AW: Re: AW: Re: AW: Transistor matching

[Magnus Danielson]

> 
> > Juergen, we just look at this from two diffent perspectives... and there 
> is no
> > point in us building up arguments against each other...
> > You asked an question and I gave my view...
> [...]
> > Currently I am quite tired of the flame-happines that exist on some other
> > lists
> > so I won't go into deep argument without a good reason. I will go back to 
> my
> > semiconductor book that I am supposed to do deep studies in anyhow...
> > If you don't feel my views on the topic comply with yours just leave out, 
> OK?
> 
> Oh. I did not intend anything like this.
> Flames, argues, nothing of this kind.

As I said... I have gone a bit touchy...

> I mean I asked a serious question, because
> I *value* your (and the list's) comments - could
> there be any doubt about that ?

Rigth.

> It happens quite often that I am not sure about some idea
> or circuit. Ok, I have thought about some 1st-order side
> effects, but how can I be sure that I have taken *everything*
> into account. You answered my question, and you gave
> me an warning of some side-effect which I thought I had
> already considerd, but not yet calculated. So now I calculated
> the maximum current change, just to be sure, and found that
> what originally was just a feeling, still held true in calculation.

It's quite often the case that people start beating each other cause there
happend to be a sligth slipp of information to start with. So, I can have read
your initial mail too fast and you migth have written it too fast... and of we
went...

> Now I thought it was a good idea to share the calculation
> with the list.
> I think there is nothing like "I am right and you are wrong" about
> it all. You helped me to to go deeper into the matter, and
> (hopefully) both our posts were of some use for some other
> list members who will also select transistors.

I hope all learn some more...

I just happends to have an semiconductor exam coming up soon so I have to be
good at it anyway.... not to say that I am now.

> I am quite sure that this "first-order-side effect" that we discussed
> is of no harm. But I am still *not* sure if there are other side effects
> that I have not thought of, so far. After all, there *must* be some
> reason that Moog used real current sources instead of resistors.
> Perhaps there is a positive feedback loop that includes heating
> effects? (Vbe getting smaller, and Ic increasing ...). I am not
> sure if it has an effect at 100uA, but at least the temperature
> coefficient would be opposite with real current sources!

I had this in my head as well. There is a definit risk that you can find
yourself either too temperature dependent or getting into an thermal runaway
if you don't watch your steps.

The Vbe has a direct linear relation to the absolute temperature of the
transistor from the approximate formula:

	              Vbe * q
	Ic = Is (exp (-------) - 1)
	               k * T

Where T is in Kelvin degrees. However, to even further confuse things you will
find that the "constant" Is is _also_ temperature dependent. Not to discuss how
it changes with Vce.

If one really wants to be picky (yes! yes!) is the Vbe voltage just a side
effect of several other values, and matching of those paramters would make more
properties become matched. But then, one could go on into infinite digging of
errors beyond what the curcuit solution would account for anyway.

The above single diode relation is an oversimplification as such.

The Moog matching description stresses the need for equally tempered 
transistors
and this is for a good reason. I can dig up much more horrifying formulas for
transistors if specifically requested, but I want a good chance to learn them
first :)

In my humble opinion would a good matching occur with looking at more than one
value. This way would certain other errors be eliminated such as bulk 
resistance. The emitter will effectively have a diffrent value inside the
silicon than we can measure from the outside, this causes a diffrent real Vbe.
Matching for different currents could reduce such errors.

> I hope you see that my aim in discussion is deeper understanding,
> and not arguing. Well, re-reading my posting with a sentence
> like "Who needs an exact current of 100.00 uA ??" makes me
> think if I have hit the right tone. I am not a native English speaker,
> so sometimes I am happy when I can express the technical
> stuff fairly like I intended - sorry if politeness comes short.

Well, I am not an native English speaker either, so that make two of us.

> I never intended to be rude. I value your comments, and I hope
> we will still get them in the future. Peace?

JUERGEN AND I CAN HEREBY DECLARE PEACE!!!

Hmm... not that we had much of a war, but anyway... minor bashing maybe...

About my comments:
As long as they are received well they will continue to come. I have had 
several
nice replies today on earlier comments and helps... simple once, but never the
less.

I'll get back to you all on transistor matching when I have digged deep into
my semiconductor book... how does that sound?

Cheers,
Magnus