PopularFX
Home Help Search Login Register
Welcome,Guest. Please login or register.
2024-10-04, 06:11:41
News: If you have a suggestion or need for a new board title, please PM the Admins.
Please remember to keep topics and posts of the FE or casual nature. :)

Pages: 1 2 3 [4] 5 6 7 8 9
Author Topic: DSRD pulse generator  (Read 199818 times)

Group: Professor
Hero Member
*****

Posts: 3472
Not bad.

What is the polarity of the pulse train at pin 2 of your MOSFET driver?
Also, could you post a scopeshot of the current flowing in the DSR Diode ?
   
Group: Guest
Hi exnihiloest,

very impressive, one question concerning the 4429.
This is an "inverting" MOSFET driver.
So this means that the output of it (and the gate of the MOSFET) is always positive except for the pulse duration, meaning it keeps the MOSFET conducting all the time except for the pulse time, Right?

Regards Itsu

Hi Itsu,

The TC4429 is inverting. I send it an inverting pulse (high level except during the pulse, I should have precised this point in my previous post), so that the gate MOSFET is provided with a positive pulse and the MOSFET is conducting only during the pulse.


   
Group: Guest
Not bad.

What is the polarity of the pulse train at pin 2 of your MOSFET driver?

negative (I mean from high level to zero)

Quote
Also, could you post a scopeshot of the current flowing in the DSR Diode ?

I'm working on. I just realize that the capacity of the probe which was connected to the drain to see the back emf, plays a great role. As the back emf was to large for my scope, I put a resistive divider between drain and ground to connect the scope, but in the same time the DSR pulse reduced from 850V down to not even 400V.
So I will add a variable capacitor between drain and source and will come back soon with the scopeshots.
Update: effect not confirmed. Probably a misinterpretation from me of the signals from scope channels 1 and 2 due to their mix on the screen.

« Last Edit: 2012-11-17, 10:38:46 by exnihiloest »
   
Group: Guest
Here is the DSR signal on the terminals of R4. I increased the duty cycle to obtain a brighter trace and reduced the peak voltage to 600V. To get the voltage at the diode terminals, the scope voltage scale must be multiplied by around 100 (x10 for the probe and x10 for the resistive divider R3/R4). Note the short time scale (20ns) showing that the pulse is around 6-7 ns at the base:


The back emf was 475V, exceeding my scope screen.


Here is a view of both the back emf at the MOSFET drain (low trace) and the DSR pulse (high trace), the MOSFET being powered at a lower voltage (10V instead of 30) for both traces to be clearly viewed separately at the screen. This screen shot is just to see the time relation between the DSR pulse and the back emf.
The back emf is here only 180V while the DSR pulse is 280V. There is no direct proportionality between the back emf and the DSR pulse. The better DSR pulses are obtained when they occur during the falling edge of the back emf. Each time the drain voltage is changed, both L1 and the input pulse width must be adjusted to maximize the pulse. Here I have retuned neither L1 nor the pulse width between the two shots but just reduced the drain voltage.


« Last Edit: 2012-11-17, 15:12:01 by exnihiloest »
   

Group: Professor
Hero Member
*****

Posts: 3472
Here is the DSR signal on the terminals of R4.
Unfortunately voltage measured across R4 does not represent the current flowing through the DSR diode.
   
Group: Guest
Unfortunately voltage measured across R4 does not represent the current flowing through the DSR diode.

Not much time now to return to my previous setup and to put a very low value resistance between anode and ground to measure the current. I have modified the transformer with tighter windings and a secondary of larger section. In spite it is better designed for HF, the pulses are much weaker (150V). So I'm trying to identify the key points that enhance the pulse. They seem to be so numerous that I don't even know if I would be able to duplicate my own setup with different components!  >:(

   

Group: Professor
Hero Member
*****

Posts: 3472
Not much time now to return to my previous setup and to put a very low value resistance between anode and ground to measure the current.
Without the current through the diode, we are blind.
   
Group: Guest
Without the current through the diode, we are blind.

I don't know how to measure. I have tried with 0.33Ω in series with the DSR diode anode. Either the resistance is inductive, or it is not low enough: the output signal changed drastically. Then I tried with a current sense transformer (FIS 115) in series with the anode but the output was ringing at low frequency.
A suggestion?

Now I have solved some problems and clarified important points. My new transformer windings work as the previous. I just had a bad contact around the monostable and the pulse width had changed, letting me believe that the winding was critical. It's not. The ratio must be: primary 5 or 6 turns, secondary 2 turns.
I've tried many configurations by adding variable capacitors or inductances, as it is the habit in HF for impedance matching. None improved the functioning.

There are four important points. Point (1) is L3 (see reply #73). It is the most important. The adjustment of L3 ferrite changes the output pulse from less than 100v to more than 800. Point (2) is the pulse width around 350ns and point (3) the drain cc voltage. These three possible adjustements depend on each other. If you tune L3, you must retune the pulse width and the voltage, and so on...
The last point is the MOSFET. The 2SC2543 (500v and only 30W) seems to be very fast and produces pulses up to 800v. It's my today configuration. This MOSFET under 30 v draws 60 mA only. The IRF640 (200v 125W) limits the pulses to 600v. With a 2SK2850 (900v 125W), the output pulses are around 700v.
It seems that a drain voltage higher than 30v doesn't improve the output pulses. For me we just need to have enough power to charge the junction, and to have a back emf as high as possible to help the diode to open after it is charged.

Update: I just realized that the pulse limit was due to the back emf which is limited by the internal protection diode of the MOSFET. So in order to increase the back emf at the secondary of the transformer, we only need to reduce its ratio. I changed the primary winding to 3 turns instead of 5 or 6, and still 2 turns for the secondary. Bingo! After retuning L3 and the pulse width to near 400ns, I get 1000v DSR pulses. The MOSFET is now drawing 80mA under 28v (pulse frequency: 30 Khz). I think it can be now optimized again. Back to work...

« Last Edit: 2012-11-18, 17:54:39 by exnihiloest »
   

Group: Professor
Hero Member
*****

Posts: 3472
I don't know how to measure. I have tried with 0.33Ω in series with the DSR diode anode. Either the resistance is inductive, or it is not low enough: the output signal changed drastically. Then I tried with a current sense transformer (FIS 115) in series with the anode but the output was ringing at low frequency.
A suggestion?
Only a non-inductive 0.1Ω current sensing resistor.
The most fundamental law of the DSR Effect is: The forward charge (QF) must be equal to the reverse charge (QR).
For newbies: Charge equals to current integrated over time ( Q=i*t )

Thus, in the DSR diode the forward conduction time (tF) must be longer than reverse conduction time (tR) and to keep the forward and reverse charges equal, the forward current (iF) must be smaller than the reverse current (iR), in order to satisfy iF*tF = iR*tR.

That's why it is so important to observe the current flowing in the DSR Diode (which can be quite high) - the higher the iR, the better
Also tF cannot be too long or the charge carriers have the time to reach the other side of the PN junction and the DSR Effect disappears.


Idealized current vs. time flowing through a diode exhibiting the DSR Effect.

Belkin at al, state that the forward pumping time must be shorter than 1/10 of the lifetime of charge carriers.
Table 2 on page 8 of Belkin's paper lists these lifetimes (tzh) for various diodes.
« Last Edit: 2012-11-19, 17:38:32 by verpies »
   

Group: Administrator
Hero Member
*****

Posts: 3943


Buy me some coffee
I am wondering, this DSRD pulse looks very much like the delayed pulse i found.
I would use 2 fet's to drive 2 bifilar coils, each coil has a diode across it just as in the DSRD config, i fire 1 pulse and wait a variable nS before firing the 2nd coil, there comes a certain nS delay that causes a large pulse to appear.

[youtube]http://www.youtube.com/watch?v=uMmtSpgAij4[/youtube]
   
Group: Guest
I am wondering, this DSRD pulse looks very much like the delayed pulse i found.
I would use 2 fet's to drive 2 bifilar coils, each coil has a diode across it just as in the DSRD config, i fire 1 pulse and wait a variable nS before firing the 2nd coil, there comes a certain nS delay that causes a large pulse to appear.

It's not clear the way your setup is functioning. If the 2 coils were 100% coupled and their resistance negligible, you would have exactly the same signal at their respective terminals, and imposing different voltages would be impossible (like a shortcut). So the possibility to drive each coil with a shifted signal is due partly to the resistance and mainly to a coefficient of coupling less than 100%. So there is not much difference with taking the difference of the outputs of two independent coils.
Such pulses should be rather wide when connected to a 50Ω load. The interest of the DSR is that the output circuit of the diode has an excessively weak inductance, ns compatible, unlike a pulse coming from coils. Another interest is that the DSR principle is a capacitor providing a current pulse from a voltage while coils provide a voltage pulse from a current. From a practical viewpoint, the former needs high impedance loads to use the energy in a short time while the latter accepts practical loads like 50Ω or less, which moreover are compatible with transmission lines.

« Last Edit: 2012-11-20, 12:41:56 by exnihiloest »
   
Group: Guest
...
Belkin at al, state that the forward pumping time must be shorter than 1/10 of the lifetime of charge carriers.
...

It makes sense. For the better DSR pulses, I observe that the input pulse width must be around 350ns. Now it is also stated that the released charge is not only that one of the diode acting as a capacitor, but also due to the current released by the coil when the diode opens: 5ns pulses on 50Ω implies that C only should be around 200pF which is 4 times more than the junction capacity of my FUF5404. Even if its capacity is more with a direct voltage than reverse, and a dC/dt effect plays a role when the current reverses, I don't think that it could explain the difference.
I will try to measure the current this evening (european time).

   

Group: Administrator
Hero Member
*****

Posts: 3943


Buy me some coffee
Quote
It's not clear the way your setup is functioning. If the 2 coils were 100% coupled and their resistance negligible, you would have exactly the same signal at their respective terminals
The coils are wound side by side, i cannot remember now but about 15-17 turns, their resistance is very low, one side of each coil is connected together and connected to a +V supply, each coils other end is driven by a fet to ground.
So one fet pulls 1 coil down to ground for about 100nS then get's opened then comes along the other fet after 200nS(Varies) and connects other coil to ground at which point you get the sharp large pulse occur, each coil has a UF4007 across, i was wondering if maybe at a certain time delay that the diode may be in a recovery mode that allows the DSRD effect to enhance.

My other theory was that if you drive a coil with 2 pulses, each pulse spaced in time to the period of resonance then the amplitude becomes much larger as the impedance gets larger.

For instance if a coil is pulsed with 2 pulses 100nS apart and the coil has a resonant frequency of 10Mhz, just 2 pulses are needed to make the coil think it is being driven into resonance, so at 90nS delay the amplitude is less, only at 100nS is the coil being driven into resonance.

Anyway sorry for going off topic i thought it maybe worth throwing this at you just in case it explained my pulse.
   
Group: Guest
I have tried to reduce the number of parameters acting on the functioning and came to a schematic without transformer, and even without ferrite:



At the begining, L1 and L2 can be wound with spaced turns and adjusted with a ferrite core. Then their inductance can be increased by compressing the turns, and the ferrites can be removed. It doesn't a matter, there is no difference, so we can conclude that the ferrites were not saturated and play no role here except increasing the inductance.

Here are the results.
The pulse width is around 6.5ns and the measurement is made at a repetition rate of 20 Khz.

Under 20v, 61mA, i.e. 1.22W only for the MOSFET power consumption, the pulse peak voltage is around 600v under 50Ω. This mean an instantaneous peak power of 6002/50=7.2KW.
Now with my radio receiver I observed that the harmonics are at least 20 dB down the fundamental near 77 Mhz. They are negligible. So we must not consider that the pulses are square, we must rather approximate them with half-sine. Consequently their rms value is 600/√2 and the rms power in the 6.5ns period is: 6002/4*50=1.8KW.

The instantaneous peak power is not instructive. Only the rms power allows us to know how much energy is carried by the pulses: E = rms Power*6.5ns.
It is also interesting to evaluate the mean rms power. The duty cycle is 6.5ns/(1/20Khz) = 0.13*10-3. The mean rms power is therefore 0.13*10-3*1.8KW=0.23W. The input power being 1.22W, no overunity  :(. In fact the efficiency is higher due to energy in the ripples after the pulse. The load heats significantly although it is made of two resistances in series of 23.5Ω, 1/2W, followed by the third of 4.7Ω for the output divider.

Above 20v, the consumption increases but not the pulses.
With 2 diodes in parallel or with 3 diodes in series, I didn't notice any improvement. With two diodes in series, under 28v, 94mA, i.e. 2.63W for the MOSFET power consumption, the peak pulse voltage is around 800v under 50Ω, meaning a rms power in the 6.5ns period of: 8002/4*50=3.2KW, and a mean rms power of 0.4W. Nevertheless the load resistances become hot, so we can consider that my power calculus is underestimated (probably because of insufficient pass-band of the scope) and the mentioned powers should be considered as minimum.

What I may conclude:
- 600v/50Ω pulses are easily reached with this setup, provided that the diode is as good as the FUF5404.
- L1, C2 and the pulse width are the critical parameters, and they depend on each other.
- L2 can have to be also adjusted but is of less importance
- the MOSFET must be fast and able to handle high voltages (500-900v) rather than high currents
- the better DSR pulses are obtained when they are synchronized with the falling edge of the back emf

   
Group: Guest

19% efficient.  Where could 80% of the energy go?


   

Group: Administrator
Hero Member
*****

Posts: 3943


Buy me some coffee
Great work again Ex

Quote
19% efficient.  Where could 80% of the energy go?
I would imagine something should be getting hot? would be interesting to know where the waste goes  ^-^
   
Group: Guest
would be interesting to know where the waste goes  ^-^

To the apple tree.  It's OU  :D
   
Group: Guest
19% efficient.  Where could 80% of the energy go?

- heating the MOSFET
- heating the diode
- radiating EM
- heating the load outside the 6.5ns pulse duration (during the ripples, and also a part of the direct voltage that "loads" the diode during the 420ns period is also wasted in the load resistance).

I didn't see this setup as a possible OU device, but as a tool to output powerful current pulses that could be used elsewhere to "shake" anything in new experiments. Pulses of 600v/50Ω=12A at ns rise/fall time, is something unattainable by ordinary generators or by back emf.

   

Group: Administrator
Hero Member
*****

Posts: 3943


Buy me some coffee
Quote
I didn't see this setup as a possible OU device, but as a tool to output powerful current pulses that could be used elsewhere to "shake" anything in new experiments. Pulses of 600v/50Ω=12A at ns rise/fall time, is something unattainable by ordinary generators or by back emf.
Indeed i fully understand that.

I was just trying to understand the losses a bit better  O0, doesnt matter as you say it's a good acheivement with a simple circuit.
   
Group: Guest
Still playing with DSRD...
I observe strange effects with old diodes probably dating the 70's. Unfortunatly these diodes are not marked. They need a longer time of pumping 1->3ms so I needed to modify my monostable generator to exceed its previous limits.



One of the diodes, this one with the alligator clip, outperforms the FUF5404 in term of voltage, but not in efficiency nor sharpness. It produces 1200v pulses on 50Ω. I burnt it. I had another one of same type which gives the same results, but I limited the pulses to 800v. It's astonishing to see these old diodes which obviously are not fast recovery diodes and were probably used for mains voltage rectifying, to produce pulses with width around 15ns at voltages that are surely exceeding what was specified for their reverse voltage. I think that the reason for DSR diodes to exceed the specified reverse voltage is due to the current direction. In a normal case, an reverse voltage is applied and the diode acts as a load which absorps the current. During the DSR pulse, the diode is the generator, so the current is provided by the diode in a reverse direction. The two cases are totally different and so, the maximum voltage of DSR pulses can't be estimated from the datasheet.

The second diode gives a stranger effect. After a long time of pumping as the previous one, this one gives a negative pulse limited to near 200v. I don't understand what happens, it's surely depending on the physical implementation of the junction and needs surely physics for explaining. The pulse appears far after the back emf, after its ripples, at a time when the input signal from the MOSFET drain is back to zero.

So if you have old elecronics thingamajigs in your drawers, don't throw them, you may have fun with them...   ;)

   

Group: Administrator
Hero Member
*****

Posts: 3943


Buy me some coffee
I wonder if old transistor junctions could be tested as well, would that also work?
   

Group: Tinkerer
Hero Member
*****

Posts: 3055
Quote from: exnihiloest
So if you have old elecronics thingamajigs in your drawers,
don't throw them, you may have fun with them...    ;)

Hmmmmmmm.  Still thinking about that one...

In my "drawers" you say?  ;D  >:-)  :-[


---------------------------
For there is nothing hidden that will not be disclosed, and nothing concealed that will not be known or brought out into the open.
   
Group: Guest
   
Group: Guest
I wonder if old transistor junctions could be tested as well, would that also work?

I have also tried base/emitter and base/collector junctions of some old silicium NPN (2N3055, 2N2219). There is no DSR effect. Now it's not ended, more transistors must be tested, PNP also, unijunction, germanium...

   
Hero Member
*****

Posts: 1579
Drift Step Recovery Diodes seem to be the best way today to produce high power pulses (up to GW) at ns time rise, and with simple means.
I am not up to speed in these matters but is this not exactly what Bob Boyce and his remarkable 101 plate HHO electrolysers need
for "cracking" water?
See page 34:
http://www.free-energy-info.co.uk/Chapter10.pdf
   
Pages: 1 2 3 [4] 5 6 7 8 9
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2024-10-04, 06:11:41