[soundNERD]

I finally got around to building the LM4881 amp (if anybody remembers I asked about it before). I didn't dare test it with my $100 headphones, so I just connected it to a cheap 2 inch 4 ohm speaker I bought for $2. I connected it to the 9V battery and using my RioVolt SP50's headphone out as the source, and barely got any sound. So, I tried a 6V lantern battery, and it worked and I got sound. But the sound was distorted and had no bass. Each amp, though, delivered the same sound, which I guess is a good thing. Another thing I noticed is that there is DC on the output, as the speaker pulles in about 1/8 inch if the speaker wires are + to + and - to -. If they are switched, the speaker moves out. Could this be because I don't have a output cap on the circuit? If it is needed, what would be a good value? Something around 200uf? I don't have one because when I tried a .5uf cap, it turned into a volume reducer and even larger bass cut. How do I not lose the bass from the CD player on the input cap and the output?

Can the LM4881 even run properly with a 4-ohm speaker? I didn't see anything below 8ohm in the datasheet. Could this be the reason for the distortion?

I guess I should say what circuit I am using. I attached the circuit. The entire thing is powered from a LM7805. The input cap is a non polorized .01uf cap, the input resistor is a 10K, and the feedback resistor is a 33 or 35K cap, I don't remember which. Also, I used a 1 uf cap on the bypass pin. Finally, the shutdown pin is connected directly to the gnd, no connection through a resistor to the + supply.

Thank you for helping me, Mike

 

[soundNERD]

heres the circuit

 

[jeffreyj]

Quote:

Originally posted by soundNERD I finally got around to building the LM4881 amp (if anybody remembers I asked about it before). I didn't dare test it with my $100 headphones ... Another thing I noticed is that there is DC on the output, as the speaker pulles in about 1/8 inch if the speaker wires are + to + and - to -. If they are switched, the speaker moves out. Could this be because I don't have a output cap on the circuit? ...

It's a good thing you didn't test it with your $100 cans, because your cavalier omission of the output capacitors would have definitely been bad news for them. This type of amplifier (ie - one that runs off a single supply voltage referenced to ground) internally biases its inputs to half the supply voltage. This means that its outputs are also sitting at half the supply voltage when quiescent. DC blocking capacitors are required at the input and output terminals of such an amplifier; they are not optional as they are on amplifiers powered by dual polarity (or "split") supplies.

How much capacitance is needed is determined by a simple equation which I've personally posted here about a dozen times, but what the heck, here goes again:

C = 1 / (6.28 * R * Fc)

Where C is in Farads, R is the load impedance in Ohms and Fc is the "cutoff frequency" in Hertz (which is the point at which the voltage level will be reduced by 3dB). The farad is an inconveniently large unit so multiply by 10^6 to get the answer in uF. As a general rule, for high fidelity applications you want to set Fc at one-tenth the lowest frequency of interest, or 2Hz in for audio. For mid-fi applications, an Fc of 20Hz is typical. As you will see from the worked example I will give you, it is no small task to achieve high fidelity reproduction with a single-supply amplifier:

Fc = 2 Hz
R = 32 ohms (eg - Grados)

C = 1 / (6.28 * 32 * 2)
C = 0.002488 F
C = 2488 uF

Normally, one would choose the next highest standard capacitance value (getting a lower Fc in return) but that might be impractical given the high value of capacitance required for this application. So, 2200uF or 3300uF are your practical choices here. You can, of course, rearrange the equation algebraically to determine what the Fc will be for a given capacitance and impendance:

Fc = 1 / (6.28 * R * C)

2200uF gives an Fc of 2.26Hz while 3300uF gives an Fc of 1.51Hz. My opinion? Go with 2200uF. Oh, and your only practical choice of capacitor will be aluminum electrolytic.. don't even think you'll be able to get enough capacitance to matter using film type capacitors.

 

[soundNERD]

Thanks for the detailed explanation.

I just have one more question. Why am I losing all of the bass, even without a output cap? Is it because of the .01 uf cap on the input? If so, what would be an acceptable value?

Also, any ideas on the distortion? Could the cause be the 4-ohm speaker? Or just not having the caps?

Also, what would happen if I used a 1000 uf cap instead? I don't really have a 2200 cap small enough to put on the board.

Thanks again for the help, Mike

 

[jeffreyj]

Quote:

Originally posted by soundNERD Thanks for the detailed explanation. I just have one more question. Why am I losing all of the bass, even without a output cap? Is it because of the .01 uf cap on the input? If so, what would be an acceptable value? Also, any ideas on the distortion? Could the cause be the 4-ohm speaker? Or just not having the caps?

This is why you shouldn't re-engineer another engineer's design if you aren't an engineer!

The 0.01uF input capacitor is too low by at least two orders of magnitude, yes, but the real problem is the DC offset moving the cone away from its normal resting point. So, to answer your two questions as to why the sound is quieter and distorted in the bass, consider the following:

1) the farther from center the cones moves the harder it is to move it even farther;

2) the lower the frequency, the farther the cone must move to reproduce it;

3) there are hard limits to how far the cone can move and when it hits those limits, it comes to an abrupt stop.

Quote:

Also, what would happen if I used a 1000 uf cap instead? I don't really have a 2200 cap small enough to put on the board.

The formula for determining that is in my previous post....

 

[Ajit]

I am no expert, but I recognized the circuit from the LM4881N Application Notes at the National Semi website.

What package are you using? The circuit you posted has the pin outs for the MSOP package. The SOP and DIP packages have a different pin out, like so:

MSOP SOP/DIP
Bypass 1 3
Ground 2 4
Shutdown 3 5
Vin B 4 6
Out B 5 7
Vdd 6 8
Out A 7 1
Vin A 8 2

You probably figured that out anyway, or else I guess your amp would not be working at all.

The Application Information Section of the same document deals exhaustively with the proper selection of input and output coupling capacitors, with examples ( as I recall ).

Once again, my electronic theory is bad, but when I built it a long time ago, the values I used were:

Ci, Co = 1000uF
Rf = 100K audio taper
Ri = 18K
( Gain = Rf/Ri )
Cb = 1uF
Cs = 1000uF ( since this is a filtering capacitor )

These values worked for me. Like I said, my theory is bad and someone more familiar with it like Jeff, could explain it in greater detail. In the meanwhile, the pin outs and the values I used does work. My amp worked off 2 x AA cells by the way.

Hope this helps....

 

[Ajit]

Jeez, my posting is just as bad as my theory.....

In the table I tried to put together, the pin function is followed by the MSOP package pin number ( the first number) and then the SOP/DIP package pin number ( the second number ).

Thus the Bypass pin is Pin 1 in the MSOP package, but is Pin 3 in the SOP/DIP packages.

I hope I have not confused you even more!

 

[soundNERD]

Yes, I am using the DIP package. I did notice the pinouts were for the other package, and built it connecting things to the correct pins. Will using the .33uf cap like suggested in the datasheet cut out the bass also? If so, I will just use the 1000uf caps like Ajit suggested.

Ajit, I just want to know why you have tapers for the gain adj.? Is there any special advantages to this or something? The problem I see is getting both channels equal. That also makes me wonder, how many decibel gain do I get now with the 35K resistors?

JeffreyJ, about the distortion, I forgot to say that I did notice the speaker acting strange. Whenever there is a bass sound, it would move out about an 1/8 inch for the duration of the tone, then come back in after it is done, which explains the distortion even more.

Also, Does the input resistor I used make a difference? I noiced that when I switched only 1 of them from 35K to 10K, since the 10K fit the board better, it didn't affect the sound. I am going to switch the other, but just wanted to see what difference it makes.

If it matters for selecting the caps, my headphones respond from 16hz to 20,000hz, according to the package and have a 100mW handeling. They are 32 ohm impedence (just wondering, why are headphones such a high resistance?)

Thanks for the help before and if anybody can answer any of my questions in this post

-Mike

 

[Ajit]

I used the audio taper pot as it was the only one I had lying around and it worked fine. I used it as a variable resistance, with only one end and the wiper wired. Also, pin 5 has to be tied down to the ground or else the chip goes in to shut down mode. Obviously, you have done this...

 

[jeffreyj]

Quote:

Originally posted by soundNERD Yes, I am using the DIP package. I did notice the pinouts were for the other package, and built it connecting things to the correct pins. Will using the .33uf cap like suggested in the datasheet cut out the bass also? If so, I will just use the 1000uf caps like Ajit suggested.

Mike, are you scared of a little algebra or what? The equations I posted earlier will answer this question; just plug in the numbers!?! I'll work the equation out for you once again mainly because I am killing time until a store opens...

The impedance looking into the amplifier's input is simply the input resistor value, which is 20k in the schematic shown. The frequency at which the volume will be cut in half, then, equals 1/(6.28*(0.33*10^-6)*20000), or 24Hz. This is too high of a cutoff frequency, but even worse is the 0.01uF you used, which gives a cutoff of 796Hz! No wonder bass response sucks - you've got the cone kicked out to the end of it's travel and you've strangled the input with a 0.01uF capacitor!

Try using at least a 1uF capacitor on the input, but preferably 10uF. The two main reasons you want the cutoff frequeny much lower than 20Hz are: a) so that phase shift at 20Hz is minimal; and b) so that volume loss at 20Hz is also minimal. Phase shift at Fc is 45 degrees and volume loss is 3dB - both will be perceptible.



Quote:

Ajit, I just want to know why you have tapers for the gain adj.? Is there any special advantages to this or something? The problem I see is getting both channels equal. That also makes me wonder, how many decibel gain do I get now with the 35K resistors?

Do not use a variable resistor in place of Ri (20k in the schematic, just like Rf) to change the volume! This is poor practice. What you do is set the amplifier to the highest gain you need then use a volume pot (audio taper, dual ganged) to attenuate the input signal. In other words, put a proper volume pot in front of the input capacitor. Radio Shack sells a fairly decent 100k dual gang audio taper volume control for something like $3.00.

Now it's quiz time: what will be the cutoff frequency of the amp if you use 1000uF output caps with your 32 ohm cans?

 

[soundNERD]

No, Im not. I just got a little confused with changing the decimals to a cap value. anotherwords, how you got 2200 from .00002200.
never mind, though, Ill figure it out.

Thanks

 

[PeterR]

Quote:

Originally posted by soundNERD anotherwords, how you got 2200 from .00002200. never mind, though, Ill figure it out.

That would be 0.0022

the µ in µF (or uF) stands for 'micro'. 1µF= (1/1,000,000) F = 0.000001 F.

 

[soundNERD]

I tried it now with the headphones, and it sounds pretty good. There is still some bass not there that is there in the output, but other than that, it works pretty good. The only thing I noticed is that the speaker moves out again about 1/8 inch then back to the normal position when it gets power. It doesn't seem to affect anything though. Also, I get a loud, annoying pop in the headphone at loud volumes. I have to check, but I don't seem to get it with just the speaker. Any ideas on the cause?

Thanks again, Mike

 

[soundNERD]

Just thinking, could the gain be too high? since the schematic shows a 20k resistor and I am using a 35K?

 

[soundNERD]

One more thought... the supply has only a .5 mfd cap (in place of the .1 in the schematic) and no other filter capacitance. Does it matter since my source is a 9V battery? I don't know how well the rejection is on this amp.

-Mike