Vibrating art

[cloudywindy]

I am doing research on the effect of vibrations on objects of art, for example, in transport, problems with construction right next to a museum. For experiments in a lab, I have been trying to amplify the signal of a standard scientific function generator using an amplifier which then goes to a speaker. One was a consumer Sony stereo amplifier, another, a bass guitar speaker with built in amplifier. It seems that the amplifiers don't like this. If I run an experiment for more than a half hour, they die. (The speakers are ok.) I'm not an electronics whiz, but I remember something about impedance matching. Do I need to put something between the function generator and the amplifier?

 

[audioguru]

Audio amplifiers are designed to play music that is never a continuous tone. With music then the level keeps changing so the average level is about 1/10th the maximum level.

Use an amplifier that has at least 4 times as much power as you need and keep its volume control turned down so that its output level is half of its maximum level.

 

[alec_t]

Something's clearly wrong if you're burning out stereo/guitar amps. Over-driving them perhaps? Too high a supply voltage?

Do I need to put something between the function generator and the amplifier?

That depends on what the output voltage and impedance of the function generator are. You should be able to get both from the gen's spec, or else they could be readily measured. The amps shouldn't need more than a volt or so of signal to drive them to max output.

 

[JimB]

It seems that the amplifiers don't like this. If I run an experiment for more than a half hour, they die. (The speakers are ok.)

Just so that we understand completely, you are saying that the amplifiers fail completely and never work again,
but,
the speakers do work OK after the amplifiers have failed.

I'm not an electronics whiz, but I remember something about impedance matching. Do I need to put something

In some circumstances, impedance matching is useful, sometimes essential.
However, just to drive an audio amplifier from a function generator the impedance matching does not really matter, you can ignore it.

What could be a problem, depending on the output voltage from the function generator you could be overdriving the input stage of the amplifier.
How high is the volume control det on the amplifier?

Also, how loud are your "vibrations"?
It could be that you are running the amplifiers beyond their ratings and overheating them, at which point they fail.

Be aware that many consumer grade audio amplifiers are grossly overrated, most cannot provide anything like their rated power for more than a few seconds, if that.

JimB

 

[dr pepper]

I think the problem is within the output stage of the power amp rather than a problem caused by input impedance mismatch, sig gens tend to have low z output, and power amps hi z commonly 47k.

As allready mentioned power demands are greater with a continuous wave applied to the input, also some power amps might not like square wave signals (even a sine wave badly distorted resembles a square wave), square waves can make the amp unstable due to the harmonic content, a lot of mosfet amps have a zobel network this is to dampen any hf oscillations, I suspect that in your case you've fried the components for this and the amp has gone into oscillation and toasted the output devices, but this is just an approximation before anyone shouts.

By the way you can get vibration transducers, designed to be fitted to dancefloors or even armchairs/car seats, these can be driven from an audio amp, one of these might be usefull for your application as they are designed to transmit audio vibrations into an object.

 

[cloudywindy]

Thanks for all of your tips. We actually do test with rock music (no joke), as a realistic simulation of what happens in or near a museum doing a concert.

But the function generator work is for the more fundamental research, finding out where the resonance of an object is, and then finding out how long it takes at that frequency before it shows damage (technically, the word is "fatigue damage"). Those are longer experiments, so it does seems to make sense that the amplifiers don't like a continuous signal (we're using sine waves). The one consumer amplifier we were using actually shut itself off the first time at about half of it's maximum level ("protect" function); it died when we tried again just below that level for too long. And the guitar amplifier also worked for an hour or so before it died.

Our speakers still work. I'm now considering one of the bigger D.A.S. sub-woofers, the Action 18A model. That actually has a whole lot of power (rating 1000 w max, 500 w continuous). For us it's not a question of how loud, but how much vibration load, but it essentially comes down to the same thing.

So if I understand it, I should stay at half maximum output (volume control) of the amplifier to protect it during extended testing?

Using a vibration transducer is a thought, although I wouldn't I still have the same problem with the function generator - amplifier setup?

 

[alec_t]

Seems like you need more powerful amps. What frequency range are you using?

 

[cloudywindy]

Seems like you need more powerful amps. What frequency range are you using?

We're looking at frequencies below 150 Hz. Alot of the first mode resonances we come across fall between 25 and 75 Hz.

 

[dr pepper]

Amplifiers tend to work harder at those sort of freq's for various reasons.
You'll need to derate the amp to keep them from tripping or blowing, using the volume control isnt an effective way as you could overcome this by winding up the input signal.
If you use a power amp (one without tone controls etc) then you can estimate the output power from the input level, 0.775v rms is a common input level for max output, you could start with say 1/3 max output and see if the amp gets hot or protects itself, you can get power meters to measure the output one of these might be advisable if its a large project.
A good quality true rms multimeter will measure fine up to a few hundred hz.
With a sine wave you should be able to do what you want.

 

[cloudywindy]

Amplifiers tend to work harder at those sort of freq's for various reasons.
You'll need to derate the amp to keep them from tripping or blowing, using the volume control isnt an effective way as you could overcome this by winding up the input signal.
If you use a power amp (one without tone controls etc) then you can estimate the output power from the input level, 0.775v rms is a common input level for max output, you could start with say 1/3 max output and see if the amp gets hot or protects itself, you can get power meters to measure the output one of these might be advisable if its a large project.
A good quality true rms multimeter will measure fine up to a few hundred hz.
With a sine wave you should be able to do what you want.

Thanks for your help. Sorry, I'm not very electronics oriented, just a dumb materials scientist. I understand the part about the volume control. Both the function generator and the amplifier have volume/level controls so I can't set one without thinking about what I do with the other.
Some numbers from the specs I have: The function generator (Toellner TOE7741) we have puts out "approx. 45 mVpp to 45 Vpp resp. approx. 22 mVpp to 22.5 Vpp in pulse mode. ... Output impedance approx 0 ohm, the output is short and open circuit proof." The active sub-woofer we're now thinking about using has a Nominal LF Amplifier Power of 1000 W peak, 500 W continuous, and a Frequency range (-10 dB) of 35 to 160 Hz.
If I understand you correctly, I need to start at some low input level, say 2 Vpp (controlled by the function generator) with the amplifier at 1/3, thus 1/3 volume control (333 W) on the amplifier, assuming the control is linear. Then see if the amplier protects itself over an "extended" period of time, is that correct?

 

[audioguru]

I worked for a PRO audio manufacturer for a few years. Their best amplifier had a guarantee that said, "No questions Asked, if it fails it will be repaired or replaced". Hundreds were sold and none failed.

My boss asked me to test a competitor's PRO amplifier. With a sinewave input and at its rated output power it smoked and died in about 2 minutes. I tested our best amplifier at full power for a week and it heated the office very well but did not fail.

 

[alec_t]

I need to start at some low input level, say 2 Vpp

Even that would probably be too much. Half that should suffice.

 

[kinarfi]

How are you hooking the gen to the amp to the speaker, are you using a capacitor to prevent the dc component from passing?

 

[audioguru]

Even that would probably be too much. Half that should suffice.

You should be interested in the output level, not the input level since every amplifier has a different amount of voltage gain.

The amplifier actually cools a little if it is clipping since then it is not linear but is simply switching its output transistors on and off.
I think an amplifier heats the most when driving its rated load when its power is about half of when it clips. Then its output voltage is about 0.71 times when it clips.

 

[alec_t]

You should be interested in the output level, not the input level

For the OP's measurements both input and output levels are important, IMO. If the amp input is being overdriven to a point that clipping occurs then strong harmonics of a low fundamental frequency could still be within the frequency band of interest and give misleading resonances in the artefacts being tested. And of course output level is also important for the survival of the amp

 

[audioguru]

It would be ideal to have an oscilloscope looking at the output to see if there is clipping and to measure the p-p output level. Then it doesn't matter what amount the input level is because the input level sets the output level.

 

[alec_t]

Agreed. That would aid in setting the output at a suitable level to avoid frying the amp.
Out of curiosity I ran a simulation to see how amp dissipation varies with signal input level. Admittedly the only amp model I had was of a low-power opamp, so the result may not be representative of a beefier amp.
The result surprised me, showing a monotonic drop in dissipation as sig level increased. Clipping occurred for input amplitudes above 16mV in this sim, but that point in the curve doesn't show any marked change in the slope. Odd.
View attachment 67606

 

[audioguru]

Your graph of an opamp heat dissipation is meaningless because the opamp probably did not have a load.
Connect a load then watch the opamp heat up like crazy at about half the power of when it is clipping, except it might limit the output current before clipping.

Here is a graph of a hi-fi amplifier IC that heats like most linear amplifiers:

 

[alec_t]

Your graph of an opamp heat dissipation is meaningless because the opamp probably did not have a load.

It did have a load; 660 Ohms. I didn't try the effect of varying the load.
I prefer your graph, though . More like what I would have expected. So clearly scaling from opamp to power amp isn't valid (as I suspected).

 

[KJ6EAD]

You can improve the performance of your amplifier by adding forced air cooling over the heat sinks.