Emic Amplification / Suggestions for (Simple) Audio Amp Chip

Zoot

I've got an Emic module with the onboard audio amp. The volume (with a speaker enclosure) is pretty good, but I want more volume.

I've got some L386 Audio Amps lying around, but according to the datasheets, these are actually *less* powerful than the 300mw onboard amp of the Emic.

So... does anyone have any suggestions for an audio amp chip that would be as easy to wire up as the LM386? Also, in looking over the spec for load/resistance/etc for use with the Aout of the Emic, I realize I'm a bit lost.... would I still need to condition the Aout signal for beyond a "typical application" for an audio amp?

Basically, I'm not sure what this means:

Somebody said...
Audio output buffer pin provides a single-ended analog output of the
synthesized speech for external amplification. Drives a minimum load of
5k Ω up to a maximum of 1Vp-p. The AC signal is superimposed on
approximately 1.2VDC bias and must be capacitively coupled to the
load.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Post Edited (Zoot) : 3/2/2007 3:09:55 AM GMT

Chris Savage

Zoot, the LM383 is a fairly small part count amplifier rated at 8W. I used to build these for Nintendo audio amplifiers. Take care.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Chris Savage
Parallax Tech Support

Zoot

That is a cool amp. I'm going to order up a few of those. Thanks. It also looks like I *can* drive that amp with 12v or so without distortion (albeit with reduced power out) -- datasheet shows typical Vs of 14.4 v -- my 'bot has a 12v battery which would be the Vs (max voltage at full charge from the battery is about 13.4 volts).

I just looked over datasheet for LM386-1 and realized that the mw output I was presuming is based on 5v-6v supply -- it looks like if I use 9v supply the LM386 will give me ~700mw out -- which would be a nice volume.

Either way, it looks like the input resistance for these amps is high enough that I can drive them with the Aout pin of the Emic -- but what value cap would I use for the "capacitive couping" -- I presume this means a series cap between the Aout pin and the amp circuit.

Of course, I'll have to wait for my Emic to be repaired (or replaced by me or Parallax) -- thanks for your help on that Chris.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Zoot

OK, hopefully someone with audio amp experience can help me out here, because I just don't know enough.

I built the LM386 Audio Amp circuit following Parallax's AP-8 amp circuit (see attachment below). I'm supplying the circuit with 9v, and I ran FREQOUT signals through it and it sounds GREAT. And LOUD.

Now, my real goal is to feed my FREQOUT into the AIN pin of my Emic, and then the AOUT of the Emic into my LM386 circuit. The onboard amp of the Emic isn't quite loud enough for me. I've already built the conditioning circuit to convert the FREQOUT signals into a form acceptable for the AIN pin, and this works great with the onboard amp (see add'l schematic below).

That said, I am confused about how to correctly modify the input portion of the LM386 circuit to accept the AOUT signal. The Emic manual states:

Somebody said...
Audio output buffer pin provides a single-ended analog output of the
synthesized speech for external amplification. Drives a minimum load of
5k Ω up to a maximum of 1Vp-p. The AC signal is superimposed on
approximately 1.2VDC bias and must be capacitively coupled to the
load.

Clearly the LM386 schematic has more than a 5k load, and is capacitively coupled (that means a series cap, right?). But it seems I can cut out the bypass caps and resistors since those are used for converting the FREQOUT to sine waves? But which ones? Do I just need a single series resistor and series cap before the volume control to get the right load and capacitance? If so, what values? Or am I totally off?

Thanks in advance.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Post Edited (Zoot) : 3/2/2007 2:46:55 AM GMT

Tracy Allen

Zoot,

My suggestion would be to to take Aout from the EMIC directly into the left side of the 10uF capactior on the '386 amplifier and dispense with the part of the ampllfier that comes before that. Use the circuit that goes to Ain on the EMIC as is.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com

Zoot

Dr. Allen -- thanks! Barring responses that is what I was going to try. I will wire this up sometime tonight and try it out.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Zoot

Wait, I just thought of something. If I take Aout to the left side of the 10uf cap, then don't I have less than a 5k load when the volume pot is turned more than half way towards loud?

???

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Zoot

I think I'm answering my own question here -- spec sheet for LM386 states an input resistance (Rin) of 50k -- presuming that covers me for minimum load without external components?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Zoot

FYI all -- I wired up the Emic and the LM386 as suggested above and it sounds great. Nice and loud, for both speech and FREQOUTs from the Stamp. I added a few extra caps on the input side to get rid of noise (I think this is coming from nearby PWM and RC electronics) and it sounds great. Thanks for all the help.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

SimEnzo

Thanks for the wiring diagrams... using them I was able to get my Emic working with a LM386.

The 10k ohm audio pot I got from Radioshack was causing LOT of line noise, so I tried substituting a 5k pot that I happened to have.. the 5k one works much better, but I have a couple of questions:

* What's the implication of using a 5k pot vs. a 10k pot?

* I've noticed that there's a slight high-pitched buzz that I can minimize by moving around wires or by touching the 220 uF capacitor... what sorts of things should I be looking at to eliminate the buzz in a more rigorous manner?

Thanks!

Zoot

The trim pots at Radio Shack are cheap -- I had a similar problem and eventually just used some old 10k volume pots from an old Walkman. In any case, the pots at radio shack aren't audio taper; they're linear.

Using a 10k would cut the adjustable range in half, plus you'll be sending more current through.

I had some hum -- probably due to a generally noisy robotic platform -- I put a few small value caps (.01 and .02) in parallel with the 10uf cap in front of the LM386's volume pot. Additionally I put noise suppression .1 caps on the power rails to the amp circuit which helped greatly.

Lastly, a breadboard version of the amp seemed much more susceptible to noise than the prototype board with the same circuit soldered.

But take these possible solutions with a grain of salt -- I'm certainly no audio amp expert

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Zoot

Oh, one other possibility to eliminate hum -- if you have a Stamp/SX/Prop pin hooked up to Ain on the Emic, make sure to put that pin into an input/hi-z state when you are done, or you'll get bleed into the audio signal.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

SimEnzo

Thanks for the suggestions! I'll give them a shot...

Zoot

Joe Fishback PM'ed me about precisely how I ended up amping the Emic (cheaply and easily), so I figured I'd post a few comments here as well.

I cut and pasted the two schematics together as I wired up the circuit. I will add that I ended up including decouping cap on the power rails to the amplifier which cut down on speaker "hum", but my 'bot has always been a pretty noisy platform.

Adjustment & Software: remember that Ain is UNAFFECTED by the serial volume commands that change the volume of Emic text-to-speech generation. All your FREQOUTs to Ain will be the same volume. So you want to "mix" the volumes with the pots so that your FREQOUTs aren't way louder then the loudest text-to-speech (vol = 6). Or however you like it.

I set the master amp volume so that the loudest speech sounded nice, then I adjusted the Ain volume so that FREQOUTs were a touch softer than the loudest speech. This gives me so leeway to make speech louder or softer compared to FREQOUTs.

I debated putting a digital pot on the Ain conditioner circuit which would let me change volume on the fly for FREQOUTs, but that's for another time. I also considered two amp channels -- one for Emic TTS and one for FREQOUTs -- which would allow TTS and sound simulatenously, and from different locations on the 'bot body, but I think I'll do that for a 'bot with an even louder amp

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST

Post Edited (Zoot) : 9/11/2007 4:48:05 PM GMT

Joe Fishback

Zoot,

Thanks for the quick reply. The latest diagram clears up all my questions. That is what I thought you all were talking about, but I just wanted to make sure. I did not want to let any "blue smoke" out of my unit. I am going to try the circuit today.

Thanks

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
······ Joe Fishback

-Robots are my friends-

Tim-M

Hi Zoot,

· In your latest posted schematic, it looks to me like the upper ground connection at the speaker·is a mistake...·that connection shorts the audio output directly to·ground.· If you look at the Parallax audio circuit, the upper ground connection is supplied for the external speaker option, as set by the jumper and maybe it was carried over to the new schematic unintentionally.

Tim

Post Edited (Tim-M) : 9/11/2007 2:16:21 PM GMT

Zoot

Tim -- yer right. I was a bit too quick in my cut n paste.

Corrected.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
When the going gets weird, the weird turn pro. -- HST