TDA7294 clipping

cs

Member
2005-06-04 7:37 pm
.
This is really a follow-up to my earlier post on LM3886 clipping.

Datasheet wise, the TDA7294 looks roughly comparable to the LM3886 (but can be bought very cheaply), so I wondered if anyone has done any measurements regarding clipping behaviour ? Googling didn't come up with anything.
 

cs

Member
2005-06-04 7:37 pm
.
ashok said:
You mean a scope shot of a clipped sine wave ? What frequency 1 or 10 Khz , what load ----- 8 or 4 ohms ? Have you some prior knowledge if this is a problematic area ?

Yes, a scope shot would be ideal.

I guess 10kHz and 4 ohms would be best to highlight any adverse behaviour. I'm not aware of any specific problem, but I would just like to find out before starting on my project. The amps (either LM3886 or TDA7294) will be used in some active speakers.
 
ashok said:
OK. I was going to do this anyway , I'll just hurry it up.
You are lucky . I have both LM3886 and TDA7294 boards .. still populated with parts !:D

I wonder what happens if you arrange (adjust voltage) a test where the output powers match (nearly) between the two amplifiers? Otherwise the test couldn't be a head-to-head comparison, while the application for one differs from the other.

If that's not possible, please include LM3875 in the testing as a reference point for triangulation.

So, let's see what the measurements show. ;)

I'm so curious. But, since the application differs, then I'm not sure how to apply the measurements. How to?
 

ashok

Member
2002-06-06 4:43 am
3RS
I can only test the TDA2050/7294 and LM3886/1875

Don't know what you are expecting but all boards are working ( at the moment!) and psu's are rigged up and so I can do it without much effort. A couple of weeks down the line the situation might change ( usually does !).
 
Maybe try a comparison when TDA7294 is operating on a slightly higher voltage than LM3886?

I just think its very interesting to "rig" both amps for same output power, same gain (as in signal + power = output), and then compare them.

Another comparison point, application wise, is their performance per same thermal disappation. Actually, that's almost the same comparison point that I'm talking about anyway--just a different way to see it.

Its no secret that the less powerful TDA7294 is going to wipe the floor with LM3886, IF used on the same power supply. This means nothing whatsoever, because of, signal + less power = cleaner, lesser, output. That's not an even comparison. If you try this, notice how much cooler is the TDA7294, and how much less output there is. So, there's no baseline reference for comparison, and, these useless figures are already documented. Instead. . .

What is a secret. . . What happens with a truly even head-to-head comparison between the two?

I'm so curious. ;)
 
danielwritesbac said:
Its no secret that the less powerful TDA7294 is going to wipe the floor with LM3886, IF used on the same power supply. This means nothing whatsoever, because of, signal + less power = cleaner, lesser, output. That's not an even comparison. If you try this, notice how much cooler is the TDA7294, and how much less output there is. So, there's no baseline reference for comparison, and, these useless figures are already documented. Instead. . .
No, that's not even remotely true. Look at the six graphs on page 12 of the LM3886 datasheet. The lowest THD+N for steady state signals is acheived at the point just before clipping. I don't know if this is the case with TDA7294, but considering it's true of nearly all SS amps, it probably is.

What this means is that less output has more THD+N. For LM3886 at 1kHz into 4ohms with 28V rails, THD+N is about 0.03% at 20mW output and 0.004% at 20W.
 

soongsc

Member
2005-03-26 2:31 pm
Taiwan
I do recall the LM3886 can take higher voltages, and thus probably will clip at a higher voltage. The LM3886 runs quite cool driving an 8 ohm load whereas the TDA7294 runs warmer just sitting there. I tried palying both, but using different power supplies. Liked the LM3886 better.
 
cabbagerat said:
No, that's not even remotely true. Look at the six graphs on page 12 of the LM3886 datasheet. The lowest THD+N for steady state signals is acheived at the point just before clipping. I don't know if this is the case with TDA7294, but considering it's true of nearly all SS amps, it probably is.

What this means is that less output has more THD+N. For LM3886 at 1kHz into 4ohms with 28V rails, THD+N is about 0.03% at 20mW output and 0.004% at 20W.

Not applicable.

A higher proportion of input signal to a lower proportion of power signal is clearly a higher fidelity. This gives advantage to the weaker amplifier, TDA7294, unless its voltage is raised to compensate. Please don't be offended. I wish to subtract this advantage and the nonsense with it. . . some of which includes advertisment-based specs of inaudible errata.

Instead, lets see what happens with measuring equipment.

At expected voltages for both, above 32vdc, and forced to equal output power per input signal for accurate comparison, I expect that TDA7294 will fail to outperform LM3886.

That's why I asked for equal output power per input signal at the clipping point measurements. ;)

At 28v, you can compare LM1875 with TDA7294 on output power per input signal, albeit one is warmer. That's an interesting, seperate, topic.

Hi soongsc! Yes, I think this preference does vary by voltage, but I don't know. So, that's why I want to see them measured truly head-to-head, based on output power per input, doing what amplifiers do--adding power to signal and out it goes, with what results, nobody knows. OH Darn! A case of the limricks!! ;)
 
danielwritesbac said:

Not applicable.

A higher proportion of input signal to a lower proportion of power signal is clearly a higher fidelity. This gives advantage to the weaker amplifier, TDA7294, unless its voltage is raised to compensate. Please don't be offended. I wish to subtract this advantage and the nonsense with it. . . some of which includes advertisment-based specs of inaudible errata.
No, a higher proportion to output signal matching the input signal to output signal different from the input signal is higher fidelity. This is mostly specified as a lower THD+N. While I am aware THD+N isn't everything, it still stands as the definition of fidelity. While the THD+N of most amplifiers will go up (but not much) with high rail voltage, I don't see how comparing at a standard voltage is unfair.

I really don't understand what you mean by "A higher proportion of input signal to a lower proportion of power signal is clearly a higher fidelity.". The closest I can come up with is "a higher proportion of input signal to a lower proportion of output signal is a lower gain". Similarly, I'm not sure what you mean by "some of which includes advertisment-based specs of inaudible errata.". Do you think the National datasheet is deliberately lying? Are you claiming that higher THD is inaudible at those levels?
 
I meant to use about the same gain, but pep the voltage on the lesser power amplifier (TDA7294) to equal the LM3886, before comparison. Otherwise, comparison is unwise because applications differ.

I'd like to see both at 32v or more, so that the study corresponds to general DIY power amplifier use, which is the topic.

It isn't applicable to study clipping behaviors at either 20 watts or 20 milliwatts output. That's not the topic. If the amplifier is undervolted, then it will clip because it has been mis-applied--not because either amplifier is inferior. That's not the topic either, except that such information could be caused by testing conditions, rather than an amplifier.

The only thing that I really said about the National Semiconductor or ST Thompson literature is that. . . I'd prefer to see what measuring equipment resolves.