cancelling memory distortion ?

[mikelm]

If a audio dual matched transistor pair is used at the i/p of a balanced amp and each half fed with the two differential phases of the i/p signal, will the alternate heating and cooling effects on each device cancel out due to their proximity.

can we assume that the two devices are close enough for this to occur ?

can anyone think of any drawbacks of such a scheme ?

would the cancelation cover all relevant frequencies ?

does anyone know about this kind of stuff ?

mike

[johnferrier]

Have you seen the following site (I happened on to it this morning). I think the author is a diyAudio member...

http://peufeu.free.fr/audio/memory/


JF

[mikelm]

yes Iv'e read most of it twice !

However I don't think that this idea is covered there

If it works it is a very simple remedy in a balanced approach

[ThorstenL]

Konnichiwa,

Originally posted by mikelm
If a audio dual matched transistor pair is used at the i/p of a balanced amp and each half fed with the two differential phases of the i/p signal, will the alternate heating and cooling effects on each device cancel out due to their proximity.

can we assume that the two devices are close enough for this to occur ?

Absolutely not, UNLESS the transistor are on the same chip and the chip has a low thermal intertia. I think there are way too many if's and but's in this to rely too much on this.

Originally posted by mikelm
would the cancelation cover all relevant frequencies ?

As a rule, the higher the frequency the lower thermal distortion effects as the thremal inertia of the actual semiconductor chip becomes relevant and limits temperature changes.

Sayonara

[millwood]

Originally posted by Kuei Yang Wang
As a rule, the higher the frequency the lower thermal distortion effects as the thremal inertia of the actual semiconductor chip becomes relevant and limits temperature changes.

Sayonara

Not entirely. there are two "thermal effects" going on in a semiconductor component.

One is the steady state thermal heating. Let's say that you apply a constant sine wave, regardless of its frequency, the ohmic heating will reach an equilibrium after a period of time, and die temperature will stabilize at that point. the operating point of the component will, as such, shift depending on the signal. You will hear people talking about "thermal resistance / resistivity".

Another is the transient response of a device. the temperature of a device will go up and down with the signal, or being modulated by the signal. You will hear people talking about "thermal impedance" in those cases.

both effects are thermal-induced but its mechanism is slightly different.

[mikelm]

Originally posted by Kuei Yang Wang
Absolutely not, UNLESS the transistor are on the same chip and the chip has a low thermal intertia. I think there are way too many if's and but's in this to rely too much on this.

I was thinking of the SSM2220P and SSM2210P with dual devices designed specifically for audio i/p

I was thinking that the two devices would be physically & thermally quite close.

does any one know what this inside of these things are like ?

[jan.didden]

I don't know about these devices, but I do know that the similar LM394 dual consist of a chip of many (IIRC hundreds) of individual transistors, parallelled to form two transistors. The individual transistors are physically spread over the chip for two reasons:
1 averaging out the electrical characteristics for electrical matching;
2 idem for thermal matching.
This seems to be a standard technique, so I would expect (but don't know for sure) the SSM's to be similarly constructed.

Jan Didden

[Nelson Pass]

This approach is standard in chip op amps, where the diff
transistors are made at the same spot, thus matching them,
and have the same temperature. Op amp performance
depends on such tricks, as the ability to select the components
on the chip is very limited.

[mikelm]

Thanks, this sounds quite encouraging. I'll go ahead and do some experiments.

The only further question is one of max current

All these devices quote 20mA as the absolute max
and the max voltage is around 30V

I want to use a higher current perhaps 35mA in the cct I am thinking of but only 7 volts.

Given this very low voltage will a higher current damage the devices even though the power is far below the max

If this is the case, I do not understand why this would be.

or...Will I have to paralell them ?

Can anyone shed some light on this ?

[Faber]

Originally posted by johnferrier

http://peufeu.free.fr/audio/memory/

I've tried to modify the Balanced Line Stage with the suggestions depicted in that site and the result is great in terms of stage and darkness of the scene.
In some particulars the original design is better (armonics, warm) but they are quite negligible.
I'm building a second original-design BLS, so to make more in deep analysis.

Anyone else tried something similar?

Happy new year!