Baldor Posted March 18, 2018 Posted March 18, 2018 Nice!! Now I only need to modify the openscale software to output data without units and test the capture. But as said, next two tests will be done taring with the motor, and with two different load cells, the 20Kg one I used for the last test, and the 40Kg one used for all the previous tests. All this should be done today.
stix Posted March 18, 2018 Author Posted March 18, 2018 A link to the new version was upload. When you get the chance let me know if it works for you as expected. Looking forward to seeing the results of your tests - 4 motors? 1
Baldor Posted March 18, 2018 Posted March 18, 2018 Going to test the new version with new an old data. A first glance at the new data confirms the last results. Same geometry, 75/15/10 Pine BP, two different load cells (the 40Kg one from previous tests, the 20Kg from the last one). Both tests resulted in an specific impulse over 80s with similar burn times. (Will publish later in the other thread). I'm far away from testing multi motors. There is a two stage experiment I want to do. First stage a 19mm core burner that I still have to finish the tooling for, with two 15mm rockets on top that will lit simultaneously and go their one way once the first stage is spent.
Baldor Posted March 18, 2018 Posted March 18, 2018 BTW. Do you make some kind of rolling average in the arduino? I get some noise in my data, about +-3g with an static load.
stix Posted March 18, 2018 Author Posted March 18, 2018 (edited) BTW. Do you make some kind of rolling average in the arduino? I get some noise in my data, about +-3g with an static load. My software does do a rolling average. Over 0.5 seconds, ie. the amount of samples received are averaged, but this is only done when calibrating or testing. When recording, the software records every single sample. ie. Sample Perfect. However, the Arduino Uno device, plus the circuitar module: https://www.circuitar.com/nanoshields/modules/adc/ does have the same fluctuations as your device. ie. +-3g. In audio recording terms, you would call it the "Noise Floor". So it's not really correct for me to say that my recording device has a recording resolution of 1gram, but best to describe as 1 gram resolution +-3gm. Which in reality means a resolution of 6 grams. Although when reviewing recorded data we can see definite differences, and 1 gram resolution is recorded. This is a "snapshot" example of a static test (353 grams - because that's the weight of a block of metal I have at hand) over 0.13 seconds. You can see the fluctuations (5gm in the sampled section) the cursor shows the exact sample, and shows the exact time that the force value was recorded: My software enables you to scrutinize your data, sample perfect. It's great for recording, it's also great for analyzing and testing your equipment. :) Edited March 18, 2018 by stix 1
Baldor Posted March 18, 2018 Posted March 18, 2018 Tested the new version, works great, but the real weight at the end of the test happens very late. This is a hardware issue, no need to worry about it. You opened the Pandora box letting me test this software. Now I have another suggestion (Not request) It will be useful to see the difference between two data points. Place two cursors and see time elapsed, thrust difference, etc.... And also have me longing for a faster capture hardware, so I'm looking at instrumentation amplifiers, ARM Cortex M4 dev boards (A teensy 3.5 seems workable, but overkill. 16Bit diferenctial ADC up to 4K SPS, mircroSD, a ton of libraries, ...), etc....
stix Posted March 20, 2018 Author Posted March 20, 2018 (edited) . . . It will be useful to see the difference between two data points. Place two cursors and see time elapsed, thrust difference, etc.... Yeah, that would be easier. I do however have a simple work-around. Simply "trim" left/right the area of interest (as per the above example). It will show you the time, thrust etc. Although, I guess that won't show the elapsed time from the beginning of your original start point. If you explain more what it is you want, and/or why, there may be another solution. The software does not have UNDO as such, but you can always "RELOAD Graph" (bottom r/h corner). This will reload the graph in it's original state. This is, the original recording or the original imported data. In other words, the original data will NEVER be destroyed (except for a hard disc failure). So it means never be afraid of editing to check something. Hope this helps. Next you'll be asking if you can overlay two graphs at once - and I'll say, that's what Excel is for Edited March 20, 2018 by stix 1
Baldor Posted March 20, 2018 Posted March 20, 2018 Its a nice workarrond, it works for measuring rise time (The time it takes for the whole core to ignite) but with only 80SPS, this measurements doesn´t have any meaning. I use libreoffice, so for me, it's librecalc instead of excel. :-)
davidh Posted March 20, 2018 Posted March 20, 2018 Daydream time... Raspberry Pi Zero W + BlueTooth + Android... Powered with a cell phone powerbank... Thrust curve + voice notes + pictures + video... Wifi + webserver on RaspPi... No wires... 2
stix Posted March 22, 2018 Author Posted March 22, 2018 (edited) Daydream time... Raspberry Pi Zero W + BlueTooth + Android... Powered with a cell phone powerbank... Thrust curve + voice notes + pictures + video... Wifi + webserver on RaspPi... No wires... Cheers Dave, you're probably half joking but yeah, I've had similar daydreams Raspberry Pi Zero W + BlueTooth + Android...I did look into Raspberry Pi, but went with Arduino - worth looking into again because it does seem like the Raspberry devices are more capable. From my readings, bluetooth doesn't seem to have enough range (10mtrs?) Powered with a cell phone powerbank...Well if you are doing "in the field" tests, then an external power supply will be required. In my case, to run the load cell amp and the Ardy - although the Ardy can be powered from the laptop usb port. Thrust curve + voice notes + pictures + video...That's a big one and I have thought about it. Ideally you would have the video behind the graph. It would be synced to fit the thrust curve, therefore you could select a point in time at you will see what the motor was doing at that time - Also a "live review" as in real time (graph re-draws in sync to the vid) would be really nice. Voice notes... meh, much easier to implement than video, but then you would have to have a multiple entries scenario otherwise you would have to search through. Pictures, easy. WiFiThat part should be easy (I think). I did look into it, and there is no reason why it needs to be tethered via a cable. All I need is a WiFi module and plug it into the usb port of Ardy. It spits out the data, and theoretically can be received via WiFi on the laptop. No Wires. The module is expensive though. One mad (or not) thing I thought about was having an "Evaluate" button. Press the button, and the software evaluates the current test against all your other (similar) tests in the system. Perhaps the "critical evaluation" is spoken to you (in the voice of Hal from 2001 a space odyssey). It would work very well especially if your name is Dave Sorry Dave, I think I've picked something up. Hang on... hang on... yes, yes, I believe I've detected a potential failure. Hope you don't mind me saying so Dave, but looking at all your latest tests you seem to be going from cato, to limp as a dead dogs donga, and your current motor is unlikely to take off. Perhaps you need to adjust the nozzle diameter or the length of the fuel? Some humans seem to get fulfillment and satisfaction from activities such as knitting, pilates or stamp collecting . . . Just saying . . . Edited March 23, 2018 by stix 1
Baldor Posted March 22, 2018 Posted March 22, 2018 I think I will go with a teensy3.5. Why? Fast, fast, fast.... All done at hardware level, no need for I2C, DAC from ADC to memory.... All the important things are programed at hardware level. https://cdn.sparkfun.com/datasheets/Dev/Arduino/Boards/K64P144M120SF5RM.pdf Integrated delta sigma 16Bit ADC up to 4K SPS, enough for 1g precision in a 40Kg load cell. (Being realistic, a little less precision) Video: I prefer to use a dedicated camera, 240FPS. How much will it cost, in dollars and processing power, to add a 240FPS cameea to a Pi? Similar to arduino development system (I hate arduino, I'm a PIC guy) :-) Things as buttons, encoders and screens should be solved and I will not to reinvent the wheel. If I want WiFi, I will use an ESP32 as the center of the build, but the Teensy have a better solution when you have to make your tests at the limit of your WiFi. A microSD slot not dedicated to load an OS. No Os, of course. I like dedicated solutions, but I lack the courage to attempt a cortex m4 board from zero, so the Teensy seems a good compromise. Only need to find a good amplifier design (Something around a INA128 maybe: http://www.ti.com/lit/ds/symlink/ina128.pdf ), add an screen, some buttons and an encoder. (I love rotary input devices in my designs)
stix Posted March 30, 2018 Author Posted March 30, 2018 (edited) All that sounds good, but now it's my turn to say "way too complicated" The purpose of this post and my (prototype) device is to put together some basic and easily obtainable parts. With a bit of soldering, you have a cheap, accurate and usable device for recording the performance of your rocket motors. This has lead me on an off-tangent quest to also develop software that interprets the data. It's taken a couple of years, but I'm getting there. [EDIT] Well a couple of years part time - probably on average 10hrs per week, which equates to around 1000hrs . Well... I'm obviously not that smart Edited March 30, 2018 by stix
Baldor Posted March 30, 2018 Posted March 30, 2018 Without the software, the hardware is irrelevant, Stix. For me, the teensy seems a reasonable replacement for the arduino. It's more expensive, more capable, have the same programming environment, and have a very capable ADC, so you can forget the I2C external ADC. When it starts to complicate is when you start to dream. Why not use a capacitive touch screen? There are libraries just for this!! It have a mSD card slot, why not use it and forget the laptop? Hey, it have an OTG USB port!!! Why not use it with a pen drive and forget the microSD? Hey, it have a lot of spare cicles, why not make the calibration, taring, and all the calculations on it? An I can draw it in the touch screen!!! When you start this path, you start the same multiyear development cicle than with your software. BTW, I analized the noise in the openscale, and in the load cell itself, and I'm not happy. At least the three less significant bits should be discarded (Something I already assumed but not checked). So instead of a 24bit resolution, i have a 18-19 bit resolution. Not bad at all, but some serious thought is needed for filtering the amplifier input in future developments, and specialy, in the power supply.
stix Posted July 4, 2018 Author Posted July 4, 2018 And again, time marches on... It's been over 2 years since my initial post, and I'm still not finished. Nevertheless, this is the software screen at the moment. At least we now have icon buttons on the r/h side of the graph window to assist in editing the graph. Also, a nice, but pretty much useless feature is that the recorded graph can be "replayed" in real time. I'm currently working on the "help" system and that is becoming the hardest part - a real big and long slow task - but it has to be done. Then, before any official release, some basic altitude calculations have to be shown in a simple manner. I envisage it all being finalised within 3 months, but going by my previous "it won't take that long" scenario, then more likely 6 months. I think my average time spent on this is 8hrs per week - so I guess that's around 1000hrs spent - interspersed with some other pyro activities just to keep my interest up. I am committed to finish this before the end of this year (2018). Cheers.
stix Posted October 23, 2020 Author Posted October 23, 2020 (edited) Maybe I'll get to finish this soon. Life is complicated in 2020. According to my previous post #64, I was meant to finish by the end of 2018. !!!!!!... ouch!!! The Software is almost finished - it just needs the "Altitude Calculations" part. On the Hardware side, it's all good to go. If anyone is interested, let me know. Cheers. Edited October 23, 2020 by stix
stix Posted October 23, 2020 Author Posted October 23, 2020 Oh yeah. Please be aware that it's not like I've been working on this for the last 4yrs or so. I had other things to do.
stix Posted February 14, 2021 Author Posted February 14, 2021 The hardware part is coming together. Hopefully I'll have the final working version together in the next few weeks. It's almost become a pointless exercise, but I may as well try to finish what I started.
SeaMonkey Posted February 14, 2021 Posted February 14, 2021 Some of us are watching with great interest. Do not be dismayed. Your contribution has value. We are patient.
kramrocket Posted February 25, 2021 Posted February 25, 2021 I am too Stix, great work you have done on the device and software !!!
stix Posted February 25, 2021 Author Posted February 25, 2021 Thanks kram. I don't know why I keep saying things will be finished in the next few weeks. No chance!! The next few months is more likely. The hardware side has been developed and working as expected - this part just has to be put into an appropriate box with connection plugs (easy ). The software needs some basic altitude predictions which will take longer than I think.
stix Posted February 4, 2022 Author Posted February 4, 2022 Hello Stix, any update on this? Thanks!Thank you for your interest. Things are all good, and I will post images soon. 1
stix Posted February 12, 2022 Author Posted February 12, 2022 (edited) .I'm probably repeating myself, but for clarity there are three components to this solution: [1] Test Stand Apparatus (ver 2.0)[2] Electronics Module (SRS 50/100 Experimenter Series)[3] Software (SRS Impulse Recorder) ---------------------- [1] Test Stand Apparatus (refer to attached images) The original design has been modified to mount the load cell underneath the main plate. Therefore protecting the somewhat delicate load cell, without need for the clumsy brass covering from the previous version 1: [Ver 1] * The above design used an Arduino and ADC converter that needed an external power supply. ---- With the NEW design, force is now transferred through a hole in the main plate, via a 'transfer cylinder' attached directly between the mounting plate to the load cell underneath. Correct level (if desired) can easily be obtained by adjusting the front two knobs and referring to the bullseye level mounted top front and centre. [2] Electronics Module (refer to attached images) The 'Electronics Module' consists of a SparkFun OpenScale development board. The voltage from the load cell is amplified and converted to digital via an onboard ADC (analog to digital converter) chip (HX711). The module also contains firmware, which I wrote, that converts the incoming data to whole grams and outputs this data at 100Hz. Meaning the electronics module is capable of delivering 100 samples per second at a resolution of 1 gram (0.035 ounces). Not too bad!! The module connects to, and is powered via USB, therefore eliminating the need for an external power source. Connection to the load cell is done via a terminal block and wires are simply screwed in. I thought this was a better way to go because a dedicated 'connector' plug would be a real pain if you needed to change the load cell after a RUD moment - or alternatively, use any other load cell you like. Simplicity is the key here. Ultimately the electronics module will be mounted upside-down underneath the main plate at the front, therefore protecting it from a RUD and also ensuring that wiring integrity is maintained. [3] Software (refer to attached images) I think I've discussed this many times enough in this thread. The Software interprets the incoming data, and draws a graph based on the 'real-time' performance of the tested rocket motor. Other known data (launch weight, propellant weight) can be input (and adjusted), then performance & results are shown, the main interest being Total Impulse and Specific Impulse. The above is not the latest iteration, but you get the idea. FUN FACT: The software has a "replay" feature which allows you to 'replay' the rocket motor performance, and re-draw the graph in real time, over and over. No big deal, but it's fun to see because you most likely never get to see it that way. Interesting, but likely useless - but fun. ------------------- Further thoughts: Well, if anyone is interested I can post the Software on this forum. It's taken years to write and it's still not finished in the way I would like. Nevertheless it works great, but the reality is that it's too much effort for me to try and come up with some licensing/copyright structure and fee (iLok/Dongle) - which would really piss people off - including me. So I just have to let that part go. However, if anyone is interested in the electronics module, then there would be a cost. The electronics module combined with the software allows you to calibrate your load cell within the software which is very convenient if you have to replace your load cell. The integration of the electronics module and software should not be under-estimated. The test stand itself is just a manifestation of what "I want" a test stand to perform and look like. It's easy enough to get a load cell and knock up something very usable. BTW. If anyone has good ideas for a "fire-proof" type paint or coating, I'd appreciate your comments. Cheers. Edited February 15, 2022 by stix
Recommended Posts