To improve HF quality I'd suggest additional decoupling across the rails. Use low ESR electrolytics and keep them as close to the chip as possible.
Depends how much power you need - for 10W or so I have found TDA8932 the most transparent yet. But I haven't tried all the chip amps.
The lower voltage schematic at post #224.
You may be able to trim the gain barely slightly lower than that.
The cause of disappointing sonic quality is NOT inside the chip. There is some other locale for that difficulty.
This an ancient standard miller design pushing fets; so, performance isn't automated, but it is available. It can do what you ask very easily, but you might have to do a better job with the asking.
Additionally, that voltage is low for it, so there's less compromise, and therefore there's greater capacity for high fidelity audio, rather easily beating its competition.
Aside from power circuit, what is likely to have gone wrong is mad pinout causing poor layout. There's some easy patches for that problem. Actually, there's also some easy patches for power circuit as well. Yeah, we can patch both. It is old hat, because we have to do that just about every time. Its just nice that it is easy to do.
Thank you!
Although I cannot predict the layout/stability/gain parameters of each different board and situation, I do try to hit the ballpark.
That will vary by voltage, load, and layout. It is an older model chip and so the process doesn't include room for an endless amount of Constant Current Source circuits to accommodate all situations ideally, automatically (that won't fit). Therefore it is reliant on trimming the resistor values according to need.
It surprised me too. I haven't got bored with it yet. Thank you for mentioning that thing. I use it often (in SE mode).
Since they don't sound very different, the difference of 20W versus 90W is also the difference in room size. (limits are 44W versus 254W, peak music power, tested in real club).
There's a cost difference as well; but, if you have a bigger room, then you're on the right page. I can say that TDA7293 (modded as indicated here) does a bigger and more forceful presentation, with a very similar, pleasant tone (so it is relevant to crank it up).
I really do like both of these things.
And I don't like either of them bridged.
Last edited:
There are new offers at ebay a mono amp with rectifier and 3300 µF electrolytics with TDA7293....it costs less than a frozen yoghurt ice in the city....
a challenge for a DIY to make a monitor speaker with 2 of these boards. Of course the active crossover design cannot be done without acoustic measure equipment and state of the art is DSP ... but the challenge here is as good as possible and as cheap as possible .
From an older workout done by Philips is is known that a monitor needs no more than 40 watts for woofer and tweeter to reproduce all kinds of music according to a amplitude vs frequency statistics, not a problem for these amps. The listening distance is 1 meter.
A monitor further needs equalizer to compensate acoustic properties of room and a frequency dependent limiter. JBL has all this at a reasonable price but...its not diy
a challenge for a DIY to make a monitor speaker with 2 of these boards. Of course the active crossover design cannot be done without acoustic measure equipment and state of the art is DSP ... but the challenge here is as good as possible and as cheap as possible .
From an older workout done by Philips is is known that a monitor needs no more than 40 watts for woofer and tweeter to reproduce all kinds of music according to a amplitude vs frequency statistics, not a problem for these amps. The listening distance is 1 meter.
A monitor further needs equalizer to compensate acoustic properties of room and a frequency dependent limiter. JBL has all this at a reasonable price but...its not diy
The lower voltage schematic at post#224 above http://www.diyaudio.com/forums/atta...yle-no-lossy-emitter-resistors-tda7293dan.gif has parameters for high fidelity with TDA7293. It is 45W (and requires a slightly lower voltage transformer).
Elsewhere in the thread is mention of relocating the gain divider under-board and some power filtering too, as well as how to fit it.
If you got a unit board, power supply on same board as amp, there is means to patch it. However, if there is only 3300u per each rail at the power supply, then it doesn't support speakers larger than 5" approximately. So, you might still be looking at adding a power supply board. And, in that case the product isn't as inexpensive as it looks.
Small speakers are less efficient.
Big speakers are more efficient.
Low power amplifiers can drive any speaker big or small provided it has the correct impedance to suit the amplifier.
A low power amplifier will need LESS output power to produce MORE sound from a big efficient speaker.
Use big speakers with low power amplifiers.
Daniel is posting rubbish advice:
Big speakers are more efficient.
Low power amplifiers can drive any speaker big or small provided it has the correct impedance to suit the amplifier.
A low power amplifier will need LESS output power to produce MORE sound from a big efficient speaker.
Use big speakers with low power amplifiers.
Daniel is posting rubbish advice:
I see a lot of dual and even triple TDA7293 kits on ebay recommending 8Ω speakers. I'm trying to understand the point of more current potential into 8Ω (apart from over-engineering or perhaps to run safely at slightly higher voltage by raising the apparent impedance) since you'd be voltage limited into 8Ω rather than current limited... Would it not make more sense to use (for ex. a 3xTDA7293 board) into a 2 to 2.6Ω load? with approx 35v?
Last edited:
you see, the thing is a 8 ohm speaker can have a dip in its curve even under 7 ohm.
to get hihger dynamic headroom limiting the supply voltage is..not considered a good practice. paralelling 2 of them does indeed ease the load on the chip, but high current surges with a high supply voltage and a "less plite" woofer do call for a 3rd paralell chip to make sure you can have high rail voltage, enough output current, and stay well within the SOA of the chips.
since these chips can work in a master/slave mode with ease, it makes perfect sense to do it thisway.
to get hihger dynamic headroom limiting the supply voltage is..not considered a good practice. paralelling 2 of them does indeed ease the load on the chip, but high current surges with a high supply voltage and a "less plite" woofer do call for a 3rd paralell chip to make sure you can have high rail voltage, enough output current, and stay well within the SOA of the chips.
since these chips can work in a master/slave mode with ease, it makes perfect sense to do it thisway.
Daniel has quite a lot of good advice but no one has the monopoly on good advice.
However to damn all Daniel's advice as rubbish is unwise.
More current potential into 8 ohms turned out to be more linear output. This costs slightly less at the power supply (for same performance), and is somewhat more entertaining--both aspects regarding the bass. Also, the more linear Parallel amplifiers shout less (because linear), thus putting you a little step farther away from needing a tone control.
Makes sense. Can you please confirm that I 'should' be able to drive a 2Ω load with 3xTDA7293 in paralleled and +/-36v? (each chip effectively seeing 6Ω)? HAs anyone tried this?
Thanks Daniel
When all of the other details are missing, and despite that, there was a need to relate a minimum size/capacitance power supply to a given size speaker, here is one way to estimate it: Make a guess at what size output cap the speaker would need for a single rail amp (perhaps actually do it) and then double that figure--twice an output cap requirement is the per-each-rail requirement for supply (grossly estimated).
Of course we shouldn't be satisfied with minimums.
What I was doing is illustrating that the onboard supply was too small for full size speakers; and, likewise too small for full size performance.
Can't do that without some facts; so, I used them, although there were very few indeed.
It is true that this sort of estimate can't tell you when your supply is big enough; however, it does fairly at telling you when the supply is sure to be too small.
That estimate's bigger brother looks like:
The power supply capacitance acts as a delay timer and the transformer acts as a charger--the bigger the capacitance the slower, and the bigger the transformer the faster the charge. So, we should not have more capacitance than can be recharged to Full voltage in-between average bass beats, since too much capacitance would suppress the performance of the, probably, too small transformer.
And now, the rest of the story. . .
Combine the two estimates together, and then for a full size speaker you arrive at 12000~18000uF on average and need to buy big enough transformer to recoup full voltage before the next bass beat hits on average track. If the voltage has not recouped on time, the transformer is too small and the audio quality is minimized.
Escape hatch: Array of caps has lower esr and thus charges slightly faster than solo caps. The price difference arrives at enclosure size, because array is bulky. See post#1.
This isn't the end of power supply measures, but only the beginning. A supply absolutely must pass practical measured needs before going any further. Draw the line there.
Sir, I think you know that if the supply doesn't do at least as well as I can estimate, then it just won't do. That was the purpose of the estimates.
To others reading, we will be in need of finer measuring techniques when determining when a power supply Will do. And, please don't assert that claim until after it has passed my basics.
Edit:
Now, wasn't the shorter post nicer to read than this bit?
Last edited:
Yes. Even a pair can do that, but a triple can do it better.
Even with paralleling for greater linearity and capacity, the problem with heavy loads, is the the efficiency will go down. That cost will arrive at the power supply. You'll have either a financial cost or a performance cost. The proportions are the same as the inverse square law.
Would you like to spare the details so I can imagine a more efficient approach? It probably won't cost more.
I'm absolutely open to suggestions. This is to be used to power mid-bass for a once-in-a-blue, basement party speaker system which can either be wired for 2Ω or 8Ω... I figure with +/-36v, 2Ω is the best solution but I'm not married to it. The midbass drivers are very efficient and will HPF steeply at approx 50Hz so I don't imagine I need too much current to get very loud.
At that point I could make do with estimating all of the details except for the size of the basement.
The thing is that ears are made of flesh and the output potential is beyond the scope of most basements, unless it is unexpectedly opulent, such as big enough for at least 2.5 airplanes, and no cops nearby, unless they are not even partially but really very completely deaf.
This much output is not fake but then we have the opposite problem of it being really hazardous if the space was too small (and too small is just about airplane hangar).
Please report the amount of space?
P.S.
I did outright launch a window out during test; and, yes the tone was still pretty, but the expense was a bit annoying (the window cost more than the power supply and the expense for both was pretty bad). It does have the power. It is an aB miller amp, old school, so the transformer and supply costs are also worth a mention. If modded as indicated, it is chillaxin so the efficiency wasn't too bad, the efficiency was practical and it didn't blow the breakers. But, the thing is, that I asked for a lot, got almost a third more than what I asked for, and that much audio just isn't safe inside a small area.
Last edited:
Even though it was a gross estimate (due to lack of other information), it is still true that supply size can relate to speaker size. Because workload!
On a plotted chart from information on Nico's huge power supply thread; and, if transformer size wasn't a limiting factor, then the amount of work to be done had mostly to do with the speaker to drive.
We did have to discharge the supply to accomplish a given bass beat and then the transformer did have to re-charge the supply, QUICKLY, not only before the next beat but also well before more clarity was expected.
In other words, my previous comment said that the supply was too small unless the speaker was too small as well.
Sure, it was a sloppy estimate, but I don't think that it was off the mark. I did try it with both 4 ohm and 8 ohm speakers. Either the shortfall's alleviation of loading or the more efficient 8 ohm speakers, didn't make much difference to a properly rigged Parallel TDA7293. The 8 ohm speakers did a prettier job; but, the 4 ohm speakers didn't compromise that much fet outputs too badly either. There wasn't much difference, so that wasn't the limiting factor.
This amp is a bit needy on the supply; so, when I reported that the supply was insufficient for anything but tiny speakers, that really is what the limiting factor was.
The problem is not rubbish; but, that it would make very little sense for me to describe what the deal was.
Here's another try anyway:
This thing is needy on the power supply, and a tiny onboard shortchange is absolutely certain to disappoint in all cases.
Well, I hope that communique worked.
What he said is that there's room for improvement. And, because I wasn't spoon fed the details, he also said that I should figure that out.
It is a very nice compliment indeed, and I'm working on it.
The problem with refining this is that it will be more specific and with the costs of much less likely that anyone would build it to spec.
How many people would hear the benefit of that? One, Two, or Three?
But, he's right that we might as well try for better.
Room for improvement?
My first guess on that is to add the Thile-Small output filter components.
That's because if we stabilize with something other than the main gain, we could turn the gain down very slightly, and with the benefit of slightly higher resolution.
It is likely worth the cost of the added parts.
That bit wasn't difficult to guess. But, I'd like to request help in coming up with the next step; so, please comment.
My first guess on that is to add the Thile-Small output filter components.
That's because if we stabilize with something other than the main gain, we could turn the gain down very slightly, and with the benefit of slightly higher resolution.
It is likely worth the cost of the added parts.
That bit wasn't difficult to guess. But, I'd like to request help in coming up with the next step; so, please comment.