So here is the scenario:

I am building a dedicated amplifier (with built-in crossover, 3 way design) for some special speakers I recently built.

For the mid/highs, I really love the sound of the TDA8932 chip amps. Incredible sound for the $$.

But these chip amps have a couple issues:

1) They are very sensitive to the supply voltage. Most run them from linear supplies since they easily "pass through" various bits of noise, etc from switching supplies.

2) They really start to run hotter-than-I-like once the voltage creeps over 20-21 volts or so.

This entire amp is powered by a single-rail 24 volts supply. And it is a switching supply at that. This raises BOTH problems mentioned above: the chip runs pretty warm AND you get some high-frequency interference that can be heard.

To solve this issue, I created a cascade of high-current silicone diodes (4 in a row) to drop the 24 volts down to around 21.2 or so. Just after each diode, I have a 440uF capacitor between the output of that diode and ground. This solves both issues: it drops the voltage down just enough that the chip runs FAR cooler (yes, them couple volts makes all the difference!) AND it serves as a really good noise filter. All the switching noise I had is now gone.

These amp boards are only around 35 watts max - and at the voltage (and 8 ohm load) I am running... they will never be able to supply more than 18-20 watts total. This is only for mid/tweet.

I have read that diodes used in such a fashion are typically not the best choice - but can not think of a better solution.

A DC-DC buck converter makes the noise problem WORSE.

A LDO regulator would need to have a minimum voltage difference of around 3-5 volts. Say I set the output voltage to go from 24->19 (playing it safe).. 20 watts divided by 19 volts is around 1 amp. So the LDO would have around 1 amp passing through it with a voltage drop of 5... That would be 5 watts of heat, that would be a bit too much - no? Or is that reasonable with a LM317 or such?

The diode setup seems to work pretty good. The input voltage should never change (its a regulated switching supply!). The diodes I am using are some run-of-the-mill 6 amp silicone diodes. They seem pretty stable. A similar 1 amp (max) at .7 volts is less than 1 watt [per diode] and seems reasonable for a large diode [rated at 6 amps] to handle it seems.

Another thing to consider is what I am powering: 20 watt hifi amp boards. This is max output. I am not using these in a PA setting, typical household use - small-to-mid-sized-rooms. Typical music power will be much less most the time. These amps are behind a 24db active xover - they never see anything over 180hz. So all the real demanding high-current action is off their plate.

Should I consider something else?

Any sugestions would be welcome. Thanks!

 

This solves both issues: it drops the voltage down just enough that the chip runs FAR cooler (yes, them couple volts makes all the difference!) AND it serves as a really good noise filter. All the switching noise I had is now gone.

It works. Move on the the next problem.

 

A LDO regulator would need to have a minimum voltage difference of around 3-5 volts.

That would be a regular non-LDO regulator. LDO (low drop out) are more like a 0.5 to 1V margin.

 

That would be a regular non-LDO regulator. LDO (low drop out) are more like a 0.5 to 1V margin.

You see why I am asking questions here to the people who know better? lol....

I am no EE. Just tinkering with speakers and home-grown DIY amps.

joeyd999 thinks I should more on the the next issue.

That's good enough for me if nobody else thinks this will cause me issues down the road.

 

The trouble with diodes is that the supply voltage changes with current. This could result in distortion of the audio, though I don’t know if it is enough to be noticeable.

 

I would lean towards using an LDO with 1V or less dropout voltage.
That would provide good filtering with a more stable output voltage.

But if the diodes work for you, then I seen no big reason to change.

 

I would build a voltage regulator with a transistor, a couple resistors, and a programable voltage reference. I've used this many time where noise is an issue and a linear regulator is not appropriate for some reason or another. It gives you stable voltage with whatever current your system needs. With audio amplifiers, this can be sizeable.

 

You can use a Capacitance multiplier circuit with A Darlington transistor to reduce the ripple.

 

While the supply voltage may be constant, the load current is not. The use of diodes or resistors will result in a soft voltage being supplied to the amplifier, which equal distortion.

I would use a LDO voltage regulator for a simple and lowest distortion solution.

Since the amplifier is supplying only the mid/tweeter, that is typically about 20% of the total audio power. So, if at a reduced voltage of 19 volts, the amplifier can only produce 20 watts, 20% of that means that the expected current draw will be about 0.5A. With a 5 volt drop across the regulator, that means you must dissipate 2.5 watts of heat. That should only require a small heatsink. I would verify the maximum current draw before selecting a final regulator and heatsink.

Whether you use diodes, resistors, transistors, or IC regulators you still have to get rid of that 2.5 watts.

 

An option not mentioned yet would be to add a series filter that would also drop the voltage a bit. A large filter choke with a heavy laminated steel core will block a lot of noise from the 24 volt supply. Probably a shunt voltage regulator to keep the voltage at the chosen limit when the amp is not drawing much current.
NO, it will not be as efficient, but it can sound good.

 

I don't have much experience with LDO regulators specifically, but with the much more privative LM317 type regulators they are VERY quiet regulators when the output capacitor is sized properly. If the capacitor is too LARGE, the internal feedback is defeated and they pass noise like a wire. I ask the question - Do LDOs behave similarly?