Helped needed for a small chipamp

[wavycertified]

Hey everyone i'm new to the forum. please correct me if i'm wrong. I'm building a small 4W lm2877 based amplifier and need some help. to eliminate output cap, i decide to use +/- supply configuration as shown in the datasheet here: http://www.national.com/ds/LM/LM2877.pdf first off lm2877 has an open loop impedance of 4Mohms, does that mean i don't have to add a buffer in front of the amplifier since its input impedance is so high, which is equivalent to a buffer? also, from the schematic, there was only one 0.1uf cap in the positive rail, should i also add one to the negative rail? last question, what value should i use for coupling cap and feedback cap? i calculated the value for feedback cap and its around 16hz at -3db point for 5uf. furthermore why was the feed back capacitor reduced to 5uf instead of 10uf in the single supply configuration?

thank you guys so much in advance for any input.

[sregor]

Originally Posted by wavycertified

Hey everyone i'm new to the forum. please correct me if i'm wrong. I'm building a small 4W lm2877 based amplifier and need some help. to eliminate output cap, i decide to use +/- supply configuration as shown in the datasheet here: http://www.national.com/ds/LM/LM2877.pdf first off lm2877 has an open loop impedance of 4Mohms, does that mean i don't have to add a buffer in front of the amplifier since its input impedance is so high, which is equivalent to a buffer? also, from the schematic, there was only one 0.1uf cap in the positive rail, should i also add one to the negative rail? last question, what value should i use for coupling cap and feedback cap? i calculated the value for feedback cap and its around 16hz at -3db point for 5uf. furthermore why was the feed back capacitor reduced to 5uf instead of 10uf in the single supply configuration?

thank you guys so much in advance for any input.

Input impedance will be about the resistance between the + input and ground - in the national sheet it is the value of the volume control - 100k ohms. So a buffer shouldn't be necessary. Definitely had a cap to the negative rail (in addition I would add at least 100 uF cap on the board as close the the IC as possible although it negates KISS)

input coupling cap - I uses as large a film cap as I can find - with the values given both input and feedback circuits have -3 db points of 16 hz - I would probably try to make at least one of them a lot lower - depends on speaker extension. (this would definitely be a try it and see if it makes a difference experiment.)

The split supply actually has a lower cutoff frequency than the single supply because the feed back resistor which determines cutoff is 2K instead of 510 ohms. The gains are also different (50 versus 200) and increasing the value will help the low end in the split supply, but not in the single supply as it should be limited by the output coupling cap. So, making it bigger may help if the speakers have enough low end to make it noticeable. YMMV.

[wavycertified]

Originally Posted by sregor

Input impedance will be about the resistance between the + input and ground - in the national sheet it is the value of the volume control - 100k ohms. So a buffer shouldn't be necessary. Definitely had a cap to the negative rail (in addition I would add at least 100 uF cap on the board as close the the IC as possible although it negates KISS)

input coupling cap - I uses as large a film cap as I can find - with the values given both input and feedback circuits have -3 db points of 16 hz - I would probably try to make at least one of them a lot lower - depends on speaker extension. (this would definitely be a try it and see if it makes a difference experiment.)

The split supply actually has a lower cutoff frequency than the single supply because the feed back resistor which determines cutoff is 2K instead of 510 ohms. The gains are also different (50 versus 200) and increasing the value will help the low end in the split supply, but not in the single supply as it should be limited by the output coupling cap. So, making it bigger may help if the speakers have enough low end to make it noticeable. YMMV.

Thank you for your suggestion i will add 100uf cap to both rails. I also did a bit of research on the forums, seems like the common value for input coupling/feedback capacitor are 4.7uf/47uf and 2.2uf/22uf. What do you recommend?

[sregor]

Originally Posted by wavycertified

Thank you for your suggestion i will add 100uf cap to both rails. I also did a bit of research on the forums, seems like the common value for input coupling/feedback capacitor are 4.7uf/47uf and 2.2uf/22uf. What do you recommend?

For this application it shouldn't make much difference. If it is used for a head phone amp, would definitely make both larger. In general, since it's easier and cheaper, I would make the input cap larger and then increase the value of the feedback cap if I needed more base extension. Part of the whole DIY / design thing is to determine which changes make the most (usually cost effective) differences. As I suggested before, try different values to see if they make any difference. YMMV Good luck.

[AndrewT]

the single supply schematic (Av=200) shows 0.1uF with 1M on the input and 510r with 10uF on the lower leg of the feedback loop.

The dual polarity supply (non inverting split supply) shows 0.1uF with 100k on the input and 2k0 with 5uF on the lower leg of the NFB loop.

Neither schematic shows any RF interference attenuation.
Neither schematic shows any component numbering.
Neither schematic is Fig numbered.
Typical sloppiness from some of National's datasheet writers.

Let's look at the high pass filters that the four combinations of components give
component values . RC . . . . . . F-3dB
1M + 0.1uF . . . . . 100ms . . . ~ 1.6Hz
510r + 10uF . . . . . 5ms . . . .~ 31Hz
100k + 0.1uF . . . . 10ms . . . .~ 16Hz
2k0 + 10uF . . . . . .20ms . . . .~ 8Hz

The high pass filters are all over the place. I see no logic in the choice of RC values. All I can guess is that they have generally chosen high resistance values to allow minimum cost capacitors to be used.

The single supply is limited by the F-3dB of 31Hz giving an LF bandwidth of ~60Hz. But the NFB cap has significant AC across it and this increases distortion.
The dual polarity supply is limited by the F-3dB of 16Hz, giving an LF bandwidth of ~30Hz. This time the NFB cap has virtually no AC voltage across it.

I see these choices of typical component values as useless for beginners reading this datasheet.
You need quite a bit of background knowledge to recognise any of these "errors" and thus beginners will end up building poorly performing amplifiers.

National should be showing good typical circuits and superlative performance circuits to show how good their chipamps are and thus generate extra sales from all the satisfied builders.

[AndrewT]

Which version dual or single polarity do you want to build?
We can give a better schematic with appropriate component values.

[wavycertified]

Originally Posted by AndrewT

Which version dual or single polarity do you want to build?
We can give a better schematic with appropriate component values.

Thank you for pointing that out, I did not realize the circuit was sloppy until I saw the numbers you have shown, and how much I still have to learn down the road.
I’m building the dual polarity version. The reason why I chose this chip is because it has low distortion and can be powered with dual polarity supply. Even though the amplifier only has 4 watts of output power, I still want to use quality components to get good sound. Could you please give me a better schematic with appropriate component values if you don’t mind? Thank you so much

[AndrewT]

Can you post a copy of the dual polarity schematic with all the components identified?

[wavycertified]

Originally Posted by AndrewT

Can you post a copy of the dual polarity schematic with all the components identified?

Yes definitely

[AndrewT]

topleft (input):
5uF change to 50uF
2k change to 1k0
topright:
100k change to 27k or 30k or 33k
bottomleft:
0.1uF change to 1uF
100k stereo(2gang log law pot) change to 30k fixed resistor
add a 10k stereo pot before the 1uF cap.
add a capacitor in parallel to 30k about 330pF to 1nF depending on your taste in treble roll-off.
add a series resistor about 1k0 between the 1uF cap and the top of the 30k
add a 200r or 220r in series to pin4.
add a connection (twisted as a pair with the signal wire) from the bottom of the pot to the RCA socket barrel.
add a connection from the bottom of the pot to the 30k and 1nF
add a connection from the bottom of 1nF to the bottom of the 50uF,
add a series resistor (about 10r) to the bottom of 50uF and run a connection to the main audio ground.
middle:
move the 0.1uF decoupling caps to right next to the power pins (one on +ve and one on -ve) connect the zerovolts of these two caps together. add a 100uF cap to each supply (one to +ve and one to -ve.) again connect the zero volts of these two caps together. Add a connection from the 0.1uF zero volts to the 100uF zero volts and then connect this to the main Audio Ground. Connect the output Zobel to the main Audio Ground. Connect the speaker return to the main audio groun. Connect the PSU zero volts to the main audio ground. connect the Chassis to the PSU Zero Volts. Connect the PE(third wire in the mains cable to chassis using a strong permanent mechanical fixing.

Draw a new schematic showing all these changes and post here for checking and any further additions.