Would DLH amplifier, by nature of having Rush cascode, be inherently less susceptible to thermal memory effects?
DLH Amplifier: The trilogy with PLH and JLH amps
DLH Amplifier: The trilogy with PLH and JLH amps
In the Rush circuit, the current goes through two Vbe in series. Their non-linearities add. The Rush circuit is less linear and is more thermally sensitive than a single transistor. There are only a few examples where it is used as a differential input stage.
In the most common LTP differential circuit (needing a constant current source to bias the transistors), two Vbe are in parallel. It works as a push-pull, when current increases in one branch, it decreases in the other. There is a compensation of linearities and the whole is more linear than a single transistor.
Last edited:
Yes it is, both are differential circuits with high impedance inputs.
Both are even similar to a single device which is mainly driven by the voltage across its base-emitter junction (but has a low impedance input, the emitter).
Member
Joined 2009
Paid Member
I would be interested to see the math behind this - I have never been able to fathom any inherent advantage regarding signal distortion between the two options. They are both simple emitter-coupled amplifiers forming a differential amplifier. One is simply a 'folded' version of the other.
As an aside, I've never found there to be any difference in their distortion behaviour when simulated in Spice where I've employed them as error amplifiers in a feedback amplifier. Perhaps their behaviours at large signal swings is where the differences show up - the LTP performs poorly with large signal swings, so I use it when I have high feedback factors and the LTP experiences a relatively small differential signal.
The dc-offset / thermal issues do not recommend the Rush, which is why I've never used it myself over the Singleton. I've found my better sounding projects to have used a Singleton input over an LTP - I credit Hugh Dean with that insight.
From about 1978 some hi end CFP amps have the predriver inclusive VAS always powered on to get a stable temp when the current amp is powered on. That requires three PSU altogether ( an auxilar for remote control). With relatively little mechanical effort one can place all semiconductors which make up predriver in a kind of thermostate which is powered from auxiliar supply. It keeps junction temp at 70-90 °C . As "heat" propagates by diffusion junction temp after power on predriver will of course vary but this simple thing is an improvement at almost no cost
This is about a no improvement because the thermal capacitance of the driver junction is far smaller than all other thermal capacitances involved in this.
Another wrong invention from lack of correct understanding of thermal systems.
No significant improvement over a well designed thermal set up.
Another wrong invention from lack of correct understanding of thermal systems.
No significant improvement over a well designed thermal set up.
I am using it in my high-voltage regulator T-reg.
I was wondering if you could use this as the input stage for a CFA amp, with the feedback going to the 2nd base instead of the 1st emitter. Would that still retain all CFA advantages like gain-independent bandwidth?
Jan
The thermal issues you noticed tell that the devices of the Rush circuit do not compensate their non-linearities between each other, contrarily to what happens with the push-pull arrangement of the two devices of same polarity in the common differential pair.
One of Gerard Perrot's main concern was to avoid thermal modulation of the input stage transconductance by the signals. So clearly, the two transistors Rush circuit is not an appropriate choice if we consider this modulation as having a significant effect, which it is still to be proven in feedback amplifiers, even with a singleton input stage.
This is a figure from Arto Kolinummi's book regarding the input stage thermal distortion.
Here we see that with a cascode, there is overcompensation which is about the same as the original deviation. Bootstrapping works much better. Which in my book is intuitively expected because bootstrapping keeps operation conditions better constant than a cascode.
Jan
Here we see that with a cascode, there is overcompensation which is about the same as the original deviation. Bootstrapping works much better. Which in my book is intuitively expected because bootstrapping keeps operation conditions better constant than a cascode.
Jan
Last edited:
Assuming you addressed me? Thermal distortion comes from slow heating of the device which changes the parameters. Cascode keeps the device voltage wrt ground constant but not the device Vce, and not the Ic. The graph I posted showed that there actually is a reverse effect.
A bootstrap keeps the Vce constant and that appears to work much better, as shown.
The transistor doesn't know whether we call something 'Vas' or 'pupukaka' so that makes no difference ;-)
Jan
No significant effect on a yet to be proven memory distortion.
It will reduce heating of the case. It won't reduce thermal modulation of the junction.
That is a matter of thermal time constant at the junction and a correct modeling of this thermal system.
A Rush circuit is a misommer if you call it a cascode. Note the fundamental role played by the Sziklai pairs in Perrot's patented circuit. Consider the whole circuit, not parts of it, to understand how is reached its aim to avoid thermal modulation by the signal.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.