Running PS near Max....Good Idea?

[tommytube]

I was throwing together a little amp using the ST Micro TDA2040V chip. Sounds fantastic with the little app note on the datasheet. Just wondering about my PS. I've hooked up a 12.6-0-12.6 transformer I got from Radshack and am now getting close to +/-18.6V and the chip's max is supposedly 20V. I'm afraid if it's used in a house with slightly higher voltage it'll raise up to 20V or a bit more. Even a surge might go up to the MAX. So should I do anything about it? Is this too close to MAX?

Any ideas for dropping it down easily? A resistor divider would work, but it seems so inelegant......but maybe the KISS principle is OK here. I don't want to limit dynamics or do something that contributes in the audibly negative direction.....

thanks for any ideas!

Tom

[cpemma]

A resistor divider wouldn't work, it's a variable load.

Is that an open-circuit voltage? Seems a bit high from 12.6V AC, may drop with the amp coupled up. Test with the amp idling.

[Bazukaz]

Hi,
some tricks can be made with standart diodes.each has a voltage drop of around 0,6V. So ,by connecting 5 these in series , you get 3 volts drop.
Another option that i have used successfuly is unwinding some secondary windings of trafo.

[!]

Originally posted by tommytube
[B]I was throwing together a little amp using the ST Micro TDA2040V chip. Sounds fantastic with the little app note on the datasheet. Just wondering about my PS. I've hooked up a 12.6-0-12.6 transformer I got from Radshack and am now getting close to +/-18.6V and the chip's max is supposedly 20V. I'm afraid if it's used in a house with slightly higher voltage it'll raise up to 20V or a bit more.

Keep in mind that I'm not looking at this specific datasheet right this moment, but... WHen they say 20V max, you don't have to worry because you're a little under that, because you are UNDER the max.

Now another consideration- when are you measuring the voltage? You'll have to have an output over 25% to get meaningful voltage numbers, the closer to idle/no-output the chip is, the less important if it's drifting up to the max voltage value.

Even a surge might go up to the MAX.

I encourage you to completely ignore this possibility. IF you think surges are common enough to warrant consideration, design that into your supply. Surges aren't just 10V higher, could be 2X the nominal voltage if not a lot more and ANY typical amp would be overvoltage if using an unregulated PSU when some surges are encountered.

So should I do anything about it? Is this too close to MAX?

You should make sure the chip is 'sunk well. Voltage ratings are guaranteed _minimums_, not a "hit this exact number and the magic smoke escapes" number. If your supply, hooked up to the amp were continually over that 20V figure it would be reason to do something, but currently it is not.

Any ideas for dropping it down easily? A resistor divider would work, but it seems so inelegant......

I hate the word "elegant" because it's nonsense. Sorry, I'm not trying to be offensive but elegance has nothing to do with amp design except the case. The design goal should be specified and only the parts needed to achieve it, used. NO more, no less. I guess I'm a no-BS designer, if I "only" wanted to drop the voltage for example, I wouldn't build a jung super regulator or a dual tracking integrated chip regulator supply, I'd just use a pass transistor diode arrangement. Not that I'm advocating this over any other topology, but it is what meets the design goal of only voltage limitation that I specified. So, specify exactly what your requirements are, including available space, difficulty in building or testing, and cost... not necessarily here (though we might be helpful... sometimes less so, sometimes moreso), but in your mind when planning it at least.

A resistor divider is a horrible solution because it wastes so much power and creates so much heat. Not only that, but it does so when similar power waste could accomplish something useful, like regulation. If you want a regulated supply, go ahead and do it. Having 18-odd volts on a 20V chip isnt' in itself a reason to do it.

but maybe the KISS principle is OK here. I don't want to limit dynamics or do something that contributes in the audibly negative direction.....

First, what output power do you need? Voltage is directly related to that, into any given impedance speaker(s). I'd have to assume the required output isn't very high or you'd be using another chip... since in the end the least costly part is the couple dollars difference between one chip or another. LOL, I've made that mistake before too, thinking "heck, I already have two LM3886 just lying around, it'd be cheap to build an amp out of them". Then a couple hundred dollars later... but I tend to overbuild, a minimalistic one could be cheap I suppose but if it's worth doing then it's worth doing right.

Simple can be good though, your chip probably has fairly good PSRR, so as long as the noise isn't HF, you should be ok. Providing your transformer can deliver the required amps at 18.x V, and you have a suitable heatsink, it would seem you have an ideal transformer, not a problem per se.

[AndrewT]

Hi,
is 12.6Vac the open circuit voltage of the transformer?

Is the 18.6Vdc the open circuit voltage of the rectified and smoothed output from the unloaded PSU?

Try loading the PSU output with 40R and 20R to draw about 9W and 15W from the PSU. Measure the voltages while delivering currents.
The load resistors will get hot very quickly, so buy power resistors or make up a dummy by paralleling 600mW resistors. 390R would be a good start.

[tommytube]

I'll try and answer some of the very good questions being asked.

I am using a fairly large heat sink, so this should be quite adequate. I was having problems with the chip leads breaking as I installed the entire amp/heatsink into a smaller box, so I'm in the process of redoing part of the part to chip interface much better with better strain relief on the pins. Luckily this part is only $2.70, so it's easy to live with broken pins. In the midst of the redo on the amp, I was trying to address the power supply uncertainty in my mind.

The open circuit voltage of the transformer was something over 12.6, more like 13.4Vac. When I tried it out at work, the DC output of my supply was under 18V, like 17.5V....but at home my AC is a bit higher and it's coming out at 18.6 loaded. This is with the amps hooked up and running. Like I mentioned before, great transparent sound, much better than I expected.

So I believe my heatsinking is more than adequate....leaning toward a bit of overkill. I just wondered if a friend got this amp and used it at a different place whether a higher input AC would fry the chip.

Alternatives might be to go with a 6.3V-0-6.3V transformer I have. The current one is rated at 3 amps.

Again, thanks for all the input and don't hesitate to give the straight skinny. I want to learn how to do these things correctly. I've done a few amps before, but this is my first scratch build, point to point one w/o a PCB.

Tom

[demogorgon]

as has been nemptioned before, 20V is not the point where the chip blows.

the lm3886 has a limit of 42V in use. i'v been pushing it over that limit sucessfully. had to go up to +- 50V to blow it

your current setup is not a problem, unless you make it out to be

-Marius

[tommytube]

I don't have a problem......at least on the PS front.

I did load down the supply with a couple 100 ohm / 20 watt resistors and got the output below 18v......(it's sitting around 17.85V fully loaded).

Output from the Radshack tranny is 13.75 instead of 12.6V fully loaded, unloaded it's over 14.

So I won't sweat it. Thanks for the input.

More voltage will mean more power anyway, so it'll be nice to run it at it's potential.