What is the smallest switching time of a transistor ? !:confused:

[Ali Inam]

Main Question or Discussion Point

What is the smallest switching time of a transistor ? !

 

[pallidin]

Not sure of an answer, but this site has some interesting info:
http://www.electro-tech-online.com/general-electronics-chat/1848-fastest-switch.html

 

[Bob S]

I have used unsaturated switching npn (I think 2N5770) transistors (in a differential pair with a common current sink) switching a few milliamps at 0.75 volt, and achieved ~ 1 to 1.5 ns rise and fall times into 50 ohms.

http://www.hamtv.com/pdffiles/2N5770.pdf

The data sheet claims a 20 ps collector-base time constant.

Bob S

 

[bjacoby]

What is the smallest switching time of a transistor ? !

What is the clock speed of the current fastest CPUs?

 

[Born2bwire]

What is the clock speed of the current fastest CPUs?

Clock speed is not a direct indicator of switching speed. The switching speed of a transistor is much much faster than clock speed.

It depends on the type of transistor, the specifications that you need to satisfy, the package (switching on chip or with the package) and whether you are asking about off the shelf or what is only available in the laboratory.

In 2005, I know that the world's fastest transistor was just over 600 GHz. They should have surpassed 1 THz by now as the technology that allowed them to break 600 GHz was expected to bring the speeds into the 1 THz range.

 

[bjacoby]

Clock speed is not a direct indicator of switching speed. The switching speed of a transistor is much much faster than clock speed.

It depends on the type of transistor, the specifications that you need to satisfy, the package (switching on chip or with the package) and whether you are asking about off the shelf or what is only available in the laboratory.

In 2005, I know that the world's fastest transistor was just over 600 GHz. They should have surpassed 1 THz by now as the technology that allowed them to break 600 GHz was expected to bring the speeds into the 1 THz range.

Sure clock speed isn't quite a direct indicator, but it's close enough for physics is it not? The actual switching speed may be faster but on the other hand CPU clocking will have rates in excess of discrete transistors because they are smaller and closely wired to neighboring parts.

On the other hand I'm not sure if the OP was asking what the current fasting switching rates are or if he wanted to know the theoretical minimum switching times. I'm not sure how to answer the last question.

 

[Studiot]

Surely the answer is it depends.

It depends on the circuit configuration as well as the transistor.
Switching off takes longer than switching on.
Do we count switching circuits where the transistor is not actually switched to increase speed or are we going to insist on full saturation?

 

[Born2bwire]

Sure clock speed isn't quite a direct indicator, but it's close enough for physics is it not? The actual switching speed may be faster but on the other hand CPU clocking will have rates in excess of discrete transistors because they are smaller and closely wired to neighboring parts.

On the other hand I'm not sure if the OP was asking what the current fasting switching rates are or if he wanted to know the theoretical minimum switching times. I'm not sure how to answer the last question.

The clock speed isn't anywhere close to the switching speed because it is largely design dependent. The clock is dictated by the longest critical path which is going to depend upon a lot of factors. The speed of switching, the number of transistors in series, the desired error rates, race conditions, length of transmission lines, pipelining, the transistor topology (NMOS, CMOS, etc) and the latency from memory operations (cache->RAM->Disk). For example, the same CPU could be easily clocked 1 GHz or 500 MHz depending on how you set up the pipeline.