Search the web and found this similar to the Inverse piezoelectric effect. This effect happen to most materials.
Basic principle of operation
http://www.ceramtec.de/intl/e4/ta=e4*240/la=en/ta0=ind*3/ta1=gb*108/ta2=e3*420/le=3/
Basic principle of operation
http://www.ceramtec.de/intl/e4/ta=e4*240/la=en/ta0=ind*3/ta1=gb*108/ta2=e3*420/le=3/
I think Yamaha or something made a feature about puttting rubber blocks between heatsink fins to stop ringing decades ago.
They did not distinguish whether this was acoustic feedback caused HS fin ringing, or transistor/current caused.
I don't think they made actual claims about this being a benefit, except for saying that these rubber blocks stopped the HS fins ringing.
Just like a phono stylus sending direct audio into the room, so does a heatsink - I prefer not to have these sounds in the room alongside the speakers sound.
Eric.
They did not distinguish whether this was acoustic feedback caused HS fin ringing, or transistor/current caused.
I don't think they made actual claims about this being a benefit, except for saying that these rubber blocks stopped the HS fins ringing.
Just like a phono stylus sending direct audio into the room, so does a heatsink - I prefer not to have these sounds in the room alongside the speakers sound.
Eric.
Yeah my T03 Mosfet amp does it too, although I hadn't narrowed it down to the mosfets.
http://www.diyaudio.com/forums/showthread.php?postid=284415#post284415
Above was a post where I mentioned it...... Was driving into a dummy load of about 7.5 Ohms, also amp would have been close to clipping, got much louder if I drove it into clipping.
The weird thing was it also started doing it when I was doing tests using rmaa (but only when there was no input signal.... I put my multimeter (in freq mode) on the speaker output and it was registering 56Khz. If I unpluged the sound card from the amp the sound went away and so did the 56Khz..... All very weird
Tony.
http://www.diyaudio.com/forums/showthread.php?postid=284415#post284415
Above was a post where I mentioned it...... Was driving into a dummy load of about 7.5 Ohms, also amp would have been close to clipping, got much louder if I drove it into clipping.
The weird thing was it also started doing it when I was doing tests using rmaa (but only when there was no input signal.... I put my multimeter (in freq mode) on the speaker output and it was registering 56Khz. If I unpluged the sound card from the amp the sound went away and so did the 56Khz..... All very weird
Tony.
"What have you been smoking, send me some."
Jim W.
I really don't appreciate your comments here on this site!
Since you made that rather insulting post its pretty obvious that alot of other people have also heard this same effect. I heard the effect loaded but some amps would also sing somewhat if unloaded. Though usually not as loud.
Mark Gulbrandsen
Jim W.
I really don't appreciate your comments here on this site!
Since you made that rather insulting post its pretty obvious that alot of other people have also heard this same effect. I heard the effect loaded but some amps would also sing somewhat if unloaded. Though usually not as loud.
Mark Gulbrandsen
It is not a thermal effect. It is not due to the devices being lateral or vertical. The conduction paths of the silicon differ in each case by mils..
Thermal: the thermal time constant of the junction is 10 to 100 usec..through the die is 1 mS..to the case..10 to 100 ms..to the heatsink..10ms to 2-3 seconds depending on the isolator.
Bonding wires will move from lorentz forces at all frequencies...amplitude based on current..when the freq gets up into the tens of kilohertz, the wires can resonate, which will lead to fatigue failure( this can be seen on autopsy by slip deformation on the surface of the wire, you will see the clean surface of the wire become rough in a crystalline fashion near the break.
I would have a hard time believing the bonding wire movement would be heard outside the case..and, noise from an encapsulated IC certainly does eliminate that.
Leaving movement of the wires leading to the device, or movement of the actual device on it's isolation mount.
Cheers, John
Thermal: the thermal time constant of the junction is 10 to 100 usec..through the die is 1 mS..to the case..10 to 100 ms..to the heatsink..10ms to 2-3 seconds depending on the isolator.
Bonding wires will move from lorentz forces at all frequencies...amplitude based on current..when the freq gets up into the tens of kilohertz, the wires can resonate, which will lead to fatigue failure( this can be seen on autopsy by slip deformation on the surface of the wire, you will see the clean surface of the wire become rough in a crystalline fashion near the break.
I would have a hard time believing the bonding wire movement would be heard outside the case..and, noise from an encapsulated IC certainly does eliminate that.
Leaving movement of the wires leading to the device, or movement of the actual device on it's isolation mount.
Cheers, John
That does still not fully rule out thermal causes. If the sound is generated thermally on the die then it just has to find a way out of the package.
Though I also opt for magnetic or piezoelectric effects. This would in turn also mean that it might be good to pay a little attention on mechanical resonant behaviour of an amp's construction.
So maybe we should even copy the "Spruce amp" ?!
Regards
Charles
Mark,
I was just kidding with you, just because I never heard the effect does not mean it does not exist in some form. It seems that some people are talking about TO3 cans and others are talking about heat sinks, which could be two different issue or maybe not. This could be something like a sympathetic oscillation caused by the inductance in the DC power leads with the TO3’s and coupled with the transformer when under load.
Some one that has experienced the problem and can repeat it should look at the issue. I just have never heard or experienced this. Having said, I believe this is real but is not caused solely by the bonding in the TO3’s.
I was just kidding with you, just because I never heard the effect does not mean it does not exist in some form. It seems that some people are talking about TO3 cans and others are talking about heat sinks, which could be two different issue or maybe not. This could be something like a sympathetic oscillation caused by the inductance in the DC power leads with the TO3’s and coupled with the transformer when under load.
Some one that has experienced the problem and can repeat it should look at the issue. I just have never heard or experienced this. Having said, I believe this is real but is not caused solely by the bonding in the TO3’s.
phase_accurate said:
That does still not fully rule out thermal causes. If the sound is generated thermally on the die then it just has to find a way out of the package.
Though I also opt for magnetic or piezoelectric effects. This would in turn also mean that it might be good to pay a little attention on mechanical resonant behaviour of an amp's construction.
So maybe we should even copy the "Spruce amp" ?!
Regards
Charles
The die has a TCE of 3.3 ppm/degree C. A reasonable sized power die will undergo temp excursions in the 10 C or so range maximum at the surface of the die for frequencies that are faster than the die to header time constant.
A reasonable die is 250 mils by 250 mils.
.25 inches * 3.3parts *10 e-6 (per million) * 10 degrees C =
.25 *3.3 *10e-5 = .825 *10e-5 = 8.25 microinches.
Hmmm..that seems pretty small.
Also, don't forget..that movement is lateral..the die width and length is moving 8 millionths of an inch sideways, not up and down. Since that movement is symmetrical about the center of the die, there is no coupled vibration which is going through the die attach material (solder or braze).
I'd personally discount the die thermal as the possibility..
For a to-3, it's possible that the lead pins are make of kovar, which is a magnetic matched expansion material used for going through glass seals, but mrf said he heard it on a plastic encapsulated IC, which is purely copper based construction.
Cheers, John
PS...Boy, do I love this crap...
Oh, almost forgot this calc of vertical movement.
Since the die temp excursion is 33 ppm (assuming 10 C), the vertical movement (assuming even HALF the die thickness is being used) would be 10 mils times 33ppm, or .33 microinches.
And to consider the silicon as a reaction mass against which the die size change can push, a quarter inch square of 10 mil silicon VS a 60 mil thick copper nine times the area, 54 times the volume, about 100 times the mass..the case of the package will move about .0033 microinches, or 3.3 billionths of an inch as a result of silicon thermal expansion mass reaction.
Still think it's the external reactions...but I may be wrong..
Cheers, John
Since the die temp excursion is 33 ppm (assuming 10 C), the vertical movement (assuming even HALF the die thickness is being used) would be 10 mils times 33ppm, or .33 microinches.
And to consider the silicon as a reaction mass against which the die size change can push, a quarter inch square of 10 mil silicon VS a 60 mil thick copper nine times the area, 54 times the volume, about 100 times the mass..the case of the package will move about .0033 microinches, or 3.3 billionths of an inch as a result of silicon thermal expansion mass reaction.
Still think it's the external reactions...but I may be wrong..
Cheers, John
first of all, I have been using mosfets for over two decades and have never heard it.
2ndly, is it possible that other parts of the amp are singing?
3rdly, the construction of the mosfet is very much like a capacitor. so if capacitors can sing (which I have herad), maybe mosfets can sing too.
2ndly, is it possible that other parts of the amp are singing?
3rdly, the construction of the mosfet is very much like a capacitor. so if capacitors can sing (which I have herad), maybe mosfets can sing too.
jewilson: I understand you have some strong feelings here, but the personal insults are totally out of bounds and against forum rules.What !!!!!!!
"3rdly, the construction of the mosfet is very much like a capacitor."
That is not even close to accurate. First off you would have to specify the type of capacitor since electrolytic, metalized film, and film and foil capacitors have different construction even among similar types. Winding tension, outer tape, and whether or not the cap is round or oval in cross section. Different monolithic ceramic capacitors have greatly different piezo electric characteristics. Which type of cap in particular do mosfets resemble? None of them would be my answer. I have heard audible noise from amplifiers when driving high currents into resistive test loads. There are a variety of mechanisms that can cause this noise. This is not that rare an occurrence as most people who have run high current test into resistive loads. What is all the controversy? This is high school physics stuff and is not that complicated or mysterious.
"3rdly, the construction of the mosfet is very much like a capacitor."
That is not even close to accurate. First off you would have to specify the type of capacitor since electrolytic, metalized film, and film and foil capacitors have different construction even among similar types. Winding tension, outer tape, and whether or not the cap is round or oval in cross section. Different monolithic ceramic capacitors have greatly different piezo electric characteristics. Which type of cap in particular do mosfets resemble? None of them would be my answer. I have heard audible noise from amplifiers when driving high currents into resistive test loads. There are a variety of mechanisms that can cause this noise. This is not that rare an occurrence as most people who have run high current test into resistive loads. What is all the controversy? This is high school physics stuff and is not that complicated or mysterious.
sully, if the heating starts at a surface region, the time contants don't enter into it. You can heat and cool a surface region pretty fast, and that can couple to the surrounding air (look up Rosencwaig's work on the photoacoustic effect, for example). I'm not saying this is the mechanism- I think some sort of effect like what you speculate or similar is more likely- but surface effects can be funny and they don't depend on heating up the entire chip or even the entire junction.
SY said:
Agreed in principle..but only for the top of the chip.
On the bottom of the case, the heat pulses (no cooling pulses here) will have been low pass filtered out the wazoo..with a time constant in the tens to hundreds of milliseconds, the filtering is strong.
But I will admit...pulses of hot air pinging the underside of the to-3 case like an ink jet printer would be pretty neat...the to-247 case of course, does not have that air..
Cheers, John
- Status
- Not open for further replies.