piezo ceramic project

[hoola]

If anyone has had experience in dealing with microwave suppression in a system similar to mine, please advise of what sources you had for both suppression and reflection materials and tips on how to apply them. It would be funny if what the Shawyer engine uses for thrust,  proves to be one of the Woodward engine's  waste products.

[hoola]

does anyone know if piezos are capable of direct microwave transmission? I can't find anything online regarding this, but I was hoping for some comfirmation  of that, which would explain what happened in the first high output test with the melted chocolate last Monday. I knew early on that some RF came out of the area near the stack, having occasionally placed an AM radio near and picking up hashy noise peaking at about the mid dial area  (1mhz) and trailing off at the high end. I didn't think any more about it till now, as the distance was short, having to set the radio within a few inches and just guessing the stack was the source, didn't seem important at the time if it was the stack or something else, harmonics from the audio generator, perhaps, but very unlikely. At this point I was driving the stack at only 25khz, barely above audio. Perhaps the microwaves, if they indeed exist, came from the lead wires to the stack, and only generated within the stack, but having no way to get out physically, till they hit the input lead wires, which turn also into transmission antennas. I guess the thing to do now is to go back to low power experiments specifically to try to determine the source of any RF that is generated.  Reducing the supply back down to 200V should limit the radiant energy to a few milliwatts of whatever is coming out of it. Hopefully, shielding the input wires to chassis ground will solve the issue, and be a quick, easy fix.

[hoola]

I am thinking about enhancing the NFB circuits by using the unused 7014 output ports to "chase after" each piezo as they are serially stimulated.  At this point, the  five NFB circuits are all turned on at once, at the end of the fifth piezo stimulation, during retrace. If I were to add on 4 ancillary NFB trunk lines, one for each NFB circuits of the first four piezos, to be activated individually just after each  of these first four piezos gets stimulated, the NFB circuits would "chase after" each pulse, quieting it's aftermath on the fly.  When the fifth piezo is done stimulation and the normal collective triggering of all of the NFB circuit occurs during retrace , there should be much less noise in the system in general for it to have to deal with, most of it having be negated during the course of the sweep, not waiting till after the last of the scans occur with all the chaotic echoing that has accumulated in the stack.  This could lower the counter thrust, raising efficiency and reducing noise, which  could help stabilize the magic waveform after I go back to high voltage tests. I am still waiting on a microwave detector, that's on order and have begun accumulating microwave doors from salvage to take out the screens to overlay the confinment box in a sort of Faraday cage.  I have also added a 36pf cap in series with a 27K resistor across each piezo directly at the connections, and placed two ferrite beads on each input lead for added suppression. The RC combo acting as a low pass filter, absorbing frequencies much above 200khz. Current low power tests  altered to accomodate this new arrangement are going well, as the overall frequency and duty cycles are only slightly affected. Still running at 188khz.

Edited June 4, 2020 by hoola

[hoola]

I have found the "chase" function to help with stabilizing certain waveforms, but still figuring out the pattern, mostly turn off that function until a useable waveform established, then switch them on separately and see which one or combinations help or detract. I got the microwave leak detector and have found no emissions, but having the suppressors across each element, no way to tell without removing them to see if microwaves ever did develop, which I may do at some point when I tear down the stack and inspect it.  I have been having other issues with any supply going much over 300 volts in that I am achieving a much higher pulse or sets of pulses, but the noise is also much harder to control, and probably swamps any thrust signal should it show up and the waveform collapses too quickly to try a scale balance test. I have gone back to 300  volts as I am getting a weight gain of 2 to 3 milligrams reading on the scale upon activation with reasonably stable waveforms at that voltage. I have a "jeopardy" button attached to the wall, then swung into the chassis for physical isolation. This button turns the NFB circuits on and off. When off, the lack of NFB should allow the pulse echos to nuetralize and eliminate any possible directional thrust. I have noted that on occasion there is a "tick" when I turn off the NFB accompanied by an additional increase in observed weight above the 2-3 mg level by another 2-3 mgs.. I think this is due to the pulse pattern just completed upon shutoff, hence no additional noise from the next scan period, thereby extra thrust. This could be how the Woodward team is getting thrust, as they say it largely occurs upon the end of the stimulation periods. (They report having blown up stacks with running them to the point of fracture.) The stack is mounted on the short end of the level and is pulsing order down, so the observed thrust is in the correct direction as the stack should be getting lighter, allowing the long end contacting the scale to register the increase. I have found that the wires from the chassis to the stack are a main source of unstable readings, plus the pivot point on the level arm seems an issue also. Upon setting up the scale, up to several minutes of readings bouncing around of plus or minus 5 millilgrams are noted. I have found that placing a speaker on the underside of the bench directly below the scale/lever assy helps reduce this issue. The speaker is driven by an FM radio on an empty channel, giving white noise. This seems to vibrate the entire assembly out of a tendency to stick to any particular spot, and now the instabilities are lowered to plus or minus two milligrams,  and settle down to a stasis point more quickly. The next step is to mount the stack upside down, which should make the stack heavier, giving a scale indication of a a lower value.

Edited June 15, 2020 by hoola

[hoola]

Upon upgrading the scale assy, I have found that the lever arm was oscillating at a very low frequency, in the order of seconds, and that my observed thrust readings were due to riding that resonace up and getting appartent thrust, then getting reverse readings, followed by no readings. This explains the two months of false readings that occured in my normal testings. However, the rare anomolus "click" upon shutting down the nfb system, say one out of a thousand tries, which always read 2-3mg weight gain, is the only potential lead I have to proceed with. This only happened a few times, and while no possibility seems for continuous thrust, perhaps a pulsed on/off control can be added that can allow the system to build up thrust over a period of ten or so cycles, then at the right time of shutting off the scan at the appropriate piezo, the "tick" sound can be recreated and see if it coorelates to a  release of stored energy that will give the weight increase observed previously. There is a "reset" control on pin 15,  currently unused,  that will halt the count, then restart at output 1 again (labeled as output 0). I will draw up a  circuit to interrupt the scans at regular, controlled intervals, and make the scan end point selectible on which piezo is the last scanned. I have observed that the scan pulses are sometimes higher when some of the piezos are turned off, depending on drive frequency, and PWM settings, so it seems possible that the max stored thrust may be available before the last piezo it driven. I have shifted over to approx 44khz as clock frequency from the previous 188khz. If anyone golfs, the click sound I am refering to is the sound of a perfectly hit long shot on the driving range which doesn't need to be hit very hard to get great range.

Edited June 23, 2020 by hoola

[hoola]

The reset based circuit has created a pulsed scan schedule as hoped, and I am getting 2-3mg weight gain on the scale when activated with the jeopardy button, and when released, the weight quickly dissapears back down to the starting indication.  Testing over a period of 2 hours has has shown a reliable force seems to be developed. The jeopardy button is no longer controlling the nfb systems, which have been put permenantly on, it now contols activating the reset pulse contol system. I have to use the 800v supply to get these readings, and have had to replace the output transistor load resistors as they were getting too hot with this new arrangement and have replaced them with slighlly higher resistances and doubled the wattage. They still run hot, as well as the output transistors themselves, so tests are limited to 20 mins or so with a 20 min break for things to cool off as the unit is in a closed off area to prevent air current disturbances. A strange feature has developed along with this change in that touching or even getting near the chassis, jeopardy cord or button causes wild swings in the scope traces and just getting anything near the stack casuses max reaction of the odd behaviour.  Placing the microwave leak detector near (2") from the stack  causes this behavior to manifest most . The "click" sound is now a steady hash of clicks, and perhsps future tests will confim my initial findings

[hoola]

Further tests have shown similar weight gains occur with supply voltages in the 375-400v range, no changes noted with 300v and below.

[hoola]

The issue with the jeopardy button and overall proximity sensitivity has been resolved. The pulse from the last piezo and it's delivery to the reset pin must necessarily go thru the button, which the inductance and self capacity of the cabling affects the waveform drastically as it is moved or held. A relay needs to be installed to isolate the system from the button. I have temporarily removed the button from the circuit and simply push the needed connection together with an insulated probe and get the desired weight gain without the previously mentioned artifacts and the stability of the readings are further increased. Tests after this change are showing weight gains of 2-3mg with as little as 250v supply. Increasing the supply back up to 500v has provided up to 10mg of weight gain when activated, but that is an outlier with the desired waveform still rather unstable over long periods as the stack warms up,  and not easily reproducible, so that represents an exaggerated reading that may be readily attainable with further improvements. Oddly, the nfb circuits have little effect upon the new pulsed waveform and turning them  up, down or off seems to has no apparent effect on readings and very little effect upon scope waveforms, so the nfb circuts may not be needed in a final product. I am finding that turning off piezos 2 and 4, and just running 1,3,5 seems to deliver the cleanest trace and highest weight gains, at least as the machine is hooked up at present. Still around 44khz but am finding good results up to 52khz clock freq. The spec sheet shows the piezos to have an individual resonance of 25khz, so finding good results in the range of twice that seems logical.

Edited June 26, 2020 by hoola

[hoola]

The relay is in and the anomolous sensitivity issue is gone with the jeopardy button  hooked back up.

[hoola]

An ususual weight gain is noticed even on low voltage  if I sweep the freq generators level output. I call this the PWM control as this drives the 7014 chip, which in turn drives the first stage of the five individual piezo amp legs, which determines the duty cycle of those respective stages. This creates a weight increase,  just by sweeping manually with the external level pot.  When I switch the drive wires around and should be seeing a weight decrease, and it increases also. (Sweeping the freq conrol in a similair fashion has no effect on weight.)  I am in the process of moving the tests to a warehouse with a slab concrete floor, and has widowless small rooms for isloation, and to falsify these readings if indeed they are  errors of proceedures and scale instabilities.  I am not sure even that a weight gain in the scale reading cooresponds to a thrust, I presume it does. How else it could be accounted for? I apologize for not uploading pics of the waveforms and apparatus, my chromebook doesn't recognize my phone, but am working on getting that done in the next week or so.

Edited June 27, 2020 by hoola

[hoola]

Today I have found that the trigger function of the wavetek 182A can function as a pulse control, and does so better than the new circuit I just added to do so. So now, that system is turned off, as well as the nfb circuits, both of which seem uneeded at this point. The nfb circut does make some minor changes, but there seems to be some natural phase cancellation of the unwanted reflected pulse going on at specific frequencies of clock freq and trigger speed may make those circuts  unecessary for now.  I have found that a 110khz clock freq along with a 70hz pulse rate  from a standard audio oscillator set to square wave triggering the wavetek is a sweet spot that gives a 10mg weight change to the scale, and when the drive is removed, the weight change dissapears, indicating a potential thrust that is now well above the noise level of the scale and limitations within the system. A possible error may be happening if RFI is affecting the scale directly from the stack which is quite noisey, but it is about 10" away from the stack at the opposite end of the lever. My new setup in the warehouse has two small rooms adjacent to each other, and the driver unit, generatoris, and all attendant power supplies will be in one room and will feed drive to the stack and scale in the other room through a small opening, which will have RFI shielding on the wall surfaces. A camera will view the scale, so the stack room will be the only thing in it. I will install shielding on the scale itself to try to minimize any errors there. 

Edited June 28, 2020 by hoola

[hu??]

I am impressed by the effort you put into this. Especially the way you drive the piezo stack can make a huge difference to the simplistic way Woodward is doing this.

 

However, having followed Woodward's efforts for about 25 years (yes really!), i am also a bit skeptical about the results: Woodward manages to only show about 5 to 20 micro newton, and even that is highly debatable because of measurement problems. Vibrations, temperature differences, electric and magnetic fields, etc can very easy give false signals.

So i would advice to keep on the good work but also to be very skeptical to any 'thrust' signal and try everything to debunk your test-setup.

 

[hoola]

^{thanks for the good input hu??....yes, I have pretty much determined that the thrust measured so far is from the audio standing waves causing a column of air pressure below the stack, as I re oriented the stack to be well above the platform and the thrust disappears, but with some odd aspects that need to be resolved later, but for now,  I presume some actual thrust is there  but near the same level with woodwards team, down in the micronewtons. I have had to air cool the output stages running 1KV into the supply to the output transistors. Pushing the piezos too hard causes excessive noise in the system,  overheats the machine, making all relevant issues more difficult. My plan now is to lower the voltage back down to 300V.  The machine runsr cool at that level, and can easily  deliver a stable signal of my thrust candidate. i will next build a five parallel stage linear to be driven by the equipment with this lowered supply voltage. This will allow the linear to do the  heavy lifting to drive a second stack  of identical piezos which will now be the proposed thrust element in the system, the existing stack to be used as five separate signal sources.  I have looked up transistors that can handle high voltages and currents and have found nothing so far, so will proceed with 6JE6 horizontal output tubes which can handle plate voltages up to 5KV.  The 6BK4 was used as a HV reg in early color tvs, and operated as a triode with a normal plate voltage of 25 KV, so if the "piezos diriving piezos" concept proves workable,  and no thrust seen in 6JE6 drive mode, higher voltages will be tried with 6BK4s until the breakdown voltage of piezos is reached.}

Edited July 13, 2020 by hoola

[hoola]

It has occured to me that there could be some advantage to grinding off the normal electrical contact surfaces from the faces of the piezos, and moving them to the inner and outer ring surfaces, which are normally electrical insulators. This would allow a central rod attachment for grounding the stack, and move the drive signals to each exterior ring surface. The advantages would be no intervening poly insulators or contact materials within the stack, enhancing shock transfer between elements. The physical distances between contacts would be doubled from approx 5mm to 10mm, perhaps offering a higher breakdown voltage, and an overall simplified assembly and shorter stack length. The central ring contact could be the full width of the body, as all are contacted electrically anyway, but the outside rings would need some gaps between them, so would necessarily be somewhat narrower, to prevent mixing of drive voltages between the edge contacts. The voltage leakage  between the edge contacts would be an issue to deal with, and the main problem which I see with this idea.

Edited July 22, 2020 by hoola

[hoola]

A re-introduction of a thinner poly washer between piezos would solve the leakeage problem with a minimum of shock transference loss. The poly washers would only have to be thick enough to reduce leakage to a tolerable level. thus allowing a full coverage of the outer ring surface, which should be to some advantage in electrical energy transfers. I will  grind off the normal contact material of one and apply aluminum paint to the inner and outer ring surfaces to see if this idea is feasible.

[hoola]

I have not removed the normal conductors as there is a small boundary between the edges of these layers and the outer and inner ring areas. I plan on leaving the normal contacts and adding the new inner/outer edge contacts since I can keep them electrically separated about 1/16th" With low voltage experiments this should offer sufficient isolation. Before modifying the original contacts, it seems advisable to see how the piezos respond with varying the placement of the stimulation voltages, between the regular contacts vs. the new ones, and then stimulate both sets concurrently, and then switch phases  between the two sets to see how the piezos will behave in these novel situations.   

Edited August 1, 2020 by hoola

[hoola]

I received the conductive pen and am finding the conductivity is rather low on the Chemtronics CW2000 nickle based pen. It might work well enough to try to see if the alternate contacts can be done here, but I suspect the paint will not properly bond atomically to the ceramic, and will offer a weaker response or perhaps a mere capacitive transfer of energy. I have attempted to register with steminc, the manufacturer, with no good results. Somehow it doesn't recognize my inputs, but have sent two emails  through their website as an unregistered guest to request a small custom lot of the alternate contacts I want with no response as of yet. If anyone has had experience with conductive pens as to particular quality and ease of use, and has a particular best choice, please let me know. I tried to add multiple layers, with poor results. After the first layer dries, a second layer added has the rather harsh solvents that accompany it, dissolving the first layer, making a clumpy mess of nickle flakes that fall off, and not giving me a good result of a smooth low resistance contact. I will try a phone call this coming week and see if something can be arranged with  factory applied alternate contacts for a small run. I see they offer a similar ring piezo with no contacts, which might work better on trying to add my own if no alternative is available from the factory. If no alternative is available, the outside of each surface will be crimped and soldered with a narrow copper thin sheet for individulal  inputs and the interior of the ring coated also, with a slightly oversized steel spring expanding to maintain outward tension on the interior, coupling all the stack elements with a common ground.

Edited October 11, 2020 by hoola

[hoola]

Steminc, the manufacturer of the piezo materials has agreed in principle to make the altered contacts, but would need a minimum  order of 300 pieces, which is beyond my resources at this time, so I will proceed with altering the contacts as described.

[hoola]

today new youtube videos are up that claim a low power/ high thrust including videos of traces accompanying observable movement of the latest version of the the woodward device. They are claiming thrust up to the millinewton range.  These are from SSI APW 2020 zoom meetings just completed of several speaker with many interesting side comments. I have found that the conductive pen i have tried does work, but requires an entire pen to do just the outside perifery, so will get some aluminum or silver paint and try to make the contact change economically feasible.

[hoola]

the nickle paint I ordered off amazon is much more useful as the amount of 12ml is sufficient for the job of coating new contacts on all 7 piezos. The paint seems to be of the same consistency of the pen, which is rather clumpy and dries flaky, but does provide for low ohm contacts. I have finished one piezo and it does respond to an input across the new contacts. A 10V  p-p  1KHZ sine wave across them is nearly as loud as to attaching to the original tab faces.

[hoola]

Am having some trouble getting the paint to stick, but they should all be ready in a day or so with experimenting on how to cure them.  One mod I have made to the existing stack is to parallel the first buffer piezo  (normally disconnected) to the first active piezo's input  This couples it's pulse in tandem,  offering a larger initial pulse, creating a slightly larger overall trace with no observable increase in noise. The last piezo is the end pulse "turnaround element" buffer, and as such can only be used as the scope signal source, at least as the stack is currently configured.

[hoola]

Three of the piezos have been more difficult than the other four to get the paint to stick well to the inner surface. I cleaned and repainted this three and used a normal hair dryer set on high heat, low air to attempt to get them to be bonded well.  As I was holding the dryer I began to notice small sparks emanating from the piezos. At first I thought the dryer element was arcing slightly, but quickly saw that the sparks were from the outer edges of all 3, which have had their new contacts on for several days and seem sturdy,  sparking over to the original contacts, on the faces. This went on for a minute or so, of these tiny white sparks. I stopped the blow dryer for a few minutes and then resumed the same settings blowing onto the piezos and saw no further sparks. The new outer contacts had been kept about 1/32" gapped to the original ones. This is where the sparks occurred, at random points of this gap junction. Upon checking the piezos with an ohmmeter, the new, still uncured inner contacts were quite shorted to the orginal contacts, as the painting process is rather messy, and the intervening edges much be cleaned up to provide the small gap after full curing of 24 hrs. So, it appears that the only gaps in the piezos that weren't shorted out were between the outer new  contacts to the originals. Perhaps some chemical reaction of the solvent being rapidly dried has an unusual electrical side effect.

[hoola]

I have found that the original face contacts can be easily removed with a fine grit abrasive paper. The removal process reveals that the piezo faces were not quite flat, so the removal process results with flatter surfaces,  which should increase shock transference.

[hoola]

Initial hookup of new stack gives larger voltage scans of very different appearances  from previous stack and responds to generally higher frequencies of scan rates. 

[hoola]

I discovered a wiring error that went undetected until hooking up the new stack yesterday. The  firing order for piezo 3 and 4 had been reversed when I added a quick disconnect plug between the stack and the driver chassis, when I moved out to the warehouse last spring. This explains my negative results all summer. Upon correcting the error, the new stack still didn't seem to provide a trace that seemed interesting, so I hooked the original stack back in and now am getting weight gains or loss of up to 8 milligrams on a consistent basis. The maximum weight change is an increase of 8 mg occuring when turning off piezos 1 and 5. At that point I was driving the stack at 88KHZ with a trigger frequency of 31 KHZ. Small variations in drive vs. trigger freqs determine whether the stack becomes lighter or heavier, although the 88KHZ main drive frequency seems the most stable trace, which makes sense since the resonant frequency is  listed at 44KHZ, Oddy, when I purchased the piezos,, I was sure they were listed at 25KHZ resonant. There is a resonance at the frequency, which may indicate that the overall resonance may have more than one peak. I still suspect that the acoustic pressure is a possible source of the readings. I am in the process of attaining a vaccuum chamber to eliminate that possiblity as it is rather loud, but some of the higher readings seem independent of the volume of the tone. There is a "singing sound" that coincides with the 8 mg reading that is rather pleasant, unlike the harsh normal tone that usually occurs when the stack is giving lower weight changes in the 3-4 mg range.

Edited November 18, 2020 by hoola