Of course you are correct, Edmond. However, in the real world, bipolar transistors have many limitations due to nonlinear beta, exponential transfer function, nonlinear capacitance, and early effect, and perhaps some other effects. The only advanced way to use them is followed by design procedures put forward by Barrie Gilbert and other advanced designers.
Jfets have their own limitations as well, but they can be quieter in difficult situations, such as with an input pot in series with their gate, and they have low base current. Complementary jfets do a great job of distortion cancellation as well.
Jfets have their own limitations as well, but they can be quieter in difficult situations, such as with an input pot in series with their gate, and they have low base current. Complementary jfets do a great job of distortion cancellation as well.
Edmond Stuart said:
You are messing up a few things but essentially you are correct and this is the model originally developed by Ebers&Moll. Ebers-Moll and Gummel-Poon (which is essentially current oriented) models lead to exactly the same results. Reason why Gummel-Poon is preferred is that it allows a "black box" approach. In the Ebers-Moll approach, the transistor model is more 'grey' than in the Gummel Poon model because it begins to describe and breakdown the operation of the transistor into several elements/regions/types of operation. The Gummel-Poon black box approach is also the reason why it is used in Spice.
I strongly disagree though with the view of a BJT as a gate leaking FET. The two devices have completetely different physics and conduction mechanisms both at large and small signal levels. The small dignal models may look similar (after all, the BJT model is similar to a JFET with finite rb'e) but the devil is in the dependency of the transconductance on the bias point. At large signal though, the BJT and JFET models and behaviours are wildly different.
And here are the mods for the unbalanced version. Please notice that this circuit has the same gain, instead of half of it. A disadvantage is the need for a three pole switch. The distortion of the input buffers is much lower of course, as now they see a much higher impedance.
Cheers,
E.
Cheers,
E.
Attachments
syn08 said:
Hm... a model is a model is NOT a model (opposed to your rose).
Enfin, here they are:
*** N-Channel JFET
.MODEL 2SK170BL NJF (AF=500.504m BETA=62.7612m CGD=19.8997p CGS=24p
+ KF=.002229965f LAMBDA=1m RS=8.03465 VTO=-420.565m)
*** P-Channel JFET
.MODEL 2SJ74_BL PJF (AF=500.257m BETA=46.351m CGD=106.132p CGS=73p
+ KF=.0005386405f LAMBDA=1m VTO=-409.414m)
Please, give me also your models.
Edmond Stuart said:
.subckt P2SK170 D G S
+ params:
+ BETA=59.86m
J_P2SK170 D G S J2SK170
*
.model J2sk170 NJF(Beta={BETA} Rs=4.151 Rd=4.151 Betatce=-.5 Lambda=1.923m
+ Vto=-.5024 Vtotc=-2.5m Cgd=20p M=.3805 Pb=.4746 Fc=.5
+ Cgs=25.48p Isr=84.77p Nr=2 Is=8.477p N=1 Xti=3 Alpha=10u Vk=100
+ Kf=111.3E-18 Af=1)
.ends P2SK170
.subckt P2SJ74 D G S
+ params:
+ BETA=92.12m
J_P2SJ74 D G S J2sj74
*
.model J2sj74 PJF(Beta={BETA} Rs=7.748 Rd=7.748 Betatce=-.5 Lambda=4.464m
+ Vto=-.5428 Vtotc=-2.5m Cgd=85.67p M=.3246 Pb=.3905 Fc=.5
+ Cgs=78.27p Isr=129.8p Nr=2 Is=12.98p N=1 Xti=3 Alpha=10u Vk=100
+ Kf=26.64E-18 Af=1)
.ends P2SJ74
Adjust BETA for the desired Idss value. The subckt values for beta are the default values.
Don't buy it!
Sorry PMA, I don't buy it. I have built excellent sounding circuits with only U440 JFET as no complementary part is available.
Also many ICs have a non-complementary frontend and can sound good.
Also the 2SK389 and the 2SJ109 are not truely complementary; f. a. they have grossly different capacities.
PMA said:
Sorry PMA, I don't buy it. I have built excellent sounding circuits with only U440 JFET as no complementary part is available.
Also many ICs have a non-complementary frontend and can sound good.
Also the 2SK389 and the 2SJ109 are not truely complementary; f. a. they have grossly different capacities.
Btw-2
A few months ago when I was exploring for a new design I was looking into a similar servo loop that is essentially running on the IPS common mode rather than interfering with the signal path. The overall AC (distortions, etc...) experimental results were nothing to call home about so I gave up this whole idea. The servo idea could be though interesting.
A few months ago when I was exploring for a new design I was looking into a similar servo loop that is essentially running on the IPS common mode rather than interfering with the signal path. The overall AC (distortions, etc...) experimental results were nothing to call home about so I gave up this whole idea. The servo idea could be though interesting.
Wavebourn said:
the problem with constant current + constant voltage
is always = no gain.
And you know what they say:
- no gain
- no pain
So, we can replace your constant 'amplifier' with one WIRE,
which is the ideal 'no pain' device
Now, if we want our transistors to perform gain = amplify
we have to take at least a tiny bit of pain.
We have to make our transistor work not in one single point
of the current voltage diagram showing transfer curves.
Now, to make this transfer operation point very small .. close to one single x,y point
is of course a good way to ensure almost constant transfer conditions.
But the smaller this very tiny line is .. the less gain we get.
Which in turn means usign more devices, each with a tiny gain
----------
My approach has always been Simplistic, like some Nelson Pass amplifiers.
Keep it simple = As few devices as possible.
When doing this, we go for more gain in each device
while trying to make at least one parameter rather constant.
for example current, or for example voltage.
Depending on if you use MOSFET or BIPOLAR or TUBES
they are a bit different in what parameter we try to get more constant.
Because different semi conductors have different characteristics vs. current and/or voltage gains.
JFET does not behave like one BC550C in all aspects.
As you may know.
And this is what a good designer will know
and make his circuits work well using this very valuable knowledge.
Lineup
Re: Btw-2
I really hope you didn't designed this kind of BS by yourself. You cannot correct an offset (by definition a differential thingy) by fiddling with common mode signals (provided that it's 100% common mode and 0% differential) Who invented this?
A servo is always interfering with the signal path, otherwise it cannot remove an offset.
syn08 said:
I really hope you didn't designed this kind of BS by yourself. You cannot correct an offset (by definition a differential thingy) by fiddling with common mode signals (provided that it's 100% common mode and 0% differential) Who invented this?
A servo is always interfering with the signal path, otherwise it cannot remove an offset.
syn08 said:
Guys,
A question, why do you parameterize around beta? I would do it around Idss. In an ideal model beta is pretty much constant. If you take
Toshiba's plot from the 2SK170 datasheet at 1mA (for instance) just about any part you grab will have 15mS gm ie. constant beta. I know the graph does not fit that well to the square root function on these short channel devices but I consider that a second order effect.
Like so...
vto = -sqrt($IDSS/whatever),beta = whatever
Attachments
- Status
- Not open for further replies.