DIY printhead

I updated the wiki page, on the ToDo list "The build a drive incorporating an arduino or add pwm to existing box" and I put a link to my my message showing the Arduino + Mosfet + piezo.


VDX Wrote:
-------------------------------------------------------
> Hi Casainho,
>
> ... i've edited the link to the document - you
> have to insert a space after the documents name or
> the separator is part of the name and the link
> won't work ...

Thanks. Viktor, wana join us? ;-)

Edited 1 time(s). Last edit at 08/17/2010 06:26AM by casainho.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

Hi Casainho,

i'm following your progress and maybe i can help you with some related infos ...

But actually i have several other projects running, so my time/ressources-limits are 'hitting the edge'

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Found this device, SP4501, google the datasheet, its a piezo ic, any use to us.

johnrpm Wrote:
-------------------------------------------------------
> Found this device, SP4501, google the datasheet,
> its a piezo ic, any use to us.

That IC looks like it boost the Vcc up to 100V, on the Cint capacitor. Basically is a way to drive the piezo with "high voltage" other than using the transformer.

Do we need such IC or transformer? -- before you wrote on wiki: "Amplitude... this effects the distance the jet travels, and again appears pretty linear in response, going from 0 to 110 volts on the dial results in the jet traveling from 0 to 150mm. (damage can occur to the bimorph above 110 volts)"

So I went and tested/verified that I have a piezo "making noise" with 24V. 24V would give for 30mm as for your experiences.

I think you should try again with the transformer circuit you have, and see if you can print. After try the same print head but with lower voltage like the 24V circuit.

I guess 100V would make piezo contracting/expanding more than 24V and so make higher pressure on fluid. We need to find if piezo/print head really need hight voltage to be able to print. And define the needed voltage and fix it over the experiences we will make in future.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

The 20mm piezo's arrived today, When the 25mm clear acrylic rod arrives I intend to make a selection of bodies, with shallow, deep, flat bottomed, conical cavities, although I have no means of measuring the effects other than how well it works, (assuming it does)

But first I have a plan so cunning that you could put a tail on it and call it a fox, (black adder)
I want to make the piezo disk serve as the nozzle, imagine the current design upside down and the nozzle hole in the centre of the disc.
I drilled the disk with a carbide pcb drill, thinking it would crack the ceramic and spoil it, but apart from some roughness it looks ok, I connected
it to the drive and it worked as normal, well as normal as I can see without some exotic equipment.
the next step it to use the idea by Slavko and have a thin wire and a blob of solder to make the nozzle hole.
I may not work but is worth a try.

Edit, If it does work it may be a self cleaning nozzle?.

Edited 2 time(s). Last edit at 08/17/2010 05:01PM by johnrpm.

Johnrpm said: Agreed, the H bridge looks like it will give a lot of control further down the line, but we need to walk before we run at this point and understand some fundamentals first, I think your electronic skills are far better than mine, so I would love to see your circuit design, if you are willing to share.

I'm afraid my electronic skills can be most accurately described as cookbook electronics. I'll be very happy to share anything I come up with (or more likely, run across). At this point I really don't have anything to share, as my driver circuit consists simply of 2 mosfets and two 1k resistors, just like the circuit you described building.

Here is the key finder enclosure as a OBJ 3d model, it seems that it is designed to maximize the sound output from a low voltage supply, so may indicate the shape for our printhead cavity?.piezo_enclosure.zip
Sizes in a jpeg

Edited 1 time(s). Last edit at 08/19/2010 05:39AM by johnrpm.

... i have some drivers for the piezoleg-nanopositioners form www.piezomotor.se , which are sourced with 12Volts and convert into 48Volts and steer 4 bulk-PZT-'legs', so they bend and stretch accordingly to a 4-leg walking gait.

I think you can simply apply one of the outputs to a piezo-dish and it would start to oszillate.

In contrary to the common drivers, which start with a slow frtequenzy and accelerate to a specific maximum, we've got a special firmware - here i can activate a special mode, where i apply a voltage controlling a 'slow' oszillating with variable frequency without acceleration, so this would be better for ink-jetting ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Thank you Viktor, those drivers look very good.

At present we have no more than proof of concept, and lots of work to get it plug and play, but I can not help thinking ahead,
do you think it possible to detect from a feedback signal whether the printhead is printing fluid, I was thinking that the drive could go through a set routine to establish optimum frequency and voltage automatically at startup, and be able to detect
problems when printing?????????????????.

I know I am getting way ahead of myself here.

P.S now that the acrylic has arrived I can make the test housings, I shall shine a laser pointer through the acrylic to see if
it shows up turbulence better.

... it should be possible to detect if the oszillator runs in air or in fluid - air has nearly no mass and is compressible, so the resonance-frequency or the amplitude at not resonating frequencies should be much higher than with fluid ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I have tried the active nozzle idea, drilled the piezo disc, suspended the tungsten wire through the hole and tied on a 12mm nut to tension it, soldered the hole, filed down the solder blob to make it flat then pulled out the wire, then realized I had to seal the active side.
I used nail varnish as the sealant but could not risk blocking the nozzle hole, I turned an acrylic body to hold it and set it up to try.
All seemed well at first, it takes a while to extract the air before a jet will form on a new head, after a few minutes the unsealed area around the nozzle started to smoke, the ink was causing a short circuit locally, the bimorph cracked and that was the end of that.

The conclusion is inconclusive, due to my lack of foresight in forgetting to seal it properly before pulling out the tungsten wire,
i shall revisit this later, and try to cut an area of the crystal material with a diamond bit to allow proper sealing, a bit like a washer.

Wetting on the front face of the nozzle is inevitable, and effective sealing is a must.


I may have two bimorphs with a spacer between them, the front one being the nozzle and the rear one acting as a sensor.

Johnrpm, you might take a look at using a ring style piezo device. It would allow you to construct the active nozzle without needing to drill through the piezo.

steminc appears to sell small quantities for reasonable prices (although the international shipping cost is pretty high).

More references

Yang style using a piezo disk (this will look really familiar)

Simple Piezoelectric Droplet Generator

On-Demand Droplet Generation

Effects of a thin, flexible nozzle on droplet formation and impingement

Tube style devices using Ring/Cylinder piezos

Positive pressure drop-on-demand printhead for Three-Dimensional Printing

Microdrop Generation

The "Microdrop Generation" book appears to provide very detailed instructions for making tube style devices including the construction of glass nozzles. There may be some overlap here with the effort by rocket_scientist to develop glass nozzled for plastic extrusion. I found and ordered a relatively cheap used copy earlier today.

Progress, or the lack thereof

I've had little success getting the print head to produce either drops or streams. I've tried driving the piezo with higher voltages and with faster rise times, but the only results I have to show for it at this point is a couple of shorted out mosfets and a burned up SN754410 h-bridge chip. The mosfets were destroyed because I rigged up a transformer between the driver and the piezo and then turned the supply voltage up too high - the kickback voltage from the transformer destroyed the mosfets.

I have seen a lot of pretty little blue flashes scattered across the top of a small piezo when driven by too high a voltage from the transformer. I believe each flash represents a bit of the piezo material being destroyed. In the end there was not much left of the piezo material on that particular device.

Madscifi,
Well done for finding all that info, They look very similar to what I did, and I thought I was being origonal,( nothing new under the sun), at least it proves we are on the right track.

I am testing different acrylic bodies, (your excellent idea) , I am having a lot of problems getting rid of entrapped air, even a small bubble when subjected to vibration becomes froth, and dances around as I fiddle with the frequency, it seems to want to migrate to the inlet pipe, when I do eventualy get it running the fine nozzle,(.050mm) produces a really tiny jet.

Untill I can solve the air entrapment I can not do comparisons, maybe the high frequency is de-gassing the ink?????,
with the larger nozzle, (.2mm) the air could probably escape easier.

My feed pipe is 2mm bore, a lot smaller than yours, may have an effect?, keep trying and dont give up, you are my only partner in this.


P.S the good news is that the 20 mm piezo discs work, and at about £3 for 25 good value.

The two of you may be the only active builders but there are lost of us lurkers really interested in the project! Thanks and keep up the good work.

Johnrpm, I think you are probably correct about the size of my feed line opening being too big. In addition, I had the "clever" idea of making two openings in order to make getting the air out of the chamber easier, which at this point I think was another mistake. It is probably just contributing to the dissipation of the pressure inside the chamber.

Don't worry too much about me giving up anytime soon. The problem is interesting and a working device has the potential to be very useful. Progress will, no doubt, be slow due to other projects and a day job, but it will, also no doubt, be made.

I was going to try ordering some of the 20mm piezos that you purchased from Rapid, but I checked the international shipping cost and I just cannot stomach paying 50 GBP shipping for 3 GBP worth of parts. Which is funny because I might be willing to pay three USD apiece for 25 of them... Fortunately, several of the local distributors appear to have comparable devices so I'll order something comparable from one of them.

Your suggestion that the vibration in your device is degassing the ink might be a good one. Take a look at this blurb concerning the degassing of liquids using ultrasonics. Perhaps you should try intentionally degassing the ink before filling the device?

I have a vacuum pump so could build a rig for degassing, I bought the ink from poundland some time ago, its for re filling printheads, I have used 2 bottles of magenta and am using yellow now, I hope that we do not need to de-gass as it will make things less appealing, but at the moment I can not find where the air is coming from, sometimes its best to walk away for a while and come back with a fresh outlook, a work friend has just got married, so tommorrow I am carving some wood from a wedding photo for him on the x2 cnc conversion, will make a change from printheads for the day.


Arvin, thanks for the support, I would really like the reprap community to have a plug and play device that can handle various materials, the active nozzle for instance came about from thinking about printing silver loaded paint, 3d pcb's , bio printers, the list goes on.

This morning after an hours rummaging, I found a small bleed valve, fitted it to the head and blead all the air from within, whilst fiddling around to get the jet going air appeared, thee are two possible reasons for the air, my fittings are not air tight or the ink is de-gassing, I shall refit with silicone and try again when time permits.

I have made three heads, all the same but different depths, , also I would like some opinions on a circuit for the lighter driver, will post later, stuck for time just now.


Good news, I sealed all the fittings with instant gasket, and the problem appears to be solved, the head works down to the lowest frequency my drive will do, about 1KH and up to the max around 8KH, also down to 30 volts and up to 100 volts,
I am an idiot for not using sealant before, but in mitigation I applied moderate pressure with the syringe and no leaks, so wrongly assumed a good seal, maybe high frequency vibration has its own rules, or am I looking for excuses?.

So am now hopefully back on track, onwards and upwards.


P.S the blead valve is actually a miniature pneumatic flow valve.



I was going to try ordering some of the 20mm piezos that you purchased from Rapid, but I checked the international shipping cost and I just cannot stomach paying 50 GBP



That does sting, I am not sure if there is any advantage to the 20mm discs, the frequency is higher so the displacement should be smaller, which is something to look at down the line when we are trying to squirt ??? picolitres per pulse to get a small dot size.


I have been reading some of the papers you have linked to, lots of information, a lot of the questions have been answered already, I shall finish reading them and have a rethink if necessary.

Edited 6 time(s). Last edit at 08/23/2010 10:18AM by johnrpm.

---

Random Precision

Going back to the head after 24 hours, it fired up straight away, no problems, it always needs a few extra volts for a second or two and it will stop and start from then on, but I read that the researchers had the same problem, the only issue at present is the jet is not perpendicular, it is at about 10 degrees, I smoothed the front with fine wet and dry, maybe a tiny particle embedded in the solder is deflecting the jet ?, or a micro burr.
The wire, as mentioned previously is .002" (.050mm) but a point mentioned in one of the papers is shrinkage of the solder, so the hole may be slightly larger, another interesting item is the signal profile.put simply a long suck time and short squirt time, the arduino can do this but would it take much to have a circuit relieve the arduino of the task and receive a logic high or low and produce the signal profile, with some presets for fine tuning,
or is the arduino capable of running the reprap as well as the printhead.

...or is the arduino capable of running the reprap as well as the printhead.

The standard Reprap controller uses two microcontrollers, one for the xyz motors and one to control the plastic extruder. The one that controls the plastic extruder would certainly be capable of controlling the piezo head. It might be practical to run it directly off the main controller, but it would depend on the specifics of the pulse profile needed.

At this point I would like to take a step back and attempt to replicate, as closely as pratical, your new print head (as detailed in testheads.jpg). I want to do this because I've not been able to make the variation that I made work. I have a list of questions I'd like to ask in hope that the answers will help me duplicate what you have done. Some of the answers would probably be obvious to someone even poorly versed in machining, but I'm definitely not.

1. What material is the current printhead made of?
2. What did you use to attach the piezo to the body?
3. Can you provide a link to the 20mm piezo that you are using?
3. What is the diameter of the ink feed hole and where did you drill it?
4. Can you provide any more information about the bleed value and how and where you attached it?
5. Can you suggest a minimum voltage across the piezo that will reliably start the head working?

The following questions assume that that you are using a 555 timer and a transformer to drive the piezo.
6. Are you driving the transformer directly from the 555 timer or is there something between the two components?
7. Can you provide any details concerning the transformer?
8. If you own an oscilloscope, could you measure the rise and fall times of the voltage across the piezo?

Based on what I've been reading, I don't think the two fets and two 1K resistors pseudo h-bridge is a good choice for driving the piezo (easy to say after neither of us have been able to get it to drive a print head usefully). Everything that I've read indicates that the pulse rise times should be pretty fast, preferably a microsecond or less. The 1k resistors in the pseudo h-bridge give me edges closer to 50 microseconds. Although smaller resistors would, of course, improve the edges, there are better approaches to solving the problem. And as far as low cost solutions go, your use of a transformer has a lot going for it. A true high voltage and fast linear amp would provide the most flexibility, but that is necessarily more expensive. Combining a fast low voltage linear amp and a transformer might be the most flexible, reasonably priced approach.

Yesterday I received the copy of Microdrop Generation that I ordered last week. It looks like it will be a very useful book. Most of the information in it is geared toward tubular style drop generators, but a lot of the information is applicable to any drop generator. The book suggests transformer based drivers as a low cost solution to the high voltage problem, hence my comments earlier.

I'll definitely be attempting to build a tubular style head as it has some advantages over the disk style design - the liquid to be dropped only ever touches glass and, due to the design, getting all of the air bubble out is trivial. The biggest disadvantage that I'm aware of is that the piezo necessary is not one of the low-cost disks that can be easily purchased from any large electronics distributor. Since the availability of appropriate piezo devices is so limited, I don't think a tube based device is the best design to pursue as a general purpose Reprap accessory. Rather, it is more of a specialized device for use when the normal drop generator is incompatible with the liquid to be dropped.

Edited 1 time(s). Last edit at 08/26/2010 12:44AM by madscifi.

Madscifi,
I shall answer all your questions in chunks, by editing this after getting the info or pictures to explain things.


1. What material is the current printhead made of? 25mm acrylic rod, (see picture )
2. What did you use to attach the piezo to the body? superglue
3. Can you provide a link to the 20mm piezo that you are using? here
3. What is the diameter of the ink feed hole and where did you drill it? M6 to take a fitting (photo) the tube is 2mm/2.5mm approx, I think its used in aquariums

4. Can you provide any more information about the bleed value and how and where you attached it? opposite the inlet (photo)
the bleed valve has a M5 thread, its really for compressed air, to regulate the flow.to a cylinder or such

5. Can you suggest a minimum voltage across the piezo that will reliably start the head working?
To be conservative I would suggest 70 volts, less may do the job but 70volts is more or less sure to start it

---

The following questions assume that that you are using a 555 timer and a transformer to drive the piezo.
6. Are you driving the transformer directly from the 555 timer or is there something between the two components?
A 555 goes to a tda2030 amplifier, then to the transformer, the transformer was from some scrap equipment so have no data, also it was built almost 3 years ago, (a lot of water under the bridge) I have looked for any info I may have had at the time but can find nothing.
I am useless at electronics but understand that when a voltage is removed from a coil, the collapse causes a large spike,
I wonder if this is playing a part in driving the piezo, because I am having little success with 30 volts by mosfet etc, it does not seem to have the ummph, apparently old valve radio transformers are good, they work at higher frequencies, middle one circuit with some modifications looks like it would work?.

7. Can you provide any details concerning the transformer? see above

8. If you own an oscilloscope, could you measure the rise and fall times of the voltage across the piezo?

Now your asking something, I do have an ancient osciiloscope but only use it to see the type of signal and as a visual voltmeter, I will do my best and get back to you.

---

Based on what I've been reading, I don't think the two fets and two 1K resistors pseudo h-bridge is a good choice for driving the piezo (easy to say after neither of us have been able to get it to drive a print head usefully). Everything that I've read indicates that the pulse rise times should be pretty fast, preferably a microsecond or less. The 1k resistors in the pseudo h-bridge give me edges closer to 50 microseconds. Although smaller resistors would, of course, improve the edges, there are better approaches to solving the problem. And as far as low cost solutions go, your use of a transformer has a lot going for it. A true high voltage and fast linear amp would provide the most flexibility, but that is necessarily more expensive. Combining a fast low voltage linear amp and a transformer might be the most flexible, reasonably priced approach.

I was under the impression that transformers were simple, boy was I wrong.

---

Yesterday I received the copy of Microdrop Generation that I ordered last week. It looks like it will be a very useful book. Most of the information in it is geared toward tubular style drop generators, but a lot of the information is applicable to any drop generator. The book suggests transformer based drivers as a low cost solution to the high voltage problem, hence my comments earlier.

Tube piezo materials were very expensive, they may be cheaper now,

---

I'll definitely be attempting to build a tubular style head as it has some advantages over the disk style design - the liquid to be dropped only ever touches glass and, due to the design, getting all of the air bubble out is trivial. The biggest disadvantage that I'm aware of is that the piezo necessary is not one of the low-cost disks that can be easily purchased from any large electronics distributor. Since the availability of appropriate piezo devices is so limited, I don't think a tube based device is the best design to pursue as a general purpose Reprap accessory. Rather, it is more of a specialized device for use when the normal drop generator is incompatible with the liquid to be dropped.

When I first started, I envisaged a pen plotter type system, squirting viscose materials onto a substrate, maybe two "pens"
the fine one drawing the border and a thicker one for the infill, apart from ink the current design would be usefull for solvents,
fusing polymer powders, but would not have the power to squirt higher viscosities.
Thats why I am going over to the lighter spark units, although I do not know for sure yet, I think they will have more power, but need more power to drive them, I am building a 500 volt drive at present, Drop on demand is good for rastering but vectoring is better suited to a steady stream imho.

One possible way to drive the spark units would be a pulse generator feeding into an audio amplifier then to a coil,
the pulse generator should not be to difficult, the audio amp from a car boot sale, and the coil could be wound by hand,
old pc skeakers had a small amp, but not sure if they are big enough, velleman do a 30 watt amp in kit form I think, I shall have to look into coil winding?.

Edited 7 time(s). Last edit at 08/26/2010 04:18PM by johnrpm.

---

Random Precision

Thank you. Your answers will help a lot. I do have a new question: it looks like you made the nozzle section out of brass and screwed it onto the body of the print head - correct?

I'm afraid my knowledge of transformers is very limited. I've mostly worked with digital devices, and other than the standard power supply transformer I have pretty much ignored transformers as I have never needed them. The book I got recently does, however, have a suggested transformer for a simple high voltage driver. It is constructing using 60 (22 AWG) and 300 (28 AWG) turns on a toroid (part number 5961001001). The toroid appears to be made by Fair-Rite and appears to be equivalent to a FT-114-61 sold by Amidon. I'll buy a couple and try testing them out with the circuit from the book.

Speaking of standard power supply transformers, I tried using one to drive the piezo with a simple transistor circuit and it unsurprisingly did not work well. It boosts the voltage just fine, but the pulses became triangles after passing through the transformer.

Tubular style drop generators do not need to use tubular piezo elements, which is a good thing because I don't know where to get tubular piezo elements at anything approaching a reasonable price in small quantities. The book discussing a variety of designs and at least one common design uses a disk or ring style piezo. While the ring piezos that stemic sells are probably not ideal, they look close enough to what is discussed in the book to make it worth trying a couple.

Yes the nozzle is brass, screwed in with a small O ring to seal the face, the plan was, to have the same nozzle with different bodies to evaluate any differences, but the papers you linked to answered my questions, " larger chambers damp oscillation" so higher frequencies can be used, this also applies to longer feed pipes.
That is assuming I have understood it correctly, there is little point in spending time repeating their work, especially as I do not have the equipment to see subtle effects, only big differences.
Looking at the transformer it appears to have multiple tapings, so could be an autotransformer.

When time permits, I will do another video, (as bad as the first) of the head, then for the time being draw a line under the bimorph head, and concentrate on the spark units.

Thanks for the info on the coil, would it be the same if wound as a standard air coil, also what is the output and input voltage, or should I say ratio.

would like one of these machines
toroid winding

Tubular style drop generators do not need to use tubular piezo elements,
Could you do some sketches of other configurations?.

P.S are we allowed to post circuits from books.

Edited 2 time(s). Last edit at 08/28/2010 06:25AM by johnrpm.

---

Random Precision

Here are STL and OBJ 3d models of the spark units, spark.zip also some images.

I intend to make a housing to hold a pair at first, with some provision to preload onto the tube, then try the bank of three to see if that works, the tube will be anything I can get hold of, from stainess to rubber.

As mentiond before, when connected to the existing drive, very small amplitude can be felt compared to the bimorphs, but are virtualy unstoppable by pressing hard onto them, I am building a drive to output more power to see how far they will go, rather than risk blowing up my existing one, also they should work at higher frequencies, like the bimorphs they are cheap enough to blow a few up without worrying.

---

Random Precision

The input/output voltage ratio is defined by the ratio of the number of turns in each coil, so in this case the output voltage is 5 times the input voltage.

I don't think an air core would work well. I think the number of turns would need to be much greater than with a ferrite core and apparently there are other issues as well (see Transformer).

Microdrop Generation discusses two piezo configurations other than tubular (cylindrial).

Slab, or rectangular, style is still a big chunk of piezo with a cylindrical hole down the middle. Consequently, the piezo material is going to be just as expensive and specialized as the cylindrical designs and the drive voltage required is much higher (according to the book) than needed for cylinder or disk style piezos.





Disk style uses a significantly smaller piezo. My assumption is that the rings sold by Steminc would work, although that is conjecture at this time.





The simple pulse circuit from the book looks as follows.





Note that I may have the orientation of D2 wrong. Also, D2 is considered optional and its use depends on whether or not you want to squash the negative voltage across the piezo when the pulse ends.

Update 2010-08-31: It is currently believed that the specific design above will not work with Yang style bimorph-based print heads.

---

I've been spending a lot of time looking for examples of piezo drivers and trying to figure out from the articles that I can find what electrical attributes are actually needed for driving a print head. A lot of the designs that I have found are high voltage but rather slow, good for controlling a deformable mirror on a telescope but not so good for driving a print head. Microdrop Generation has a lot of good information, but the details are specific to the tubular style of print head and so may not apply to the Yang style drop generator. The book's design for a fast linear driver is probably really clever but I'm still trying to wrap my head around it. In any case it would be rather expensive to produce as it depends on multiple, isolated, high-voltage DC-DC converters.

Specifically, the book suggests the following:

• Suggests that a minimum drive bandwidth of 1 MHz is required.
• Suggests that rise and fall times tolerances are rather large, where large is undefined.
• Pulse width: .5us - 10us for large volume disk/ring designs, and 5us to 20us for short cylindrical designs
• Rise times: "rise times of up to about a couple of microseconds do not seem to affect the operation much."
• Fall times: not really specified, but in the case of viscous liquids times in the few to few 10s of microseconds can decrease drop size and increase stability. Suggests that "very long fall times seem to suppress drop ejection."

Update 2010-08-31: It is currently believed that the above recommendations do not apply to Yang style bimorph-based designs. Certainly the pulse widths are too short by a factor of about 1000, at least according to all of the literature that we have read.

The relationship between pulse voltage, pulse width and fluid height in the tubular style head is all over the map. For a given fluid height the voltage required can vary over a 10 to 1 range as the pulse width is varied from less than 1 microsecond to 50 microseconds. In all cases charted, as the pulse width changes from less to 1 microsecond to 2 or 3 microseconds, the voltage required drops very significantly. After that, the voltage required becomes unpredictable and varies significantly as the pulse width changes. Voltage requirements based on fluid height also vary all over the map, although once the fluid is deep enough, the range drops and narrows significantly.

In some cases the ability to produce a double pulse for each drop is useful, specifically 1 to 3 microsecond pulses with a delay of 4 to 30 microseconds between the pulses. This may be something that a transformer based design does not do well.

At this point I think the circuit that Casainho published earlier (or a minor variation) would work well as the basis for a simple pulse driver. It certainly would allow for much faster rise times than the circuit configuration suggested by Slavko.

Although I don't have the suggested toroid yet, I did manage to find a FT50-43 toroid locally, so I plan on giving it, in conjunction with a driver that looks a lot like Casainho's, a try later. I'm hoping that the increased voltage and rise times of a driver based around the toroid will enable my current print head to at least show some hint of progress.

Edited 2 time(s). Last edit at 08/31/2010 11:19AM by madscifi.

Madscifi,
Thank you for sharing all that info.

[*] Suggests that a minimum drive bandwidth of 1 MHz is required.
I may be talking rubbish here, but I wonder if the bimorph, as it is a metal disc, can physicaly move at very high speeds compared to other types of pzt???.

I bought a function generator yesterday in kit form and got it working this morning, it can produce, sine, square, saw,triangle waves and goes from 1 to 37KH, on the scope the waveforms are perfect, so I connected a small 4 watt amplifier then a 240 to 6 volt transformer and a bimorph, the waveform at the bimorph is virtually the same whichever wave is selected on the function generator?, I think nophead said they act like capacitors, or is it the amplifier, it seems to happen at all frequencies so should not be over driveing the amp, its probably that inductance thing, any suggestions.

I have a laser ignition board by coherant from the skip, it has a pwm IC going into a small high frequency transformer then into a large toroid that outputs 2kv, I may desolder the toroid to put a meter across it, I also have a small laser drive that may be good for parts,

I shall have a go at the circuit, thanks, I think I may try and get a flyback from an old TV and rewind it, they are getting rare now though.

Slavco did manage to produce dod, but I could not get any output from his circuit, probably my error, or his head design is better, I did hope that the push pull would improve things

---

Random Precision

...but I wonder if the bimorph, as it is a metal disc, can physicaly move at very high speeds compared to other types of pzt.

That is a very good point. I just looked at the resonant frequency and capacitance of the ring peizo I'm ordering - the resonant frequency of the rings is 50-150 khz, as opposed to 3-6 khz for the bimorphs, and the capacitance of the rings is 350-1020 pf as opposed to 10000-20000 pf for the bimorphs. I assume the effective values for the rings will change when attached to a tube, so I'm not certain what the effect will end up being.

It may be the information in the book I got is completely wrong for the bimorph driven devices. The book does not claim to be applicable to anything other than tube style devices and I've just been hoping that it would be broadly if not specifically applicable.

A quick review of the papers I can access concerning bimorph driven heads:

Simple Piezoelectric Droplet Generator states the pulses where less than 10 ms and 30 volts, except when starting up (in which case they were somewhat longer or greater).

On-Demand Droplet Generation discusses using 3 ms pulses where the voltage changes from -60 to +55 volts.

Effects of a thin, flexible nozzle on droplet formation and impingement discusses 5 ms pulses.

If the electronics for the bimorph designs need to generate pulses in the milliseconds as opposed to the microseconds range, then the design I posted yesterday from the book is not going to work. It does not have the current capacity required to generate a high voltage pulse of even a millisecond into a bimorph.

Curiouser and curiouser. Anyway, acrylic rod is on order, as are a variety of electronic parts and piezos. When they arrive I'll get on with the task of building a new head.

---

Update

I went back and set up a 12 volt power transformer with the 120 volt side attached to the piezo and the 12 volt side attached to a single mosfet driven by a 3 khz square wave. I get a something very close to a sine wave on the piezo side, but I'm also seeing an effect on the water streaming out the nozzle. I've set things up so that there is too much water pressure and water is naturally streaming out the nozzle. The nozzle is pointed down, but due to some flaw in the nozzle the water is flowing out at an angle. The angle of the stream is consistently modified by changing the voltage level of the signal. It is a likely a long way from being able to control either a stream or generate drops on demand, but it is definitely progress.

It seems I've been chasing the wrong goal in a driver, at least for the bimorph design. Thank you for pointing out the difference between the two types of devices.

Edited 1 time(s). Last edit at 08/29/2010 07:51PM by madscifi.

Great news, glad your persistance has paid of, your mains supply is 120volt whereas in the uk its 240volts, so transformers here would have a higher ratio, I dismantled my drive to give you some info on the transformer, it has multiple tappings but the resistance is 1.3ohm input and 300 ohm output , I know this is not a true indication because they could have used different size wire for different coils, but if its an autotransformer may be an indication I think.

I suppose that 12 volts from a modded PC power supply or similar should be our target, to keep things simple, like the laser drive, a transformer feeding into another transformer may be worth considering, also trigger transformers and pulse transformers look interesting.

---

Random Precision

I bought a function generator yesterday in kit form and got it working this morning, it can produce, sine, square, saw,triangle waves and goes from 1 to 37KH, on the scope the waveforms are perfect, so I connected a small 4 watt amplifier then a 240 to 6 volt transformer and a bimorph, the waveform at the bimorph is virtually the same whichever wave is selected on the function generator?, I think nophead said they act like capacitors, or is it the amplifier, it seems to happen at all frequencies so should not be over driveing the amp, its probably that inductance thing, any suggestions.

Based on what little I know the transformer is going to act as a bandpass filter, the bandwidth and corners of which is going to be a function of the transformer's construction. Since you are using a power transformer that is designed to work at 50 or 60 Hz I'm not surprised to see the transformer acting as a low pass filter for the range of frequencies you tested. This will make all of the waveforms from the function generator look pretty much like sine waves. Also, as Nophead said, the piezo acts mostly as a capacitor, which decreases the upper frequency response even more. A wider bandwidth transformer can certainly be found or constructed that would allow more high frequency energy through, although at this point it is unclear if that would provide any actual advantage when driving a bimorph style head. Conjecture - you might try an old tube amp output transformer?

Using a transformer to drive the piezo has both good and bad points. Its biggest advantage is that it is an inexpensive way to get high voltage impulses (or even voltages in the lower 20 to 40 volt range when using a 12 volt power supply). On the other hand it limits the pulse profiles that can be delivered to the piezo in terms of both high and low frequency content. That is not to say that a good or even the best profile will not be doable with a transformer. You've demonstrated that a stream of ink can be controlled with a transformer, so it known to be a useful approach, and it is a lot less complicated than using any other type of high voltage circuit.

Currently, I'm thinking that I will probably attempt to construct a high voltage bipolar linear amplifier that can drive the piezo directly since it would allow experimenting with a greater range of pulse profiles. On the other hand, if someone comes up with a good head/pulse profile for doing drop on demand before I get around to building such an amplifier, I probably would not bother. In the end I hope a head design can be arrived at the requires 30 volts or less at the piezo since mixing high voltages with liquids does not sound like a very good idea for a device designed to be constructed by a lot of people with minimal electronics knowledge. At the moment I can only just see a bump of water from the nozzle of my current head when driven by a 300 volt pulse, assuming the moon is in just the right position and so on, so a better head design is most pressing for me.

I suppose that 12 volts from a modded PC power supply or similar should be our target...

I agree that it would be best if any design could be powered from a 12 volt supply. However, lower current supplies for voltages up to about 48 volts are not too costly.

...I dismantled my drive to give you some info on the transformer, it has multiple tappings but the resistance is 1.3ohm input and 300 ohm output...

Thank you. I took measurements of the power transformer I'm using and it was 2.1 ohms and 148 ohms, but as you say it is difficult to get too much from that measure alone. It is a rather small transformer, designed for only about half an amp, so it would be interesting to try one designed for more current.

---

Also, that toroid winder is neat. Someone needs to design a laser cut-able or reprap-able version. It need not be fast, I'm willing to wait. Winding by hand gets tedious very quickly.

Edited 1 time(s). Last edit at 08/31/2010 02:48AM by madscifi.

Interesting info on tube amp transformer winding. here

What about voltage multipliers here

Question, could we condition the output of the transformer in some way, a clipper or such???.


The thing that started me using a transformer was the cycle buzzers, thats how they had done it so I copied them, they were very loud.
I have seen trigger transformers used also, which may be cheaper.
I know some chaps that play with high power lasers, they abondoned expensive drives in favour of good audio amps, 1000watts etc, then into home made coils, I will ask them for advise.

---

Random Precision