SPDR ( SURF PI DIGITAL REVISION )

[waltr]

Hi Moodz, how did the house move go.

What software tools are you using for the PIC32 programing?

[moodz]

Still unpacking ... I use a pickit3 and xide. All coded in C. The pic hardware is brilliant for generating all the timing. This chip I'm using ATM has 5 pwm for the TX ,Rx and audio.

[waltr]

Still unpacking ... I use a pickit3 and xide. All coded in C. The pic hardware is brilliant for generating all the timing. This chip I'm using ATM has 5 pwm for the TX ,Rx and audio.

thanks and the unpacking never seems to end.
What I meant is the tools to write and compile the code. Guessing MPLAB X with C32 compiler.

I have been using mid-level PICs (12F, 16F & 18F) for years which is what is running the my Hammer Head based detector.
Have been working on theory for improvements of signal determination so looking at PICs with Faster ADCs and processing.
Would not mind using a PIC that others here use so easier the share code. I just do not have a compiler for these yet so looking to see what you are using and/or recommendations.

[eclipse]

You can give it a try with MPLABX 5.10 with XC32 v2.10 and Harmony 2.06 + Pickit3.
I'm using this with PIC32MZ and it's good enough for timing and ADC/PWM sound generation.

[Mdtoday]

thanks and the unpacking never seems to end.
What I meant is the tools to write and compile the code. Guessing MPLAB X with C32 compiler.

I have been using mid-level PICs (12F, 16F & 18F) for years which is what is running the my Hammer Head based detector.
Have been working on theory for improvements of signal determination so looking at PICs with Faster ADCs and processing.
Would not mind using a PIC that others here use so easier the share code. I just do not have a compiler for these yet so looking to see what you are using and/or recommendations.

Hi Waltr,

I've been using the DSPIC33EP512MU810 for about 5 years and a combination of MikroC compiler for dspic (for legacy designs) and XIDE and version of C32 compiler for later designs.
I avoid using non transferable libraries such as those in Mikro C unless its a one off and i get lazy, so have built up quite a few of my own along the way..for external and internal peripherals.
Mind you, there is a lot of support and users out there now with Mikroelectronika compilers so its not a bad option at $250
The Microchip IDE is kind of split a bit as standalone or cloud based with the introduction of Harmony but the Microchip option in my opinion, is still a good way to go.
As for the particular chip, I am actually in progress of swapping designs over to use the same PIC32 for the reasons both yourself and Moodz have said. PWM and timers are great in both DSPIC and PIC32 though
Basically going from a 100 pin chip to 28, so happy about that too..and making more sensible use of both analog and digital circuits has helped make that move even easier.

cheers

Mdtoday

[waltr]

Thanks everyone for the replies and suggestions.

Years ago when MPLAB X came out I installed it and it broke the older MPLAB so avoided it. MPLAB 8.6 worked perfectly for the PICs I was using.

Last night I decided to give X a try again and downloaded it and the XC compilers. Hey, It seems to work fine and not destroy the legacy MPLAB, yea.
Then I found some dsPIC30 source code Moodz posted in another thread (thanks Moodz) so created a project to compile this as for learning to tools. All looking good so I think I have the solution.

I do like the 28 pin DIPs and seems almost any of the PIC24, PIC32 or dsPICs will work well for a detector I'll buy a few of the PICs Moodz is using.

[waltr]

First: The new IDE and XC compilers from Microchip work and I have learned how to use them. Also now have a C code project for the PIC32MX250F128B chip that Moodz is using here.

Next: Learned what Moodz is using in the PIC32 to generate the TX & sampling pulses. That is the OC (Output Compare) modules in continuous pulse mode. I think it would be good for anyone here that wants to use Moodz MagPi code to know how this works (without too much detail).
1- A Timer runs and is reset when matching the PR compare value. PI cycle period. This time is used be all OC modules so timing of all outputs are coherent.
2- The OCxR value causes a Rising Edge at the output pin when Timer matches this value.
3- The OCxRS value causes a Falling Edge at the output pin when Timer matches this value.
4- These two values are from the Start of the Cycle (Timer = 0). So any output edge can be anywhere within the cycle period.
5- If OCxR < OCxRS then output is Low idle and pulse is High (the way Moodz shows on the MagPI document).
6- If OCxRS < OCxR then the output is High idle and pulse is Low (this is what my HH2 uses).
These OC modules puts all the timing into Hardware and allows software do do other things. This also allows for a very easily configurable pulse timing.

Now for understanding some of the SDR001 schematic (pdf in post #6 of this thread).
Moodz has three sampling outputs, S1, S2 & S3. S1 puts the pre-amp into the V- diff amp.
S2 & S3 analog switches do the same thing, puts pre-amp into the V+ diff amp. When I was studying the schematic I wonder Why do that?
Well the answer is partly in how the Sample pulse are generated. The OC modules can only create One pulse per cycle. But most PI circuits require at least a third sample for EFE. This is what S3 can be used for.

Moodz,
Neat design but I think there are a couple issue you missed.
Lets assume we wish to do a Three Sample scheme that does EFE and GEB.
The Per-amp inverts so First sample, S1, goes to V- input of Diff-amp.
Next Sample, S2, goes to V+ of diff amp. So far all good.
GEB will be a longer S2 sample time. Still good.
However, the counter the longer GEB time and do EFE we need a sample into the V- of the diff amp just before TX pulse starts.
The S3 analog switch can not do this (however, there may be a sampling scheme that needs a second V+ sample).

There is an unused analog switch and a fifth OC module as well as unused pins. If would be good if these are used for an S4 pulse.
An option would be to Jumper select which analog switch S3 goes to. Then the experimenter can select U4B or U4A for S3.

I do like the idea of a universal processor/code (MagPI) for an experimental platform. The way you have the interface that allows easy changing of timing is great and way easier than rewiring 555 timer chips.

Last: Why the PIC32MX2xx chip with USB?? There are PIC32MX1xx chips without USB and have two more usable pins.
With small pin number chips the challenge is to adjust what resources are used and which pins can be used.

[moodz]

Fair analysis...i used the USB chip with the idea that it could be used to update the firmware without a programmer .... The non USB chips are good too.
Now i will give you a teaser .... There is a fourth sample pulse available to cancel EF and other interferance like mains hum ... It's the TX pulse itself .. but not during tx.

[waltr]

Fair analysis...i used the USB chip with the idea that it could be used to update the firmware without a programmer .... The non USB chips are good too.
Now i will give you a teaser .... There is a fourth sample pulse available to cancel EF and other interferance like mains hum ... It's the TX pulse itself .. but not during tx.

Ok thanks. Always interested in the reason for a design decision.
More to think about.

Here is my teaser. No analog integrator and TX ON sampling.

[Mdtoday]

In my designs I chain timer interrupts with EF and first sample triggered from TX pulse,followed by rest in the chain. Each timer can be proportional to other timer settings or fixed. as to sampling switch, a break before make SPDT from front end amp output to either side of integrator.. using dspic. The front end uses a simple blanking switch...

The other way I do timing is use a tick timer and event counters on devices without enough resource.... horses for courses I guess

cheers

Mdtoday

[moodz]

Ok thanks. Always interested in the reason for a design decision.
More to think about.

Here is my teaser. No analog integrator and TX ON sampling.

Ok so each sample channel has its own ADC ( or muxed ADC ) and the tx on sampling is not with a monocoil ? ( ie needs a tx / rx coil of some sort ).

Ref my teaser .... the tx modulates the unwanted signals like EF , hum frontend noises etc .... so the solution is to coherently demodulate the unwanted modulation and subtract from the input leaving only the target signals using the same carrier signal ( ie the TX waveform itself ) ...thus the fourth analog switch is used to do this.

[waltr]

In my designs I chain timer interrupts with EF and first sample triggered from TX pulse,followed by rest in the chain. Each timer can be proportional to other timer settings or fixed. as to sampling switch, a break before make SPDT from front end amp output to either side of integrator.. using dspic. The front end uses a simple blanking switch...

The other way I do timing is use a tick timer and event counters on devices without enough resource.... horses for courses I guess

cheers

Mdtoday

Yep, I've done it both way depending on which processor I'm using.
The PIC32MX has Output Compare hardware that do all this. These are pretty versatile and once setup they create the pulses without software intervention.

Ok so each sample channel has its own ADC ( or muxed ADC ) and the tx on sampling is not with a monocoil ? ( ie needs a tx / rx coil of some sort ).

Yes, an almost balanced IB coil. I have some details in my HH2 thread on the coils (a concentric and a DD).
https://www.geotech1.com/forums/show...ake-on-the-HH2

I have a few PCBs I designed for this PI detector and have been experimenting with TX on sampling. Post #92 in my thread discusses idea for TX on sampling and Iron discrimination.

Ref my teaser .... the tx modulates the unwanted signals like EF , hum frontend noises etc .... so the solution is to coherently demodulate the unwanted modulation and subtract from the input leaving only the target signals using the same carrier signal ( ie the TX waveform itself ) ...thus the fourth analog switch is used to do this.

Sounds intriguing if not a Bipolar TX (which is another way to do EFE and reduce noise).
I have to think on this idea.

[moodz]

Below is the root of the modulation problem ....

A 50 hertz magnetic hum is present in the vicinity of the coil in the spectrum plots shown below ( blue traces ). The signal is measured at the o/p of the preamp. The 50 hertz primary can be easily removed with a pass filter or even fancy dsp .... the 50 hertz modulation present on the spectrum representing the target signal ( starting at 10 khz in this instance ) cannot be so easily removed. The green trace is the same signal but with TX sampling enabled ... no Hum, No EF , No noise ....no modulation.




I have used 50 hertz here to illustrate the problem ... however any low frequency noise including EF will be modulated onto the target spectrum.