NEW !!! Full differential PI front end.

[moodz]

Nice work moodz! This project is really coming along!

I had looked into FPGAs a while back, but it seemed very cost prohibitive for the general hobbyist. Watching your work with them has renewed my interest.

BTW, are you programming in Verilog or VHDL? I've learned Verilog to some extent, but have not had any experience with VHDL to this point.

Thanks hobbes ... FPGA boards and tools are alot better and cheaper than they were a few years ago. You can certainly get cheaper Microprocessor development boards however you cant do the sort of real time things that the FPGA can do .... there are boards from Digilent and Avnet that for about $75 that are perfectly adequate. Coding is no more difficult than a micro IMHO and might even be easier if you are used to building stuff in the old days from discrete logic like TTL and CMOS. I found it harder and longer to set up the Microchip PIC IDE and get coding than the FPGA development ISE from Xylinx. ( Have not tried Altera but it looks similiar )
This project uses VHDL which has a more structured format than Verilog .. however I have built up big microprocessors in the past using Verilog and it is easy to convert code from one to the other automatically or even use mixed code modules.

Moodz.

[hobbes_lives]

Nice work moodz! This project is really coming along!

I had looked into FPGAs a while back, but it seemed very cost prohibitive for the general hobbyist. Watching your work with them has renewed my interest.

BTW, are you programming in Verilog or VHDL? I've learned Verilog to some extent, but have not had any experience with VHDL to this point.

[moodz]

FPGA progress

.... The RX filter banks, the oscilloscope function and the TX pulse functionality are operational. Have not built frontend yet so the shot below shows the output from the chip with no coil / frontend connected. The purple trace is the unfiltered input the white trace is the output from the digital filter ... I no longer need my Tek CRO

Note the whole display is generated by about 6 digital outputs on the FPGA and few resistors...... basically a 'free' scope.

The TX pulse ( yellow trace ) is the MOSFET driving pulse and is 127 uS duty cycle in 512 uS.

Sorry about the flash on the screen but it takes a much clearer picture of the traces with the flash on ... must be the lighting in my 'lab'

moodz

[moodz]

Hi Paul,

Very nice... Lately just started tracking this thread.
Any success on field trials reg ground balance system using bi-symmetric differential drive system.

Cheers,
Michael.

Hi Michael,
some success ... however I am moving the whole design into an FPGA backend. I have teamed up with Tinkerer to do move the analogue processing into digital processing. I hope to do field testing shortly.

Best Regards,

moodz.

[revmaster]

One of the problems with sensitive PI detectors is earth field. If you wave a coil through a static field ( eg the magnetic field of the earth ) a slight voltage is induced that is proportional to the rate of change of flux through the coil. eg at type of generator. This leads to false target detection. However if the PI is driven by an alternating Tx pulse the receiver will see two things ...

1. Targets will respond with an alternating field as the eddy currents alternate in track with the stimulating tx pulse.

2. The earth field will produce a steady ( like DC ) response depending on the motion of the search head.

The only thing that would fool this would be if you sweep the search head at the same frequency as the alternating Tx pulse. ( if you drink lots of coffee you might be able to reach a 1000 Hz sweep speed )

The other benefits ? .... have you heard of chopper stabilised amps ? ... read up .... if you want to measure really small voltage changes like in a PI this definitely is better.

Coupled with the differential coil this system changes the PI landscape forever .... bisymmetric differential drive system schematic is published below ....

... wait till you see the ground balance system I have devised ... more soon.

Moodz

Hi Paul,

Very nice... Lately just started tracking this thread.
Any success on field trials reg ground balance system using bi-symmetric differential drive system.

Cheers,
Michael.

[moodz]

Someone I know did some work with the Digilent board and then moved to this 3A board for $49:

http://www.em.avnet.com/evk/home/0,1707,RID%253D0%2526CID%253D46501%2526CCD%253DUSA%2526SID%253D32214%2526DID%253DDF2%2526LID%253D32232%2526PRT%253D0%2526PVW%253D%2526BID%253DDF2%2526CTP%253DEVK,00.html

It doesn't have the display but way cheaper...
Also another friend mentioned an Actel Igloo Icicle board for $49 that may be lower power.

Good find ! ... only look through Avnet now and then .... this board would make an excellent PI controller engine ... it hasn't got much RAM though for bigger apps.

[bklein]

Someone I know did some work with the Digilent board and then moved to this 3A board for $49:

http://www.em.avnet.com/evk/home/0,1707,RID%253D0%2526CID%253D46501%2526CCD%253DUSA%2526SID%253D32214%2526DID%253DDF2%2526LID%253D32232%2526PRT%253D0%2526PVW%253D%2526BID%253DDF2%2526CTP%253DEVK,00.html

It doesn't have the display but way cheaper...
Also another friend mentioned an Actel Igloo Icicle board for $49 that may be lower power.

[moodz]

Are there a down-loadable FFT module for FPGA?

Have a look here ... lots of stuff .... http://www.opencores.org/projects



Moodz

[WM6]

Seriously though if you are an old hand you would remember the Z80, 6809, 68000 etc .... these are all available precodet

Are there a down-loadable FFT module for FPGA?

[simonbaker]

I will post a cheeky reply .....

A microprocessor is a pre-programmed FPGA.

Seriously though if you are an old hand you would remember the Z80, 6809, 68000 etc .... these are all available precoded for FPGA platforms ... they just run alot faster than the original chips plus all the peripherals like UARTs etc that occupied board space are in the chip.

Regards,

Moodz

In my next life I'm going to learn that stuff... wicked slick...

-SB

[moodz]

I see. I guess I just wouldn't look forward to implementing a "microprocessor" in a FPGA. Just because you CAN do something doesn't mean that you should.

But I'm an old software guy from way back so I'm more familiar with programming microcontrollers. Thus that would be my preference. It sounds like FPGA isn't nearly as expensive as what I originally thought so that makes a lot more sense. To each his own I guess.

Thanks for the explanation.

I will post a cheeky reply .....

A microprocessor is a pre-programmed FPGA.

Seriously though if you are an old hand you would remember the Z80, 6809, 68000 etc .... these are all available precoded for FPGA platforms ... they just run alot faster than the original chips plus all the peripherals like UARTs etc that occupied board space are in the chip.

Regards,

Moodz

[mts]

I see. I guess I just wouldn't look forward to implementing a "microprocessor" in a FPGA. Just because you CAN do something doesn't mean that you should.

But I'm an old software guy from way back so I'm more familiar with programming microcontrollers. Thus that would be my preference. It sounds like FPGA isn't nearly as expensive as what I originally thought so that makes a lot more sense. To each his own I guess.

Thanks for the explanation.

[moodz]

Why wouldn't you use a cheap microcontroller instead of a FPGA chip costing $200? FPGA is fine for implementing discrete logic functions. But they aren't that good for algorithms that require "programming" such as FFT's and such. You can get a microcontroller with a built in ADC, memory, and even various peripheral support for around $5. The programmer costs $49. Once you have the programmer you can buy various different chips with differing capabilities so the possibilities are virtually endless.

Maybe I missed something earlier in the discussion as to why you are considering FPGA over a microcontroller. If so, I apologize.

I was using a cheap microcontroller ( dspic4102 ) which is fine and more than adequate for driving a PI detector however it is not ideal as a sandbox for developing more exotic functions like descrimination in the digital domain.
... the FPGA chip itself is about $10 - $20 ... the boards with RAM, FLASH, ethernet, power supplies, switches, LCDs etc cost $59 - 200 or more depending on application.
Apart from the fact that you can 'build' a microprocessor in an FPGA or a number of them simultaneously if you wish one of the fundamental limitiing factors of a uP is that it can only execute instructions sequentially. Insruction 1, Instruction 2, Instruction 3 etc etc. On the FPGA it would be something like ON CLOCK EXECUTE ( inst 1,2,3 ) .... ie all the instructions would execute in one clock. In other words a FPGA can trash a microprocessor applied to a SPECIFIC APPLICATION. On the other hand a microprocessor will probably trash an FPGA on a GENERAL APPLICATION. ( ie you would probably not write a word processor on a FPGA )
It is interesting that you mentioned FFT calculations ..... An FPGA will trash a CPU at FFT no matter how optimised the code in the microprocessor is. This is because something like FFT is ideally suited to parallel hardware that the FPGA provides. Also you can set up multiple parallel FFT process units in a FPGA to further increase speed.

An FPGA is no more than a digital breadboard .... it allows you to program not just conventional microprocessor code but to rewire the whole circuit if you wish. You cant do that on a conventional microprocessor board.

If anyone remembers space invaders ... the arcade game .... an enterprising individual mapped the original game logic board ( ie the chips ) and ROMS into the Spartan s3e FPGA as mentioned above. So if you ever want to take a break in the field ... you can play the original game .. sounds and all. www.fpgaarcade.com/games.htm The point I am making here is that one minute your box could be a metal detector ... the next it is a game box all without changing a single component or wire.

Moodz

[mts]

Why wouldn't you use a cheap microcontroller instead of a FPGA chip costing $200? FPGA is fine for implementing discrete logic functions. But they aren't that good for algorithms that require "programming" such as FFT's and such. You can get a microcontroller with a built in ADC, memory, and even various peripheral support for around $5. The programmer costs $49. Once you have the programmer you can buy various different chips with differing capabilities so the possibilities are virtually endless.

Maybe I missed something earlier in the discussion as to why you are considering FPGA over a microcontroller. If so, I apologize.

[WM6]

So far I am still hammering the code together so not ready for release 1.0 however hope to share progress with the group shortly.

What of frontend schematic you apply? Something new or one of here posted?