-
Categories
-
Platforms
-
Content
..right, so we are saying that there is still some "miller effect" in the optocoupler transistor even if it has no base resistor?I'm sure that's true. The reason is the reduced change in the voltage across the transistor in the optocoupler, which reduces the Miller effect greatly and causes much faster response in the optocoupler. This is the reason for the cascode design in the first place.
but still I believe CC is better than CE
It must be my computer then putting cascade instead of cascOde.
Cascode
cascade
cascade
I doubt it's the web site. I've never had a problem with it. Do you have some kind of automatic spell correction feature installed in your browser?I spell "cascode" like that because otherwise the web page changes it to "cascade".
I believe the Miller effect is relevant, and will apply equally whether the output is taken from the collector or the emitter...right, so we are saying that there is still some "miller effect" in the optocoupler transistor even if it has no base resistor?
Right so far.The following concerns using a casc0de stage to improve opto feedback bandwidth..... http://www.onsemi.com/pub_link/Collateral/AND8273-D.PDF
...on page 9, just above fig13, it does say that the cascode is used so that "looking" up from the opto's collecter, the impedance seen is less as its "looking" into the emitter of a BJT biased ON.......
No, because the performance differences between common emitter and emitter follower stages are only relevant when the input signal is applied to the base WITH RESPECT TO THE POWER RAILS. This requires a connection to the base. With an optocoupler with no connection to the base, there is no difference between collector output and emitter output, because the signal (the photocurrent) is always applied between the base and emitter (this is the closest analogy), whether the emitter is grounded or not.from this you would then conclude that CC is better than CE.....because even though CC doesn't involve a cascode stage, it DOES involve looking up from the collector of the CC'd opto and seeing a low impedance, because from the opto collector in CC, you "look" up and sees the supply rail, which is a low impedance to AC.
No. There's a good reason why cascode improves the bandwidth. The voltage change across the collector-emitter of the phototransistor is greatly reduced. This is the whole idea of the cascode circuit, whether it's used with a phototransistor or not. The reduction in voltage change across the transistor means that there is less voltage change to be coupled from the collector to the base through the Miller capacitance. It makes the Miller capacitance almost irrelevant. The phototransistor is being used as a variable current sink, driving a load with very low impedance (the common base stage). The effect of the Miller capacitance of the other transistor (common base stage) is also hugely reduced because the base grounded for AC signals. (Usually a decoupling capacitor is added between base and ground, if the base is driven from a voltage divider as shown in the ON Semi app note).So I would conclude that if the cascode improves bandwidth over a CE connection...then so too must the CC connection
conclusion, I say CC is better
I almost find myself thinking Steve has discovered this, and has found the reason for it too.
If so, I guarantee all that the result will be a revelation for the world of SMPS.....most of this type of info is so secret that it just doesn't exist in books.
I certainly think linear.com would be interested to see your results.
The LT4430 datasheet is now very intriguing on pages 16, 17 and 18.
I am of course, impelled to believe that there is no difference between CE and CC connection after seeing your experiment.
It looks like you're comparing the turn-on time with the turn-off time. In the second picture you have the scope connected with reversed polarity across the load resistor. This means that the first part of the scope trace, from which the edge rises upwards, corresponds to the phototransistor being ON, and the rising edge corresponds to it turning OFF. The first picture shows the opposite. The Miller capacitance and charge storage are the reasons why the transistor turns OFF more slowly than it turns ON.I did a similar experiment with a BFX43, base left open. Maybe I had the setup wrong but I followed what Steve was doing apart from having two power supplies.
A difference of 4 µA in a 2370 µA signal is 0.17%. This could be easily caused by your power supply warming up, or your battery discharging slightly, or the imperfect repeatability of your multimeter, or the device warming up, or other factors.Also did another experiment if your interested with a OP593 which shows an increase in current out of the emitter compared to the collector. There was a small debate about this also. This device does not have a base lead.
It looks like you're comparing the turn-on time with the turn-off time. In the second picture you have the scope connected with reversed polarity across the load resistor.
A difference of 4 µA in a 2370 µA signal is 0.17%. This could be easily caused by your power supply warming up, or your battery discharging slightly, or the imperfect repeatability of your multimeter, or the device warming up, or other factors.
No, if that's true, then it can't. But if you're running the phototransistor side from a 9V battery (which is fine and a good idea), it's easy to repeat Steve's test. Just connect the scope across the load resistor in each case. Here are the node lists for each configuration.I did this because I didn't have a an invert function on the scope. Both traces were captured as I connected the power to the IR emitter by hand, so this can't be turn off can it?
You know there's no electrical connection to the phototransistor in an optocoupler, right? The phototransistor circuit is completely isolated and self-contained.I did an earlier experiment wit the BFX43 and used a precision meter, this showed and increase of 10uA. But the meter is to big to get in the picture. The PSU had been on all morning as I had been using it for other work.
So it does look like the only way to get improved stability at faster transient response with opto feedback is to use a cascode stage (windows 8 is still writing 'cascade')
In the post #49 I show two other ways to connect up optos to get more facilitated stability and response.
Of course, we all know that light is an EM wave......we also know that the entire concept of electric current flowing in a circuit is nonsense...but at low frequencies, it is an approximation that we can get away with....at microwave, we have to consider it as it really is, a transition of EM waves.
So if EM waves are coming from the base, then to me that is like an electric current anyway, so I believe that Arouse1973 is right in many ways.
What I have been saying all along, perfect.
BTW I have emailed a Physics contact I have in Australia to get an unbiased accurate answer.
I also have a few contacts at Linear Tech, unfortunately I can't email Jim Williams as he has passed away sadly. But I will see if I can get another opinion.
Thanks
Adam
Also did another experiment if your interested with a OP593 which shows an increase in current out of the emitter compared to the collector. There was a small debate about this also. This device does not have a base lead.