Measuring Switching Times of NPN Transistors

I watched an excellent video by W2AEW on transistor switching times and realised I had much to learn.  Here is a link to the video: W2AEW inspiration.  Practical experimenting is by far the most effective way for me to learn so I set up a testing circuit.

Alan explains the theory and his circuit for testing switching times here: Explanation

This my circuit which includes a square wave source:

This is the 74AC14 (hex inverter with Schmitt Trigger input) square wave generator.  Note a 100nF decoupling capacitor on pin 14 (Vcc) and all unused inputs tied to ground to reduce the noise on the square wave output.

Output is 5V peak to peak at 3.482KHz.   The rise time is 2.7nS.  This very fast version of a hex inverter is required to provide those very fast edges otherwise the true transistor performance may not be detectable.

Here is the test rig.  There are female headers with turned pins to insert transistors under test.  There are two 100 ohm resistors in parallel in the centre but these are not connected - only required when using a 50 ohm source generator.

The connection points for oscilloscope probes are made from Dupont female pins bent at 90 degrees - these are a good fit for my probe tips.  The ground connection made from solid hook up wire core turned tightly around the probe ground then slid off and soldered in place.   The probe inserts through the coil with the tip entering the Dupont pin making a robust connecting with the shortest possible ground connection to minimise ringing.

Transistor switching on

Here is how switch on appears on the oscilloscope.  When the input voltage goes high there is a short delay before the collector voltage starts to fall (delay time).  Then the collector voltage falls and when it reaches 10% of Vcc it is considered to be fully on

Transistor switching off

Here is the transistor switching off.   When the input voltage goes low the base current goes negative but the collector voltage remains low until the charge is removed from the base / emitter junction.  This is the storage time.  When the charge is fully removed the collector voltage starts to rise and when it reaches 90% of Vcc the transistor is fully off

Results

All measurements were made with Vcc = 5V

These are the measured times.   Two versions of a '108' are 5 times slower to switch off than the two switching transistors.   Switch on times are similar for all 4 devices.

The method provides consistent results within the accuracy of setting the cursors on an oscilloscope.

Next I wanted to see the effect of using your circuit with a pair of diodes to speed up switching.

The modified circuit is this with two 1N4148 diodes added and the base resistor split into two:

 and these are the results:

Whilst switching times are slightly reduced there is a huge reducing on the storage time for all devices, sometimes by as much as 80% reduction.  Wow, that is huge.

I have been unable to find any reference to this circuit or a description of its mode of action.

Next the effect of a resistor between the base and ground (Rbe) was investigated.   This provides a path for the stored charge to ground and should speed up switching the transistor off.

This is how the results look as Rbe is reduced.  There is a small increase in the switch on time of 8nS from 36nS to 43nS.  Toff improves from 353nS (no Rbe) to 173nS (Rbe 1k)

Here are the results:

Adding the two diodes prevents saturation occurring and so reduces storage time (and total switch off time) more than any value of Rbe.

A combination of Rbe of 1k and two diodes achieves a switch off time that is 6 times shorter than with neither of these additions.  There is a 9nS increase in switch on times - small but 25% increase!

References:

W2AEW YouTube Channel - highly recommended source of endless interesting and useful educational videos.  

An index to all videos by W2AEW