How to make a Power BJT switch transistion time less than 1micro-second

Thread Starter

Melvo

Joined Feb 19, 2018
19
Hello All,

I am trying to make a Power BJT transistor switch have off-on and on-off transistion times less than one micro-second. At the moment I have a fall-time in the nano-second range(~8ns) and I have a rise-time in the range of ~2 micro-seconds.

I am trying to design the circuit such that this rise-time is below 1 micro-second. I am using a TIP41C Power BJT. When first connecting the circuit, the fall-time was ~20ns which I was able to decrease to ~8ns with the usage of "speed-up" capacitors.

I was trying the speed-up capacitors in order to decrease the amount of time it takes the base to discharge its charge thus decreasing the amount of time from on to off. Also, to provide a low impedance path to the base for the high frequency content of the pulsed trigger to trigger the base earlier which seems to work fine and decreased the off to on time. I am still not sure exactly how to size these capacitors because the transistors datasheet does not give a junction charge.

In the schematic the 5V source is a 1Hz, 10% duty cycle, 5Vpp square wave. I have attached the schematic, transistor's datasheet, and my two scope captures below. I have also attached the datasheets of two transistors which may work for my application(2SC5200, MJE15032) the reason I think they will work is because they have a transistion frequency of 30MHz as opposed to the TIP40C's transition frequency of 3MHz. I am also hoping they will have less input capacitance, but can't say for sure.

It is probably important to note that these measurements were taken on a 50MHz scope.

Any help is greatly appreciated.

Thanks
 

Attachments

danadak

Joined Mar 10, 2018
4,057

crutschow

Joined Mar 14, 2008
34,201
What is the actual load you want to drive?
Obviously you don't need a power transistor for a 22kΩ load. :confused:

Here's the LTspice simulation with a Schottky Baker clamp, a larger input resistor, a smaller input capacitor, and a smaller collector resistor.
The rise time is now <100ns.

upload_2018-4-28_1-5-34.png
 
Last edited:

Bordodynov

Joined May 20, 2015
3,174
What is the actual load you want to drive?
Obviously you don't need a power transistor for a 22kΩ load. :confused:

Here's the LTspice simulation with a Schottky Baker clamp, a larger input resistor, a smaller input capacitor, and a smaller collector resistor.
The rise time is now <100ns.

View attachment 151471
crutschow,
I think that it is more correct to take into account the resistance of the signal source. Even if you use a powerful driver, a more correct result will be if the account is taken. Then the results of modeling will be more adequate to reality. I unfortunately did not.
Melvo, which signal source did you use?
 

Thread Starter

Melvo

Joined Feb 19, 2018
19
Hello Dana,

It looks like one person there is having the same issue as I am. I will see what worked for him/her and try it out.

Thanks
 

Thread Starter

Melvo

Joined Feb 19, 2018
19
What is the actual load you want to drive?
Obviously you don't need a power transistor for a 22kΩ load. :confused:

Here's the LTspice simulation with a Schottky Baker clamp, a larger input resistor, a smaller input capacitor, and a smaller collector resistor.
The rise time is now <100ns.

View attachment 151471
Hello Crutschow,

Thanks for your reply. I tried the schematic you have given with the rise and fall times increased(see plots below with 22k load). Also, the reason I am using a 22k load is because I have a cheap power supply(Korad 3005D) which has a voltage drop when loaded heavily with a 200ohm load.

In the plots you will see that I used 12V instead of 5V. This is because this is for automotive applications which use a 12V battery.

I cant find any specification on the internal resistance of the power supply. The power transistor is there because I want to be able to pass 3A continuously is need be.

I am trying to find a range of loads that will give me the rise/fall times needed because the load represents an arbitrary device that could have a range of resistance value, I just want to find what that resistance range is, so I know what loads will be compliant with the test jig.

As a interesting note, disconnecting the speed up capacitor actually sped up the transition times and got rid of the reverse conduction(I think that is what it is) before turning off.
 

Attachments

Thread Starter

Melvo

Joined Feb 19, 2018
19
crutschow,
I think that it is more correct to take into account the resistance of the signal source. Even if you use a powerful driver, a more correct result will be if the account is taken. Then the results of modeling will be more adequate to reality. I unfortunately did not.
Melvo, which signal source did you use?
Hello Bordodynov,

I have not tried your circuit yet, but I will do so after this reply and report back. I am using a RIGOL DG1022 which has a 50ohm output for the trigger and a KORAD KD3005D for the power supply.

Thanks
 

Danko

Joined Nov 22, 2017
1,827
I cant find any specification on the internal resistance of the power supply.
KORAD KD3005D power source works up to 5A. 3005 = 30V 5A. See https://www.sra-shops.com/docs/srasolder/instructions/korad-multiple-channel-power-supplies.pdf
I think that it is more correct to take into account the resistance of the signal source
With extremely low impedance of input source we can have miracle results (see simulation below).
Of course acceptable circuit should works from low power input source, like logic gate.
Fast-pulse.png
 

Attachments

Last edited:

MrAl

Joined Jun 17, 2014
11,342
Hello,

Is it just the rise and fall times or do you also need to reduce the storage time?

To turn the transistor off as fast as possible means removing the base charge as fast as possible. It is often referred to as "sweeping the charge out of the base region".
See if you can find something called a "snap off" circuit on the web.
Can your circuit stand a forward biased diode in series with the emitter? You obviously get a higher apparent saturation voltage, but the turn off is faster when the diode is also biased. Also obvious is a negative voltage at the base reduces the charge faster than anything positive. The diode (or diodes) helps that when there is no negative supply, but if there is a negative supply or one can be introduced then the turn off can be made faster.
If you need to reduce the storage time then you need to keep the transistor out of saturation. That is what a Baker Clamp is for, or at least a FULL implementation of a Baker Clamp.
 
Top