I wanted to compare the power output of these ICs. While tinkering one night I found the TDA2040 clipping earlier than the 2050, so I wanted to experiment to see what was happening.
Each was tested by feeding it a 125Hz sinewave. The signal was increased until it started to clip and backed off just a bit to get the maximum clean power of the amp. The output load was non inductive 8 and 4 Ohm resistors (tested separately). To monitor the sounds, a speaker was connected across the output in series with a 150 Ohm resistor. Also an oscilloscope was used. Unfortunately, I don't have a regulated supply so I included the sum of the split supply voltages with each measurement. Naturally, the higher the power delivered to the load, the lower the Vs became. RMS voltage was measured and power calculated.
I purchased the ICs from DigiKey except for the TDA2040 which were purchased from an eBay vendor. I can't be sure the 2040s are authentic.
The TDA2050 provides the highest output swing at a given supply voltage, but I noticed one problem with it. At levels just before clipping and into clipping, I noticed some "fuzz" on the lower swing of the waveform. It caused an audible buzz in the speaker output. Grounding was the star formation and the supply leads were decoupled with .27uf film caps near the chip. Removing the output snubber made it go away. It seems it is somehow injecting noise into the ground. Adding additional film capacitors would not clear it with the snubber in place.
Each was tested by feeding it a 125Hz sinewave. The signal was increased until it started to clip and backed off just a bit to get the maximum clean power of the amp. The output load was non inductive 8 and 4 Ohm resistors (tested separately). To monitor the sounds, a speaker was connected across the output in series with a 150 Ohm resistor. Also an oscilloscope was used. Unfortunately, I don't have a regulated supply so I included the sum of the split supply voltages with each measurement. Naturally, the higher the power delivered to the load, the lower the Vs became. RMS voltage was measured and power calculated.
An externally hosted image should be here but it was not working when we last tested it.
I purchased the ICs from DigiKey except for the TDA2040 which were purchased from an eBay vendor. I can't be sure the 2040s are authentic.
The TDA2050 provides the highest output swing at a given supply voltage, but I noticed one problem with it. At levels just before clipping and into clipping, I noticed some "fuzz" on the lower swing of the waveform. It caused an audible buzz in the speaker output. Grounding was the star formation and the supply leads were decoupled with .27uf film caps near the chip. Removing the output snubber made it go away. It seems it is somehow injecting noise into the ground. Adding additional film capacitors would not clear it with the snubber in place.
Last edited:
UPDATE: Not happy with the results, I fixed the problem (-Vs decoupling film cap connection issue) and reran the test. Clipping point is easier to determine and measure. This resulted in a bit lower output numbers (the 2050 seemed high at 19.4 watts with the 4 ohm load). I also found an older and probably authentic TDA2040. The chart has been updated with the new figures.
Even though the TDA2050 pulled the supply voltage a bit lower due to its greater output swing, the greater output swing allowed it to produce more power into the load. Even with the slightly higher Vs, the TDA2040 does not have the near the output swing. If Vs was reduced to match that of the 2050, the 2040's output would be even less (especially with 4 Ohm loads). I consider this chip to be obsolete now. The LM1875 stands in the middle, but has the advantage of higher max Vs for a higher output swing when using 8 Ohms.
Listening tests: I don't have those golden ears that some people claim to have. These ICs sound quite nice to my ears as long as they stay out of clipping. Comparing the input signal to the output, the ICs are quite transparent (I can differentiate and nearly null the signal).
Even though the TDA2050 pulled the supply voltage a bit lower due to its greater output swing, the greater output swing allowed it to produce more power into the load. Even with the slightly higher Vs, the TDA2040 does not have the near the output swing. If Vs was reduced to match that of the 2050, the 2040's output would be even less (especially with 4 Ohm loads). I consider this chip to be obsolete now. The LM1875 stands in the middle, but has the advantage of higher max Vs for a higher output swing when using 8 Ohms.
Listening tests: I don't have those golden ears that some people claim to have. These ICs sound quite nice to my ears as long as they stay out of clipping. Comparing the input signal to the output, the ICs are quite transparent (I can differentiate and nearly null the signal).
Last edited:
My findings are about 5 watts less than the datasheet curves at the same supply voltage. I'm using a center tapped 2A 25.2 volt transformer, full wave split supply with 4,400uf of filtering on each rail. I'm measuring the Vs across pins 3 and 5 and Vs at pin 4 and ground under load so I'm at a loss to explain the difference.
According to the datasheet, At +/- 18 volts .5% distortion (probably right at clipping measurement) the 2050 should do a minimum of 24 watts and a typical of 28, so it has potential. I would NOT recommend pushing these chips beyond +/- 18 volt supply with 4 ohm loads due to the thermal limitations. At this voltage and load I'd mount them direct to the heat sink with no isolation. If Isolation is needed, mount them to a heat spreader and then to an electrically isolated heatsink.
According to the datasheet, At +/- 18 volts .5% distortion (probably right at clipping measurement) the 2050 should do a minimum of 24 watts and a typical of 28, so it has potential. I would NOT recommend pushing these chips beyond +/- 18 volt supply with 4 ohm loads due to the thermal limitations. At this voltage and load I'd mount them direct to the heat sink with no isolation. If Isolation is needed, mount them to a heat spreader and then to an electrically isolated heatsink.
Last edited:
I think with these low power chips Biamping is the way to go. Shure you need two stereo amps for 2 speakers....but.... Lets say its a 20W amp (for best performance with the 75 which I am more familiar with). If you biamp and crossover at a suitable freqency, the two 20W amps will have an output ability of twice that.
It has to do with what happens when two frequencies far apart are ampified at the same time.
Lets say you are passing a full range signal to your two-way speakers through one cable to avoid all the variables. If the signal contained 1V HF and 1V low frequency content, it sums and the amp actualy has to provide 2V of potential. Splitting the frequency bands effectively doubles the power at hand. So your 4 20W amps now acts as a 160W amp.
It has to do with what happens when two frequencies far apart are ampified at the same time.
Lets say you are passing a full range signal to your two-way speakers through one cable to avoid all the variables. If the signal contained 1V HF and 1V low frequency content, it sums and the amp actualy has to provide 2V of potential. Splitting the frequency bands effectively doubles the power at hand. So your 4 20W amps now acts as a 160W amp.
Johnr66, just for tzhe sake of experiment, could You please test me something?
If possible, can You hook up a tda 2050 to a more powerfull powersuply ?
I would be intrested to see the results, compared to Your exsisting results.
Actualy i my self do not think a household needs more than 15-20 watt in stereo for a proper sound, but that is just my personal preference, actualy i kindof like TDA chips, even the 1557 and its relatives.
If possible, can You hook up a tda 2050 to a more powerfull powersuply ?
I would be intrested to see the results, compared to Your exsisting results.
Actualy i my self do not think a household needs more than 15-20 watt in stereo for a proper sound, but that is just my personal preference, actualy i kindof like TDA chips, even the 1557 and its relatives.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.