Distortion through Early Effect - mainly low Order or high Order Components ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jim Early first characterized this effect due to an effective decrease in the base width by BjT's because of the widening of the base-collector depletion region, resulting in an increase in the collector current with an increase in the collector voltage.
What kind of distortions are to expect, if the Early effect is very strong (large values of early voltage VA) ?
Is it rather even-numbered or odd-harmonic?
so as more low order or high order distortions?
Thank you for your comments.

here some informations from Mr. James M. Early:
James M. Early
 
Typical misinterpretation: because the characteristics looks linear does not imply that E.E. is a linear effect: to qualify for this, the characteristics should remain parallel for different base currents, which they don't: they all converge towards a (hopefully) single point, which happens to be the Early voltage.
This implies that the output resistance varies with current.

Early effect is the main distortion cause in the low-power, voltage amplification stages of an amplifier
 
In order to produce a voltage variation at its output, an amplifying stage needs to modulate its current; since the synthetic output resistance resulting from the Early effect varies with the level of current (as shown by the different slopes of the characteristics), it generates non-linearities... But you are entitled to your opinion...
 
Member
Joined 2011
Paid Member
Easy enough to measure. Just connect a series RC between collector and emitter. The R is effectively in parallel with the transistor's hybrid pi "ro" (incremental signal Early effect), thus you have externally modified the Early effect. The series C is simply a DC blocking capacitor. Choose C >= 1/[(6 rad/sec)*R] and you'll be fine.

Measure distortion without and with the series RC. Does distortion go up, go down, or stay the same, when you increase the Early effect by decreasing "ro" ??
 
I do ont agree. I see dV/dI = constant.
Distortion does not come from Early effect per se. This so called distortion is an indirect effect.
No matter how you turn the problem: to generate non-linearities, you need a source of non-linearity, ie. some parameter varying with the signal level, and the Early effect is the origin of this non-linearity, because it is a non-linear phenomenon.

If you model an amplifier using the small signal transistor model (current sources, resistances and conductances), there is no way to generate distortions.
Adding the Early effect makes the output conductance (or resistance) non-linear and causes distortions.

The Early effect can also be seen as a β increase when Vce increases, which is a non-linear effect, exactly like the voltage across a diode increases its conductance, just more subtle. Of course, you can try to argue that this too is an indirect effect...
 
What kind of distortions are to expect

Those distortion have an orders of domination in the amplifiers.

What can be done to reduce Early Effect in Wilson current mirror?

You need to put any common-base/gate stage inside to decouple working device collector from high output potential change.
Just use full-Wilson in the input shoulder and add common-gate at the output shoulder.
 
:cop:

Posts containing links to material under copyright have been removed.
was it not only a preview from google-books like this URL?
Analog Circuit Design: Art, Science, and Personalities - Google Books

Early effect is most easily thought of in a simplified way as the current gain of the transistor increasing somewhat as base-collector voltage increases. Its all about base current and where it goes. Early effect is not linear and the idealized model of the extrapolated Ic vs Vbe curves converging on a single point at -VA is just that - idealized. But it is a helpful approximation for insight. Most importantly, Early Effect should not be confused with quasi-saturation effects at low values of Vce.

Because Early Effect can be seen as a potentially nonlinear change in current gain with changes in Vce, circuits whose small-signal gain is less affected by transistor current gain are less susceptible to distortion from Early Effect. For example, a 2T VAS with some emitter degeneration is far less susceptible to Early Effect distortion than a 1T VAS without emitter degeneration.

A cascode has far less Early effect distortion, but it is not gone completely, since the cascode transistor still has base current that is a function of the Vce of the cascode transistor. Follow the base current! Some circuits can "recycle" the base current of the cascode back to the emitter of the CE stage that lies below, largely reducing the "loss" of that base signal current and further increasing the output impedance of the cascode stage.

The Wilson current mirror is sometimes not all that it is cracked up to be, at least in the discrete world where transistor current gains of the 3 transistors may not be matched as well as they are in an integrated circuit implementation. The Wison current mirror achieves its very high output impedance by cancellation of various base currents in the circuit. Once again, it is based on the "follow the base currents" approach. As such, it will not function as well in the real discrete world when current gains of the devices are not matched.

In a single-ended VAS, Early effect distortion will be mainly manifested as even-order distortion. In a push-pull VAS, Early Effect distortion will typically be lower and more odd-order in nature.

Cheers,
Bob
from post 38 under
The best audio amplifier books - Overview (Google books)

Thank you very much for this explanations.
 
Member
Joined 2011
Paid Member
I would like to gently remind folks that MOSFET amplifiers produce distortion through the same nonlinearity (namely: gds = dIds/dVds is not constant independent of bias) even though MOSFETs have infinite beta. There's zero base current in a MOSFET so there's no "base current error" for a Wilson current mirror to compensate. You'll have to come up with a better mental model than "Beta varies with operating point" to explain what's actually going on.
 
Administrator
Joined 2004
Paid Member
Hi Mark,
even though MOSFETs have infinite beta.
Only under steady state conditions. When you are looking at a dynamic situation you do draw gate current (gate charge). The higher the frequency or higher slew rate waveforms are, the higher the gate charge current will be.

Anyway, it just makes things more fun when you consider the dynamic conditions.

-Chris
 
The Early effect is due to the collector-base depletion region widening with increasing Vcb. As the depletion region widens in the base the effective base width is reduced. As a consequence, the transistor "sees" a higher Ic or effective gain. Sorry Bob, it's not base current - that is only the current which does not make it through to the collector.
Because the depletion region follows roughly a square root law with voltage, it is not linear, but it also depends on the doping profile of the base (as it is the behaviour in the base which is important here). So we have a non-linear increasing gain with voltage which generates the distortion.
 
Is it possible to have someone post a LTSpice file that could be great learning tool for the discussion here. I personally would be delighted to be able to play with such a circuit example, and I dare say some others too.


Thanks a lot for any positive results in advance.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.