Hi everyone. I'm planning on building a small 20+20 W stereo amplifier using two TDA2005, each one in bridge mode.
Looking through the datasheet, I found a couple of application circuits. The first one (attached as TDA2005_Bridge.jpg), seems to be the most commonly used circuit, but with the components of the figure, it has a gain of 50 dB, which I think is way too high if I want to use the amp with my phone or an MP3 player. As far as I've read in the forum, it would be better if the gain is around 30 dB.
As I understand, in this first circuit, both amplifiers are configured with the same gain, excep the top one is in a non inverting configuration, and the bottom one is inverting. Each one has a gain of around 167 V/V, and because they're bridged, the total gain is 334 V/V = 50 dB. In this case, both take the same input, the input audio signal.
There's another circuit in the datasheet, titled "Low Gain (34 dB)" (attached as TDA2005_Bridge_LowGain.jpg). The gain is still a bit high, but a lot better than the 50 dB from the first one. Except that in this one, the top amplifier has a gain of 28 V/V, and the bottom one has unity gain, but it's inverting the output of the top one, not the input signal, so the total gain is 28*2 V/V = 56 V/V = 35 dB. NOTE: R6 has the wrong value in the datasheet, it should be 1 k, not 1 ohm.
Now, I have two options: The first option is just using this second circuit. But I'm worried that one of the amplifiers has such a low gain (unity) while the other one is doing all the voltage amplification. My experience so far is limited to OpAmps, where unity gain wouldn't be an issue, but I've read around here that power amplifiers usually need a minimum closed loop gain to work properly, higher than unity.
The second option is to change the two 12 ohm resistors on the feedback loop of the first circuit, to, for example, 100 ohm, so the total gain would be 32 dB, and then each amplifier would have the gain set at 20 V/V, "sharing" the gain.
Now, the question is: ¿Which one would you think is the best option?
Looking through the datasheet, I found a couple of application circuits. The first one (attached as TDA2005_Bridge.jpg), seems to be the most commonly used circuit, but with the components of the figure, it has a gain of 50 dB, which I think is way too high if I want to use the amp with my phone or an MP3 player. As far as I've read in the forum, it would be better if the gain is around 30 dB.
As I understand, in this first circuit, both amplifiers are configured with the same gain, excep the top one is in a non inverting configuration, and the bottom one is inverting. Each one has a gain of around 167 V/V, and because they're bridged, the total gain is 334 V/V = 50 dB. In this case, both take the same input, the input audio signal.
There's another circuit in the datasheet, titled "Low Gain (34 dB)" (attached as TDA2005_Bridge_LowGain.jpg). The gain is still a bit high, but a lot better than the 50 dB from the first one. Except that in this one, the top amplifier has a gain of 28 V/V, and the bottom one has unity gain, but it's inverting the output of the top one, not the input signal, so the total gain is 28*2 V/V = 56 V/V = 35 dB. NOTE: R6 has the wrong value in the datasheet, it should be 1 k, not 1 ohm.
Now, I have two options: The first option is just using this second circuit. But I'm worried that one of the amplifiers has such a low gain (unity) while the other one is doing all the voltage amplification. My experience so far is limited to OpAmps, where unity gain wouldn't be an issue, but I've read around here that power amplifiers usually need a minimum closed loop gain to work properly, higher than unity.
The second option is to change the two 12 ohm resistors on the feedback loop of the first circuit, to, for example, 100 ohm, so the total gain would be 32 dB, and then each amplifier would have the gain set at 20 V/V, "sharing" the gain.
Now, the question is: ¿Which one would you think is the best option?
Attachments
Don´t mess with chipamp gain, it may easily become unstable.
Instead, use a fixed attenuator at the input.
To begin with, you do not have that high gain excess, at all.
Bridged amp can put out some 8V RMS (15W RMS @ 4 ohms)
As shown gain is (1000/12)*2=166X , so you need 8000/166=48mV signal, round it up to 50 mV
Since typical phone/player headphone output is between 100 and 200mV, you are quite inn the ballpark, with a little gain to spare in case you listen to a somewhat lower than average MP3 song.
At most, I would attenuate input by 2-4X , not more, so say a 10k/10k divider, or tops 15k/4k7
But build it "standard" first and experiment with your player, I guess you will be happy as-is since you have a volume control anyway, straight at player.
Please post results, the build and test may help others with similar doubts.
Sadly many solve their problems and disappear.
Instead, use a fixed attenuator at the input.
To begin with, you do not have that high gain excess, at all.
Bridged amp can put out some 8V RMS (15W RMS @ 4 ohms)
As shown gain is (1000/12)*2=166X , so you need 8000/166=48mV signal, round it up to 50 mV
Since typical phone/player headphone output is between 100 and 200mV, you are quite inn the ballpark, with a little gain to spare in case you listen to a somewhat lower than average MP3 song.
At most, I would attenuate input by 2-4X , not more, so say a 10k/10k divider, or tops 15k/4k7
But build it "standard" first and experiment with your player, I guess you will be happy as-is since you have a volume control anyway, straight at player.
Please post results, the build and test may help others with similar doubts.
Sadly many solve their problems and disappear.
"1. The closed loop gain must be higher than 32 dB."
Table 6 shows suggested part values for Gv=40dB.
For sufficiently high gains (40 to 50 dB) it is possible to put R2 = R4 and R3 = 2R1,
Gv (dB) R1 (Ohms) R2 = R4 (Ohms) R3 (Ohms)
40 -----1000 ----------39 --------------2000
50 -----1000 ----------12 --------------2000
IIRC, aside from HF instability, the real low impedance in this network also reduce supply buzz.
And I totally agree: build it with any known-good plan and TRY IT before fretting about imagined problems. May not be as bad as you think. May be easily fixed with passive attenuation (or reading the sheet closer).
I think there is a problem with the "low gain" schematic. R6 should probably be 1K and not 1 Ohm. Besides the gain issue, it is also a 11 Ohm DC load on the positive output and ~7 watt resistor dissipation. As is, I would expect the negative output to be grossly overdriven.
I'm not sure why R5 is 10 Ohms and not 36 Ohms like R3. I have to assume there are issues with real parts that make it a better choice. A resistor between +/- inputs can be used to reduce the feedback without resulting in a lot of gain, but the data sheet does not use that strategy. For this chip, the +/- inputs are a different DC voltage so it would have to be outside the coupling caps. Bottom line is take these app notes with a grain of salt.
I'm not sure why R5 is 10 Ohms and not 36 Ohms like R3. I have to assume there are issues with real parts that make it a better choice. A resistor between +/- inputs can be used to reduce the feedback without resulting in a lot of gain, but the data sheet does not use that strategy. For this chip, the +/- inputs are a different DC voltage so it would have to be outside the coupling caps. Bottom line is take these app notes with a grain of salt.
You are right, R6 is a typo, must be 1k
As of R5, its value is not critical, it can be 10 ohm or 36 ohm like R3, in any case "slave" bridger amp will behave as an inverting unity gain stage (just what the doctor ordered), its individual gain is R4/R5 but its "system" gain, as part of a whole bridged amp is minus R2/R4
As of R5, its value is not critical, it can be 10 ohm or 36 ohm like R3, in any case "slave" bridger amp will behave as an inverting unity gain stage (just what the doctor ordered), its individual gain is R4/R5 but its "system" gain, as part of a whole bridged amp is minus R2/R4
Yes, Fig33 R6 should be 1K.
To a first-approximation, the value of R5 is not critical. Lower may stay further from RF oscillation -or- give better PSRR?
To a first-approximation, the value of R5 is not critical. Lower may stay further from RF oscillation -or- give better PSRR?
If you look at it funny, R5 IS "between +/- inputs", no?