Maker Pro
Maker Pro

Voltage Amplifier

Hi,

I need help trying to power a piezoelectric patch which has a voltage input range 0f 0v-360v, where the voltage supplied to the patch is determined using a Raspberry Pi 3 board. I will be using one of the RPi's GPIO pins, so the output would be 0-3.3v depending on a code I have written so ideally when the output from the board is 0v the piezoelectric patch will recive 0v, board=2v piezo=200v, board=3.3v piezo= 330v etc.

I have looked into non inverting amplification circuits using op amps, but I can't see this being very successful even in series. I have tried using an amplification circuit just to test to amplify 3v to 9v but I have had no luck, but this could be the op amp that I am using or the way that I have connected the circuit.

Any suggestions would be appreciated! If it's possible to name the actual components that would be required and why those ones I would be very grateful.

Darren
 

hevans1944

Hop - AC8NS
You need a high-voltage op-amp. Try Apex Microtechnology PA-96 for starters. If this is a DIY project, you can also "roll your own" from discrete high-voltage MOSFETs and/or BJTs, but be prepared to "let the smoke out" of a few components before getting the knack of it.

The circuit configuration you need is an op-amp with about X100 closed-loop gain, inverting or non-inverting makes no difference except for polarity of output. The RPi provides a 0 to 3.3 V DC output, the op-amp amplifies this to a 0 to 330 V DC output. Use an op-amp with more closed-loop gain and a higher voltage power supply if you need more voltage. Note that the piezoelectric element has appreciable capacitance and will require substantial current from the op-amp if you want to move it rapidly.

This is not a project for the faint-hearted or those without deep pockets full of cash.
 
Thank you do much for reply, given the expense of that particular op amp I would definitely prefer to try the DIY option. Although as this is a uni project if the DIY option would be too difficult the op amp is an option.

I'm sorry I'm not very familiar with the components you mentioned, would you be willing to expand on your idea? If you could point me in the direction of a circuit I would be very grateful!
 

hevans1944

Hop - AC8NS
I'm not very familiar with the components you mentioned
Really? All I mentioned were MOSFETs and BJTs.

It appears (to me at least) that you are totally unqualified to design and build a high-voltage DC amplifier that will accept a low-level 0 to 3.3 V DC signal and produce a high-level 0 to 300 V DC (or more) drive signal for your off-the-shelf piezo actuators. Such amplifiers are readily available, although a bit pricey. Suggest you make Google your best friend and search for high voltage piezo drivers. Maybe start here.

I doubt anyone on this forum is willing to design the DIY piezo driver circuit you need, although we may help after we see that you have exercised some due diligence in trying to solve the problem yourself. I can tell you that it is a very expensive proposition for someone with little or no skills in analog circuit design, much less high-voltage analog circuit design. If you are in a university environment, try to enlist the help of someone in the electrical engineering or physics department who has experience working with high-voltage analog circuit design and construction.
 

hevans1944

Hop - AC8NS
WOW, that's a serious Op-amp !!
Veeery serious... and expensive at $146 each. They are now available in Europe for € 130,69. The OP should order several spares because it is soooo easy to "let the smoke out" of these hybrid devices until you learn the proper circuit construction techniques and placement of the current-sensing overload-protection resistors. The sockets aren't cheap either, but IMO absolutely necessary for a successful build.

I first started using Apex hybrid op-amps late in the last century, when their PA12 first became available. Before that, you could buy external current "boosters" for op-amps from Burr-Brown and others, or "roll your own" from discrete semiconductors (usually a BJT). It was always a real PITA to get it working right, properly frequency compensated for whatever closed-loop gain was needed, and protected from short-circuits and overloads at the output. I was so grateful for Apex picking up the task from Burr-Brown of building hybrid op-amps that I didn't mind (very much) the high price. IIRC, the application was driving large magnetic deflection coils for ion-beam x-y positioning at the entrance aperture of our particle accelerator.

We also had Apex high-voltage op-amp circuits that drove the electrostatic raster-scan deflection plates at the high-energy output of our particle accelerator, but this was a National Electrostatics Corporation design that "cheated" by using the op-amps to drive high-voltage transformers, providing differential deflection signals to opposing x-y pairs of electrostatic deflection plates. Negative feedback was used to linearize the transformer outputs, which needed to be triangle waves. Think of a traditional CRT oscilloscope on steroids with 3 MeV heavy-ion beams instead of an electron beam. This particular gem wasn't working when I arrived on the scene in 1996. Someone had burned out the op-amps by raising the deflection voltage high enough to short out the coaxial cables feeding the deflection plates.

Sometimes students take on thesis projects involving electronics instrumentation without a clue as to what is required for success. As a technician in the early years of my career, working for the University of Dayton Research Institute, I was often handed the job of getting their projects off the ground. Sometimes this was an impossible task if they insisted on duplicating the work of someone else, usually without understanding what it was they were trying to duplicate. The problem wasn't limited to just students either. Professors would ask me to replicate an experiment published in a paper, usually a peer-reviewed journal, using the exact same components and equipment the author scrounged up ten or fifteen years prior to publication. By the time I was handed the task, most of the equipment was obsolete or unobtainable. It was quite a chore, not always successful, to explain to the learned individual how the published results were obtained, what was important and what generic substitutions could safely be made to obtain the same results. Thankfully, after I graduated, I left that environment to work on a set of entirely different problems.:D
 
How quickly is the output voltage changing? Are you driving the piezo with a varying, but essentially DC voltage, or with some kind of periodic waveform? Also, what is the minimum rate-of-change you can stand, especially when decreasing the output voltage? And finally, do you have any information about the peak output current needed?

My thinking here is that you might be able to get away with a high voltage linear regulator rather than a full-blown opamp, or opamp with a booster stage. Whatever it is, it needs an effective gain of 100. Be prepared for a noisy output, but slowing down its response time will decrease the noise.

ak
 
Hop,
You are always a wealth of knowledge that comes with a lot of experience.:)

If I'm not mistaken APEX was producing Hybrid RF Power Amps since the 70s(or earlier),
they were the "go to guy" for quick RF solutions that work back than.
Looks like they are still alive and kicking ;)
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
If you're at a university then they may already have some high voltage power supplies that can be controlled by an analog voltage, or even a dedicated high voltage on amp (I picked up one of the latter on eBay for a song).

The device I have would be happy in a rack, and offers a couple of fixed and variable gain with outputs up to +/- 1000V.

Here is an example (very pricey)
 

hevans1944

Hop - AC8NS
Here is an example (very pricey)
Holy moly, Batman! That one is a BARGAIN at $325 for THREE channels. And an ex-NASA pull too! If I were the OP, I would jump on that if it is still available. There is no way you can build even one channel (successfully, from scratch) for such a low cost. Pricey indeed! Pricey cheeeap!

Hop
 

hevans1944

Hop - AC8NS
...If I'm not mistaken APEX was producing Hybrid RF Power Amps since the 70s(or earlier), they were the "go to guy" for quick RF solutions that work back than.
Looks like they are still alive and kicking ;)
I don't think it is the same company. Apex Microelectronics Inc. is (and always has been) based in Tuscon AZ and was started in 1980 to satisfy a (then) niche market in high power, high current, op-amps. I don't think they were ever involved in RF amplifiers. Of course Apex is a common enough name, so you are probably right about that.

Apex Microelectronics has had its share of struggles to remain a viable, profitable, business. It was once sold to Cirrus Logic for $42,000,000 on July 24, 2007, but I believe the principals pooled their resources and bought it back for $26,000,000 in 2012.

Apex was started by a group of former Burr-Brown engineers, who up until then had specialized in hybrid, high-performance, op-amp "bricks" competing with the likes of Philbrick-Nexus. That market soon gave way to integrated circuits which Burr-Brown (correctly) decided were more profitable. B-B was successful enough in making IC analog devices (op-amps and DACs mainly) that Texas Instruments purchased them in 2000 for $7.6 BILLION USD.

Hop
 

hevans1944

Hop - AC8NS
I should take some photos from inside it. It's remarkably simple.
Please do! The devil is always in the details though, simple as that may be.

Getting an op-amp with a bi-polar ±1000 V DC output isn't exactly rocket science, but it does require some good electrical engineering design. If it can be done while delivering one milliampere (or more) of current with a moderate bandwidth, it would be a useful piezo-positioning transducer driver.

If I were trying to design something like this today, I would probably investigate the possibility of using a heavily loaded Cockroft-Walton voltage multiplier running at 50 kHz or more to get a high DC output voltage with fast response. Even so, a closed-loop bandwidth of ten to one hundred hertz might be the best that could be easily achievable. An advantage of using a C-W multiplier is only high-voltage diodes are required to produce the DC output. However, you do need TWO multiplier chains to produce positive and negative outputs, and heavy loading of the output to obtain fast response... so not very efficient.

For better efficiency, perhaps a series or shunt regulation scheme with stacked MOSFETs is viable. Optical isolators, operating in a linear mode to get the gate biases proper, could be practical... I am not really prepared to get involved with high voltage analog electronics again, but there has probably been some really good stuff developed in the last twenty or thirty years. So maybe I will Google around the Internet to see what pops up...

Hop
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Please do! The devil is always in the details though, simple as that may be.

Your wish is my command...

Hmm unfortunately it's only 0-1000V, it's not bipolar (sorry about that).

Here's some photos:

IMG_20171209_134128.jpg

IMG_20171209_134816.jpg

IMG_20171209_134144.jpg

IMG_20171209_134647.jpg

There's nothing much exotic in there.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
There's several of these on eBay in the US at the moment. Two have buy-it-now for $25. It's got to be cheaper that rolling your own...

Edit: apparently these were used for precision control of peizo devices, so maybe they're exactly what the OP wants.
 
Last edited:

hevans1944

Hop - AC8NS
There's nothing much exotic in there.
Except for the high-voltage power transistors (BU 209) with BV(br)ces = 1700 V DC.

These "horizontal deflection" transistors have become somewhat rare since television sets no longer require high-voltage fly-back transformers for horizontal deflection of the electron beams in color CRTs, as well as kilo-volts of acceleration potential derived as a by-product of the horizontal deflection circuitry. I recall using a similar transistor in the 1970s to control the current supplied to a small helium-neon laser, all part of a (successful) attempt to stabilize the laser output for some arcane purpose, long forgotten now since the funding ran out.

I still have somewhere the cast aluminum Bud box I built the circuitry in, which featured extremely difficult-to-find, multi-pin, miniature bayonet-style connectors (male cable plug, female chassis mount). This allowed the HeNe laser to be unplugged from its separate power supply and my control box to be inserted. There were other more conventional connectors for an external photo-diode that sampled the laser beam with a beam-splitter as well as line-power input, but those two MIL-SPEC connectors gave me the most grief finding them. Finding the connectors was absolutely a requirement, as the owner of the laser (it was on loan for this task) was adamant that we could not make any modifications to their 'laboratory quality" laser before returning it to them in pristine, original condition. Which, of course, we did... not long before that particular model became obsolete and no longer supported by the original equipment manufacturer.:D

... apparently these were used for precision control of peizo devices, so maybe they're exactly what the OP wants.
Good possibility! Most PZTs work just fine with one bias polarity. In fact, the devices cited at the website given by the OP are specified to operate from -60 V to +360 V DC. So, he might have to "offset" the output to mechanically "zero" the device and then work plus or minus from there, but certainly worth investigating if bi-polar output is not required. Good find, Steve!
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Good possibility! Most PZTs work just fine with one bias polarity. In fact, the devices cited at the website given by the OP are specified to operate from -60 V to +360 V DC. So, he might have to "offset" the output to mechanically "zero" the device and then work plus or minus from there,

That's exactly what the"bias" control on the front panel does :)
 
Top