• Sensor board

    100dollarhacker02/02/2024 at 21:56 0 comments

    Until now we made the QFM (quartz fork) module and we used signal generator to create a sine waveform with specific frequency and measure the output using scope. 

    Now we want to do it programatically. We would like to have a device that will gnenerate frequency as a command from Arduino board and meassure the output using Arduino. 

    To create a signal we will use AD9833 signal generator board. (For now as a external PCB board with few capacitors and crystal clock). AD9833 can output frequency with 0.1Hz precision. Arduino can provide I2C that will set AD9833 to the desired frequency. For output we wil look for zero phase as this is less error prone (at least it's looks for me now). 

    The next step would be to add xor and integrator to find the phase difference.

    For now we can see it how it changes.

    Now, I've added two comarators and XOR gate and got nice two peaks for each half cycle as expected. 

    Now we get digital output of the peaks. Since output is digital we can simply sum the time output is high relative to low. The frequency where it's most low is the main frequency. 

    The last part would be a converter of AD9833 which output is 5-0v to voltage needed by QFM module which is 5-(-5)v.

    -----+-------------+------------+--------

    Ok, got AD9833 connected and running. The amazing thing I didn't know previously is that it can get float numbers which means we can set frequency of sub Hertz. For example here I've set 10.3Hz

    The last part will the LM324 op amp to convert 0.6v of AD9833 to 5-(-5) volts swing that will be input to QFM.

    Ok, got all components together working...

    The output is bit warped but I think good enough to me. (It should be sine but it looks like triangle)

    That's all for the electronic part. 

    Next stage would be to put them on PCBs. Working on breadboard is not convenient 

  • Soldering SMD - take 2

    100dollarhacker12/24/2023 at 17:01 0 comments

    This is a the point where I continue the 'Soldering SMD'. Last time I had to go back and fix the double rail power supply problem. Now it fixed and we may continue. First lets check the attenuator 

    Looks good we got about 1Vpp. 

    I found something strange or bug in the design the guarding ring is not coated so terminals of quartz fork may touch it. 

    Thus may lead to short circuiting. Hmmm...

    Now lets add all other components like OPA656 and another opamp.

    Had some problem of putting to much soldering paste and all pins was glued together. But some playing around solved the problem.

    Let's plug it in.... I will continue tomorrow it's getting too late...

    The test was not successful, once I connected everything I got only noise. Since I don't have too much testing point I tested the output of attenuator. I was surprised to get there no signal. After while, I checked the power lines no power! What ?! Once disconnected power was back. The problem is I got shirt circuit somewhere. Found some problems in my soldering but still got no output. Checked attenuator output, still none. It *was* working before, how we went back?

    Now I don't know what to do. Should I try to debug while I don't have accurate board design. (The one I got once I bought the board is blurry) or should I design one myself? 

    After couple of days I've tried to debug this PCB. I've decided to build my own version hopefully once I will start building I will understand what is the problem with the previous one. 

    It uses the same schematic as the previous just without the capacitor compensation thing. 

    GCode seems ok.

    I've tried to make it not to dense, otherwise it hell to connect all the timy components.

    Printed the circuit using 15° 0.1mm v-bit.

    The quality is really good and still there always some parasite connections that created because of the copper debris. After some cleaning and connectivity check it looks OK. 

    Next step is solder chip by chip and check that each step is working properly.

    Careful placement of components

    Gives lean result

    I guess it's 'you reap what you sow'

    Let's see if it's working as expected, it should get input from signal generator and attenuate it by 100. 

    After all the connections it's still not working. Changed to other AD8676 I had. Still same problem. Without power the resistors seems to propagate the signal. But once power applied everything goes down. 

    I've checked the power voltage it's only 4.6 and the minimal is 5v I was afraid to use 5v because of OPA656 it needs max 6 volts and I wanted to be on the safe side.

    Let's try to increase voltage to 5.2v by using 430KOhm instead of 300KOhm.

    ----------------------------------------------------------------

    Changed from 300KOhm to 430KOhm. The voltage is +5.6 and -4.6. That should be enough for our little omamp to come back to life. But what I get is only noise. 

    I am clueless....

    I ran some simulations, with this configuration and 4v sinus input I we should get 40mV. Maybe the  power rail high noise obscure it?

    I'll have to test with old fashioned power supply wich doesn't have this switching noise.

    Yeah, that was it. Changed to old style power supply and got nice 50mV sinusoidal wave.

    It's starting to get complicated as usual

    It took me few long hours to connect OPA656n

    Fingers crossed, if this will work we can see that electrical part of AFM as done.

    Added output resistors, this way connecting oscilloscope would be easier.

    Powered on and .... NOTHING

    It just doesn't work... Started testing each board independently and power seems to work fine provided nice 5.4v and charge pump board creating positive and negative voltages until I connect the latest one. Once connected negative rail goes to zero. It turns out that the new board is short circuiting negative rail to ground. 

    I am pretty sure I've checked it. Funny that...

    Read more »

  • [Re-]designing PCBs

    100dollarhacker12/10/2023 at 05:47 0 comments

    The design is simple, the idea is to stack everything over standard Arduino board layout. 

    From Arduino I need only SCL,SDA and GND. 

    So, I will get one board for Arduino. Another one for power supply. Power supply will generate 12v and 5v only.( The double rail voltages will be created at each board that needs it). It's tempting to make the 12-(-12) and 5-(-5) voltages on the power board and route them to needed boards, but I am afraid long lines will add noise. Plus we prefer each board to has it's own voltages, this way we reduce inter component's noise.  

    Another observation is to use analog ground planes and not standard GND ones. 

    Some chips needs special layout like charge pump, so special care should be taken when designing them. 

    Another advantage of using stackable boards is each user potentially could choose which sensor to use.

    Ok, let's start with the main board which provides rectified voltages of 12v, 5v and analog ground.

    And it's getting bit messier on Arduino shield...

    Converting it to GCode and sending to CNC/Candle software

    Fingers crossed 

    [Hour later... and lots of noise]

    After two hours of soldering I got this:

    It's bit ugly but does what it suppose to do.

    5v and 11v. Wait what? Why 11? It should be 12v. 

    One thing I've learned is to use wider pads for soldering otherwise connectors just breaks.

    Single to double rail, here I come...

    Anyway, the input should be 14.5v to 27v.

    Now let's start with input. We should get -5 to +5 voltage for OPA656N to work.

    Here is the schematic for generating two rails from one, using LT3260. Most  of the time took me to learn how to import LT3260 into The result

    The routing/placement will by difficult 

    ---++++----

    Actually it wasn't that difficult. I've used the one provided with LTC3260. I got this,

    Let's zoom in..

    Now it's time to print and test it.

    I know this is suboptimal, the vias are huge and I don't use smaller than 0603 capacitors/resistors. 

    And this all because I create PCBs manually, no dual copper layer etc. 

    On the other hand, I can change those designs in matter of hours 🙂

    First looked good by after some examination it turns out to be bad, on GCode level.

    The second one got broken bit in the middle. And third got wrong height for some reason.

    But fourth was much better, 

    More than I could ask from old and cheap CNC machine.

    Checked the circuit for conductivity, one of the pads was short circuiting. Tried to fix it and bricked the PCB. Arrrr ...

    Another three attempts ended up with two broken bits and one successful print. Made smaller circuit so printing would be faster.

    It looks worse than the previous one but electricity wise it is ok.

    Added all the components, boys it only looks easy.

    Soldering paste made things easier but components so tiny, it's crazy!

    Checked all the lines, looks ok, except one. I hope I could fix it without bricking the PCB again. 

    The good news it's working!! 

    It's looks like I've made all possible mistakes one could think of. 

    First I managed to rotate power supply pins. Luckily I don't need Arduino inputs. Then from all capacitors I forgot charge pump capacitor.

    After connecting those it still didn't work. I thought first that I burned the IC by connecting wrong polarity. Luckily I got -5v Vo (not LDO-) so it encouraged me to look for a bug. Then I found that the problematic line was not problem with IC pins but near capacitors. Finally another problem was pin that was not properly connected, so when I checked it with Fluck it looked connected, but actually it was floating. Resoldering the pins gave finally the long waited results

    With stable output voltage 

  • Building my own PCBs

    100dollarhacker12/06/2023 at 14:44 0 comments

    It may sound like going two steps back, but it worth it. One of the problem with such projects is that they're hard to reproduce. It is hard to do such project with breadboard. I've tried and burned my charge pump(single to double rail dc-dc converter) since I moved wrong resistor in the jungle of resistors on the output stage.

    So, I needed to go back to the beginning and do it right. I thought of printing delicate SMD connector using laser module and on 3D printer but it creates dangerous fumes, arrr... Then I've stumbled on video about creating SMD with CNC 3018, WHAT?! Is it even possible? I had to check it out.

    My old cheap 3018 CNC, had controller damaged so I had to replace with even cheaper CNC shield v3 that was laying around idle. After connecting/compiling/installing/tuning I finally got my CNC back.

    Ugly but doing it's job.

    Now I started to play with height map, it's important for such delicate components like 0.5mm pitched SMDs. 

    After some tuning on the fly, I got my first SMD connector.

    This is a test of LTC3260 0.65mm pitched layout

    It's promising especially given the fact that it was made with old and bit broken bit. Results can be improved even further. 

    Now, PCB can be created in matter of hours and I don't have to wait two weeks for each PCB iteration.

    It was worth the time invested 😀

    After inserting new/sharp bit I got quite satisfying results

    It good for development, the final board will be printed using one of many fabs out there, that print PCBs.

    Now let's run example with 0.5mm pitch

    Not bad for standard 30° bit. I guess 20° or 15° can make even better resulta. 

    In order to use high precision for my PCB, I used DRV8825 motor drivers. With setup of 1/32 step. But it got it's price, drivers got super hot and I got automatic shutdown. To prevent this I printed this nice box and vent to cool down everything. 

    Another necessary step back from my cause. 

  • Soldering SMD

    100dollarhacker12/01/2023 at 20:52 0 comments

    Got the components and the boards, fhew, it took so much time! 

    Last time I tried to solder tiny SMD components it was so messy, I thought I will try easier method of solder paste. 

    Step one find a victim

    Let's say this tiny cute 500 ohm resistor of 0603 size code and some PCB board I don't need anymore.

    Put some solder paste... 

    Oops, I got him all covered and hopefully not lost. Now I started my hot air gun, and it's become quite funny since the resistor started to swim inside the paste. Although it funny on demo, it won't be funny once it displaced on real PCB board. Maybe I should use slow air stream. 

    Lower air speed and 250°c done the job

    Not as pretty as I wanted to be but I think it will do for now.

    Take 2, less paste, wider air gun nozzle and I got much better result. 

    Let's try two simultaneously, if this goes well we can try the real thing.

    Ok, sanity check ✅ let's put some components together just before the big mess...

    And soldering....

    Not bad for first try... Oh, wait a minute I got resistors switched... The first one should be the 47K... Arrrr...

    I took me two hours soldering and de-soldering to get to 

    Not bad but it's only half of the board... Alas!!!

    At this point we got only attenuator IC, I thought I would be a good idea to test it. We already know what shows be, we did it in the early stages of this project back then it was breadboard and single rail. Now we have a nice PCB and double rail for power. Instead of making new one we can decrease change pump from 12v to 5v. Easy....

    It's just change one resistor for each rail. Positive changed to +4.5 but the negative is still on -12v. What?! Why?! 

    After few hours of poking each pin and trying to understand what is going on, it looks like it is broken. 

    It's quite annoying to go back and fix already working parts. 

  • Don't re-invent the wheel

    100dollarhacker11/10/2023 at 05:20 0 comments

    I thought that it could be a good idea not to re-inventing the wheel this time. I have a great post from Dan Berard. In his post https://hackaday.io/project/5713/logs he explain how to do the last electronic piece that I don't have. The amplifier that will amplify the tiny Quartz fork movement and convert to standard voltage signal, signal which could be picked up by standard ADC and then to some kind of DSP.  

    Sounds easy, right? Well here things started to go little bit south. I bought the board it's only 5$ for 3 pieces. (I don't have machinery that could do sub millimeter PCB boards so, sounds like a good deal)  bought all the amplifiers just to find out that I got it wrong and I bought the wrong layout... Arrr!!! Plus I don't know what sizes of the resistors/capacitors.  

    I paid for the board like week ago and still nothing... 

    I guess I should buy nice laser and print such circuits by myself, it will take an hour ... 

    Arrr!!! Hate to wait....

    -------------------------------

    Still waiting to board and components to arive... meanwhile I thought it could be a good idea to 'reverse engineer' sensor board. Although you got the schematic it's unclear how and where to place each SMD component on PCB. 

    So I placed every resistor/capacitor from schematic: (an yeah.. I got kids ... 😂)

  • Mechanical fine movement

    100dollarhacker09/30/2023 at 19:37 0 comments

    I know it looks silly but I started to prototype using some cardboard. This way I can get look and feel without hours of 3D printing.

    And I already got some interesting insights. For example the spaces between the pillars can be greatly reduced without touching the moving stage base/enclosure. 

    It's only a POC made from few layers of dense cardboard but it's working and it amazing. It does moves a tiny actuator (a toothpick which is already broken) that can be seen only by microscope, since movement is so tiny.

    And now same thing 3D printed. Simple and does it's purpose. Now let's put four of this under the base.


    Now I need four of them to hold 'the table' on which we will place nano scale needle.

    TA-DA!! The nano probe holder

    And now on pezo-mechanical bed

    Actually this stage was done poorly as the glue was not silver glue. And hot gun glue that holds the pizo-electric devices is distributed not equally. I guess I will have to redo this stage later on as part of improvements.

    Some piece of math was missing. I mean if I move the piezoelectric up on ine side and down on the other side, how much nano probe will move? 

    Now let's connect all together the RaspberryPi based microscope to see probe's height, fine movement mechanics and electronics.

    (On the screen you can see magnified toothpick, tangsteen probe will come later) 

    Turn on the power and nothing 😞 it doesn't move... 

    More debugging with scope and reprogramming firmware to make move faster.

    Finally I can see my toothpick moving up and down (run it with 1080p settings otherwise it's blurry) skip to 6th second. 

    I think I am done with this part! 😄

  • Coarse mechanical part

    100dollarhacker09/24/2023 at 23:32 0 comments

    I was examining the mechanical parts given by some available opens sources. Most of them has only Z-axis manipulator. It's OK when you don't care where you land, for me it's crucial.  So what I've saw used 'Adjustment Screw' to adjust the height of stage compared to the  afm-probe. Plus the price tag was awbout 33$ for each. (and we need three of those). 

    So I decided to go on a different path. I am familiar with OpenFlexture project, it's an open source 3D DIY printed microscope. The provide both microscope and moving stage with submicron accuracy. 

    It has both the moving tray and a placeholder to put RaspberryPi based microscope. 

    We will need this microscope to accurately place the nano probe without bending/breaking it. (Also it has place to hold the LED) what else can I ask? 

    Now I have to figure out how to mount RaspberryPi based microscope/camera and I am done. Easy peasy, right?

  • Piezoelectric controller

    100dollarhacker08/16/2023 at 04:12 0 comments

    This basically DAC which takes input from I2C. In my case this is Arduino nano as a master.

    All other connections are the output and other configurations of the DAC chip. (Like reference,gain etc.)

    The next step will be making Arduino and DAC talk over I2C and watch for change on A,B,C and D outputs.

    Actually, this stage of I2C or Arduino to DAC went smoothly even beyond my expectations.

    All I had to do is to disable input code (since we are not going to use same input) and change LDAC pin (since we are using nano) and vaila! 

    Here is simple code 

    And the output:

    The next step is to connect converter from 0-5v to  (-12) - (+12)v but it's very technical and I don't expect any trouble with that.

    As expected, connecting the op-amp was quite straightforward. Although I found bug in OpenAFM schematic (or my coping it), the gain of DAC should be connected to 5v so the total output should be from 0 to 2*Vref. Plus connecting the power supply we get this:

    The output is not simetrical around 0v but this can be fixed by changing reference voltage for op-amp. Voltage is not fully rail-to-rail, but it can be fixed by taking bigger resistors as feedback. The problem is that I have that noise (the thick line) I believe it's due to the fact that I use bread board and not PCB with suggested layout.

    I think I done with this part of the project 


  • Building power supply

    100dollarhacker08/08/2023 at 10:37 0 comments

    Started with charge pump as it was most tricky part in my opinion in power supply circuit.

    Other power supply components I've already worked with such as LM7805 and they are pretty straightforward.

    This oddly looking thing is the LTC3260 with all capacitors and resistors connected

    And the result is negative voltage compared to GND.

    The next steps are creating downsteping from 24v to 12v and 5v. 

    As expected it was quite straightforward

    In my case it gets 24v DC and convert to 12v DC and 5DC.

    The outputs of 12v and 5v seems very flat signals, although I am not quite sure how to measure it.

    Well that's is another milestone 

    Since there is a lot of space on this bread board I thought to dense both the power supply and charge pump. This way 12v, 5v, and -12,12v from the same board. 

    Got messy just the way I like it 🤣

    The schematic is much simpler