High voltage nanosecond pulse generator

Thread Starter

kender

Joined Jan 17, 2007
264
Colleagues,

Could anyone suggest a schematics or a reference for a high voltage pulse generator? I need a 500V pulse 10ns long or sorter. Negative spike from 500V to 0 and back to 500V would work. I have a schematic for an old (1977) pulser, but the SCR that's used there has been discontinued. I know how to generate a nanosecond pulse at the logic levels. But what would be a good topology for the high voltage "switch"? MOSFETs have gate capacitance that needs to be charged within a few nanoseconds. SCRs (which I, unfortunately, don't have experience with) have their own problems too.

Enlightenment will be appreciated!

- Nick
 

Thread Starter

kender

Joined Jan 17, 2007
264
SgtWookie said:
Sounds like you may be looking for IGBT's.
Thanks for this tip! I'll look into IGBTs.

What sort of load do you need to drive with this pulse?
The load is going to be an ultrasonic transducer. 50 to 300 Ohms with some capacitance to it. I don't have a datasheet for the transducer at the moment, but I will know what the load is soon.

How about a spark gap?
I don't know how. I wish I knew.
 

Ron H

Joined Apr 14, 2005
7,063
You need to consider the transient (capacitive) current requirements as well the resistive load. Remember I=C*dv/dt. If you need 5ns rise and fall times into 100pF, for example, that's a peak current requirement of 10 amps. Not trivial. If all the capacitance is in series with the resistance, then you just need to consider the resistive component.
 

Thread Starter

kender

Joined Jan 17, 2007
264
kender said:
Ron H said:
What sort of load do you need to drive with this pulse?
The load is going to be an ultrasonic transducer. 50 to 300 Ohms with some capacitance to it. I don't have a datasheet for the transducer at the moment, but I will know what the load is soon.
I've clarified what the load is going to be. 50 Ohms.
You need to consider the transient (capacitive) current requirements as well the resistive load. Remember I=C*dv/dt. If you need 5ns rise and fall times into 100pF, for example, that's a peak current requirement of 10 amps. Not trivial.
My plan is to charge a large capacitance from a relatively small power high voltage supply. Then this capacitance will fulfill the peak current requirements.
 
Last edited:

SgtWookie

Joined Jul 17, 2007
22,230
Thingmaker3's spark gap suggestion might be the easiest (and cheapest) solution.
It's generally accepted that 3,000,000 volts are required to jump an air gap of 1 meter near sea level.
This translates to 76.2v per mil (0.001"). So, a spark gap of .0065" (rounded) should cause the 500v supply to ionize and jump the gap at right around 495v. The duration of the spark would then depend on your load impedance and the size of your capacitor.
 

Ruptor

Joined Apr 26, 2009
44
Hi
I know this thread is old but I just saw this circuit and thought it might help others to build a more modern version of a nano pulser.

http://www.case.edu/cse/eche/ESA2008_Proceedings/H3.pdf

Incidently using gas breakdown in a spark gap requires that the gas breakdown voltage be exceeded it doesn't matter what the gap is to start with the voltage has to exceed Vbrk. Reducing a gap does not make it break down at a lower potential.
 
Last edited:

CDRIVE

Joined Jul 1, 2008
2,219
Thanks for this tip! I'll look into IGBTs.


The load is going to be an ultrasonic transducer. 50 to 300 Ohms with some capacitance to it. I don't have a datasheet for the transducer at the moment, but I will know what the load is soon.


I don't know how. I wish I knew.
Ultrasonic Transducers are not designed or intended to receive high voltage nanosecond pulses. What are you trying to do?
 

Thread Starter

kender

Joined Jan 17, 2007
264
Ultrasonic Transducers are not designed or intended to receive high voltage nanosecond pulses. What are you trying to do?
I was trying to "whack" the transducer with a spike (practical approximation of a δ-function) and have it ring at its resonant frequency. This was a project for one of my clients, the client was on the acoustics/transducer side of this project, me on the electrical side. According to him, the medical ultrasound scanners use this method to generate pings with repetition rate up to 2kHz.
 

CDRIVE

Joined Jul 1, 2008
2,219
I was trying to "whack" the transducer with a spike ..According to him, the medical ultrasound scanners use this method to generate pings with repetition rate up to 2kHz.
Sorry about that. I thought you were referring to a depth finder transducer. Let me talk to my co-worker, he works on the Ultra Sounds at the hospital.

While were on the subject, here's a tip for all the young fellows on a tight budget. Check with the Engineering, Bio-Med, IS and Communications departments at your local hospitals. You would not believe the amount of electronics (both working and not) that get tossed out the back door. Because of environmental sensitivity we have changed our policy and this stuff no longer goes in a dumpster. It now sits in the Doctors parking garage until a recycler picks the stuff up.
 

Thread Starter

kender

Joined Jan 17, 2007
264
Hi
I know this thread is old but I just saw this circuit and thought it might help others to build a more modern version of a nano pulser.

http://www.case.edu/cse/eche/ESA2008_Proceedings/H3.pdf
Thanks for the article! I've seen that article during the original development last Fall. It was of some use - I was considering that approach as a back-up. I had a pretty tight spacial constraints. My pulser had to fit inside tube 25mm ID, 75mm long, including an HV power supply. Humongous FET and humongous gate driver would barely fit. Also, both me and my client frowned at the fact that IXYS stuff is expensive and hard to get.
 
Hi
I know this thread is old but I just saw this circuit and thought it might help others to build a more modern version of a nano pulser.

http://www.case.edu/cse/eche/ESA2008_Proceedings/H3.pdf

Incidently using gas breakdown in a spark gap requires that the gas breakdown voltage be exceeded it doesn't matter what the gap is to start with the voltage has to exceed Vbrk. Reducing a gap does not make it break down at a lower potential.
I tried the circuit according to the paper. The pulse width was 7ns. But the Peak-peak voltage was only 348mV, and the low voltage was around 500mV.

DEIC420 requires that the input high voltage be at least 3.5V, low voltage less than 0.8V.

How can this circuit be improved?
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
Did you have 0.1uF bypass caps (poly metal film or ceramic) across each of the IC's power/ground pins? If not, they are needed.

At such short time periods, circuit layout will be critical. You won't be able to breadboard it successfully, as the parasitic L and C's will be far too high. The use of the old 74LS IC's is quite dubious; they're really only good up to ~30MHz or so.
 
Did you have 0.1uF bypass caps (poly metal film or ceramic) across each of the IC's power/ground pins? If not, they are needed.

At such short time periods, circuit layout will be critical. You won't be able to breadboard it successfully, as the parasitic L and C's will be far too high. The use of the old 74LS IC's is quite dubious; they're really only good up to ~30MHz or so.
so PCB must be used to test to circuit? :eek:
if bypass caps was used, the output from 74LS74 could reach 3.5v? :confused:

The circuit was proposed on EDN magazine in 2000. i only need 200HZ repetition-rate of the nanosecond pulses.

Here is the photo from the paper. Do you know what are the two black components used for?
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
so PCB must be used to test to circuit? :eek:
Yes.
if bypass caps was used, the output from 74LS74 could reach 3.5v? :confused:
At those speeds, it's doubtful.

The circuit was proposed on EDN magazine in 2000. i only need 200HZ repetition-rate of the nanosecond pulses.
The problem is the rise/fall time, not to mention the propagation delay through the 74LS series themselves. Off the top of my head, I think it's somewhere around 15nS -20nS for even the SSI gates. LSI's will have even longer prop delays.

Here is the photo from the paper. Do you know what are the two black components used for?
It's hard to say. They might be tantalum caps, they might be diodes. I can't make out anything on the top of them.
 
How about using cortex M3?

When VCC pin of 74ls74was connected to 5V supply, nearly all the pins have high voltage level. GND was tied to the ground. What went wrong?:confused:
 
Last edited:
Top