DSPECT ... A development tool for metal detectors ( and other stuff )

[moodz]

I wanted a tool so I could look closely at the frontend of metal detector design. It turns out I had it on the bench right in front of me.

1. An FPGA board
2. A 24 bit ADC board.

BONUS .... No Soldering at all.

The hardware is about the same cost as a good PC soundcard.

So plug-em together and what do you get ... DSPECT ( a play on Digital Signal Processing ) and Spectrum.
Heres the blurb sheets

[moodz]

... finding those pesky mains harmonics with 0.3 Hz resolution.

[Aziz]

Hi Paul,

this is way way a much better tool compared to my hand written USB sound card tool.
Good job.
Aziz

[moodz]

... thanks and here is the actual GUI file that talks to the hardware. ( if you have the hardware )

It is just a single html file .. no apps / no webserver.

The only criteria is that is only built for Chrome or Edge browsers ( I didnt test it on Edge ).

If you want to play with it ... just unzip the attached file and open it in chrome and click demo on the top bar. ( usb wont work unless you have the hardware ).
The update rate and features in demo mode is exactly the same as the hardware.

I have not released the FPGA code yet .. .still adding features.

you can click and select on the spectrum trace as well as using the control panel.

If you want more features let me know.

[boilcoil]

... .The only criteria is that is only built for Chrome or Edge browsers ( I didnt test it on Edge )...

It works perfectly on Edge in demo mode.

[pito]

compare it with this ? Visual Analyser 26.10 no additional hardware need, just PC

[Aziz]

Hi pito,

compare it with this ? Visual Analyser 26.10 no additional hardware need, just PC

no, Visual Analyser is a quirk development software.
It can not be compared.
But it is useful for simple LCR meter measurements. On the other hand, there are way better measurement tools for LCR meters available.
Aziz

[moodz]

compare it with this ? Visual Analyser 26.10 no additional hardware need, just PC

...the main problem with soundcards is that the TX and RX signal generally cannot be phased locked at the sample clock rate. Then there is noise and bandwidth...the hardware I am using is 1.25 MHz ...that is fast enough for direct PI sampling. It only takes a couple of lines of code to make DSPECT into a high performance VLF detector.....mainly because that is what it really is.
In addition to that all of the DSP is done in the hardware not the GUI ...the GUI just displays it. The GUI is HTML based so it runs on just about anything...it's not windows locked. So mainly I built it because l needed very good performance to investigate metal detector frontends.

[Aziz]

Hi all,

this brief technical analysis summary lists some noise sources on USB sound cards (see AI PDF).

It doesn't list vendor specific noise sources of course.
On my Sound BlasterX G6 (@192 kHz SR), the upper frequency range (48 - 96 kHz) is completely noisy and practically unusable (up to +20 - 30 dB more noise).
This limits my frequency range up to 48 kHz.
Aziz

[KingJL]

... thanks and here is the actual GUI file that talks to the hardware. ( if you have the hardware )

It is just a single html file .. no apps / no webserver.

The only criteria is that is only built for Chrome or Edge browsers ( I didnt test it on Edge ).

If you want to play with it ... just unzip the attached file and open it in chrome and click demo on the top bar. ( usb wont work unless you have the hardware ).
The update rate and features in demo mode is exactly the same as the hardware.

I have not released the FPGA code yet .. .still adding features.

you can click and select on the spectrum trace as well as using the control panel.

If you want more features let me know.

There seems to be an anomaly with TX F... setting the frequency to 5.00K shows 5.00K in the TX F button, but shows 5.010K N=499 in the slider panel area... there seems to be no way to actually set it it to 5.00K N=500.

[moodz]

There seems to be an anomaly with TX F... setting the frequency to 5.00K shows 5.00K in the TX F button, but shows 5.010K N=499 in the slider panel area... there seems to be no way to actually set it it to 5.00K N=500.

You can enter exact values using the keyboard rather than the slider ( arrow keys should adjust also ).

The TX F slider moves in 1,000 logarithmic steps across N = 4…8191, about 0.77% per step. Around 5 kHz, N=499 and N=500 are only 0.2% apart, so the slider can’t land on 500. If you type 5k ((on the keyboard), N=500 is set correctly and the tile shows it. The line above the slider is the bug: it recalculates N from the slider’s rounded position instead of showing the real value. TONE F and SPAN have the same rounding in that line.

I’ll fix both:

When the slider hasn’t been moved, the line shows the actual value.
The TX F slider gets 20× finer resolution, and the arrow keys step N by exactly ±1 (Shift ±10).

The corrected file is attached.

[moodz]

The next rather obvious step is just to allow DSPECT to connect to the sound card in the PC as an option.

So here it is .... if you select "SOUND" in the top menu it will popup a selection window. Default works most of the time.

Again its all running in the browser so should work anywhere ( I tested linux, windows and MAC ) but apparently it will run on Android too ( but the screen will be very tight ).

Also this version has a lockin amp built in for coherant demod. So that means you can use sound out as TX in a metal detector that needs coherant demod .. maybe I have not tested it extensively and soundcards prob dont make good detectors

The screenshot below shows the very poor motherboard soundcard in my linux box with the sound card popup select nowhere near the performance of the AD7760 ...
The attached zip contains the browser file. Just unzip and open with Chrome or edge ... others may work.

[moodz]

using a better soundcard ... an arturia soundfuse 2 ... the sampling rate is 96khz so we have a bandwidth a bit south of 48 Khz and a noise floor of -135 dbv... enough for MD work

[moodz]

Upto about 48Khz the soundfuse 2 usb soundcard is about the same performance of the AD7760.
Below we zoom into a 100 hz span to detect the 5 khz transmit carrier which is on but we just put a wire near the Tx and Rx so a bit of leakage carrier is seen at 5Khz. ( ie connected by air )

The lockin amp is doing its job as we see the noise floor fall to south of -150 dbv. We are on maximum zoom so each trace takes 20 seconds but the resolution is .046 Hz .... I am impressed.

no good for realtime metal detectors but great for chasing down shielding and switchmode noise issues.

[Aziz]

Hi Paul,

nice App.

For real-time metal detecting approx. 100 times / s measurement is ok. Even 50 times/s is ok (slow motion version).
Instead of Lock-in amp demodulation, Goertzel will do it slightly faster.

With an ultra-low-noise amplifier, the USB sound card version will do a good job.
Aziz