Project Logs

Collapse

• Data acquisition

  100dollarhacker • 12/24/2024 at 22:55 • 0 comments

  All the work comes down to this moment, data acquisition. The idea is to get data representing the height of tip as a function of x,y. 

  Unfortunately, it's not consistent. I go up and down 2000 steps and I see frequency change. I would expect frequency to change at certain height give or take some noise. Well, I see something similar and the noise is of about 200 steps. Than comes the trouble, if I continue doing it say four five times, the threshold frequency changes qnd goes up. Tried with tip and just clean quartz fork. Added tire to remove vibration and turn off all electrical appliances around. Still same result.

  I suspect it's the rubber bands in 3D-printed xyz stage.

  I've ordered cheap Z-stage on aliexpress.  https://www.aliexpress.com/item/3256806481789030.html?src=google

  Got this beauty 

  This beauty is really small, so I need to rebuild all the 3D mechanics again for this miniature size. It will be third time I 3D-print this part... 🤕

  I really fed up from this project, instead of printing I've just place hard metallic piece. It's ugly hack but hey that's my name 😉

  Let's see if the creep comes from all the plastic or 3D printed Delta stage.

  Still got the creep, now trying different types of quartz forks, maybe it's because of soft metal the quartz fork connected to the tiny PCB board. 

  Basically it's same fork now glued to PCB. Other option is one of the fork's prong glued to the tiny PCB. Fingers crossed...

  'qSensor' didn't work, one I couldn't find frequency other I could see barely something. 

  The regular with base glued to PCB one didn't work for some reason ( maybe it got bit of glue on it) other was working and I could find the base frequency. Then I managed to break it. 

  Ok, made another three quart forks (just to be prepared if anything breaks). Made a few tests, still have the same problem. 

  It's not the xyz stage, it's not quartz fork (now it's rock solid). What else could it be?

  It's not USB connection used little piece of PCB to make it stable. It's not electronic problem rechecked with scope.

  What can it be? I have no idea... 😧 

  Ok, after reading some papers and specs of piezoelectric materials it looks like even the expensive ones have some creep (about 10%) and hysteresis. Moreover, it changes with time... 

  So I wrote simple piece of code which basically goes up until passes some threshold and then goes down the trick was to make it slow so piezoelectric element will get stable. 

  Ta-da first time data...

  It stable enough for now with 400 steps diff. I will remind that those readings were done in noisy environment, without vibration dumping and without shielding. 

  Another setback, although I can achieve precision of 400 steps. It's changing with time, in other words if I do two tests one after another I get close results, but of I take a long break between I got results not even close. 

  I suspect that this is mechanical, as I've applied too much glue to hold the electric wires when connected piezoelectric elements. Plus used regular electric wire instead of one thin wire

   I will try with using tiny wires and soldering paste and see if it's the problem 

  Made another mechanical part version 4.0 I think. 

  For now I have used AliExpress micro positioner, only because it's smaller and easy to obtain. I should go back to OpenFlexture Delta stage, since it more accurate ( I think, don't know how to measure it) and can be easily motorized. 

  Used old piezoelectric discs to try to learn soldering. It turns out brass part is easy to solder to quartz almost impossible without ruining it. 

  So I will use conducting ink(scribepen)+super glue to connect 0.3mm wires. 

  The result was quite good and even my kids were unable to break it. Let's connect the real one and hope for the best. 

  Started adding thin wires (ground)...

  Read more »

• From DIY to 'turnkey' project

  100dollarhacker • 11/12/2024 at 05:26 • 0 comments

  TL;DR - 

  Now you can buy all electroincs of this project (boards with components assembled) just by two clicks on PCBWay. No prior knowledge of electronics required. Plus, you can buy all the bodyworks using 3D-printing services. All is left is some gluing and some minimal soldering that everyone can do. 

  
  
  The longer version - 
  
  The idea is that this project will be accessible to all, not only electronics hobbiest. One of the problem with is that this project has lots of electronics. I was wondering, how someone without  soldering skills (no, it's not easy as it looks like) and good electronic knowledge could make this project?

  In perfect timing (burned most of components by applying wrong voltage) I've been contatcted by a guy from PCBWay who was willing to support the project for a honest review. [Oh well, now I'm biased 😜]

  It turns out that PCBWay and other comapnies are providing something that called PCBA. Which stands for PCB Assembly. Wait! What?! It's fantastic I could provide my schematics/boards and people could get assembled board with all the components. 

  So, I wasn't laizy and checked other companies all had the same answer put the files here (some demanded to convert to some bizzare formats) and engineer will return to you. Big NO,NO! I want people to push the button and get those boards. 

  I've used OSH Park, they have realy nice sharing option and it's supper easy to buy boards from them but they only provide PCBs.

  For example, JLCPCB provide something close to that, you need to provide project files, BOM file and CPL file, then it goes through process of choosing components. In my case it chosen non-standard components and price was dobuled (plus fee of $3 on each non standatd component). After going each component and choosing the basic one price dropped to something reasonable. After that it took me to other window where components were placed for some reason all my ICs were in wrong orientation. So that had to be fixed. The price was the same as PCBWay give or take 10% (but it provides you two assemled boards). I've tried to ask if only one board could be provided but nope. For seasoned hardware developer it could be a good solution but if I want to share it with people who just want to build AFM with minimal electric knowledge it just not an option.

  This is the point where I got realy amazed by PCBWay [Maybe I'm bit biased here] they DO have a posiblity of sharing the project, with all components. 
  

  To show how things may go easily bad was my experience with OSH Park. I got QFM board shared project in OSH Park, but they provided only the PCB (no assembly) and I failed to solder all the components properly (yeah, took me a week to realize that) and they don't provide the full schematic before you buy. I ended up building my own PCB layout just to make it work.

  Biased? Maybe, but until now couldn't find any other service that provided PCB and component assembly which could be easily shared.  

  We can conclude it with the following: You got mainly three options.

  • Mill PCBs and assembly by yourself [fast]
  • Buy PCBs and assembly by yourself [cheap]
  • Use PCBA services (such as PCBWay) to do it for you. [easy]

  Now, lets look at the solution in terms of time and money.

  Milling yourself is the fastest (and cheapest) but you have to own CNC machine and the knowledge to make PCBs. (It can be quite tricky). So timewise it will take you few minutes to build PCB board and few hours to assembly if you have good soldering station (plus good hands) and knowledge to solder 0.5mm pitch (distance between each leg of the integrated circuit) components. 

  Buying PCBs can save you the hussle of building ones but it takes about 2-3 weeks to get to you (unless you pay much more for expedit shipment, then it will take only 7-10 days). You still need a good soldering station (I use hot air gun and tiny tip iron) with magnifier...

  Read more »

• Final tests

  100dollarhacker • 09/24/2024 at 18:07 • 0 comments

  I've added new electronics and better mechanics. Now it's easier to position probe as I like and I can easily change probes (no soldering needed).

  Probe #1

  I got it non simetrical for some reason (mainly because it's hand made and not CNC). but it's WORKING! You probably ask yourself why it's important, well because I was afraid that longer lines will add resistance and stray capacitance which will kill oscillation (like in clocks) but it didn't. It works fine! Now it's time for more compex probes.

  Probe #2 - quartz fork without the case

  I know we already tested it, but I prefer to go slowly but surely. 

  We can see that it's stable without the quartz fork encapsulation.  

  Sanity check done ✅ 

  Let's start more compex probe designs.
  

  Probe #3 - Quartz fork with Tungsteen wire (sharpened by NaOH)
  

  If already wrote about how to create supper sharp Tungsteen wires (this project logo :) ) but once I glue it problem begins. If there is too much glue on the fork's prong it kill oscilation. If too little it doesn't hold the Tunsgesteen wire properly. Glue plus wire kills oscilation sometimes only the glus's weight kill oscilation. like in thie case
  
  Probe #4 - One of the prongs connected to the base. 
  

  Couldn't find the frequency. Well I've found it but it was too high I am not sure if it's the main or side lobes.

  Probe #5 - I was totally clueless of how to continue. 

  Then I recalled that I own really thin Tungsten wire, of 0.012mm wide. It's so thin it breaks easily in your hands. But now I got more experience and nothing to lose. I managed to push some wire into syringe. Since it was so tiny I thought to use 0.1mm wire to paste the glue (instead of toothpick). 

  The glue was strong enough but I had another problem. Since the wire was so short I couldn't etch it. Luckily (yeah, I feel lucky today) I managed to cut it with scissors. (It's impossible to do that with 0.1mm wire, it breaks the glue) 

  Now I've tried to scan the frequency and TA-DA! I got stable frequency 

  Now it's time to lower the probe and see if it successful and we got AFM or all this project goes to garbage. 

  Stay tuned 😉 

  P.S.

  I found fundamental design flaw in my design. I've noticed that sometimes the difference I got is near zero and sometimes near maximum of 5k. 

  The reason is that quartz fork is oscillator with its frequency and phase.

  That's why sometimes I get no signal, it depends on the phase between input and output of quartz fork. There are many ways to solve it but for now I will just restart the system and hope for the best.

  After some rethinking, it looks like we have software solutions. AD9833 is able to change it's frequency programmatically. So we can start normally look for high and low. If not successful, change phase by 90° and try again. Once right frequency found try to get better results changing phase, until looked.

  Update:

  Started another test, failed as the usb metallic cover was on the way. Few power moves the pliers and it ok. 

  Now we had some issues with frequency. I didn't had mechanisms to set frequency. Now I do. Continuing to the test of soft landing this time with real (not yet sharp) Tungsten probe. 

  Update II: 

  Was able to do first measurement with Tungsten wire. 

  The procedure is run 'range' which runs coarse frequency change of 1Hz and range of 200. This finds the frequency where maximal/minimal eclipse occurs. 

  Now we set new base frequency by 'bf' and run 'land' function steps of 100 (I ran with 1000 and crashed into sample). If fails to land change high of micro meter movement (open flexture Delta stage). 

  Update III: 

  Added 'ph'ase function which changes current phase. It helped to change convolution result from ~120 to nice ~ 30. No matter what I did I couldn't go below 30. 

  Going step by step, I thought it would be a good...

  Read more »

• Make lemonade out of Lemons

  100dollarhacker • 08/12/2024 at 06:37 • 0 comments

  I need to redo sensor board as I've glued a tiny pcb piece to the sensor board which now I cannot remove. I guess I will have to create the board again. This is really annoying! 

  Although it highly annoying I thought I might improve my design. Now I have to solder Quartz Fork each time I want to check something. The thin connectors are already done, after much soldering. 

  Now, I found a wat to fix it. I will use USB socket. this way I can connect different forks without need to re-solder the board. 

  It's a win-win situation since every quartz fork I've made needed some kind of PCB to hold it. It's imposible to hold quartz fork with tweezers.

  First schematic with USB quartz fork holder.

  Got USB connectors, stay tuned 😄 

  This is the idea, 

  (I know the connector is upsidedown, it's just a preview). The tiny PCB's thickness is standard  and it site quite solidly in it's place. Definitely a nice lemonade solving such big problem in such simple way.

  Now we can test different ideas for quartz fork and see how they respond without resoldering (and breaking main board).

  I've tried to print this PCB at home with laser (it gives much better results than CNC-mill). But for some reason it always under etched or overetched.

  I gave up and ordered PCB from China. This time paid extra 10$ for fast shipping.  One day after ordering and it didn't even started yet. 

  Although it's annoying to wait those boards are have all the fixes I had to do in the previous version. This is important since at some point we want to make it PCBA, which means people who wants to build AFM, could order not only PCBs but all the components assembled. 

  Got the PCBs let's get started...

  Ok, found some time to solder first part, the attenuator. 

  I thought it would be a good idea to check. Got 40mV bit noise but as expected. (For some reason I thought it should be 400mV)

  More components added, 

  Tested looks good once in right frequency the output is quite high 

  Added all components including usb port that will be used to hold the probe. 

  And that's how it connected to the frame

  I hope the probe low enough so nothing will interfere it's work.

  Added new frame v3:

  With smaller plastic that holds the screws, I got better movement range of coarse mechanical mover.

  Now it's time to make all kinds of probes and see which one is the best for us. 

  
  

• First data from microscope

  100dollarhacker • 08/01/2024 at 19:10 • 0 comments

  Well, you probably expected 3D view of some surface, you should wait a bit for that. This is graph of Z-axis and you can see the Van der vaals forces. Which is how this microscope works.

  The X axis in this graph is 100 steps in Z axis of AFM microscope. The Y axis in this graph represent PLL mismatch. If my electronics would be perfect it should be 0, but there is noise. Once the sample gets near the surface it's frequency changes. Thus, changing PLL output from about 200 to 2000 mismatched samples. 

  
  I got three different graph (different resolution 50 and 100 stesps) and they all have this structure which is coharent with theoretical ones you can find in scientifical papers about AFM. 

• Vibration reduction

  100dollarhacker • 07/24/2024 at 01:31 • 0 comments

  I've faced this problem before. When I had a 100x objective on OpenFlexture and I was looking for fine details of ICs, there was some vibration. I've tried to turn off appliances but it didn't help enough. It looked like the electricity making this interference. I didn't know what is it and how to remove this interference. 

  Now I know that it was simple acoustic noise and the easiest way to reduce it is using iner tube and brick. (In my case box with rocks) 

  Nothing fancy box with rocks and iner tube. 

  At first I put a two glasses of water one on the box and one on the table. Vibration was more powerful on the box one. 

  It was counterintuitive for me, it took me some time to realize that waves inside a glass are low frequency what I want to see is how ot will behave under the microscope.

  Without vibration reduction 

  You can see how it gets really blurry once I knock on the table. On the other side with vibration reduction, I got

  You can see that this simple but effective mechanism is working. 

• Enter the probe

  100dollarhacker • 07/23/2024 at 15:38 • 0 comments

  I have made the probe with tiny Tungsteen wire etched to be very sharp. It's time to connect and turn on so long waited microscope. 

  I've glued the quartz fork to a tiny pcb

  Which in turn was glued to the QFM board.

  Looks promising but nothing happened. I've tried to search for frequency but nothing. 

  Maybe I've applied to much glue and killed oscillation? 

  Now let's make the probe with wire connected on the top instead of connection in the middle. This way glue will not kill elasticity of the quartz fork 

  So instead of gluing one of the sides I am trying to glue Tungsteen wires on both sides. Since I don't want to etch both sides, it would be faster to glue tiny piece of Tungsteen on one side and etch the other. 
  

  It turned out that cutting is breaking the fork :( 
  
  But it does possible to glue tiny piece to a fork. That's what I am going to try next.  

  I've tried different combinations, I don't know what I'm doing wrong but once I apply glue it stops behaving like crystal. Maybe I should apply glue only on the ends of the quartz crystals fongs.

  Although I managed to glue Tungsteen wire it was always so difficult. Now I found a method to simply connect both qrustal fork and tiny wire by using microscope! 

  Calibration goes like this, first place quartz fork and some helper and make sure it seen in microscope ( height should be adjusted) then place playdough (yeah, I got kids) and put syringe inside it inside syringe should be tiny Tungsteen wite. Change microscope plate height so wire became visible. 

  Now only thing is left is to apply glue and be patient.

  
  

• Unexpected funny problem

  100dollarhacker • 07/17/2024 at 01:31 • 0 comments

  I am in the middle of final setting. And then I discovered a serious problem. 

  I DON'T HAVE ANY SAMPLE! 

  It may sound funny but I can't use any surface. I mean I can use any surface but it useless. It will not tell me if there is a lot of noise or maybe it doesn't work at all and there is some garbage/noise that looks flat...

  I've recalled that someone used CD-ROM pits to test his device. So I found some old CD-ROM. 

  Under the microscope it looks like 

  I've focused on the numbers each CD ROM disc has before actual data. 

  Are those the pits and grooves? Let's take a close look.

  It does look like the grooves. Let's look at the data

  It's not clear her from this picture but there is no pits but  straight parallel lines.

  This is a writable CD-ROM. Let's take a look at read only CD-ROM.

  Here we can see the pitches and the data itself looks like:

  Here we don't see the groves only the pitches. Those two CD-ROMs can be used to varify if our AFM is working 

• Software and first tests

  100dollarhacker • 07/10/2024 at 00:59 • 0 comments

  Now for the software, I am using some of the pieces of old software but most must be rewritten. For example using floating point to represent voltage in Arduino (?!) is total waist. 

  The next step are just like the first steps in moon landing.

  First step, lets make a 'hard landing' - we will run full range sinus and look how frequency will change. We don't use quartz fork  with probe but the simple quartz fork. We don't mind banging the quartz fork.

  Second step, soft landing we will lower the quartz fork gently until it reach the bottom. Once it stops we can go up and down and see if this consistant.

  Third step, scannig of the sample

  Well the first step includes building my own software plus libraries for example AD5696 DAC library from Analog Devices. 

  I took me two days to figure out why it's not working. I changed the address bits in hardware but forgot to upade GIT about. 

  Now I got the ability to get quartz fork/probe up and down with resolution of 16bit. 

  Now let's setup the frequency generator AD9833 and try to hit the surface qnd and see what happens. 

  FIRST TEST - Using coarse movement mechanism I managed to lover the QFM less than millimeter from surface. This mechanism is basically a screws with springs they able to lower the QFM board along with QF (quartz fork) on it. The intermediate movement is more complex, it OpenFlexture delta stage with ability to move in XYZ with resolution of about micrometer. 

  So first thing was just to 'crush' QFM into the delta stage that is right underneath quartz fork. As a result PLL output went from stable 180/5500 to 2000/5500

  This cheap 10$ long distance microscope shows every upsidedown. 

  Next test is to bring the quartz fork close and use piezoelectric (fine movement) to gently touch. 

  We can see how both PLL output and frequency changes as a result. 

  To conclusion, now we were able to crush into the samle holder and manually soft landing. 

  The next step will be fully automatic soft landing. 

  Yoohoo! Was able to land softly!!! 🎯

• Connecting everything together

  100dollarhacker • 06/14/2024 at 02:07 • 0 comments

  We have created nano scale probe using Tangsteen wire, but we have never glued it to active Quartz fork. 
  

  Once we have QFM, we could connect the electronics on PCBs, [PCBs makes it more mechanical and electrical stable].
  

  After electronics done, we could connect the mechanical part and do the first experiment. 

  First thing first, Quartz Fork, it looks like an easy step but twice I've tried twice I got my Quarts fork filled with glue, I don't think it will do any good. On the second time I got idea to turn qfm upsidedown but now both qfm and syringe got in glue... 

  One interesting thing I've learned is to use a non-flexible 'helping hand' as it more rigid and has less movement.

  After successful gluing using syringe and microscope 

  The next step was to etch a nano scale tip. After few hours of etching I got a feeling than I don't use Tangsteen wire  after short google-seach I found out that I was sold some other wire
  
  Got a new 0.1mm Tungsteen wire (I hope it Tungsteen) third time I scream :) 

  Managed to glue it right this time

  After etching it in NaOH solution we get

  If we zoom in, we can see the tip

  I don't know what I have done wrong to get this strange form but it's not bad for first time  :) 

  It's a frist time I am integrating data path from Arduino back to PC. The idea is to be able to measure frequency. After some minor changes (pullup resistor and sosftware chagnes) I was able to be able diffrentiate 1Hz quite easily. The output changes almost twice from 730 to 1200. 

  Now I am able to connect all boards together and get both frequency sweeper and 16vpp piezoelectric controller.

  The next stage would connect it on top of mechanical stage and write software that could both change piezoelectric voltage and sample frequency in one simple software.

  The coarse movement for now it's just made of standard screws and nuts connected with hot gun glue. In the future it's will be part of 3d printed frame

  On the bottom of the quartz fork module is the quartz fork.(Not seen in the picture)
  

  At some point I've realized that thinner wire must be used instead of breadboard jumper wires. 
  

  Plus, the eight wites of piezoelectric discs must be converted to five wires with all four ground connected. For this I made simple PCB converter.  

  Connecting everything now looks easier and stronger build.

  Now, all together 

  
  

  Although this is a second video of toothpick dancing under the microscope. This time we use 10$ camera with long working disatnce. All component soldered to PCB instead of using breadboard. All components were tested and noise was reduced. 

  Another aspect is understanding of how much it will be able to move in real life. In other words how close we should place the sample in order to pick up the surface.

  The next thing was adding copper 0.3mm wires instead of breadboard jumper wires. This way readings should less influenced by those wires.

  Afterwards I've glued everything together and added springs to reduce movement 

  Here is the full microscope 

  One thing I didn't do is to add quartz fork with Tungsteen wire. 

  That's because I don't want to break/bend it. For now I will do the first steps with just quartz fork and once it works probe will be added as last step when scanning could be used.

  ----  THIS MINIMAL VIABLE PRODUCT.  ----

View all 30 project logs