Maker Pro
Maker Pro

How to ignore pulses less than 700ns?

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
Hoping you guy's can give me some ideas.

I am building a sysyem that will drive multiple, 3300V/1500A IGBT's.

The driver circuit has a Fiber Optic input for the Gate drive, and a Fiber Optic output for the status and fault information.

The status information is a 700ns pulse, and any longer than 1.5usec means a fault condition.

The driver was designed to interface to a controller, but my design is discrete logic.

I don't care about the status pulses, only the fault conditions, so I need a circuit that
will ignore the 700ns pulses.

Using a Microcontroller is out of the question for this design. The system will be generating pulses of up to 3.5MW, so it is not an environment where I can trust a microntroller.

Anybody got any ideas?
 

Electrobrains

Jan 2, 2012
259
Joined
Jan 2, 2012
Messages
259
For detection of the pulse, you could use a simple pulse discriminator (see principle diagram of core circuit below).
The short output pulse could be saved through setting an R/S flip-flop.
Simple_Pulse_Discriminator.GIF
 

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
Thanks for the answer Electrobrains! That might just do the trick.

I will breadboard something today. Looks like my accuracy would be based on the specifications of the R and C.

I will let you know if this works.
 

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
Works great!

Sometimes you get so wrapped up in working so many years, you forget
the basics.

I was afraid it wouldn't be accurate enough for the 50-100ns range, using resistors and capacitors I just had laying around....but it works fine.

Although I did have to jiggle with the calculations a bit to arrive at the voltage threshold I needed to trip he Scmhidt Trigger.:D
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I looked at this last night and thought that it could be solved more simply with a single schmitt trigger inverter.

I decided not to post the comment, but it may be of use if you need a lot of these.

Essentially it is connected as a pulse extender, however it is connected to extend the ground state rather than the pulse. For a little additional utility, you can follow this with another pulse extender so that you get a minimum length pulse whenever your input pulse exceeds a certain time limit.

I think it is a little better than electrobrains' answer as his may be subject to creating runt pulses on the falling edge of the input waveform in some cases.

Look at the last page of the Fairchild AN-140 for the circuit. In your case it needs to have the diode reversed to stretch the low logic level.
 

Electrobrains

Jan 2, 2012
259
Joined
Jan 2, 2012
Messages
259
Works great! ...
Just adding the 3 other gates in the IC package would give you a memory function.
After processing the error routine, you can send a reset pulse to let the circuit run again.
The Schottky Diode D1 can be omitted if your input pulses are not very close to one another (low duty cycle).

Steve, I don't see how the circuit would produce runt pulses...
And even if it could, the memory function would give a clean error signal.

Simple_Pulse_Discriminator2.GIF
 
Last edited:

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
Just adding the 3 other gates in the IC package would give you a memory function.
After processing the error routine, you can send a reset pulse to let the circuit run again.
The Schottky Diode D1 can be omitted if your input pulses are not very close to one another (low duty cycle).

Steve, I don't see how the circuit would produce runt pulses...
And even if it could, the memory function would give a clean error signal.

Simple_Pulse_Discriminator2.GIF

We think the same way! If you were in Massachussetts I would be interviewing you for a job!

Thanks Guy's!:D
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Steve, I don't see how the circuit would produce runt pulses...

Forgive me. I was thinking fuzzily.

Still, you can do the same with a single Schmitt trigger inverter :)
 

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
You guy's will be glad to know that we tested the circuit today, and it worked just fine.:D

A few points I learned:

Since the fiber optic output was 0-5V, I immediately converted it to 15V to give me a low impedance drive for the RC Network. (In this environment I always use CMOS logic at 15V to raise the "noise Floor")

I used a 1% resistor, and a Mylar Film cap (1nF) into a Schmidt Trigger Nand Gate so my threshold was up at about 6.8V.

I used a Nand Gate, because I was also able to use the remaining gates in the package for the S-R Flip Flop as suggested to latch the fault condition.

I talked to the Engineer from the company that makes the IGBT Driver, and found out that the status pulses were never over 700ns, and the Fault condition would be a minimum of 1.5us.

I set the R to trip at approx. 1us, and everything works just fine. (No potentiometer, I hate the damn things. Potentiometers are just an admission that you built a sloppy circuit!)

I tested the circuit with a Precision Pulse Generator, and it worked to within +/- 40ns!

So now, the question was how far will it drift with temperature?

So I grabbed the heatgun...you gotta "Push it to the Limit":D

I got to the point where I was 100ns off of the switching point before the solder joints started melting, and everything went to ****.... so I guess it's going to work OK!

Want to thank all you guy's for your input....wish I could buy you all a beer!
 

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
I also posted this problem on some other forums, and all I got were suggesting for using a Microncontroller...but some things can't be trusted to a Microncontroller.

A 3.5V, or even 5V microntroller can latch up....and when you are switching an IGBT at 1400V and over 2000A, you are not gonna just "Let the smoke out", you are gonna have an explosion!

Sometime, the simplest solution is the best!
 

Electrobrains

Jan 2, 2012
259
Joined
Jan 2, 2012
Messages
259
Thank you for the positive feedback!
Quite often you don't get any response after presenting a solution. That's a bit discouraging (I use to wonder; "did that guy get electrocuted or not?")

Micro controllers sure have brought a lot of flexibility and new opportunities, but as you say they indeed are not as stable as the "old" analog and digital solutions (I have experienced that many times).

Concerning the one-gate solution from Einar and Steve, I must admit that that circuit has one level higher ingenuity than the NAND gate. I think I would use that solution if I had a space problem.
Commercially seen, if I had space, I would probably choose my NAND solution. It might come out a little cheaper and have less "trouble sources". Only 3 components rather than 6 (or 7 depending on reset polarity).

Anyway, it's amazing what you can do with those Schmitt Triggers. I really like them!
Once, long ago, I built a 2 motor wiper controller for buses here in Switzerland. There were several adjustable on- and off-delayed timers and some other logical circuitry. Everything built around just one CD40106. It worked wonderfully - till winter season came with salt water from the street and condensed water from the air....:(
Well, that was my first free lesson in practical automotive electro engineering!

Thanks for the beer invitation!
Unfortunately my US visit frequency is very low pass filtered and I was never in Massachusetts (my business contacts are in PA). But if you're once in Switzerland, instead I'll invite you for one of my wife's coffee macchiato inventions in her small café (cafeprisma)!
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
I think if GonzoEngineer ever gets electrocuted by one of his projects, all they'll find are a pair of smoking boots.
 

GonzoEngineer

Dec 2, 2011
321
Joined
Dec 2, 2011
Messages
321
I think if GonzoEngineer ever gets electrocuted by one of his projects, all they'll find are a pair of smoking boots.

Not Gonna Happen.

When I was a Soldier, I had to respect the dangers of the tools of my trade.

It's the same thing today.:D

And an update:

Since I had to replicate this circuit for 10 IGBT Drivers, I did go with Hex Scmitt Trigger Inverters, and Dual 4013 Flip Flops to keep the parts count minimal

I tend to modify my designs as I layout the board.....:cool:
 
Top