Exploring the Future Potential of AI-Enabled Smartphone Processors
Electrostatic kick report
1. Kick Report
Draft report
By Roberto Notte
v. 1.0
02/25/2013
Each test is daily documented in a preliminary easy format
2. Kick Report
Kick Report
A brief kick’s
hystory
Draft report
The Kick
During the far 2006 I joined firstly to OU forum where many guys
astonished by Steven Mark TPU demonstrations tried in every
possible/impossible way to replicate the secret original device.
After a 5 years of efforts no definitive retro-engineering possible.
Today the problem still exists… providing that it is not a fake.
Since 2006 Steven Mark
astounded the world
presenting and
demonstrating his 15”
3KW TPU (Toroidal
Power Unit) apparently
As per leaked SM info, basic to TPU operation is the Kick. It is
just such small pulse that triggers the TPU to start .
delivering 220VAC
voltage . Since then
thousands of
Object of this research is to find the way used by SM to generate
researchers tried
the kick.
without fortune to
replicate his design.
1
3. Kick Report
01/20/2013
From: http://en.wikipedia.org/wiki/Switched_capacitor
The ‘Flying Capacitor’
…The simplest switched capacitor circuit is made of one capacitor C and two switches S1 and S2 which connect the capacitor with a
given frequency alternately to the input and output. Each switching cycle transfers a charge from the input to the output at the
switching frequency
. Recall that the charge q on a capacitor C with a voltage V between the plates is given by:
where V is the voltage across the capacitor. Therefore, when S1 is closed while S2 is open, the charge stored in the capacitor CS is:
When S2 is closed, some of that charge is transferred out of the capacitor, after which the charge that remains in capacitor CS is:
Thus, the charge moved out of the capacitor to the output is:
Because this charge q is transferred at a rate f, the rate of transfer of charge per unit time is:
Note that we use I, the symbol for electric current, for this quantity. This is to demonstrate that a continuous transfer of charge from one
node to another is equivalent to a current. Substituting for q in the above, we have:
Let V be the voltage across the SC from input to output. So:
So the equivalent resistance R (i.e., the voltage–current relationship) is:
Thus, the switched capacitor behaves like a lossless resistor whose value depends on capacitance CS and switching frequency f…
What exposed is of course the ‘standard’ knowledge but in the case I want to present, the output
charge is not connected to any direct load but rather is simply used to charge 15cm of wire used as
antenna. Such wire is not resonating to anything so the charge on it can only interact in a way with
surrounding environment.
2
4. Kick Report
What I found is that allowing to charge the wire sequentially to both C sides does indeed solicit
the local environment to re-balance the energy giving birth to a 20ns pulse that can be captured
by a near lets name it ‘collector’ wire.
I strived to find any other possible external source of such pulse but I have not been able to identify it.
1st t est
What it follows is my first implementation to try duplicating the kick effect observed by Marco with
his quite different setup . I have retained that the T. Bearden classic ‘flying cap’ setup would well be
taken into advantage but with a twist: the antenna connection and relative induction into an external
collector coil.
It is clear that the circuit as drawn, after the initial transient that charge the cap (surge limiting
resistor is a must), does not provide any system closure to 150V PS providing no switches K1,2,3,4
overlap.
The antenna wire is simply connected/disconnected sequentially (break before make logic) to both
side of C capacitor. The cap itself is recharged for 20msec for every cycle. As I have seen there is no
need to recharge the cap as it does stay charged even after 24 hours. This imply that I could get rid
of picoreed k1, k2 and associated CH1 just starting the circuit by using a pre-charged capacitor!
Perhaps simply feeding back the kick to cap could realize a self runner.
The basic sequence as detailed in schematic 1 could be:
1- close momentary the paralleled reeds k1 & k2 in order to charge the cap
2- open, till next start sequence, the paralleled reeds k1 & k2 to isolate the cap
3- close momentary reed k3 to connect +C side to wire
4 - close momentary reed k4 to connect -C side to wire (not used in the first test)
5 - repeat cycle
The Data pattern generator may be programmed to do all the necessary switching allowing as well
to vary the small delays necessary to realize the 'brake before make' pico reeds operations. Once
programmed the sequence may be run in closed loop at whatever repetition rate. Picoreed using is
nice because that small devices are able to reach easily 100 ore more switching/sec. and perform like
an ideal switch.
An interesting operation explanation (about Marco’s finding) has been provided by k1ngrs:
“The process of charging up a coil surface with a charge of either +ve or -ve will require an
amount of energy to be transferred by the capacitor but I am not sure if that alternating
between the +ve and -ve sides of source the capacitor will reduce its total charge after many
alternating connections.
What may be happening is that the medium between the 2 coils (air) acts as a dielectric and
3
5. Kick Report
transfers a charge directly to the output coil which is consumed by the load or the surge of
charge to the surface creates an electrostatic shock wave (if there is such a thing) or an
electromagnetic wave.
I think some tests need to be done with 2 coils tuned for perfect resonance with each other and
for capacitor pulsing applied at this resonant frequency.”
I observe tat:
The picoreed coils can radiate a bit.
The picoreed actuation coil could induce on his contact every time it gets a +4V pulse. Anyway I
avoided bouncing effect by paralleling the suggested diode. Well, to check for this effect should be
enough to connect the existing available shield between coil/contact (pin 9).
Even though the capacitance between the wire and the coil is very low it still could be the cause of
the effect.
100ms
Data Pattern Generator
1
2
1
2
CH2
2
Picoreed
CELDUC
D31A3100
20ms
GND
CH1
Charge C
CH1
1N4148
4
CH3
6
3
3
CH2
6
Connect C positive
5ms
CH3
2
3
4
4
4ms
Connect C negative
Solenoid/350t
+
7
14
1
C
Straight wire
Lenght=25 cm
3
150V P.S.
-
7
14
8uF/630V
R=5.6K
2
4
probe1
probe2
Differential input
oscilloscope
Figure 1 basic test circuit
Setup:
Vps = 50VDC (DC power supply voltage to picoreeds k1,k2)
F= 10Hz (cycle rate)
T=20msec for C charge (picoreed k1, k2 ON duration)
t=5msec pulse to Antenna (picoreed k3 ON duration)
Tdelay=4msec (delay time between picoreed 1, 2 and 3) (picoreed 4 always opened)
4
6. Kick Report
Figure 2 Data Pattern Generator setup: picoreed1,2 and 3 commands
Picoreed used k3 (C pulsing with only + side); picoreed k4 is left alw ays open.
Capture coil: solenoid 350t, 1.6cm diameter, Cu en 0.15mm wire closed on R=5.6K
Antenna: wire length 25cm inserted full, partially or external to capture coil
Figure 3
From left to right you can see picoreed k1, k2,k3,k4 and the 3 input commutation commands jacks
coming from Data Pattern Generator (the first jack commands bot 1 and picoreed).
The C is actually composed by 8 x 1uF caps in parallel. Those caps are MKS qualified at 630V (
model: 36527 p654).
5
7. Kick Report
Coil is actually a Kacher coil I already had. It is closed on 5.6K. The shown differential probe
connection it is absolutely necessary (high CMMR) to see the kick…otherwise too much 50Hz
induction.
01/21/2013
Several interesting facts appear.
1 – while antenna wire full inserted into coil (for all its length) the coil does increment by about 200%
its ability to capture local RF (HF stations, WIFI, ….). Extracting the wire for ¾ the coil shows only
marginal ability to capture local RF fields. The kick appears to be the same for both conditions.
2 – The kick does appear to be around 0.2Vpp appearing of course every 100msec. The kick does go
positive and negative. It starts with very low amplitude and within 30sec it does reach 0.2V
positive…then cycle repeat and it decrease till zero and then to negative…it is like the kick sums up
on 50Hz base…and the sliding due to fact that I have used a precision 100,000Hz clock on data
pattern generator. The 50Hz on grid is not so correct/steady.
3 – With 25cm antenna wire full inserted into capture coil that is the waveform captured
differentially across R
Figure 4
4 – Extracting for ¾ the 25 cm antenna wire or laying it externally ad a 5cm distance:
6
8. Kick Report
Figure 5
The kick are evident and I checked it with PS varying from 10V to 150V: few differences seen (later I
discovered that there is a non linear relationship between the C potential and output kick
amplitude). Probably I should use 500V or more. The problem is that picoreed are rated 100VDC
max, so (Ok, it is also evident that there is not any ddp between contact themselves …as long as
there is not a circuit closure)?
I checked also that setting to 0V the PS the kick effectively disappear…but re-appears with only 510V applied.
The C itself is re-charged every 100msec …hence it stays always at max PS applied: this implies that
I could use a longer timing.
I must now do same measure but using different capture coils in order to see what parameters are
better to obtain higher kick’s amplitude.
Energy in a capacitor = 1/2 x C(farrads) x V^2
E = 0.5 x 0.000008 x 50 x 50
E = 0.01 joule
OK so to charge my capacitor of 8uF to 50V requires 0.01 joule of energy.
So therefore 1 joule = 1 watt for 1 second then for 20 nsec …very low power!
Please do note that the technique I've implemented is the so called 'FLYING CAP' (generally used in
chopper DC amplifiers): it is clear that any kind of input I use for charging the cap, nevertheless the
cap is then completely 'disconnected' from the charging source...NO WAY the source tampering
with the output (well, except trough the picoreed plastic body having electric rigidity=1.5KVAC).
7
9. Kick Report
What seems really important is to constantly keep the local environment unbalanced...so assuring a
constant flow of energy: here I have tons of possible ideas. Of course that's only a preliminary
hypothesis. About background noise: I've had to switch OFF all eco lamps in the room (each lamp is
actually a transmitter) but at least in my case the 'capture coil' I'm using is to be considered as small
antenna as it is 300t single layer on a 1.6cm diam. PVC tube.
I do not know if you have realized that switched wire potential INCREASED by five/ten fold the RF
capture ability of the coil itself ;) IT SEEMS THAT I HAVE DISCOVERED AS WELL A NOVEL
METHOD TO IMPROVE RADIO RECEIVERS ;)...I'd like to hear your opinion.
02/22/2013
Two pulse test
In this test I used both K3 and K4. After received a new set of picoreeds (the previous melted for my
error) I've been able to run the test. I do confirm that for PS=50VDC (charging voltage on cap C) and
antenna wire inserted into capture coil for 5cm, I obtain a ±2V kick every 100msec (the chosen
repetition rate). I do confirm also that said kicks are 50% modulated in amplitude (AM) with 2 - 3 sec
timing.
Figure 6
The kick's amplitude is dependent on PS voltage even if there is not a strict proportional amplitude;
for 50V I have +-2V, for 150V I have +-8V.
8
10. Kick Report
Figure 7
The Kick's detail for PS=50V is the following
Figure 8
The medium kick's width is about 20nsec. It is clearly difficult to obtain such wave as repetition rate
so low.
Programming the generator for 2 more pulse in both channel 3 and 4, as expected I obtain as output
2 positive and 2 negative pulses
9
11. Kick Report
Figure 9
The picoreed k1 and k2 only experience much of a current pulse during the very initial charging of
C (surge current). During subsequent time as the C remains charged the recharging current is almost
not existing. The natural C discharge time is hours… nevertheless placing a current limiting resistor
of 1K should protect for the initial surge. The following pic shows the updated circuitry with C=1uF:
more than enough for this test.
100ms
Data Pattern Generator
CH2
2
Picoreed
CELDUC
D31A3100
1
2
1
2
GND
CH1
Charge C
20ms
CH3
CH1
2
1N4148
4
CH3
6
3
3
CH2
6
Connect C positive
5ms
3
4
4
4ms
Connect C negative
+
1K
Straight wire
Lenght=25 cm
14
7
1
Solenoid/350t
C
3
1K
-
7
14
1uF/630V
R=5.6K
2
4
probe1
probe2
Differential input
oscilloscope
Figure 10 Basic circuit – updated1
This way of operation: namely charging the antenna wire alternatively from C positive and negative
side led me to pass from 200mV kick to 2V kick:
THAT’S A MAJOUR STEP AHEAD. The ZPE ramifications in my opinion could be
astounding!
10
12. Kick Report
01/24/2013
Capture coil load test
With reference to basic circuit I checked the effect of different load resistor R while using same basic
setup. Reassuming Vps=50V, wire inserted for 5cm into coil and differential measure, I have:
1. R= 5.6K ->Vout= +2/-2V
2. R= 470Ohm -> Vout= +1V/-0.3V
3. Rising Vps to +150V and R=470 Ohm ->Vout = +3V/-1.5V max
In following pic you can see the output across the R=470Ohm with evident AM modulation with
T=about 3sec.
Figure 11 Output
I managed also to make a current measure on R lead with surprising results.
Figure 12 Waveform captured with Tek Current probe
This is the scenario I see:
11
13. Kick Report
- Pickup coil & R load are 100% floating and not connected to anything, the only exception being the
scope's two High impedance probes. At what potential is that circuit? Well, lets suppose is at 0V
level (because of the probe connection that with time brings the circuit to GND level).
- Suddenly I transfer some charge to antenna: IE for a 5ms the antenna potential is raised from 0V to
+150V. Indeed there develops a ddp between the antenna and the pickup coil that is at 0V, hence
some current should flow even if minimal, in effect I'm using flat DC so the initial transient peak
current comes only trough air dielectric (in effect C remains charged for at least 4 hours of
operation) I'm speaking about at least 1000- 10000MOhm (current injected in this way less than
1nA).
- After some complete cycles the C is completely charged and do not require any more current from
PS. Nonetheless it does continue to deliver transient charge to pickup coil and load...at least for
4hours.
- I put a tek Current probe on R lead and measured (for R=470Ohm) +58mA and -80mA. The big
question is: FROM WHERE DOES COME SUCH POWER?
Considering that the Current probe transfer function in the case is 2mA/mv, the scope reading is
+28mv hence 28*2=+56mA while the negative peak is 40*2=80mA. Anyway as the pulse duration is
only 20ns then the associated energy is very low.
While R=5.6K, PS=variable, T=100msec, antenna wire over the coil for ½ coil horizontal length I did
measure the following output for PS = 10V to 150V. I remember that in this standard case the kick
output is composed by two kicks (corresponding to K3 and K4 actuation): the first negative and the
second positive. I remember also that said amplitude are amplitude modulated (30%) with a 1020sec repetition time, hence following measures are run taking note of only the max values reached.
8
4
Kick amplitude
2
kick (V)
PS
voltage 1st pulse 2nd pulse
10
-1,2
0,5
20
-1,8
0,8
30
-2,4
1,1
40
-3
1,5
50
-3,9
2
70
-5
2,8
100
-8
4
150
-10
7
2nd pulse; 7
Kick amplitude
6
0
-2
10
20
30
40
50
70
100
150
-4
-6
-8
1st pulse; 10
-10
-12
Figure 13 Kick Output
12
14. Kick Report
So, it appears that there is a non linearity at about 70V – 100V. At that point the output does increase
significantly its rate of increase. It would be interesting to rise the PS up to +500V in order to see what
happens.
It remains to say that both radiation & scope or probe issues must be taken in account. Let’s examin e them in
any possible details:
- Contact fast rise time - true - but I had mixed results while trying to measure it in circuit. I suppose I must do
it with an external picoreed while switching a resistor. The problem is that the kick rise time I saw using the
differential mode 200MHz scope is 1ns (I do not see any bouncing has only a 15% undershoot and only one
ringing of about 10% of the peak signal (on the pulse trailing edge). So, the pulse width = 20ns. Doing the
same measure using the 1GHz digital scope I obtained different results: 120ns rise time and 1.3us fall
time....Hmmm...both scopes are more than 20 years old and suffer of some malfunctions hence at this point
I'm not so sure about that measures. Surely I'd like to get a new scope!! Please consider also that I used the
inverse diode across the picoreed coil. NO mercury wetted contacts.
- 150 volts in 1ns that is still 150000 V/us: that's it, I only stress the fact that the pickup coil does see only 3cm
of antenna...of course it does capture the radiation coming from the other 22cm. All in all I've seen that
radiation contribute is at least 100 less than the 3cm contribute. Remember the inverse square law with
distance for the EM radiations. It remains that my good(?) Tek P602 current probe measures up to 50-80mA
into 470Ohm load resistor! from where it does come? The current measure way does eliminate any external
field! It goes without saying that I could eliminate the antenna wire and slide the coil directly over the K3, K4
junction wire (and making such wire no more than 5cm long.
- Differential measure are very difficult mainly due to 10Hz pulse rate and nanosecond rise time pulse.
- Every piece of wire excited by a 'Dirac delta' does radiate. In my case, in base of my RF knowledge, the
radiation on K3, K4 picoreeds does exist but it is very low: probably its contribute in nano/micro ampere onto
a few cm of a possible receiving wire. In the object case the K3,K4 junction wire is at least 15/20cm far from
the pickup coil...hence the contribute very low, not significant to 80mA seen on R load.
- The 25cm wire I used certainly has its own characteristic impedance and radiates at his resonance
wavelength even if working at only 10Hz and relying on the broad band spectrum generated by the kick it
does magnify only the very high portion of spectrum: this implies that its amplitude very low.
- For what I've seen, there are at least two problems associated with probes: 1 - the necessity to reject the
50Hz noise coming from grid and bench apparatus, 2 - the rejection of RF signal permeating the local
environment. In the first case CMRR easily gets rid of 50Hz issue. RF issue do not pose a problem as its level
about 1/100 of the picked up signal (8V). Current probe confirmed such fact: the wave forms are very similar
(with that of differential voltages): actually I do not see any RF feed trough even if going at more sensible
scope's input settings.
- In this case I do not think that SPICE simulation possible. Si I will not spend precious time on PSPICE.
I do not exclude that radiation contribute could be messing everything. I urge anybody to confirm at least my
8V kick measure on 5.6K load resistor.
13
15. Kick Report
Ciao
01/25/2013
Test with different coils & wires
The following series of tests is aimed toward finding the best coil to use for this kind of ‘potential’
transfer circuit. Of course I’ve used a set of coil I already had.
1. BIFILAR PANCAKE COIL (R=0.47Ω, L=170uH). I tested both electrical & magnetical coupling
with the wire antenna. In this case I found that best result is using mag coupling forming 1 turn
with the wire antenna and locating it near the pancake. Overall the output, for PS=50VDC and
R=5.6K, is lower than what obtained with the Kacher coil used in the first test.
2. SIMPLE ONE WIRE PANCAKE COIL. In this case as well the best output is with mag coupling
but result is lower than Kacher coil.
3. 150t 2cm diam SOLENOID. Lower than Kacher coil.
4. 100t 1.6cm diam LITZ WIRE SOLENOID. Antenna inserted for only 1cm. Output is same of
Kacher .
Figure 14
5. Test as before but using a 10cm 200 Litz wire as antenna. Output is ±10V for PS=150V. Antenna
wire loose coupled to coil and external (to coil).
6. Test with toroidal small coax coil. The antenna wire connected to coax external shield and output
of a 2t collector closed on 560Ω.Output is ±15V for PS=150V
7. Test using a toroidal coil + 1 turn Moebius coil. Antenna connected directly to one side of toroidal
single wire coil. Load of 470Ω across Moebius junction. Output +15V/-30V for PS=150V, Current
on R is about ±60mA max.
14
16. Kick Report
Figure 16 Current on R (2mA/mV)
Figure 15 Test with Toroidal coil + Moebius coil
It is clear that best coil is the N.5 and N.7. In the latter there is a caveat as the antenna is directly
connected to toroidal winding and Moebius. Well, its all floating...but who knows?
Disconnecting power supply
With reference to Basic circuit diagram (fig. 10) while device in operation I disconnected the PS (both
wires) and found out that device still running on itself. In effect the capacitor C does not have a
discharge path except trough : dielectric air between his own leads, picoreed plastic body and
dielectric air between antenna wire and caption coil surface. It is to be noted anyway that the caption
coil is 100% floating with no possibility at all to close the circuit to C.
I left the circuit operating on itself and it appears that the C (1uF initially charged to 150V) very
slowly discharges: after 3 hours it halved the output amplitude (to be confirmed): yes you hear well
– w hile continuously dissipating energy on R +15V/-15V on 470Ω load resistor.
Making a tentative loop-back
Tentative #1 Single pulse feedback
100ms
230VCA
Charge C
20ms
k1
k2
k1
k2
Data Pattern Generator
Connect C positive
5ms
CH1
GND
CH1
CH2
2
CH3
2
2
Picoreed
CELDUC
D31A3100
4
CH3
9
6
6
9
9
6
9
Connect C negative
6
Caption coil
Solenoid/80t – Litz wire
1K
14
7
K1
230VCA
C
150V P.S.
Straight wire
Lenght nserted=3.5cm
Litz – 200wires
K3
1K
-
4
4ms
S1
PS ON/OFF switch
+
3
3
CH2
2
7
14
K2
1uF/630V
MKS type
R=560Ohm
S2
K4
probe1
probe2
D1
Differential input
All diodes 1N4148
oscilloscope
Figure 17 Loop-back #1
15
17. Kick Report
Figure 18 Feedback diode and S2=OFF
Setup:
Pickup coil N.4 (fig.13, 17)
PS = 150V
T=200msec
PS switch S1=OFF
Load switch S2=OFF
After pressing momentary the S1 switch in order to charge C, I left S1=OFF so PS disconnected and
circuit self-running. The output (no load) is +40/-10V peak max.
Figure 19 Output (no load) loopback simple
16
18. Kick Report
Note:
the negative kick does appear rectified in positive region.
The AM effect is still present with an observable T=5sec.
After ½ hour the situation remains the same. What happen for subsequent time is that some pulses
are missing, then the circuit auto recovers and starts again; each time at a lower level. After 3 hours
kick’s amplitude reduced to +3V.
Putting S2=ON (connecting the R=560Ω) the output voltages stabilizes to 10 / 15V that respect the
2V for the pickup coil N.4 is a big improvement.
Changing R to 5.6KΩ -> kick jumps to +40 and -15V.
(Setup: k3 and k4 command pulse duration to 5msec and disconnected PS after pre charge C to
150V).
Figure 20 Kick output with Tentative #1 and R=5.6K
Tentative #2 - Double pulse feedback
After several different test the best loopback proved to be the following:
17
19. Kick Report
230VCA
100ms
Data Pattern Generator
GND
CH1
Charge C
20ms
k1
k2
CH2
2
2
CH3
2
k1
k2
CH1
2
Picoreed
CELDUC
D31A3100
6
6
9
9
6
9
4
CH3
6
3
3
CH2
9
Connect C positive
5ms
S1
PS ON/OFF switch
4
4ms
Connect C negative
Solenoid/80t – Litz wire
+
1K
14
7
K1
230VCA
K3
C
150V P.S.
1K
-
7
14
K2
1uF/630V
MKS type
Straight wire
Lenght nserted=3.5cm
Litz – 200wires
R=5.6K
K4
probe1
probe2
All diodes 1N4148
Differential input
oscilloscope
Figure 21 Tentative #2 double pulse feedback
Here I captured the output waveform across R=5.6K after disconnecting the 150V PS (S1=OFF).
Figure 22 Output kick with 2 diodes loopback
It is evident that the negative pulse has been rectified and appears in positive quadrant near the
normal kick. MAX POSITIVE KICK OBSERVED = +40V. AM still present but limited to less than
20V (AM=50%).
18
20. Kick Report
Note: in these Tentative feedback, there is no more 100% galvanic separation between C and pickup
coil. In fact in this case there is a little current flowing as C does discharge itself within 5 seconds
(while left disconnected from PS).
It is anyway worth to note that the test’s aiming not to setup a selfrunner (that could be considered a
by-product) but rather to auto generate kicks in KV range.
Mosfet’s Switch v ersion
01/30/2013
In this implementation I replaced the picoreeds with optical isolated bidirectional mosfet switches
from now on called SSR. Said modules are equipped for this test with IRF820 mosfet hence the max
voltage is limited to xxx it is clear that using the STP xxx I could easily use up to 1.2KVDC as source
of potential.
Command waveforms , F = 10Hz
230VCA
100ms
Data Pattern Generator
Tek DG2020
Charge C
20ms
SSR1,2
CH1
Connect C positive
5ms
Tek I/0 unit P3420
CH1
GND
CH3
CH2
3
3
CH2
4
CH3
4
4ms
Connect C negative
S1
PS ON/OFF switch
+
110VCA
Straight wire
Lenght over coil=3.5cm
Litz – 200wires
PIckup coil
Solenoid/80t – Litz wire
10K
SSR1
SSR2
3KVDC P.S.
FLUKE
C
10K
1uF/630V
MKS type
-
R=560Ohm
S2
SSR4
SSR2
Current probe
D1
+
Probe1
Tek P602
probe2
-
+12V PS
All diodes 1N4148
Differential input
Floating power supply for SSR
Oscilloscope
Tek 2232
Figure 23 SSR test circuit version
Note: it is important that SSR2,4 really fitted with bidirectional features.
First test indicates that using PS=50V, F=1KHz, I measure ±2Vpeak on 5.6K load. Tested it for F up to
10KHz without noticing any difference in amplitude. Rising PS To +150VDC I measure +10V/-8V on
5.6K R load. I noticed that running at different F and PS=50V the voltage measured across C is:
19
21. Kick Report
F
10KHz
5KHz
1KHz
500Hz
100Hz
10Hz
C voltage
48.7V
49.1V
49.4V
49.5V
49.5V
49.5V
So it’s clear that using SSR there is some more load on C: hence it discharges quicker. Consider that
for every cycle there is a charging pulse (20msec @ 10Hz, shorter proportionally with F rising).
01/31/2013
I tried firstly to get a rise/fall time measure:
Figure 24 Kick at f=1KHz, PS=150V
From pic it is possible to see that rise time not measurable (it’s a vertical line) so less than 1ns, fall
time about 115ns. Middle duration=20ns, Amplitude=8Vpeak. It is also evident a typical 50 MHz
component overlapping all the pulse, its undershoot and also for the subsequent small oscillations..
I verified the two probe differential calibration by joining the tips: the output is less than 1V so much
less than the observed waveform.
The main observed effect is on current on R (5.6k) load: THERE IS NO CURRENT OBSERVABLE. I
did the measure several time but always with same result. Well, it seems to me weird so I think to
remake that measure at different time.
I measured as well the C charging current on + side = 0.6A peak, an minus side = -0,2Apeak. Current
going into SSR 3,4 is measured as ±0.2A peak. Of course I’m referring to a pulse much like the kick.
20
22. Kick Report
02/09/2013
Serial cap arrangement
Figure 25 Serial cap charging breadboard
The aim of this test is to stress the environment with pulses having different amplitude in order to
solicit each time a rebalancing flux.The circuit shown has been practically built using SSR as
switches: implied is the fact that in output only low energy as previously checked. In order to look
for max amplitude feedback diode is used.It is important also to note that switches sequence must
every time invert the polarity. The sequence start closing for 5msec SSR0: this charges the 3
capacitances to full battery voltage (approx. +150, +100, +50V). After 2msec the sequence starts:
SSR1, SSR4
SSR2, SSR4
SSR3, SSR4
Figure 26 Data pattern generator setup: 100bit, clock=1KHz, 1msec/bit
21
23. Kick Report
The full actuating sequence is arranged in 100 memory locations and run with a variable clock (for
example 1Hz- 1MHz). Selecting 1.0KHz clock, each memory location lasts for 1msec. In 100 memory
locations I’ve allocated 2 full sequences:
SSR0, SSR1, SSR4, SSR2, SSR4, SSR3, SSR4, SSR0, SSR1, SSR4, SSR”, SSR4, SSR3, SSR4…repeat
SSR0
SSR1
1N4148
+
-
DC
SSR2
+
-
SSR3
+
-
R=5.6K
antenna
SSR4
L=80t Litz
Typical sequence
S0,
charge C
S1, S4
S2, S4
S3, S4
S0,
charge C
---
Figure 27 Serial cap charging, all C = 1uF/630VL
Connecting differentially two probes across R and using a 3KHz main clock and 150VDC, I’ve an
output of ±200Vpeak
Figure 28
From upper fig, I note the presence of 2 kinds of bursts: 1 st with t=0.15msec, 2 nd with t= 200usec. I’ve
then tried to look at the bursts as in following pics. I note that bursts are similar to typical Kick
previously reported.
22
24. Kick Report
Why these bursts appear is not clear. They do not have any clear pattern nor they seem to happen
randomly. Some time pulses sum up and amplitude exceeds the 150V charged applied voltage. The
SSRs themselves do not have such artifacts.
Figure 30 Burst
Figure 29 Burst
I’ve tried as well to see how about charging a cap after FWBR the output:
1N4148
SSR0
SSR1
Cout
+
-
DC
SSR2
+
-
SSR3
+
-
antenna
L=80t Litz
SSR4
Typical sequence
S0,
charge C
S1, S4
S2, S4
S3, S4
S0,
charge C
---
Figure 31 Charging output capacitor using kicks
What happens is that Cout charges up to 50V within 6 seconds.
Changing the switching frequency I’ve been able to check the DC output:
F (Hz)
10
20
30
Vout (V)
+25
+30
+35
23
25. Kick Report
40
50
70
100
150
200
+37
+38
+38
+40
+44
+44
Mercury wetted Relays switcher
Figure 32 Switcher mounted on PCB with ground plane
Having experimented the critical picoreed use due to easy contact melting for input voltage > 150V, I
retain useful to complete this study setting up a ‘mercury wetted contacts’ relay version in order to
avoid the cited issues and look for eventual differences due to 100ps switching time. My chose went
to possible vintage new units available on ebay. There, after some search, I found suitable new
CLARE - HGSM5181 mercury wetted contact SP2T. They sustain up to 500V operation and up to
200Hz switching speed without any need of clamping coil diodes as mercury assures intrinsic no
bouncing. The bad side is that relay actuation is much more complex as you do need to supply 2
pulses (to the 2 coils) in order to have a pulsed contact. That simple fact does make necessary to limit
the circuit to a max of 3 relays (6 coils to operate). So, switching only the + power supply lead as in
following schematic.Circuit operation is smooth no problems verified even when power supply set
to 500VDC.
Note: The antenna wire only 3.5cm long. The whole unit does seem to radiate much lower than the
picoreed breadboard circuit. To remember is the fact that PS – lead is connected to home ground
by the pow er supply itself.
First test run using same picoreed settings in order to be able to spot eventual differences. In
particular I used a 5msec starting pulse (SW1) to charge the cap C, a 3msec inter-pulse gap and a
3msec SW2-3 pulses.
24
26. Kick Report
Figure 33 Command pulse timing
CH1
CH3
CH2
CH4
NO
R=1k
Power supply
150-500VDC
NO
C
C
SW2
SW1
3.5cm antenna
R=1k
L=80t Litz
C=2uF/630VL
5.6k
SW3
NO
C
CH5
All relays CLARE HGSM5181
Current
measure
CH6
Figure 34 Mercury wetted Relays switcher
Figure 35 Operative switcher (vertical board position)
25
Diff V
measure
27. Kick Report
Initial test
02/18/2013
For the initial test I have used the same settings for picoreed switcher version: Clock data pattern
generator = 1KHz that divided by the 100bit memory gives 1msec/bit and 100msec sequence = 10Hz
rep rate. The power supply initially set to 50VDC and timing as in fig. 33. Antenna & pickup coil as
shown on fig.35. While running the scope time base to 0.5sec I firstly note that there is still some AM
on output pulse height but with less modulation deepness: in this case perhaps only something like
10%. Pulse amplitude range: ±2Vmax.
Figure 36 AM on pulse (PS=50VDC)
Figure 37 Output pulse detail (PS=50VDC)
The measured current (on R load) is betw een +5mA / -1mA.
Rising the power supply voltage to +150VDC,
Figure 38 Output pulse detail for PS= 150VDC.
26
28. Kick Report
The output pulse amplitude rises to +5V and -5.8Vmax while current is ± 20mA like in following pic:
Figure 39 Current measure (2ma/mv)
Connecting the loopback diode, it roughly doubles.
Note: Disconnecting the 150V power supply + lead the current on load continues to be supplied for a
long time. Disconnecting also the PS – lead the delivered output current halves but then remains
constant for a long time.
Pancake coil Antenna
In this case I wanted to check flat pancake behavior and spherical antenna to help capturing local
environment energy.
From schematic it is possible to see that I also lowered the load resistor to 470Ohm and connected
also SW4 switch to enable the feedback and hence see the output difference.
It is important to say that in this design NOTHING IS IN RESONATION, antenna included, so the
net effect on load is in a sense not depending on frequency range used on switches SW1,2,3. Ok,
that’s the limit of this kind of implementation where max switching speed limited to a max of
200Hz. It must be said also that I generally tend to use it a 50Hz just to see in any modulation effect.
Well, there is modulation effect even if much lower than picoreed case in same conditions. Anyway
the self modulation appears to be only about 10% of the amplitude and it’s frequency appears to be
near random where probably 2 components: 1st long time about 10sec , 2nd short time near
500/100msec…
Central clock used for this test is 2Khz hence real clock applied to circuit is 2000/100= 20Hz. The 2
pancake coils, as possible to see in fig. 41, are completely overlapped, I tried also various gaps but
best results only if the two units near together.
27
29. Kick Report
SPHERICAL antenna
Diam=20cm
aluminum
CH1
CH3
CH2
CH4
NO
R=1k
C
NO
C
OD=10cm
ID=2cm
Overlapping=100%
SW2
SW1
Center lead
Power supply
150-500VDC
Outer lead
R=1k
L=25t
pancake
C=2uF/630VL
SW3
NO
C
470
L=25t
pancake
CH5
All relays CLARE HGSM5181
CH6
D1
Current Diff V
measure measure
SW4
Figure 40 pancake and spherical antenna version
Figure 41 Breadboard
The current measured on 470Ohm load is in Fig 42.
I measured +40mA and +160mA (PW=20ns). The no load voltage output is +100 and 150V…hence near to driving Power supply.
28
30. Kick Report
Figure 42 Current measured on 470Ohm load (2mA/mv vertical scale)
So, it is clear that using pancakes the output available power is roughly x4. The spherical antenna
contribution is also evident as it does increase and regularize the output pulses.
Sadly the output net energy is still too low hence for itself could be used only in SM scenario where
kicks are used only to initialize the process. It is not useable for self energy amplifying or generation.
I wait for any suggestion or observation... welcome.
The unit for itself is almost self sustaining due to floating design and no DC circuit closure. The
circuit does closure on itself via stray capacitance only...so using 2uF cap it does produce kicks for
hours...but it progressively go toward zero as normally expected.
Even if this approach is ‘aperiodic’ it is nonetheless interesting that using 2 equal pancake coils I've
been able to get output pulses having almost same input cap charging voltage: in my case 150V. I've
current as well...up to 160mA into 470Ohm load ...P= R*I^2 = 12W peak...the problem is that pulse is
very narrow about 20nsec hence the pulse area integral is near zero...in fact I've tried to charge a 1uF
cap and at 10Hz rep rate it takes about 6 seconds... Interesting is also that the circuit able to self run
for many hours...providing the kick output.
In the system I'm using, pulse amplitude multiplication ...seems not possible. A way around could
be to make interact two or more asynchronous Kickers. In that way it should be possible to have
random pulse summing...like described by SM. Still the difference remains that all is heavy limited
by the low repetition rate used. Best should be to go in 10-50KHz range…
29