I have been playing around with Ltspice and have a question regarding using emitter degeneration on an input long tail pair. It seems that by dropping the value of resistance the distortion also drops. I would have thought it should increase??
I am using very linear transistors to start with and was just wondering two things. If using transistors with good linearity to start with, what advantages are there still to be had by emitter degeneration, and why would the distorion from the simulator less without them. Is it because of the gain increase?? Something of a noob here. Been reading up, but the simulation tells me different to what I ( think I ) understand.
Thanks.
Paul.
I am using very linear transistors to start with and was just wondering two things. If using transistors with good linearity to start with, what advantages are there still to be had by emitter degeneration, and why would the distorion from the simulator less without them. Is it because of the gain increase?? Something of a noob here. Been reading up, but the simulation tells me different to what I ( think I ) understand.
Thanks.
Paul.
Emitter degeneration will generally make a simulation look worse but due to transistor differences in the real world the use of degeneration is a good thing. If you were to tweak the SPICE models of the LTP transistors to be slightly different from one another in a few key parameters you may actually see the addition of the degeneration resistors reduce distortion.
When simulating, the LTP is basically using perfectly matched devices and therefore shows somewhat unrealistic performance. As a result, the additional gain achieved by omitting the degeneration results in greater open loop gain and derives more benefit from global negative feedback.
When simulating, the LTP is basically using perfectly matched devices and therefore shows somewhat unrealistic performance. As a result, the additional gain achieved by omitting the degeneration results in greater open loop gain and derives more benefit from global negative feedback.
Ok, thanks. That makes sense. Given that my simulation results are not totally accurate, what approximate values should I be using. I'd like to use the lowest value that's practical, given I've also noticed a fair increase in noise as well using them. Is there a rule of thumb??
Thanks again.
Paul.
Thanks again.
Paul.
Member
Joined 2009
Paid Member
I found the same thing with my own simulations, the resistors act to reduce the open loop gain of the LTP and hence the distortion corrected in a feedback amplifier rises when these resistors are added. I built my first amplifier without them. I matched the LTP devices as closely as I could.
I now accept it as good practice to include them. For a basic LTP something in the range of 10R should work well.
If however, you have a high gain LTP, because you are using a current mirror collector load for example, then it's likely that higher values would be beneficial for overall stability. When I built an LTP using Sziklai pairs I used at least 47R.
Another reason to use degeneration resistors is that although it reduces the LTP gain, it extends the linearity to larger signals. This improves the bass response in particular because it is at low frequencies that most large signal transients occur. In this case, degeneration of 100R up to 1k will be beneficial but you must have enough open loop gain left over for global negative feedback and this usually means using a current mirror load.
Read up some of the stuff published by Doug Self (most of it is on the internet) for a much better explanation.
I now accept it as good practice to include them. For a basic LTP something in the range of 10R should work well.
If however, you have a high gain LTP, because you are using a current mirror collector load for example, then it's likely that higher values would be beneficial for overall stability. When I built an LTP using Sziklai pairs I used at least 47R.
Another reason to use degeneration resistors is that although it reduces the LTP gain, it extends the linearity to larger signals. This improves the bass response in particular because it is at low frequencies that most large signal transients occur. In this case, degeneration of 100R up to 1k will be beneficial but you must have enough open loop gain left over for global negative feedback and this usually means using a current mirror load.
Read up some of the stuff published by Doug Self (most of it is on the internet) for a much better explanation.
When you use degeneration what you should be looking at is keeping the transconductance of the LTP the same as before otherwise its not a fair comparison. The way to do this is to increase the tail current. Then youll see the benefits, high frequency distortion (above 1 khz) will be significantly reduced. There are other benefits too, reduces overload conditions, increases slewrate ect.
- Status
- Not open for further replies.