the ts922, which is a cheap low voltage dual rail-to-rail op amp; you could parallel the outputs to get 160mA short circuit output current.

jason

 

[vk6kro]

If the output current is to be 100 mA (presumably RMS) then the peak current will be 141 mA for a sinewave and this would produce a drop of 7.07 volts across the speaker. (141 mA * 50 ohms). On the negative peak there would be -7.07 volts across the speaker.

So you would be talking about an output signal of 14.14 volt peak-to-peak and this would probably have to come from a power supply of about 16 volts.

So, it would not be easy to do this with a supply of only 3 volts. You may be able to use a step-up transformer for the audio or you may be able to produce 16 volts DC from a 3 volt supply

 

[f95toli]

If the output current is to be 100 mA (presumably RMS) then the peak current will be 141 mA for a sinewave and this would produce a drop of 7.07 volts across the speaker. (141 mA * 50 ohms). On the negative peak there would be -7.07 volts across the speaker.

Speakers tend have 4, 8 or 16 ohm impedance. 8*141=1.128V which is doable. Is there a reason for usign a 50 ohm speaker?
Even headphones can have impedances as low as 20 ohms or so.

 

[yungman]

I think the OP can choose the supply voltage and speak to get the desired loudness.

The point is I don't think simulation help much in this case unless you can have the model of the speaker, short of that, don't get hung on the Spice simulation, just cut and try. Most important of it all, you can't judge the sound quality by simulation.

 

[axemaster]

Hmm... on a side note, I was wondering - I was hoping to use a higher impedance speaker, i.e. in the 50+ ohm range. A few things then:

-Is there a reliable relationship between speaker impedance and loudness? The speaker I have right now is 8Ω, but I found "balanced magnetic armature" speakers with 150Ω that claim to actually be louder.

-Can speakers be simulated as a resistor and inductor in series? If so, how do I find the info to set the values for them. (i.e. impedance is given, but not the breakdown)

Finally, my goal here is to find a speaker with good frequency response in the 200-3000 Hz range. Moreover it needs to be very small. The only speakers I have found so far are the balanced magnetic armature speakers made by Knowles. Are there any other speakers I haven't found yet that can do this?

Examples of Knowles speakers:
http://www.digikey.com/product-detail/en/CM-23152-000/423-1036-ND/458549
http://www.digikey.com/product-detail/en/SR6438NWS-000/423-1142-ND/2686685

 

[yungman]

I am not familiar with these small speakers, but on the larger ones, you cannot simulate with a resistor and inductor, that would be too easy! It is very complex, it has resonance peak. It is affected by the environment also. Think of it like a motor, when it run and there is no resistance, it draw very little current. but if you put resistance to the motion, the current drawn goes up. This is due to back EMF of something.

Case in point, I have a power attenuator for guitar amp that reduce the output. The design is quite complicate to simulate the speaker so it would not affect the sound of the amp. The major part of the challenge of hifi amplifiers is how it react with the speaker, that's the reason THD and all the spec don't mean anything. Take a look at the spec of some dirt cheap Onkyo or Pineer amp vs the audiophile amps, the spec looks similar, but when you listen to them, it is day and night difference.

That's the reason I said that it's no point of simulation unless you know the characteristics of the speaker. And even at that, simulate for what?! You listen to it and be the judge whether it's good enough!

I don't think impedance has anything to do with how loud the speaker with given input. It's is govern by the power AND more important of it all, the efficiency of the speaker.

But, like Vk6kro explained so well, if you require power, don't use too high impedance of a speaker if you cannot raise the supply voltage.

 

[maimonides]

The speakers you linked to are for headphone/earphone use, forget about them. (The db numbers don´t mean the same as in normal speakers).
You need size for a decent midrange, don´t go below 30mm diameter if you can. (A cordless phone phone I repaired a few weeks ago has a 36mm speaker for "hands free" operation)
Check http://www.visaton.de/en/industrie/miniatur/index.html and you will find similar models by other manufacturers if you look for them.
[Edit: Click the Art. No to see specifications and frequency response]

And when you compare technical data by different manufacturers, always check under which conditions the loudness in db was measured.

 

[axemaster]

[q]The speakers you linked to are for headphone/earphone use, forget about them. (The db numbers don´t mean the same as in normal speakers).[/q]

So what is the difference then? I have noticed that different companies appear to measure dB in different ways, and I find it very confusing...

 

[maimonides]

A headphone/earphone driver needs to produce sound pressure in the ear only, so it is tested on an ear simulator (a small, short tube, you can find the details in the datasheets in the links you gave).
A "normal speaker" is expected to produce sound everywhere around it. Apparently this needs more energy (resp. gives less sound pressure at the same energy input)
There is a quasi standard for "freefield" speaker specification (input 1W, distance 1m), but some miniature speakers are specified at 1W/10cm or something else, and just to be sure you better check the 0db sound pressure (usually it´s all in the small print in the datasheet).
I´ve looked up a US distributor for you:
www.alliedelec.com/search/searchres...s:RetailPrice|101|1|,N:106&SearchType=0&ps=60.

The speakers are cheap, just order a selection and test, like yungman suggested.

 

[axemaster]

Most of the specs are given referenced to 0dB at 20uPa at 10cm.

First of all, am I correct to convert this to -20dB at 20uPa at 1m?

Also, is this the usual dB definition? I.e. I usually see "people talking" as about 60dB, so are these on the same scale? Or to put it a bit differently, if I set the speaker to 60dB, would it be as loud as I am when talking?

Sorry if I'm being a bit difficult, I'm just trying to understand how to do the conversion... My goal is to use a very small speaker to reproduce normal human speech at full volume, at least in the 200-3000 Hz range.

 

[yungman]

I really think you over think everything. You spent quite a few days where you could have just build it in a few hours and you would have got all the answers.

I am no expert on speakers. I do know enough that speakers do have directivity as an important variable. You really want to get into all these? dB measuring at a distance do not give info on directivity. A highly directional speaker will be a lot louder at one direction and fade fast in other direction, remember the blow horn? Too many variables. Just get the speaker and the IC and build it!

I said this over and over because that is a big problem with some beginning engineers. They think, and they think, and then they think some more. They end up to be so slow and low in productivity. They hurt their career like that. The art of electronics is knowing when to dig deep into the theory, AND WHEN NOT TO. You are not designing an audiophile amp, the components you choose are for small portable AM radio quality at best. Remember, people look for an engineer, they look for result. They will hire someone that can produce a working design by card reading over someone that keep talking about theory and can't get things done! Believe me, I've seen people that that talk theory and you can't argue with them, but they can't get the job done, they got fired!

You want to dig deep into theory, choose something more involved, something like RF wireless, PLL control system, DSP....

Lastly, it you really want to dig into audio, you should start with a stereo amplifier, then go to a forum specialize in hifi audiophile. It is a different world. I design music electronics for guitars including tube guitar amp. I can tell you simulation is next to useless as human ears are so so much more sensitive than any instrument can read. Simulation almost tell you nothing other than fundamental operation of the amp. The more I get into it, the more I found I have to trust my feelings! There are so much trial and error going that circuits of some first class amps looks laughable.

 

[maimonides]

In theory you are correct with a 20db conversion factor.
But it applies only for a point source, far field conditions and negligible influence of the measuring microphone, and you cannot assume all of that. You can only use the 20db as a sanity check.
Yungman is essentially right: you are overthinking this. Specified differences in loudness of 2 or 3 db simply do not matter to you. The problem you want to solve has been solved (in so many cordless and mobile phones), so use one of the available solutions. A larger speaker will be louder than a smaller, so use the biggest one you can fit.
What do you mean by "full volume"? Quiet conversation? An opera singer at fff? (And with 200Hz you´re relying on residual hearing for the fundamental of deep male voices anyway)

And you should think about using a bridge amplifier, which will give four times the power from the same supply voltage into the same load. (Sorry, I can´t recommend an IC)

And sorry, I was a bit paranoid about the reference level for 0db SPL.
The usual db definition for "normal" speakers is 1W 1m re 20µPa. I´d consider the 10cm variant only appropriate for speakers that are actually used 10cm from your ear, which may be the intended application of many of the miniature speakers. Otherwise it´s to make the numbers apperar more impressive.

Edit: Please insert "speakerphone" where I wrote "hands free". Sorry.

 

[jim hardy]

try a LM386 audio amp it's inexoensive enough to experiment and quite versatile.

old jim

 

[Enthalpy]

+1. A complete audio amplifier is better than a buffer.

old Marc