Arduino + White's Spectrum XLT tinkering

[daniel]

Hi guys,

Most of you seem to be pretty comfortable with hardware, but I'm more of a software guy. Because of this, I'd like to use my Arduino [1] to "listen in" on the TX and RX signals on my old White's Spectrum XLT, send these signals to my desktop computer, and learn a little more about phase shifts where I'm most comfortable -- in software.

I've ordered an extra set of Switchcraft slim-line 5-pin connectors and created my "wiretap", but I'm a little worried about a couple of things:

1. What sort of voltage will I see on these wires? The Arduino can only take up to 5 volts in the analog inputs. Will it be more than that? If so, how can I scale it down?

2. The Arduino analog inputs are converted to 10-bit values (a range of 1024 values). I would think this will be granular enough to run some analysis on the signals... does anyone see a reason why it wouldn't?

3. What's the frequency on these signals? The maximum baud rate back to the desktop computer appears to be 115,200 bps. With 10-bit values, that's 11,520 values/sec. If the actual signal frequency is something like 10 kHz, we might be talking ~1 sample per oscillation... not quite enough :-)

If anyone else has used an Arduino to play with their detectors, please let me know!

Take care,

Daniel

[1] http://arduino.cc/

[Carl-NC]

Do you have an oscope? You really should start there. Directly sampling the coil signals won't net you much useful info without some pretty heavy processing, which requires both a hefty processor and knowledge of exactly what kind of processing you need to do. That knowledge is what you are trying to acquire, so there's a bit of a Catch-22.

The XLT signal is 6592 Hz; you need to also look at the ADC sample rate, which I think the Atmel processor only runs about twice that rate. So not nearly fast enough. Direct sampling requires a fast ADC and a Very Fast processor.

My suggestion is, start getting comfortable with hardware. You really can't bypass that step.

- Carl

[daniel]

Hi Carl,

Thanks for the feedback! I don't have an oscope, but I'll be watching Craigslist for one... or do you have any better suggestions for finding one?

Regarding sampling rates, I've run some experiments and I can get my Arduino's sampling rate up to 52 kHz with some tweaking (see https://github.com/gredler/arduino). Do you think that would be enough to sample the XLT's 7 kHz signal?

Take care,

Daniel

[Carl-NC]

I've bought oscopes off ebay, but you really need to be careful and know what you're looking for. The old Tek 475 are rock-solid analog scopes and generally cheap.

Quadrature sampling (4X) is a minimum, 16X or better is recommended. I personally haven't designed a direct-sampling VLF detector so I can't advise specifically, but the higher oversampling lets you do more filtering and DSP. That then dictates processor requirements.

[target]

build your own scope here ,or theres numerous small scopes around cheap

http://www.sparkfun.com/products/9484


olimexino has more power than arduino

https://www.olimex.com/dev/olimexino-32u4.html


arduino scope too

http://www.pcworld.com/article/22838...illoscope.html

[target]

heres a mini scope that sits on your breadboard

http://www.youtube.com/watch?v=Q22tB7C-bMg

[golfnut]

For a cheap processor with a repsectable sample rate take a look
at the MSP 430 range I think these often sample at 200k as a dedicated unit so you tell it the mode, and it writes the val to ADC10MEM for you.

You can get a usable dev platform called a launchpad.

http://processors.wiki.ti.com/index...._(MSP-EXP430G2)

You get
usb lead
solderable headers
32k watch xtal
processor
a spare proc
works at 3v - so can use a cell phone LCD a Nokia 5110 is a good choice from Adafruit
free IAR ide code limited. Code exampes etc.

usefull kit for $4

MSP430G2553 is a good start 10bit

::S

[hdphilip]

go with the chipkit, It works!

anything less, the math is too much to process,
anything more, the support is less.......

Philip

[mario]

Come on guys, lets be honest.

Daniel doesn't need any DSP filtering/processing. He simply wants to sample an ADC signal and feed the signal to a PC via UART for processing. In this case, hardware filtering is more than adequate to clean the signal, so you don't need a complicated software filter.

You're going to seriously tell me he needs a speedy processor for something this simple?

Daniel, assuming you're using an Atmega328P, you're not likely sampling at 52khz. The highest rate will likely be around 15khz (200,000Khz/13 IPS = 15,384khz). You do not need a fast processor for this purpose because the ADC clock is independent of the processor, but you do need a fast ADC component .

I'm not familiar with your sampling experiments, but have heard about certain overclocking techniques to boost that rate.

Anyway, there's no reason why your goals shouldn't be reached. At 15khz, you're twice the 6.6khz required for minimum sampling, so try it out and let us know.

Take the advice of these guys with a grain of salt. Most haven't actually done anything with the hardware in question, so shouldn't be discouraging you or other members from trying something they don't really know much about. Like Carl said, become familiar with your hardware first.

One last suggestion. If you're using the Arduino, you're not likely going to get a stable 10bit (all 1024 values) ADC solution because of jitter (for instance, your value will likely vary +-3). You're going to need an additional low-pass filter to get a rock-solid ADC value. To do so, you need to implement an external reference voltage source. It seems like the Arduino folk's implementation of ADC is limited to an internal reference voltage source, which gives you the -+3 jitter to lower the component costs on their board. In any case, a -+3 jitter may be irrelevant for your purpose

[hdphilip]

mario.

The XLT signal is 6592 Hz; you need to also look at the ADC sample rate, which I think the Atmel processor only runs about twice that rate. So not nearly fast enough. Direct sampling requires a fast ADC and a Very Fast processor

i agree with Carl on sampling rate..... and thats being

honest

Daniel,

The maximum baud rate back to the desktop computer appears to be 115,200 bps. With 10-bit values, that's 11,520 values/sec.

115200 is for 8 bit.. correct me if I'm incorrect....

Philip

[moodz]

the journey or the destination ?

Daniel .. if you only want to sample the signals you already have the tools ... the DSP engine is your PC and the ADC is your sound card .... if you are a software guy you will find everything you need if you google something like DSP and soundcard or soundcard Oscillscope etc. There are thousands of software both source and apps to explore every facet of VLF type signals ( phase, waveform, dft ,fft, filter, transform etc etc ) this is the DESTINATION. eg scope ...http://www.zeitnitz.de/Christian/scope_en/

If you want to learn about arduino then learn arduino ..."then" find out it wont do VLF all that well mostly for some of the reasons the guys posted above. This is the JOURNEY.

If you are a software guy and want to get to the ducks guts then the DESTINATION is for you ( ie soundcard + DSP software )

If you want to sail to china in a row boat then the Arduino is for you ... the JOURNEY (not that you wont have some fun and unless of course you are already in china. the Arduinos are probably cheaper there LOL. ).

moodz.

[mario]

mario.



i agree with Carl on sampling rate..... and thats being

Daniel,


115200 is for 8 bit.. correct me if I'm incorrect....

Philip

I'm not sure if your hardware defies the laws of Nyquist, but as far as I know Nyquist is a basic principal that applies to all types of these conversions. 15khz sampling should be adequate for a 7khz signal. I agree about more being better, but processing will be done in the host CPU.

I will concede on UART being an issue. At 15k sampling, 115kbps UART is too slow.

If Daniel still wants to follow on this project, he needs something faster.

I know the Leonardo has a high speed USB interface if he wants to remain with the arduino platform.

Also, I can confirm the atmega adc overclocking to a 30k sampling. With the right components, it can be done.

[mario]

Daniel .. if you only want to sample the signals you already have the tools ... the DSP engine is your PC and the ADC is your sound card .... if you are a software guy you will find everything you need if you google something like DSP and soundcard or soundcard Oscillscope etc. There are thousands of software both source and apps to explore every facet of VLF type signals ( phase, waveform, dft ,fft, filter, transform etc etc ) this is the DESTINATION. eg scope ...http://www.zeitnitz.de/Christian/scope_en/

If you want to learn about arduino then learn arduino ..."then" find out it wont do VLF all that well mostly for some of the reasons the guys posted above. This is the JOURNEY.

If you are a software guy and want to get to the ducks guts then the DESTINATION is for you ( ie soundcard + DSP software )

If you want to sail to china in a row boat then the Arduino is for you ... the JOURNEY (not that you wont have some fun and unless of course you are already in china. the Arduinos are probably cheaper there LOL. ).

moodz.

Yup, soundcard is all he needs, and way faster too!

[hdphilip]

the sound card is the perfect solution...........and fun to!

just be careful about were your "probing" your soundcard with respect to your coil


Philip