TOP: Improved Page Navigation 2-Feb-2011
For one thing the Krohn-Hite 5100A VCO takes Zero to -17 VDC to drive the VCO output frequency. But the O-Scope X-axis wants 0 to +5.7 volts to have the X-axis represent increasing frequency to the right. So, the O-Scope frame had to float to get -Volts into the KH 5100A and use same signal to get +Volts into the O-scope. Too confusing for a quickie hookup. To facilitate easier setup, the above Spectrum Stalker box is under construction.
This box will create 0 to +5.7 VDC triangle and saw tooth voltage waveforms to drive the O-Scope X-axis. Internal Op-amps provide offset and gain to make the O-Scope X-axis Grid calibrated in frequency. In addition, the sweep range can be bracketed via HI & LO-LIMIT front panel knobs to display any portion of the frequency range. The KH Model 5100A Generator VCO has a response that may be in the 10's of Hertz range so too high a VCO slew rate will cause the resonance bloom to shift from the correct frequency (x-position) on the O-Scope CRT.
The Spectrum Stalker internal oscillator will sweep the O-Scope x-axis from a couple Hz to about 30 Hz. In the triangle mode, the resonance bloom will be seen with advancing and declining VCO drive volts. If the VCO output frequency is not tracking fast enough, the spectrum 'bloom' will be seen to shift left and right. The higher the scan rate, the more the shift. The HI-LIMIT and LO_LIMIT knobs will permit bracketing the 'resonance bloom' to increase repetitive scan rate while using slower VCO slew rate. This trick can minimize phase shift of the apparent resonance position for increasing versus decreasing VCO frequency. Saw-tooth waveform will also cause shift of the apparent spectrum, but just in one direction which makes a better picture. May have to do a O-Scope Z-axis blanking during sawtooth retrace.
Now the experimenter is free to move the TC secondary coil up and down, changing the degree of coupling between the primary and secondary, and compare the performance of the Tesla Coil versus coupling. The Cartoon shows (absolute value of) the general shape of the 'resonance bloom' that will occur for P-S coil pairs that are under-coupled to over coupled. The ideal coupling should be the 'C' trace, slightly over coupled to broaden the resonance bandwidth. This is fascinating stuff; trust me. The schematic is shown below. Notice the '05 date, shows how far behind this project is. After wire and debug, circuit updates will be posted.
BELOW: Spectrum Stalker Circuit card stuffed with chips plus component boards, ready for smoke test. Front panel wiring is last item before power up and debug.
The triple output (+8, -8 and -18 vdc) power supply for CMOS circuitry with jumper to the circuit board is wired and tested. The circuit card (with DIP wire markers) and front panel are ready for wiring. Updated schematic with chip and component U-numbers (U1-U8) shown below. Should have the circuit wired in a couple days, maybe a day for trouble shoot and test. Then it is coil coupling measurement time!
The Spectrum Stalker is working, day & 1/2 debug; couple minor circuit changes (later on that, Rev-C). The base of the coil is grounded. The left picture above shows 3 resonance blooms. The 1st bloom to the left is the fundamental ¼-wave 340 kHz resonance spike. The second (taller) spike is end-to-center resonance that places the HV peak near the center of the coil (and the coil ends are both voltage nodes). This test was performed with an un-tuned primary driver coil excited by the Krohn-Hite VCO output. The coils as pictured looks under coupled (sharp spike), but I am not sure about that as yet; untuned-tuned may not ever show the overcoupled sidebands.
I will investigate the behavior (bloom shape) for untuned-tuned at various coil coupling (just to see). We all may be learning some physicsstuff during upcoming resonance exercises and tests. Of course, the classical tuned-tuned over coupling side-bands will definitely show up when a primary capacitor is inserted into the primary circuit (been there done that with the kluge) if the P-S-coils are over-coupled.
The Krohn-Hite VCO frequency tracks the volts dc input (SS Output) pretty good, the phase shift I previously described is visible, but just barely at the highest full BW scan rate of 21.7 Hz Triangle and 43.8 Hz Sawtooth. The second scan picture shows the scan centered on the bloom using Lo-Limit and Hi-limit settings. Scan rate was greater than 40 Hz for this partial scan. The phase shift up-freq to down-freq is minimal. In addition, there is plan to make a better bloom detect sense coil (pict was taken with a 2-turn aligator clip lead wrapped around secondary) with a ½-wave 1N914 demodulator and small filter cap to produce a dc waveform that can be recorded with a data logger for Excel charts.
Next on the list; add primary capacitance to match primary to 340 kHz secondary and configure an easy way to move the secondary co-axially closer and further from the primary to find positions for under, critical, and over coupled. Take scope pictures of that result for now. Rectified 'bloom' and Data Acquition in the Que.
Going to try to make and link a video of scope resonance 'bloom' versus secondary immersion into the tuned Tesla primary coil. I have a Sony DCR-SX40 Cam that creates huge files per second of video that must be compressed to reduced You-Tube size format. I'M still on that learning curve.
I had to temporarily break the 'closed loop' Stalker circuit at Drive Selector Switch SW2 lead to the Hi/Lo Comparators to figure out why circuit was not working correctly; it can be difficult to determine where a problem is introduced in a looped circuit. The debug was done prior to the spectrum picts above. This section is catch up on details, FWIW.
For anyone interested, changes from Rev-B to Rev-C are 4 Items as follows; An omission, SW1 on 'SELECT Triangle' left CMOS input U1-6&12 floating in the breeze. U2-3 wants to be at Gnd for Triangle, but must be at Lo-Limit wiper potential for Sawtooth waveform else Lo-Limit does not work because U1 discharges 3uF to U2-3 potential, needed extra SW1 section (SPDT to DPDT). Had to bump U4-15 capitance to 0.02 uF to get full discharge of 4.3 uF Integrate capacitor, now using 2.15 uF. Added a 0.01 uF cap across U3-6,11 to eliminate a high impedance cross talk problem. I added a 1N4148 diode from U2-7 to U2-6 to keep U2-7 from ever driving the PNP base significantly above ground (would be reverse biased).
PS: On CMOS logic, all inputs must be tied to a controlled voltage, unused inputs drift to cause uncertain state (unlike TTL, not recommended, but unused TTL gates always float to about 1.4 vdc which is seen as a logic 'Hi'.
The Spectrum Stalker is working. But, And, we need a test strategy. To accurately display the character ('Q') of the Tesla Coil primary the VCO output wants to be weakly coupled to the primary. This is a relative impedance issue. The impedance of the existing primary coil at resonance should be at least 20:1 lower than the VCO drive resistor to prevent 'Q'-spoiling thus preserve the true shape of the resonance bloom. The VCO 50 Ohm output (way too low) Will be driving the primary through about 5k ohm additional resistance.
A sense coil is designed and being built that has toggle switch selection of two or ten turns. The coil output (Low Freq RF) will be fed into a battery powered black box Op-Amp / diode demodulator (TL082C). An output switch will place either the RF or Demod-RF (dc) on an output BNC to Feed an O-Scope and for Demod, an Agilent Data Acquisition Module.
Acquired data will permit the tuned-tuned coil(s) response at various coupling distances to be displayed on a single chart for clarity. A Chart will permit easy comparison and take much less space in html than multiple JPG pictures of the O-Scope screen for multiple coupling cases.
We need an easily adjusted coaxial primary-secondary coil arrangement to permit various coupling scenarios to be tested, acquired, and displayed.
Finally, it would be good to run numbers on mutual inductance of concentric coils to see if the empirical and theoretical can be aligned. So that is the 'why' of the recent apparent hiatus. Please stay 'tuned'.
PS: I think there may be an error in Frederick W. Grover ISBN: 0-87664-557-0 P-123 mutual inductance formula. I believe equation is using winding density (n1,n2) instead of coil turns (N1,N2) in the formula mutual inductance M=0.002((PI)2)a2(n1)(n2)[r1(B1)-r2(B2)-r3(B3)+r4(B4)]. n1=N1/width, turns/length, defined on P-122. A spread sheet is written that includes M computation and Table-29 data for interpolation. I'm not comfortable with the computational results as yet.
The circuit below (updated) is the planned Sense Coil with a selection of RF or DeModulated output to O-Scope, OR DeMod output to a data acquisition module. Decided to power it with two 7.5v TR175 for 15v. DeMod Output should be able to swing 0-6 vdc. One section of the wideband dual JFET Op-Amp is used as a voltage follower to provide roughly +12.5 vdc and -2.5 vdc 'rails' to the Op-Amp. The 2nd section is wired as an inverting amplifier with a voltage gain of 1/0.636 so that circuit the output will present the same peak volts as the RF envelope (Avg Output at U2-6 is reduced by 0.636 by the Low-Pass Filter formed by 10K and 0.01 uF capacitor). The 1N914 signal diode in the Op-Amp feedback eliminates diode forward voltage drop at the output. When 'RF' is selected, the Op-Amp is turned off and the coil input is presented directly to the OUTPUT BNC to permit viewing the RF resonance bloom on an O-Scope. Four sense coil taps offer an effective 1, 2, 5, and 10x signal strength change. DeMod Panel engraving a bit smudged, laquer paint stick not dry yet (hot off the press), clean up smudges later. Wiring should be quick, probably tomorrow.
The ½-wave circuit output seen below. A mistake; the average value of a sine waveform is 2/PI, but ½-wave averaged over a full cycle is 1/PI. So the Op-amp gain should have been PI instead of PI/2 to achieve an averaged output equal to the peak value of the RF waveform. Dislikes: ½-wave requires more filtering and the circuit output impedance is high (10k ohm range). So circuit re-do to an AVA (Absolute Value Amplifier, full wave rectified) is in store, I think there is enough real estate to squeeze it in the box. The TL082C bandwidth is just shy of satisfactory (see chart below) with roll off starting at about 200 kHz, was hoping for flat to 400 kHz. Have LM6172IN Op-amps on order; Slew rate 75 v/us vs 13 v/us and BW 160 mHz vs 4 mHz, but consumes about 4 mA per amp. LM6172 also swings closer to the rails, about ½ volt overhead vs about 2 volts for TL082. Since the AVA requires another dual Op-amp, the 4TH Op-amp will be used as a voltage follower to provide low output impedance from the demodulator. And I may install two gain trim pots to balance 1ST & 2ND halves of the FW and overall gain. We'll see.
To EE circuit junkies wondering why not full wave in 1st place?, your right, I messed up. The FW will require some gain games to get desired output volts; The psudo ground generated at U3-1 puts -Vcc at -2.5 vdc and +Vcc at +12.5 vdc; therefore U2-7 gain is set low to minimize negative excursion and gain is made back up with low value of U1-3-14 against U2-1 feedback pot. An RF signal input of 10 v P-P will drive U2-7 output to -1.7 vdc including the 1N914 drop at U1-6-11. Should work, may have to tweak.
Circuit card for Full Wave (AVA) Demodulator is wired and running. See circuit above. It employs a TLC279INE4 14 pin Quad Op-amp (found some in my chip supply) which in this application appears to be flat to over 700 kHz bandwidth, see the chart above. Demodulator dc volts out will swing in excess of 5 vdc with fresh batteries. Balance and final gain potentiometer adjustments made for easy ½-wave balance and ease setting final dc output to 'RF bloom' peak volts.
I had a bit of debugging, the TLC279IN Op-amp(s) wanted to self oscillate at about 1 mHz; by-passed +inputs and the final voltage follower stage to get the circuit working as shown; the 1 mHz oscillation propensity is probably the cause of the rise in response shown in the bandwidth chart above; perhaps a 1 kohm load to ground at the output will do a 'Q' quash to flatten the response in 1 mHz range.
BTW, no O-Scope output pictures included because there is not much to see on a O-scope with CW-RF input to demodulator, need to connect systen to drive the TC; With CW-RF, Demod output is a constant dc voltage equal to the peak volts of the RF input. Now ready to use Spectrum Stalker to drive the SGTC while displaying coil gauss with the Demodulator. First need a sense coil to provide Tesla Coil RF gauss output into Demodulator. Sense Coil coming soon.
The Sense Coil (1T, 2T, 5T, & 10T) will have a BNC output and a 12' BNC to BNC cable to connect coil to the demodulator. Following that, rig up SG Tesla coil primary on guides so that the Primary to Secondary coupling can be easily positionally adjusted to seek what I believe will be ideal coupling to optimize TC output while using the Spectrum Stalker and Demodulator to produce measurable results for that effort.
The above O-scope picts show the Spectrum Stalker driving the O-scope as a spectrum analyzer. The left most pict is of (ad-hoc 10T) sense coil output connected directly to the O-scope Y-input as transferred directly through the Demod box with switch in 'RF' position. The 2ND pict is with the demodulator Box switch turned to 'Demod' where the RF bloom is internally full wave rectified, then low pass filtered; this dc signal is suitable for data acquisition.
The 3rd pict (right) is with the spectrum Stalker driving the O-scope X-axis with triangle rather than sawtooth (picts 1 & 2). Two traces are seen in pict-3 due to VCO phaseshift resulting from the scan rate being set high (to get a more uniform beam intensity for the photograph). This dual effect is a manifestation of fast VCO sweep. At lower sweep rates (for Data Acq) the two peaks almost merge; I could provide a re-trace blanking output from the Spectrum Stalker to re-trace blank the O-Scope allowing the beam to get back home without showing re-trace ghosts. NOTE: Re-trace ghost can also be seen in 2ND (middle) pict to the left of the resonance peak. The 'ghost' is frequency broadened, amplitude reduced, and phase shifted due to extremely short fall time of the sawtooth waveform.
A 10T 6 inch diameter sense coil at 10T will have about 32.6 uH inductance. The 12' RG-174 interconnect cable has a measured capacitance of 494 puffs (pF). This puts the sense coil resonance (bad thing) at 1.25 mHz which is well above 400 kHz maximum TC test frequencies (good thing); therefore, the sense coil 'personality' should be well out of the way so as to not inject error (ugly 'personality') into the test scheme results. At 2T, resonance is at 6.25 mHz.
Also note that the demodulator input is a current input rather than a high impedance voltage input. This affords circuit over voltage protection. A small downside is that the input consumes (small) energy which will slightly lower the apparent 'Q' of the Tesla coils being probed. This should be negligible as the 12 kohm imput Z is high compared to the Tesla primary and secondary (several hundred ohms). Of course, any lowering affect on 'Q' just slightly reduces peak amplitude but does not affect resonant frequencies being measured.