piezo ceramic project

[hoola]

correction...the aforementioned weight gain of 8 mg was achieved at  88KHZ and 3.1 KHZ.

[hoola]

upon considering why the new stack doesn't supply any useful looking traces, it seems possible that the piezos need to act freely, to a limited extent, and the problem is not due to the change of the areas of connection. So, I am going to place the same poly washer thicknesses into them to try to replicate the original stack spacing and torque, while leaving the altered connections, to see if the traces will then match up closer to the original.'s traces. I am also going to hook up both stacks in parallel to see how they interact. There is a lot of broad resonances going on between the individual piezos, it will interesting to see how they would all interact as a driven pair. 

[hoola]

it seems possible that any thrust measurements seen are due to a random sync between the main generator and the trigger drive generator. The trigger generator normally starts the scan process at any random piezo, so limiting any chance of seeing positive results. It seems as though, unless the scans are reliably started at the only the first piezo,  any shock wave will be much attenuated.  I have begun to use the last output, normally unused, to trigger the enable pin, which temporarily halts the scan, and with appropriate delay as provided by an R/C link between the two pins,  allows the scan to start again from the first piezo. At this point I have a 30K variable resistor in series with a .0082 cap and the traces now mimic what I was getting with the previous setup, With success of this arrangement,  the separate trigger generator will be unnecessary and the scanning sequence well regulated. During last Tuesday's positive results, I could hear a new sound that was like a theremin being played through a "drip echo" reverb unit used by surf rock guitarists. After a half hour of getting thrust readings, I shut of the machine for 15 minutes, and tried a restart. No thrust reading have been seen since then, and none of that new sound has been heard.

 

 

Edited November 23, 2020 by hoola

[hoola]

It does seem possible that a sync-up between the generators may have caused a  periodic frequency change in the main driver, causing a slight  upsweep of the signal during each scan from the basic 88KHZ. It seems worth considering to deliberately mimic this idea. The speed of the shock wave should remain steady regardless of frequency driven, and if that shock wave is overcome during mid cycle by a frequency sweep that desciribes a higher wave speed than physically allowable within the stack,  an interference between the two may offer some interesting behaviors as the signal speed "drives through" the physical speed.

Edited November 29, 2020 by hoola

[hoola]

I hooked up the alternate stack after adding the thin poly washers and found no seemingly useful traces, mostly sawtooth and sawtooth variations and saw no changes in weights of any tests so far with it. What traces that did appear were much more rich in overtones and complexities, so the idea of clamping them together directly does seem to be incorrect, at least with the current configuration. There are a few novel traces that seem to have what looks like hysteresis loops imbedded in them, rounded vacancies within some of the larger sawtooth variety. I will switch back to the original stack and drive it with the B&K 3030 sweep generator due here on Monday.

[hoola]

the new generator arrived early,  and seems to be fully functional, however with the usual dirty pots and switches. The limitation on the sweep frequency duration of only going up to the millisecond scale, is an issue and I would prefer up to microsecond. The decade switches that control that are on a separate small ckt. board and seems pretty straightforward in it's layout, so I can easily raise the upper limit on that scale,  but probably not to the microsecond level. At any rate, many novel traces are now possible with it as it stands,  and a few seem interesting.  I have had numerous weight changes, but nothing that repeats more than twice per test, which I discount as scale error. 

Edited December 12, 2020 by hoola

[hoola]

 

today I hooked the extra power supply back up and ran the B+ supply up to approx 600V and noticed an increase in minor random weight changes, which is normal with the higher voltages expressed in higher vibrations of the stack. Upon activating the time delay/enable ckt. I noticed an immediate large change in expressed weight. At the delay/enable control at it's maximum setting (largest delay between scans)  I got maximum change to the indicated weights, depending on how the B&K internal trigger control was positioned. I have read weight changes of up to 60 milligrams, repeatable, and seemingly real. I can provide  positive or negative thrust reading depending on the rotation of the trigger control position. In this situation, the higher readings were at full clockwise, the lower readings were at center, with no further changes as you went fully counter clockwise. After about an hour of this observed behavior,  I turned off the sweep portion, and went straight sine wave, and the reading protocol abruptly changed. With that setting, the weight was highest at trigger center position, and dropped  to it's lowest quickly at an abrupt change in tone of the stack as I rotated it slightly clockwise.  Under sweep mode, the changes were smooth and varied with setting as you turned it. Under sine wave, the abruptness was very pronounced. A noticeable problem occurs with turning the generator off periodically to see how the readings were changing to give me a "rest" weight, baseline. The weights do change, either up or down in appropriate fashion, but not to the degree I would have anticipated. 

Edited December 14, 2020 by hoola

[hoola]

to falsify my readings I removed the 3" hard rubber tubular interface between the scale and the stack. I left the scale and stack in the similar positions as when operating, just no physical contact was between them. Ranging the trigger control as previously described, I observed no change in the scale from 0000,  eliminating the issue of a direct dump of RFI into the scale and the lead wires I attached to it (a small pair of wires were added to provide a way to turn on and off the scale remotely, without touching the scale itself). Next, with the rubber interface still missing, I balanced the stack and it's counterweight as close as possible to  neutral, and again ranged the trigger. I saw no apparent movement, which indicates the readings have limited validity, as the observed weight changes should have been sufficient to tip the balance. This was a crude first attempt at a balance test, and I will upgrade and return to that tomorrow after continuing scale tests.  The apparatus is not properly designed for such a test, only for simple weight change detection. I normally have a 5mg overage transmitted down to the scale, which is the tare weight I have been using. There still persists the possibility that acoustic standing waves are causing the weight changes, as the sound is quite loud and I have begun to search for a vacuum chamber. I do have a large fish tank, and will be purging it with helium to see if the lighter atmosphere reduces or eliminates the readings.

Edited December 15, 2020 by hoola

[hoola]

a followup on testing today checks various on/off settings on piezo drives. I got the first 8mg weight gains last month with using only piezos 2,3, 4. I normally use that since it seemed to have some good early result, and turning on the 1 and 5 piezos seemed to eliminate the observed readings. I use all piezos now, but any three adjacent ones are necessary to get results the max. results. I turned on various single and combinations of all piezos and the results :  p2, no response, p3, no response, p4, no response. p1 and p2, some lesser response, p1 and p3, no response. p2and p3, some lesser response.  p2 and p3, andp4, max response.   p1, no response,   p5, no response,   p1and p3 and p5 no response. p1 and p2 and p3, max response. p3 and p4 and p5, max response.  Sometimes using all piezos seems to increase reading levels, sometimes not. When leaving the trigger at a stationary point, the readings change, and not just at background, levels, but in jumps of up to 15-20mgs that occur in fairly regular intervals of about 3 seconds. Still puzzled that the scale does not return to 0000 when shutting off drive to stack. There is a scale issue somewhere.  It was a $30 jewelry scale from amazon, and has been rather mistreated with bumps and  accidental overweights that occasionaly occur with months of use, plus I disassembled  it to add the remote on/off wires.  I have a container of chocolate near the stack and a microwave detector that picks up nothing. No melted chocolate so far. Overall, no responses were found since last month's 8mg anomaly (which was at the 300V level), until I raised the B+ from 300V to 600V with the new generator. I  briefly went back to 300V, and got no reading in any combination of piezos turned on. Still listening for the "surf music" and have heard nothing too similar since the 8mg anomaly. I do have an AM radio nearby tuned to a blank region, and the tonality amplifies what I hear coming from the stack, and reinforces the observation that the readings jump either positive or negative with a distinct change, or "pop" sound, of which some are like the "perfect golf swing click" as previously reported.

Edited December 15, 2020 by hoola

[hoola]

I think the load cell in the scale is wearing out and sticking, which would explain the non 0000 readings instead of returning to baseline. The scale is not designed, I would think, to absorb any excessive vibrations. It seems possiblle that  a new one would be fairly accurate for a week or so of tests, then begin to stick. So, as they are so cheap anyway, I will try one more of this same model to see if it returns to 0000 baseline out of the box using these same settings.

Edited December 16, 2020 by hoola

[hoola]

today I found that the rubber tube, which is basically an oversized weather boot for a coaxial antenna cable, which has a hollow interior, was the source of the excessive readings. I filled the boot interior with toilet paper to absorb the resonances and now I get controllable readings back down to the 2-3 mg range. The boot is of stiff rubber, tapered towards a flat bottom.  about 3" long.  I cut 1/4" off the top and cemented onto the top a small steel dome of 1/4" diameter to give the bottom of the stack a solid contact at the central point of the dome, which transfers the energy down to the scale. I have tried several interfaces between the stack and the scale and the compliance of the rubber material I thought would reduce transfers of vibrations to the scale, but I failed to consider the resonant interior space enhancing those transfers.

[hoola]

 

 

Today I redesigned the internal power supply to provide +500V to the final output transistos, alleviating the need for the external power supply for the time being. I also removed the negative feedback sub assembly as it has proven to have a negligible positive effect under the recent series of  design alterations. Recent waveforms that give weight changes show an asymmetry with  truncated tops that give weight increases, or truncated bottoms, giving weight reductions. These changes generally fall in the 2-3 mg range. I have also found that the sweep function of the generator is not necessary for weight changes, and sometimes reduces such changes, so tend to leave it off.

 

 

 

 

 

 

 

 

 

 

v

Edited December 25, 2020 by hoola

[hoola]

The redesigned power supply has proven inadequate, so have added a separate transformer driven power supply onto the main chassis just for the output transistors that supplies +600V.  I also have changed the method of the scale setup. The stack is now hanging free and the scale is positioned under the counterweight end, with a thin cotton thread hanging down to the scale with a small counter weight sitting on the scale itself, keeping the thread in a small tension of approx 6 mg. This causes the vibrations to have to travel across the mass of the rocker arm, and then down the thread to the scale, so is much more resistant to false readings. I am still getting anomalous readings in the 2-3 mg range, and these readings have been largely in accordance with what the scope traces  predict they should be, but still could be a product of acoustic standing waves which naturally would be in accord with any "real" thrust derived weight changes, so will require the vacuum chamber which I am accumulating parts for now. 

[hoola]

Completely containing the stack within acoustic dampening material may serve to falsify the anomalous readings without the vacuum chamber, and is cheap and easy to do. Attaining a second oscilloscope would allow simeltaneous monitoring of  both ends of the stack. Now, I have the end passives matrixed together with 1,2K resistors, serving as a single scope signal source. The ability to see waveforms  at both ends of the stack will allow real time adjustments to mimimize the shock energy at one end, while maximizing it at the other.

Edited December 27, 2020 by hoola

[hoola]

today I put in a DPDT switch to change the order of stimulations to go from bottom to top, instead of top to bottom. The weight changes are generally increased with the switch in "reverse" than forward, and do not reverse the direction of the weight changes.  With the new scale setup, a upward thrust is shown as weight increase in both positions.  120Khz is the new frequency of interest. Vertically assymetric traces are the most effective, which are tall "eiffel tower" looking structures. There are many fascinating traces, and everyday new ones appear that I have photographed.  Some of the more interesting ones look like iris flowers, hexagonal snowflakes, jellyfish, manta rays, dragons, the continent of India, hearts, pyramids, perfect circles, space invader aliens, and many other ones that briefly appear, but are difficult to stabilize long enough to take a pic. They appear as a string of repeating images, quasi-fractal in nature, some exhibiting  much fine detail. I apologize for not uploading them but my chromebook won't support the function of uploading from my phone, but I will persue the matter more fully and include schematics of the main chassis and stack pics later on.  Instead of wrapping the stack in sound absorbing material, I will built a hard containment box of 1/4" phenolic with a foam rubber liner on the inside, and the top of the box affixed to the balance arm. At 120Khz, ultrasonics would likely  escape a soft absorber, so a hard containment seems necessary. 

[hoola]

The hard container idea does not work to reduce the sound level, at least with using 1/4" plywood built as a prototype to get the dimensions correct. It seems louder in the container, but probably on seems so, as the lower frequencies are likely much more suppressed, leaving only the shrillest more obvious.  Upon looking into the acoustics issue, it seems that if there is no adjacent surface to reflect to, no standing waves will be developed,  so for the time being I am going to use an uneven hard surface covered in foam rubber below the stack as it is only 4" above the breadboard mount and the only close flat surface. I have considered building a similar size box of  1/4" lexan as a mini vacuum chamber just for the stack. I probably couldn't pull down a full vacuum, but might lower it enough after flooding it with helium to cause a significant drop in sound transference.  I have to wear shooter earmuffs since installing the new power supply, and I want to hear the lower frequency audio clues which I can't  now, so will port some of the scope signal to a low pass equalizer and amp it to my headphones. Tests done in the box still indicate 2-3mg of weight change, but seem slower to develop, taking approx a second, vs. about a half second previously, which could be caused by the increased mass of the assembly with the addition of the box and heavier counter weight. I suppose some of the weight of the box issue would be resolved with a low pressure helium atmosphere inside of it.

[hoola]

today I hooked up the new scale and it mirrors what the original scale does. I re-adjusted the torque on the stack and found a setting that has slightly less foot-pounds and gives a smaller trace, but gives a 5-10 mg of weight increase, translating into a positive thrust given the new scale setup.  This is occurring  when all piezos are on except for #5 and reverse order scan, at 122Khz. There is a new sound that seems to be a stable set of tones that are a perfect "A" on the guitar. It has a peak at the second fret of the G string. The tone is somewhat covered up by the high shrill sounds, but these shrill sounds seem less evident when in this  torque setting as compared to previous. The new traces are shorter and squatter, resembling short fat christmas trees instead of the Eiffel towers. I took movies of the apparatus today with real time displays, using the changes in tone to cue the watcher when the generator is on or off. I have found a way to stabilize the scale recently that involves turning up the level of the generator slowly from zero, just to be giving the 7014 chip barely enough voltage to operate. This causes a chaotic scope trace and a random vibration to appear thoughout the stack, and so "shakes" the entire assembly, which speeds the settling of scale readings to tare level. This simple trick has sped up tests, and has eliminated readings that once seemed valid. After this initial shake step of 5 seconds or so, I increase the generator output to the normal drive level to take readings.  The generator is left on for 15 seconds to ascertain  consistency of readings. After a reading is taken, the generator is set to zero, and the scale will lower, but shaking the scale as a final step is usually needed to get the reading back to tare. This reluctance to return to tare was one of the reasons to get a new scale, but the return to tare issue seems identical to the original.

[hoola]

I think that issues, such as simple cleanliness of the layer interfaces between stack components may account for some inconsistencies in traces and their relation to potential thrust development. I use nylon washers from the hardware store, with injection mold small dimples, and so offer unequal shock transfers, and as previously noted, the piezos themselves are not quite flat, plus have slightly varying resonant frequencies, which may or may not have a significance, as there does seem to be a set of sweet spots of shared resonance that makes this consideration perhaps less important than the others. The flatness issue could have some bearing on the resonance issue, as perhaps relates to torque pressure on the uneven surfaces, which perhaps distorts and delays (lowering the Q) of the shock wave.

[hoola]

at 160.8 Khz (and with no other change of settings),  a negative thrust, corresponding to weight loss of 3mg on the scale, shows up. The scope trace indicates a downward set of spikes of somewhat less size than the ones described earlier today. The correlation of upward spike traces indicating a positive thrust and downward spikes indicating negative thrust seems to be holding in today's tests.

while the 10mg of weight change equals only .1 millinewton of thrust, the question remains of what minimum amount of thrust is required in a typical sattelite station keeping mechanism (some literature indicates 25-250millinewtons ) and could this system be scaled up to provide that thrust. 

Edited January 10, 2021 by hoola

[hoola]

after having excellent results on Saturday and Sunday and Monday, doing many tests, at approx  2am Monday morning the scale (the original one) started acting erratically, adding weight in a somewhat linear fashion, without any input from the machine. I removed the scale and set it out on the bench and tested it again with a small weight and it began going up again in the same linear fashion. Seeing that the old scale had an issue, I placed the new scale in position and continued tests. None of the the following tests measured anything but random or trivial results, and I gave up at 3am. The next day, some of the results that did seem above the noise level indicated a reversed thrust signatures, as guided by the direction of the lobes shown in the scope signals.  I have had the new scale in and out of service and verified coincident readings without fail numerous times before this. I have not not registered any obvious signals of thrust since then other than very occasional trivialities and sometimes with the inverted thrust signal issue adding to the uncertainty. I documented on the phone on Sunday, when the apparatus was behaving predictably,  a real time observation of the scope signal and the scale and have 2 movies of approx 10 minutes that show constant, unfailing positive results. I had become so confident at that point, so as to make these movies and to later post them, which I still plan to do.  I have had this happen many times before, long (or what I considered long at the time) stretches of positives followed abruptly by weeks of nothing credible showing up. The debacle of the summer was when I discovered my wiring error, of which that cannot explain this, as nothing has changed in that area since then, and the failing experiments are strangely coincident with the failure of the original scale, and that the failure happened in real time, while I was sitting there, and repeating the usual  routine of making the minimum  amount of physical movements to conduct a test, the position of my chair,  keeping room temperature constant, etc. I have disassembled the old scale, found nothing obviously wrong, reassembled it, and it has worked perfectly since, and am using it, but exchanging it out with the new one twice per day. I once broke a "dead zone" by re-torquing the stack, relaxing then resetting the torque slightly differently. I did this and have had no luck in that process today. I will attain an inch/pound wrench and make the stack torque parameter repeatable and measurable and establish a range of torque values that  approximate around the center value of maximum scope trace readings, as a maximum thrust signal may occur at a less than maximum trace reading. The Nov 17 anomaly was when the trace was not very large and I was surprised that anything developed from it. 

I was afraid this would happen. I was trying to upload my movies from the phone when the email pics my friend's kit amp he just built somehow showed up here. I saw no way to delete them, so please ignore them. My apologies.

[hoola]

  correction: the day of failure was Tuesday, 2am, not Monday, also remove one "not'' from line 6 (word 7).

Edited January 14, 2021 by hoola

[hoola]

today I placed 2 hardened steel sleeves over the 1/8" bolt that has served as the central pivot point for the arm assembly.  I knew the pivot was needing an upgrade, and the sleeves certainly seems to verify that, acting as makeshift bearings, as I am now getting controllable and repeatable thrust signatures, both positive and negative. I will remove the sleeves and install the ceramic bearings I am going to order, which should further upgrade accuracy of measurements, and hopefully minimize the tendency of the arm to stick at the pivot point, which I believe is the main reason the scale doesn't always to return to tare reading at the end of tests. Since the arm moves in a such a miniscule manner during tests, I did not think the issue was as important as it apparently is.

[hoola]

Today I got in the bearings, and they work about as well as previous methods used, with 2-3mg weight changes, sluggish response, and reluctance to return to tare.  I suspect that the temporary excellent response with the addition of the sleeves was due to the interior of the sleeves having a light oxide coating that allowed the arm to move freely until the oxide coating wore down after two days.  I added graphite lock lube to the sleeve interiors after they started sticking, and that was not helpful.  Since the bearing idea has not shown promise,  I will redesign the stack to hang horizontally from  2 cotton threads tied to the ends with the signal wires dangling between them, all suspended from the trapeze above. I have found that the scale can work sitting on it's end vertically. With it's circular table removed (it just lifts off) the load cell lever arm is exposed about 1/8" above the surface, and easily accessible without further modifications other than to re-orient to a vertical mount attachment to the breadboard. I will apply a sideways torque via an intermediate push rod, directly from one end of the stack to the load cell lever arm of the scale.  With the scale mounted vertically, there is no need to alter the torque angle 90' with any extra hardware to transfer the torque with it's necessary small counterweight, down to the horizontal position. This arrangement will simplify the overall design and cut suspended weight down to just the stack and it's intermediate push rod. The suspension can be adjusted to deliver 5mg of weight to the scale under no input, keeping the push rod in constant contact with the scale.  With the reduction of mass of over 50%, the response time should commensurately faster, and without the pivot,  show a more reliable and predictable behavior.

Edited February 11, 2021 by hoola

[hoola]

Early tests with the new arrangement are going well, but the stack being hung on threads makes the assembly more susceptible to out side vibrations and air currents, plus the signal wires take time to 'relax" after removal and re-installation. The signal wires themselves are needing an upgrade, as I have used a blow dryer on low setting to relax them after redoing certain things to speed up the relaxation time, and the insulation on them seems to becoming stiffer due to these repeated heatings. Despite these drawbacks, I am now getting up to 10mg of weight increase and about 1/3 on weight decrease, similar to previous results. The response to an input is still at about 1/2 second on average, although a few quicker ones have been noticed. The scale does tend to return to tare, but not perfectly, so the wires now seem the main culprit, as the pivot point has been eliminated  I have found that I don't need the intermediate push rod, I have a rubber pad glued to the stack end which abuts the scale lever arm, and that seems to work well enough to absorb vibrations and maintain positive contact.  The previous arrangement had the weight changes transferred to the scale via a 3" long piece of cotton thread holding a small weight,  after travelling over the 1/8" x 12" aluminum lever arm (and passing the 1/2 lb counterbalance and coming though the fixed pivot point) as the thrust signal originated from the stack.  I presume this eliminated the risk of false readings from vibration.  I had placed the scale as far as possible from the stack to eliminate the possibility of electronic interference as well as vibrations. Since my readings with the new arrangement loosely mirror the previous setup, the risk of false readings from vibrations or electronic transfer seems unlikely. There does remain the possibility of acoustic standing waves causing being issue, of which would require a vacuum chamber to dispel completely, although I doubt this is an issue. The small table on the scale is gone, with only a very narrow lever arm making direct contact through the rubber pad. The surface area of the lever arm is miniscule and seems unlikely to offer a large enough surface to get a sonic transfer of energy.

[hoola]

I replaced the wires with the lightest wires I had on hand, and they are performing better on the tare issue, but still not appropriate for the need. I will not make the mistake of using a blow dryer on the signal wires again. I will get some light gauge solder wick and try that next.  That has no insulation, and being a weave, is very pliable. The new arrangement is working well, other than the return to tare issue. There have been a couple of new sounds emanating from the stack,  the clear tonal purity of a small brass bell constantly ringing, and a harsh hissing, scraping noise. These 2 new noises appear at certain settings, then tend disappear or weaken when leaving, then returning to that same setting to try to resurrect that sound.  There seems to be a "latching" effect going on, having long noticed that a certain sound or trace is only attainable by sweeping up to a certain parameter setting, and if you try to sweep down to that same spot, the effects don't develop, or are replaced by other ones.  It seems to display a certain "handedness" in this manner. Various tones seem to correlate to the thrust signals, and I can usually manipulate the readings up and down without looking at the scope, only by listening for the tonal changes as I observe the scale. The parameter change I am currently using is the voltage output from the signal generator. The scan is chaotic as it approaches the threshold  of approx.  10.5v P-P (the shake), and at approx. 12V P-P  the first clean trace and coherent sound become established, and a negative thrust is seen of 3-5mg. As the voltage is raised slowly, other tones come into play, which do not affect thrust reading much or at all until I approach 15V P-P  or so when an abrupt change of tone occurs, and the thrust signal goes abruptly positive in the range of 10-15mg. Past that setting thrust remains constant positive  until I reach the limit of the generator output, which is 20V P-P.  At these higher voltages, the 7014 IC relays these increases to the pre drivers and outputs, and I am getting to the point of overdriving them into saturation, or to damage the 7014, so I do not leave the system at the 20V P-P level for more than a few seconds.  At the 15 volt input level, all transistors are running moderately warm, and I have no problem delivering 5-10mg of thrust signal without fear of damage and have left a steady 10mg of thrust signal on for over a half hour, and results were very consistent throughout.  As generator output is now the main parameter change,  I seldom move the delay control  or to flip on and off varios piezos having found their most effective settings for this particular arrangement. I am using all piezos on,  forward scan. These responses were all at the 125khz setting.