PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2019-07-16, 21:30:37
News: A feature is available which provides a place all members can chat, either publicly or privately.
There is also a "Shout" feature on each page. Only available to members.

Pages: 1 2 3 4 5 6 7 [8]
Author Topic: DSRD pulse generator  (Read 141047 times)

Hero Member
*****

Posts: 3078
tExB=qr
The pulse width is set by the 2-7pF trimmer capacitor, but manipulating the load (here a 50 Ohm terminator plug) also has influence

The load at the time of the above screenshot was the 50 Ohm terminator plug.
But when using a 56 Ohm metal oxide resistor in series with a red led increased the pulse width / fall time to about 800ps producing a faint glow in the led.

(using the following pdf circuit on page 1108 as a guide:   http://academic.reed.edu/physics/courses/Physics332.s12/pdf/LED%20Luminescence%20Decay.pdf )
  
Regards Itsu

I use the circuit on page 1108, a) with only the coax delay line to set width, and the trigger is isolated with a tiny transformer.  For a load, I use a very large air-core coil (high impedance) and there are no reflections
   

Hero Member
*****

Posts: 1685
Both ends of your BNC signal cable (scope end and driven end) should be terminated with 50 ohms or whatever the cable impedance is to prevent unwanted reflections and ringing.

Hmmm,   the driven side is terminated with the 4.7 Ohm of the voltage divider, and i guess the scope end (set to 50 Ohm impedance) will internally be terminated with 50 Ohm,  right?
Or should i here use the trick with the BNC T with one end terminated into 50 Ohm?  And if so which side,  the 4.7 Ohm drive side, the 50 Ohm scope side or both?

Quote
The virtual open end at the scope is probably causing reflections.

again, is that end not terminated internally with 50 ohm?

Quote
This might require that your attenuator be re-arranged a bit, but not at all difficult to do.

Right, like with the BNC T with the 50 Ohm terminator?


Quote
The LED in series with the terminator resistor is not a good idea, as the line is virtually open until the threshold voltage is met, and definitely open in the reverse direction. The terminator impedance will drastically change with the series LED.

Ok, but that was the setup as mentioned in the PDF i referenced.


Thanks and regards  Itsu
   

Hero Member
*****

Posts: 1685
I use the circuit on page 1108, a) with only the coax delay line to set width, and the trigger is isolated with a tiny transformer.  For a load, I use a very large air-core coil (high impedance) and there are no reflections

Ok, the RG174 coax has a capacitance of 1pF/cm, like the RG316 i use, so the mentioned 9" would be something like 22pF.
Not sure if what they say (stops avalanching on the return pulse) is really working.
 
I can try to use a high impedance air coil,   thanks.


Regards itsu
   

Hero Member
*****

Posts: 1685

just to avoid any confusion, this is my present layout of the avalanche part of my picopulser:

Regards Itsu
   

Hero Member
*****

Posts: 1685
I received some 2n5551's, but they do not pulse in the circuit i used, see below.
I went up till 400V dc bias, but no pulse was seen.
Any idea why?

Regards Itsu


I used the circuit from the referenced PDF (see earlier) to test my 2n5551's, and finally they started to avalanche using 280V and more and having a 220pF capacitor (no coax).

First picture the test circuit used (without coax, 250Pf capacitor and 324V)
Second picture is the resulting screenshot of the scope with a nice 233V nano pulse

Regards Itsu

   

Hero Member
*****

Posts: 3078
tExB=qr
Your pulse looks great. 

You can string the transistors together for higher voltage pulses. I have used up to 16 for 4kv pulses.
   

Hero Member
*****

Posts: 1685

thanks,   

My present DC PS is limited to about 400V, so that will not work.
I still have an old HAM tube power amplifier which should have the right parts.

Regards Itsu
   
Group: Elite
Hero Member
******

Posts: 3562
It's turtles all the way down
Sorry Itsu, I missed the fact that you switched the scope to 50 ohms input, so disregard my input.

Best to try to maintain 50 ohms termination at each end of the cable and use a separate discharge resistor and a 450 ohm value to the terminator resistor. This will give you the 10x attenuator.

This could be fine tuned even further but you seem to be doing quite well now.

Regards, ION


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   
Full Member
***

Posts: 130
Busting out my 2n5551's...wonderful circuit.

Reference for proper stack configuration?
   
Group: Elite
Hero Member
******

Posts: 3562
It's turtles all the way down
thanks,   

My present DC PS is limited to about 400V, so that will not work.
I still have an old HAM tube power amplifier which should have the right parts.

Regards Itsu

Itsu

You don't need much of a power supply for the avalanche circuit as the current is very low unless you plan on a very fast repetition rate of pulses which would probably overheat the transistor.

 Those little electrostatic power supplies found in laser printers will probably work fine. I have tested some of them and you can vary the DC input to control HV output over quite a range. Trick is there are multiple inputs and after decoding the plus and minus inputs, there are other control inputs that turn the HV on and off by setting the inputs to ground or V+. It takes a bit of experimenting to find the right connections.

Alternately a Variac , MOT, rectifier and filter cap will get you there at high current capability.


---------------------------
"Secrecy, secret societies and secret groups have always been repugnant to a free and open society"......John F Kennedy
   

Hero Member
*****

Posts: 3078
tExB=qr
Busting out my 2n5551's...wonderful circuit.

Reference for proper stack configuration?

I use the circuit from Baker for stacks, first link below, with the capacitors and resistors across each transistor.
http://www.overunityresearch.com/index.php?topic=378.msg5568#msg5568

http://cmosedu.com/jbaker/papers/1991/RSI621991.pdf

   

Full Member
***

Posts: 117
Replicated the monostable nanosecond pulse generator:  https://www.overunityresearch.com/index.php?topic=1556.msg26300#msg26300

I accidentally ordered NAND gate instead of NOR, so it required a few tweaks to get the equivalent circuit working.
Tightest pulses were ~20ns, but much smoother at >50ns.  Very repeatable.

I'm impressed to get that kind of performance from a breadboard circuit, but a small PCB should perform better... :)
   
Hero Member
*****

Posts: 583
It's a nice and cheap setup, Reiyuki, but I don't see the point. Is it to power a DSRD circuit or to directly generate short pulses as a DSRD generator could do?
In the first case, short pulses are not required at the start. On the contrary, rather long pulses, 50 to 500 ns, are needed to "charge" the diode, and it is when it becomes non-conducting that the short pulse occurs ( < 10 ns).
In the second case, even if it is followed by a power transistor, the power achieved will be far from being able to compete with those of DSRD pulses, neither in power nor in pulse narrowness. Only very advanced techniques such as magnetic pulse compression can compete, but much more expensive and complicated to implement than DSRDs.


---------------------------
"Chance favours only the prepared mind."  Louis Pasteur
   

Full Member
***

Posts: 117
It's a nice and cheap setup, Reiyuki, but I don't see the point. Is it to power a DSRD circuit or to directly generate short pulses as a DSRD generator could do?

The goal in the above test was multifold, but largely to provide a stable, repeatable isolation between function generator and FET driver.  I was getting irregularities when driving with pulses shorter than 50ns, and this provided an interesting solution.  With this setup, one could get precise nanosecond pulses with an arduino or even 555 timer.



I also replicated exnihiloest's DSRD circuit described here: https://www.overunityresearch.com/index.php?topic=1556.msg26722#msg26722   

Rather than wind coils, it appears that regular jumper leads provide enough inductance to get a decent result.
However the inductance-variability and time-delay between drive and pulse was disappointing, as it puts a fairly low limit on repetition frequency (under 300kpps).


Overall I am focused on developing a more standard design for generating short HV impulses that does not require custom coils or excessive tuning (IE: something that the average EE could repeatably build).
« Last Edit: 2019-02-11, 15:50:37 by Reiyuki »
   

Hero Member
*****

Posts: 1472
Overall I am focused on developing a more standard design for generating short HV impulses that does not require custom coils or excessive tuning (IE: something that the average EE could repeatably build).
If you abhor custom cored coils then you can do this circuit with off-the-shelf coils.

   

Full Member
***

Posts: 117
If you abhor custom cored coils then you can do this circuit with off-the-shelf coils.


I tried that circuit, but was disappointed by the amount of tuning required to get a decent pulse back.
L1    Low nanohenries (lead wire)
C2   ~3.6uF
L2   1-100uH (adjustable)
D1   1n5404
Vcc  24v
Pulse ~380ns
Attached is result.

As a one-off it works OK, but I believe a replicator would have big problems trying to reproduce the circuit.  Varying the input voltage, pulse duration, capacitance, or inductances more than a few percent will significantly affect the output in my setup.

We could mitigate this difficulty by selling batches of prebuilt, pretuned boards, but I don't think my relatively mediocre 50ns pulses would be worth the development, given that >1kv nanosecond MOSFETs are already on the market that can likely surpass this performance.


It is indeed an interesting effect, apparently it's just not up my alley. C.C ;)
« Last Edit: 2019-02-11, 22:31:02 by Reiyuki »
   

Hero Member
*****

Posts: 1472
I tried that circuit, but was disappointed by the amount of tuning required to get a decent pulse back.
L1    Low nanohenries (lead wire)
C2   ~3.6uF
L2   1-100uH (adjustable)
D1   1n5404
Vcc  24v
Pulse ~380ns
Attached is result.
Then stick with the true and tried method described here and as a pulse controller use this 74VHC221 circuit with C2=22pF :
« Last Edit: 2019-02-12, 01:12:54 by verpies »
   
Hero Member
*****

Posts: 583
The goal in the above test was multifold, but largely to provide a stable, repeatable isolation between function generator and FET driver.  I was getting irregularities when driving with pulses shorter than 50ns, and this provided an interesting solution.  With this setup, one could get precise nanosecond pulses with an arduino or even 555 timer.

Once again, short pulses are not suitable to drive DSRD diodes. You must charge the capacitor/diode which will provide the pulse energy when the diode is switched off. You can't reach high power if input pulses are short, the capacitor/diode will not be charged enough.
For isolation, the TC4429 I used in the setup you've reproduced is enough. If your generator can't provide pulses in the range 50-500ns, then a monostable can be added between the generator and the TC4429. It's important to be able to adjust it at will.

I tried that circuit, but was disappointed by the amount of tuning required to get a decent pulse back.
...

I'm afraid that tuning is indispensable. Short pulses means high frequencies. If you have 6 ns pulses, the level of the harmonics being weak, it's like if you had one half-period of a 1000/6=166 MHz signal. It's like radio frequency amplification. The ouput circuit impedance must match the charge impedance otherwise very little power is transferred. This is done by tuning the ouput L/C.



---------------------------
"Chance favours only the prepared mind."  Louis Pasteur
   

Full Member
***

Posts: 117
I'm afraid that tuning is indispensable. Short pulses means high frequencies. If you have 6 ns pulses, the level of the harmonics being weak, it's like if you had one half-period of a 1000/6=166 MHz signal. It's like radio frequency amplification. The output circuit impedance must match the charge impedance otherwise very little power is transferred. This is done by tuning the output L/C.

Oh boy, I'm feeling these issues firsthand :P.  Challenges like this on top of the high-voltage tightrope are probably why there are so few successful replicators out there. C.C

I'm posting my own results to help give people behind me a gauge on the difficulty of the various paths through the forest. ;)


Then stick with the true and tried method described here and as a pulse controller use this 74VHC221 circuit with C2=22pF :
Thanks, I'll get to that eventually.  For the time-being I'm still focused on fast-switching FET gates.
   

Hero Member
*****

Posts: 1472
Thanks, I'll get to that eventually.  For the time-being I'm still focused on fast-switching FET gates.
If you don't solder the MOSFET driver right on top of the MOSFET, you can forget about fast gate rise times.
Also, supplying the driver with 20V (VGS_MAX) helps a lot and for best results use a symmetrical ±20V supply to your driver ...and if you are really ambitious, then use a gate driving waveform like this:
   
Pages: 1 2 3 4 5 6 7 [8]
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2019-07-16, 21:30:37