Get Started Working with Circuitry: Questions Answered

[Zula110100100]

I want to get started working with circuitry and I have a few questions.

1) What is "arduino"? Is it a microcontroller with a few added goodies?

2) Should I start there, or would it be cheaper(or better learning experience, or possible) to start of scratch and get breadboards and a microcontroller and other goodies myself?

3) Let's say I had an analog sensor(accelerometer), how do I store it's output? Is it best to convert to digital, and even then how does one store/output the results?

4) With an analog signal does it lose quality from traveling over longish wires? Or is there no risk there(I might not need to store the output from sensor if I can have it go through a wire to a servo.

 

[yungman]

I want to get started working with circuitry and I have a few questions.

1) What is "arduino"? Is it a microcontroller with a few added goodies?

Never heard of this! Well, I'm old too!

2) Should I start there, or would it be cheaper(or better learning experience, or possible) to start of scratch and get breadboards and a microcontroller and other goodies myself?
You first want to define what you want and then look for a microcontroller that has the most peripherals that you need.

3) Let's say I had an analog sensor(accelerometer), how do I store it's output? Is it best to convert to digital, and even then how does one store/output the results?

I used Analog Devices ADuC831 micro-controller that has 8051 core with 8 channel multiplexed ADC and two DACs. It can interface with external static RAM easily and support SPI bus. That sounds like close to what you want. You digitize the analog signal, then store the data into the external RAM, then you can either sent it out in form of digital data, or manipulate the data and sent to a DAC to get the modified analog voltage.

4) With an analog signal does it lose quality from traveling over longish wires? Or is there no risk there(I might not need to store the output from sensor if I can have it go through a wire to a servo.

For these low speed control, you need to look at the input requirement of the servo devices. From my experience the input impedance is always quite high ( over 2KΩ). You don't need high current drivers. BUT if you drive through a coax, the capacitance might cause instability to the driving op-amp. The trick is to either find an op-amp that guaranty stability when driving a few hundred pF. OR if the input resistance is over 10K, use a 100Ω in series to the output of the op-amp before driving into the coax. You can use even higher resistance if you can program the software to compansate the voltage dividing effect of adding the resistor.

 

[Zula110100100]

Hmm.. Doing a little research and it seems I could maybe just use an analog datalogger, If that were the case would I really only need a powersupply, the accelerometer and the datalogger and nothing else?

The accelerometer I am looking at is Freescale MMA7361L XYZ-axis accelerometer, operates on 2.2-3.6v outputs 3 analog signals, one for each axis, and uLog 3-Channel Analog Datalogger, logic voltage is 3.3v, but it says "Recommended supply voltage: 3.3 to 16 V" Why such a wide range, and is 3.3v fine to run it on? If not I was considering using a 9v for power so would that be enough/not too much?

Anyway, with those two components and a power-supply, it should be good to go right? And the results could be downloaded to computer?

 

[yungman]

For recording a shock, I don't think the build-in ADC is fast enough. You might have no choice to have an external ADC. I worked on shock vibration before but I am not familiar with designing accelerometer. But I can tell you the reading of one that I worked with, the edges were fast, in nS rise time range. You are not going to capture with the build in ADC of those microcontrollers. Well maybe they have faster ones since I left the industry, I doubted though.

Transient Recorder or Data logger is what you need. I designed transient recorder in LeCroy before, that would be more like it. Doing an ADC circuit design with memory is not something that you want to do unless you have no choice...speaking from one designed before. Better to buy one if you can afford instead of spending months designing one. They can show you pre and post trigger to see what's going on and the circuit is quite extensive.

 

[Zula110100100]

This is for relatively slow acceleration, I am going to attach it to a barbell. In that case would that setup work? Also, forgive my lack of knowledge on this subject, but with an Analog data logger, why does it need an ADC at all? I thought that meant it could take the straight analog output and record it?[edit] just realized you meant the ADC on the data logger probably..then the question is just is that slow enough?

 

[yungman]

This is for relatively slow acceleration, I am going to attach it to a barbell. In that case would that setup work? Also, forgive my lack of knowledge on this subject, but with an Analog data logger, why does it need an ADC at all? I thought that meant it could take the straight analog output and record it?

There are different names for the same thing, there might be detail I missed, but we used both names. There is no reliable way to hold an analog signal indefinitely other than convert to digital signal, so you definitely need ADC.

Regarding to the speed, you are going to have to read data sheets. It is not just about the speed of your signal, it is also about the sampling window. Some ADC can sample a very narrow window ( like nS window) even though it can only digitize very slow. Some ADC has a much wider sampling window ( successive approximation type usually used in those build in ADC) and you need an extra sample and hold device to sample and hold the output to give the ADC a chance to digitize. Those ADC is not going to work if the input signal keep changing while the ADC performing successive approximation.

Those build in ADC in the micro controllers is mainly for reading power supplies, some very slow changing signals, I doubted it is good enough for you. But you have to read data sheets to know. What are you trying to do? If you don't know enough about ADC and such, how do you manage to get into this design? What setup?

 

[Zula110100100]

Well...overall I am trying to learn how to build circuits and such because I am currently a Music student but I am about to switch to Physics. This project I just want to attach an accelerometer to some weights(benchpress, curls whatever) and then record the results.

This may be too big of a first project, but I have messed around with resistors, transistors and capacitors a bit on breadboards before, so wanted to try something a little harder. I like to just jump right in.

I have been doing a little research on the subject and understand a little about some of the components. like the ADC, takes a voltage and depending on strength breaks it into a series of digital pules? Other than that I was hoping to get enough info from you guys and from the net, and from sticking stuff together and seeing what happens.

"The setup" would be using just the straight analog output from the accelerometer(0-3.3v) to the analog data logger(3.3v logic). You say the problem is the speed of the built-in ADC, but for something without sudden changes in acceleration(like just lifting weights, I don't think it gets going THAT fast) would the built in suffice? It sounds like you are saying no...in which case this might indeed be too complicated to start with.

 

[yungman]

It might be fast enough if you are just talking about weights going up and down...That is human speed measure in 10's of mS.

Assuming that you can use a micro controller like what I described. They don't usually come with through hole pins. Most I see are surface mount type. How are you going to breadboard? Also, the internal logics are quite fast that make breadboard not very good.

If you only has transistors, resistors on breadboard experience, you are going to have to learn a lot more. Assume you manage to build the board. Do you know programming? Know how to load the program into the EEPROM? Then you need to debug the circuits? Unless you have knowledge on these, it is going to be tough to do on your own.

What you are trying to do can be as hard as an undergrad graduation project or something like that. This is a complete embedded processor design together with some analog op-amp circuits. There is no other ways to put it. Are you ready for that?

Even for an experience engineer, this is going to be months of full time project from choosing components to build and test, writing programs and debug. You sure you want to commit to it? Also, you are going to need equipments to trouble shoot and they are not cheap. Good soldering iron for surface mount, scope...preferably digital scope, power supply, micro-controller development kid etc. There is no chance you can get it working one shot, you are going to have to trouble shoot.

I would choose something easier, something either just amplifiers for musical instruments, or a pure digital micro-controller project. I don't know anything about buying a kid to build and experiment as the stuffs I used were in the late 70s already. Go look for a micro processor kid and experiment with it. Learn how to do programming, upload into the hardware. Baby steps! You want to do this project all in one shot, you don't know what even goes wrong when it does not work.

Someone might want to jump in and give some suggestions. I am more an RF and analog guy. I designed all those embedded stuff just because nobody else can do it and I was the manager and have to pick up the slack...out of no choice. I don't even consider I'm the brightest bulb on the tree when come to these. I just did whatever to get the job done!

 

[Zula110100100]

I know how to program decently in Java(I have programmed a few simple 2d game, as well as some a very little bit of network programming in C(very little, just to get comfortable with sockets and the like), obviously I have never used "Wiring", I was under the impression you loaded it in with a micocontroller programmer. I don't expect to build it in one shot, and I know there is much to learn, I consider myself an intelligent individual capable of learning whatever I put my mind to. I really want to have a jump on a lot of this stuff before I get to that point. I guess perhaps I should just start with something more basic, it just seemed that since the acclerometer output an analog signal of a voltage that is in range for an analog data logger was simple enough to start with. So you suggest I start with just getting a microprocessor and programmer? What brand is good to start with PIC or AVR? Or something else? Is it AVR that has a language closer to Java with like "void setup()"?

Both of the components I listed above are with holes and .1" spacing...This is why I assumed breadboard would work...didn't know speed of logic was an issue.

To be clear I am talking about these two prefabricated components, not making an accelerometer and analog recorder from scratch...as I would have no idea where to begin. But if the analog data logger isn't good enough for an analog signal within it's range then I surely need to learn more and will be starting with something simpler

 

[Zula110100100]

The components are both on breakout boards with .1" spacing. Also..Should the two components be able to work without a microcontoller in between there is no programming to do, the logger checks to see if a UART is connected when it powers up, if not, it starts sampling, if so, it waits to either get a "r" for read, or "e" for erase...this is why I assumed it would be a fairly simple project.

 

[yungman]

If you have C programming before, that will help. I read the data sheet the chip has only x y z analog output according the the force. Look at page 5 of the data sheet, it shows exactly what I was talking about that you have to input the x y z to a micro controller to digitize them. But in real life, you almost have to have amplifier circuits in between that you have to design.

http://www.freescale.com/files/sensors/doc/data_sheet/MMA7361L.pdf

Also you are going to have to program the processor to continuous reading the data, stop when receive the trigger and save the data so you can read out. Then you are going to have to find a way to communicate to either a laptop or other devices using some serial communication. You have to think through all these.

I still think you should start a little smaller, just the processor project and get used to the programming environment. This is really out of my specialty. Someone what firmware experience should continue to advice you what kid to buy as I don't know much. After you get used to programming an embedded processor, all the loading of programs, debugging the program. Then next step is to do your accelerometer project so at that time, you concentrate on the A to D conversion.

This is just my opinion, someone might have better idea how you should approach this. Like I said, I was being forced to get into this, so I just did what I had to do to get it done and no more. I am not even interested in looking any deeper!

The accelerometer might be 0.1" spacing, the micro controller usually are not. You have to look for one that is 0.1 through hole processor to do breadboarding. To me, it is more important to find a popular processor core so the experience you learn can be used in the future. In my days, the popular one was 8051 core, look into a popular one first.

 

[Zula110100100]

I agree that I should get used to a processor in addition to this project, but I believe I no longer even need a processor for this, I can hook the accelerometer right up to the "uLog 3-Channel Analog Datalogger" It does everything for me after that. In the images of this data logger, it has .1" holes along two sides. You are probably still right about not using breadboard because of fast logic you mentioned, at: http://hlt.media.mit.edu/?p=1380 they are using male and female header connections, that seems like a good idea.

 

[yungman]

You might be right, I am out of touch already. If you can do away with the processor, go for it. That's getting easy.