PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2024-03-28, 08:22:56
News: Forum TIP:
The SHOUT BOX deletes messages after 3 hours. It is NOT meant to have lengthy conversations in. Use the Chat feature instead.

Pages: 1 [2] 3 4 5
Author Topic: Development of an Avalanche Output Stage  (Read 68161 times)

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
G is your stack setup for positive pulses, i cant see how it's possible without over rating the voltage on the bottom transistors.
   

Group: Administrator
Hero Member
*****

Posts: 3198
It's not as complicated as it may seem...
Perhaps this:

.99
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
G is your stack setup for positive pulses, i cant see how it's possible without over rating the voltage on the bottom transistors.

Positive pulses - every transistor sees the full 1.5kv - overvoltage is what avalanche is all about - it doesn't hurt them - but current does.  Clunky, slow diodes and transistors work best.

By triggering the top one, each subsequent transistor sees a higher voltage and you get a "snap the whip" effect. 

   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
yeah i could see that, for some reason i was trying to work out how on earth you trigger from the bottom of the chain, i couldn't get my head around it as the base would have had loads of volts on it potentially, until Poynt pointed me to the article again you posted with the trigger on the top trany, antway looks like e don't need positive pulses so i will go with the easier negative pulse and just drive with a couple of inverters as buffers and stick maybe a couple of TVS or Zener protection diodes across the inverters to stop any mishap going back to my Digital mono chips.

Cheers, guys my head is feeling clearer now  ;D

EDIT do you still use the coax on the larger chain or do you just go with caps?
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
Caps are easy to swap out for quick tuning but need coax for square shape.  The fall sucks with caps.

No isolation? - you have been warned...

if the pulse goes through the TVS rather than the coil...oops

you can optocouple too
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
back in the saddle again...
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
So here's the first draft.

« Last Edit: 2010-06-22, 12:00:38 by Peterae »
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
Looks great!

some ideas:

run a stack for each bifilar and you can delay the trigger

if rise is too slow then change to triggering the first transistor

an avalanche diode can also be used to sharpen the output pulse

10cm coax may be a bit short and you may not rise to full voltage
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
Cheers G
Yeah i will be building 2 of these, one for each coil of the bifilar, i can then do nS phase delays exactly the same way i have so far but using much higher voltage pulses.
But if the required delay was known i guess i could just use another bit of coax, and then avoid any logic controll for a finished unit.

Plenty to try yet, trim values and coax length ect.

Could always stick 4 or 5 saturable inductors in series with the output if i need to sharpen things up, but i am hoping i wont need to do this.

We will see where it goes anyway, first i must get one working correctly.

Good news sounds like you sorted yours now and are up and running  ;D
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
blew the crap out of it

(need a ballast resistor)
« Last Edit: 2010-06-20, 20:55:34 by Grumpy »
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
i=C (dv/dt)

i=100pf (1200/100nS)

i=1.2A

i=100pf (1200/10nS)

i=12A

i=500pf (1200/20nS)

i=30A

I can manage 60A for 20nS so if i go for 30A for safety i need a cap-coax value of about 500pF

Hope i have this right.


« Last Edit: 2010-06-20, 21:27:01 by Peterae »
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
Thanks - pulsing up to 500v now
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
I need to adjust the bandwidth of the 555 down (it is about 180kHz to over 1MHz now), but other than that I am ready to go to 1500v with the little chip that could - LOL!  At 500v pulse amplitude, I measure only a few volts across the coil - which is very good.  It rings down a little after the peak and longer delay increases the ringing a little.  Pretty happy with it so far.

For my ballast resitor I used R=V/I   and for 1 ma R is 1.5 meg so I used a 1.2 and there is no change in the output voltage when pulsing.
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
OK Plan C  ;D

I am going to use a 10 cap output filter stage, and i will use link jumpers to enable each stage so it should give me pulse currents 6amps @ 20nS per stage allowing me to vary up to the 60A max value of the Zetex transistor.

   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
OK Plan C  ;D

I am going to use a 10 cap output filter stage, and i will use link jumpers to enable each stage so it should give me pulse currents 6amps @ 20nS per stage allowing me to vary up to the 60A max value of the Zetex transistor.



Try to hold it down to 10ma or less.  You do not need, want, or desire current in the pulse.  You cannot look at this from conventional transmission line theory as it does not apply.  You are pushing nature, and nature is pushing back - hard!
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
resistors ok
caps suck - will try them in series or start rolling new ones or making a cap array

tried a 1uf 1kv cap at 1kv and this is not enough capacitance to keep voltage up

ballast resistor adjusted down to 10k to keep voltage at about 1430v

one step forward and two steps back...
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
Quote
one step forward and two steps back...

Isn't that always the case  ::)

The cap arrived for my psu's, i have used 10uF 450V with 3 in series for the positive cycle rectification and 3 in the negative side.

The regulation is fantastic @ 500V, i loaded with a 60K resistor and the voltage remained 500V exactly very little ripple, thats just over 8ma and 4.1 Watts being drawn.

Very happy so far
Need to try regulation at a 1000V next

EDIT ok tested higher volts, it looses regulation above 700V when i try to draw 4 watts, so i guess i need to try something other than 2kHz drive frequency to find a sweet spot these transformers like to work at, 2 kHz seems too low anyway i would have thought it should be operating much higher than this, maybe the 2kHz was a typo on the circuit diagram and that explains why the component values on the 555 were for operating at a higher frequency.

I am going to try the value that was originally published on their diagram and retry

« Last Edit: 2010-06-25, 13:16:29 by Peterae »
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
One thing I miss about a freg gen is you can just change the freq anytime you want.  Still waiting for delivery of my new one.  You can adjust a 555 circuit but it has a set range and you have to change components to change the range.  Low freqs can be good when testing as can a single shot switch - hint hint.

I have an article somewhere of a comparison of two circuits using PFL's:  one was twice as efficient as the other and our normal av stack is the 50% efficient form.  I used the more efficient circuit with 10kv and it gave excellent results.   I had the explosions with two series gaps and three PFL's though. I am going to try this with an av stack today - just for kicks - pun intended.

See attached.

   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
put caps in series and I'm at 1460v - supply won't go higher but used to go over 1500v - not sure what happened to it, might be the resistor that adjust the output

EDIT: oops had it wired wrong - caps stop at about 750v and won't take more even in series - must be why they are "surplus"

Working up a spreadsheet now to see what I can get to capacitance-wise with the caps I have left.  All of my other hv caps are low capacitance, .01uf at best.



« Last Edit: 2010-06-25, 17:15:42 by Grumpy »
   
Group: Guest


 you could build an MMC.

I have better luck with MMC's than i do with the big expensive pulse caps

Cornell-Dubillier 942C20P15K-F are my favorite.. last forever!
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
got some old electrolytics - which I dispise - probably will short the second I hook them up

EDIT:  yep, they suck

will try with less capacitance and more ballast to limit the current  - if I can get at least 1200v I'll keep on - if not I'll put the driver on a board and make that permanent while I wait for some new caps
« Last Edit: 2010-06-25, 21:45:22 by Grumpy »
   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
pulsing at 1200v to 1350v depending on the frequency - got more caps but need to hook them up - only .005uf now but have .05uf to add

not sure if I am getting full avalanche, but I looks like it is - scope is 300mhz and probe is 250mhz - might be too slow

also not really sure of pulse amplitude do to slow scope

Need to play around with trigger and delay line

   

Group: Tinkerer
Hero Member
*****

Posts: 3934
tExB=qr
so far, everything seems a little wierd with this setup:

trigger voltage effects amplitude of pulse on scope (not sure about real amplitude) and also effect max voltage for stack

trigger rings after the isolation transformer and this ringing is also in the output right after the initial pulse - would like to filter this out and get a clean trigger

difference between 1x and 10x probe setting is only like 2x or 3x on the scope - which makes me think that the pulse is just too damn fast

also notice that pulse on scope appears twice as high when first connected than a few minutes later - kinda like when caps leak more and more when overvolted

removed ballast resistor as it is like almost no current is there

quadrupled cap across supply and very little difference from low value cap  - I think that the cap across the su pply is active when the stack is self-triggering as the voltage at the cap is all over the place

long square triggers show spike at beginning and then small square same as trigger but higher

will start to check for effects tomorrow
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
G the pulse maybe too fast dv/dt but if you are using a x1 probe i would have thought you maybe cooking your scope front end, the max my scope allows on x1 is 80V and on x10 800V much above this can do some damage (So it says in the manual), really need x100 probes for this sort of thing.
   

Group: Administrator
Hero Member
*****

Posts: 3866


Buy me some coffee
Ive just realized to get a higher voltage output all i need do is wind the power supply up.

I have been running at 12v and i can get 1200V but under load it's more like 650V for a 4watt load but i can goto 16V with the 555 and 36 with the LM339 so really i need to put a regulator in for the 555 and i would then be driving the center tapped 6-0-6 transformer with up to 36-0-36 instead of a max of 16-0-16 at the moment before damaging the 555, i just have to watch i dont exceed the 5.6A rating of the FET's which using 12 volts is only 10-50mA without a load and goes up to a maximum of 1Amp right now under load.

I am quiet impressed so far with this psu design, it should be interesting to see how it performs when i have the Avalanche board up and running, which is not that far off now.

I have attached a circuit of the finished PSU and the prototype of the Avalanche board.

The Avalanche board has a 9 stage Cap-Inductor pulse shaping circuit which is link jumper selectable from 1 to all 9 stages.
« Last Edit: 2010-06-26, 12:33:24 by Peterae »
   
Pages: 1 [2] 3 4 5
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2024-03-28, 08:22:56