Products > Crowd Funded Projects

PlainDAQ - open source DAQ module for Raspberry Pi Pico

<< < (13/13)

palpurul:

--- Quote from: dunkemhigh on August 26, 2022, 11:46:45 am ---You were going to put the ESP-Wroom-32 in for WiFi, right?

--- End quote ---

Well yes, it was the plan in the past, but right now I've got some e-mails about removing the Wi-Fi module because people can use raspberry pico W for it.


--- Quote from: dunkemhigh on August 26, 2022, 11:46:45 am ---But that also does Bluetooth so you could just slap it in and have whatever anyone wants (both at the same time if it rocks your boat).

--- End quote ---

There aren't that many people wanting to have bluetooth module on it really :(

EsPiFF:

--- Quote ---Actually yes you can. The ESP32-C3 chip is about $1. Count a few passives around this an a trace antenna on PCB, and done.
--- End quote ---

... after a 5 digit sum, to get your custom design certified, if you like to sell it. If you use a tm name like Bluetooth without a certification, to advertise it, the lawyers get after you.

When this project should show only technical possibilities, without an commercial background, for others to learn from or integrate the open source design into there solutions: then you don't need to think on safety or certification.

But when this project is intended to make a business from it (Crowd Supply) and/or should be used to teach and be used, for example by children, then safety standards REALLY should be implemented.

Here some hints:

* your github repro does not contain schematics in PDF, but native KiCad files. That makes it difficult for people to help you, by looking over the schematics. The repro got updated last month, so my guess would be thats the latest HW version.
* you should begin your design, by plan to fulfill safety regulations. Its not expensive, nor difficult, to design the product according safety regulations. This safes our children, and enables you to make money from selling products. I see, you spend much time with your own software, but over several PCB releases, still have unprotected inputs. I would make the device compatible with a standard like Sigrok, (or Labview, or GNU Radio, or Octave). They have all the features you will need years to implement by yourself. But that is your choice, you seem to have fun with it. But still: you need to implement at least the basic safety standards.
* your inputs are connected, with a 100 Ohm resistor, to the DG611 multiplexer. More then + or - 18V will kill this IC. If 220V comes on the input, smoke comes out in best case, and a splitter of an exploded resistor fly to the eye of a bystander and he is blind. Somebody suggested a 220k resistor in series with every input, what is a good start. That is a good start. But keep in mind, that the resistor must withstand the high voltage. A standard 0603 resistor may be spec`ed for only 50V. So use ether a high volt resistor, or connect multiple resistors in series. Safety standards call for 1000V. I would add at least a pair of TVS diode after the 220k resistor, to clamp the voltage to save areas  for the DG611, +8V/-8V 
* Input compensation: not required by law, but the 220k ask for compensation at higher frequencies. 10nF parallel to the 220k was suggested, but this depend on many factors, and should be measured for best results.
You should also improve your design, to not disappoint your backer. I doubt, that the current design in your github can reach the claimed Values.

* It begins, before the signal even reach the terminals. You are talking about 450kHz bandwidth, but unshielded terminal inputs? Why no coax? BNC, SMA, or audio chinch? Not even balanced input. Even Audio with its max 20kHz, need shielded cable and ether coax connector, or balanced+shielded cables and connectors. Without this, you hardly get 8 bit. So use BNC or SMA connectors
* Channel separation: you need ground between the inputs, or cross talk with ruin your 12 bit. With coax connectors, and a good layout, your can reach your planned specs.
* Your power supply concept is unpractical for an 14 bit ADC. External 5 V from USB is noisy, and the switching regulator to generate the -5V add even more poison. You could use ether specific low noise switching regulator from Linear Technology(now ADI), but they are costly. Or use low noise linear regulators after the switching regulators. I often use the Richtek RT9193 series: they are ultra low noise, have a high PSRR not only at 100Hz, but also up to a few 100kHz, and still low cost. But they are only up to 4.75V available, and only positive. TI makes special ultra low noise linear regulator for positive and negative voltages, but they are not cheap. Also ADI and other make them. Just look in the datasheet for the PSRR over frequency plot. And of course, the output noise. The RT9193 has, for example, only 100 uV RMS noise.
* Reference voltage: Why do you use a 4.096V reference, and then divide it by 2? And not use a 2.048V reference IC, like the LM4132AMF-2.0, or MAX6071AAUT21+T? The voltage divider have its own tolerances, temperature drift and will ruin your specs. 
* AD8137WYCPZ-R7... why an 110MHz GBW, 450V/┬Ás slew rate amp on this position? Such a beast needs extreme care to not oscillate, and is expensive. I would suggest you something like the MAX4462: It cost only 1/3th, has the same low Input voltage offset, easier to handle 2.5MHz GBW. But what ever op amp you use there, you should not supply it with the dirty USB-5V.  USB as power supply is good for digital things. To use it for analog parts with a high requirement on accuracy, you need to careful clean it up. Or galvanic isolate it. In your schematic is not a single ferrite bead, or common mode choke. Not even a linear regulator for the op amps. 
* My advice: You have a working prototype, that is half of the product. Make a concept for safety according the regulations, and a concept for a clean power supply. You need minimum low noise RF linear regulators, separate analog and digital grounds, ether a star ground or galvanic isolation between the analog and digital parts. Specify the PSRR - over -frequency of each regulator, and each opamp. Also study the impedance graphs from different ferrite beads and common mode cokes, to see where they can help.
* Even you like your own software solution: When you want this project to become a commercial success, or more attraction in the open source world: consider to add compatibility to standard software: the RP2040 supports tinyUSB, and tinyUSB support USBTMC.
https://www.eevblog.com/forum/projects/usbtmcusb488-class-implementation-for-microcontrollers/
https://k1.spdns.de/Develop/Projects/pico/pico-sdk/lib/tinyusb/examples/device/usbtmc/
As beautiful your own software is: you will get 10x more attention, when backers can use Sigrok with your plainDAQ.
Your users can then choose between your software and standard software.
 
* If you want, that people help you: put PDF schematics on your github, then they can help you. Otherwise your on your own. 

palpurul:
Hello I am the guy building the PlainDAQ, unfourtunately I had to take a huge brek because my main job won't allow me to work on it :(


--- Quote from: EsPiFF on October 29, 2022, 08:11:33 pm ---... after a 5 digit sum, to get your custom design certified, if you like to sell it. If you use a tm name like Bluetooth without a certification, to advertise it, the lawyers get after you.

--- End quote ---

PlainDAQ won't include any bluetooth and wifi capability I moved on from it because there is a version of pico with Wi-Fi already.


--- Quote from: EsPiFF on October 29, 2022, 08:11:33 pm ---your github repro does not contain schematics in PDF, but native KiCad files. That makes it difficult for people to help you, by looking over the schematics. The repro got updated last month, so my guess would be thats the latest HW version.

--- End quote ---
I will upload it tomorrow. The hardware is not the latest I will have to make a few changes TBH.


--- Quote from: EsPiFF on October 29, 2022, 08:11:33 pm ---you should begin your design, by plan to fulfill safety regulations. Its not expensive, nor difficult, to design the product according safety regulations. This safes our children, and enables you to make money from selling products. I see, you spend much time with your own software, but over several PCB releases, still have unprotected inputs. I would make the device compatible with a standard like Sigrok, (or Labview, or GNU Radio, or Octave). They have all the features you will need years to implement by yourself. But that is your choice, you seem to have fun with it. But still: you need to implement at least the basic safety standards.

--- End quote ---
Safety regulation issue is of the area that I am not really capable of, how should I get started to make something compliant. I haven't so much thought about making it compatible with already existing standarts, but it's a great idea, I was rushing to et something working, so I haven't had a chance to take a deep look into those issues.


--- Quote from: EsPiFF on October 29, 2022, 08:11:33 pm ---your inputs are connected, with a 100 Ohm resistor, to the DG611 multiplexer. More then + or - 18V will kill this IC. If 220V comes on the input, smoke comes out in best case, and a splitter of an exploded resistor fly to the eye of a bystander and he is blind. Somebody suggested a 220k resistor in series with every input, what is a good start. That is a good start. But keep in mind, that the resistor must withstand the high voltage. A standard 0603 resistor may be spec`ed for only 50V. So use ether a high volt resistor, or connect multiple resistors in series. Safety standards call for 1000V. I would add at least a pair of TVS diode after the 220k resistor, to clamp the voltage to save areas  for the DG611, +8V/-8V

--- End quote ---
I will add proper input protection, noted. THe new version had inputs directly connected to the opamps not the multiplexers, I had to use cheaper multiplexers because of the price and they are 5V rated.


--- Quote from: EsPiFF on October 29, 2022, 08:11:33 pm ---Even you like your own software solution: When you want this project to become a commercial success, or more attraction in the open source world: consider to add compatibility to standard software: the RP2040 supports tinyUSB, and tinyUSB support USBTMC.
https://www.eevblog.com/forum/projects/usbtmcusb488-class-implementation-for-microcontrollers/
https://k1.spdns.de/Develop/Projects/pico/pico-sdk/lib/tinyusb/examples/device/usbtmc/
As beautiful your own software is: you will get 10x more attention, when backers can use Sigrok with your plainDAQ.
Your users can then choose between your software and standard software.

--- End quote ---
That's a glden advice for me and I'd like to really thank you for this suggestion. At this stage I don't even know what Sigrok is, but I will learn it.

The technical addvices you gave me are great. I am trying make it as noiseless as possible and as affordable as possible at the same time, trying to optimize both is not easy as you know :)
14-bit version is out of question at the moment I am only building the 12-bit because 14-bit is too costly for my price range. I will keep your advises in mind when building the new revision.

Thanks!


Navigation

[0] Message Index

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod