Permissable voltage sag?

[johnr66]

I have a DIY PSU for a small amplifier. I am using a 25.2v 2a CT transformer. Using a full wave bridge, I obtain two rails. Each rail has 4,400uf of filter capacitance.

To test the PSU under load, I have 4 ohm speakers connected and driving it with sine waves. I selected a low frequency (20 some Hz) that would cause the most voltage sag. Amp was driven into clipping slightly.

Idle voltages (estimated 100ma current)
AC side: 28.0v, DC side: 35.0v

Under max load conditions (can't measure current without cutting wires)
AC side: 25.7, DC side: 29.4v

I have been told that I should select a larger transformer, but I argue that since the loaded AC side is above it's rated voltage, the trafo is fine. The dynamics of music would never load the PSU like the test anyway.

Thoughts?

[twest820]

The question seems to be whether it's OK to pull 5A RMS out of a 2A trafo. The answer seems obvious---no, unless you're trying to fry the trafo---so I'm not sure I'm understanding the question properly. If this is really what you're asking a I've found good rule of thumb is to choose a trafo with twice the VA rating of the peak load; when operating close to (but below) the VA rating the core tends to get a bit warm for my preferences.

[simon7000]

25.2 VAC should yield 34.4 VDC. 34 VDC should be 17 AMPS peak into two 4 ohm loudspeakers.

The reason your meter reads the transformer output voltage as 25.7 under load is that it is not an accurate meter! It does not do true RMS.

However a real loudspeaker at full power will double it's impedance. so if you want to be cheap get one with an 8 amp rating.

I do not want to go into an actual analysis of output versus AC current rating as that is not for this level of question and the information is easily found these days.

[AndrewT]

Originally Posted by johnr66

Under max load conditions ....................
AC side: 25.7, DC side: 29.4v

what are your max load conditions?

DC current through a resistor load,
or
Sine-wave test signal through power amplifier into resistor load,
or
music signal through power amplifier into resistor load,
or
music signal through power amplifier into speaker load?

[infinia]

music power vs sine wave power
Design the PS to supply the rails "stiffness" to support the speaker load at the lowest frequency at the highest level of your music.
If it sags at 30 Hz appreciably and not at 10 KHz by using test sine waves, what will that sound like with music playing into the speakers load?

Transformers are rated by the average heating rise above ambient, your criteria should be related to XFMR regulation at peak levels for the longest durations expected.

[johnr66]

I bought some non inductive resistors (4 Ohm) for a better test. Here are the results both channels driven just before clipping. I tried a few frequencies and the results were the same at each frequency (as expected with the resistors).

PSU voltages
AC voltage quiescent: 27.7vrms
AC voltage loaded: 25.0vrms
DC voltage quiescent: 34.5v (+/- 17.25v)
DC voltage loaded: 27.6v (+/- 13.8v)

Amp voltages
AC signal at output w/ 4 Ohm: 6.15vrms (9.5 watts)
AC signal at output w/ no load: 6.30vrms (same signal level as with the load)

[infinia]

Note) something seems wrong with your amp or test (something is limited before the output stage) b/c you said w/ no load with +/-17 volt DC rails, the amp puts out only 18 Vp-p max? I would expect to see 28V p-p
then with load same clipping point 9 Vp / 4 ohm = 2.2 A peak

1st/ check your set up with gains and expected drive levels.

1) post schematic for amp? with test equipment used

2) It would be useful to O scope the DC supply under load a)/ report ripple voltage Vp-p and b)/ avg DC w multimeter.

[johnr66]

Originally Posted by infinia

Note) something seems wrong with your amp or test (something is limited before the output stage) b/c you said w/ no load with +/-17 volt DC rails, the amp puts out only 18 Vp-p max? I would expect to see 28V p-p
then with load same clipping point 9 Vp / 4 ohm = 2.2 A peak

1st/ check your set up with gains and expected drive levels.

1) post schematic for amp? with test equipment used

2) It would be useful to O scope the DC supply under load a)/ report ripple voltage Vp-p and b)/ avg DC w multimeter.

I did not change the signal when I went from loaded to no load. I just wanted to see the amp as a voltage source and since the output changed very little, it is a pretty good one.

With 9.5 watts out and considering dissipated power, the PSU is in no danger of burning up. However, my focus has changed since I'd like to get 20 watts of output. The amp is a two channel TDA2040 based circuit right off the datasheet. With that PSU, I'm not going to get 20w per channel. Time to look into a better supply.

Something does seem wrong even with +/- 13.8 volt rails, It seems I should be getting more output swing before clipping. At least 12 watts, I'd think.

[wakibaki]

Here's another approach. You say you'd like 20W out, but this is an arbitrary number of no real significance, and there's speaker efficiency to consider too.

Is it loud enough and does it sound OK and is the trafo getting hot if you run it for a good while (representative of what it is likely to encounter at high volume in practice)? If it sounds OK, and it's not alarming to the touch, then OK.

w

[johnr66]

Originally Posted by wakibaki

Here's another approach. You say you'd like 20W out, but this is an arbitrary number of no real significance, and there's speaker efficiency to consider too.

Is it loud enough and does it sound OK and is the trafo getting hot if you run it for a good while (representative of what it is likely to encounter at high volume in practice)? If it sounds OK, and it's not alarming to the touch, then OK.

w

That is a good point. My speakers are high efficiency, large ported cabs with mid and high horns that I had to tame. They get "tinnitus" loud!

However, I plan on making a small portable amp that will see various speakers. Having the extra head room seems desirable.

The TDA2050 seems to provide a better output swing at a given PS voltage, but I need to confirm with tests. That and a better power source should get me where I want.