Digital PI schematic

[Robert Hoolko]

This is just an example to show what parts of the PI circuit could be replaced by a micro controller. I am not claiming that this design would actually work as drawn.

I just removed the parts of the original schematic whose functions could be performed by the ADuC812 microconverter, and left the rest alone. The ADuC812 is a single chip that contains an 8 channel A-D converter, 2 channel D-A converter, and an 8-bit microcontroller.

This is not the best way to connect the microconverter into the PI circuit, it was just the way that required the least amount of drawing. The + input of the amp would have to be biased so the output is within 2.5 volts of -5. The A-D input range is 0 to 2.5 volts above AGND which I connected to -5. Also the amplifier gain could be reduced.

I labeled a couple of the pins Px.x because I don't know yet which I/O pins would be best to use. Not all I/O pins are created equal.

I connected one of the analog inputs to the output of the amplifier and the other 7 inputs to pots to be used for adjustments like frequency, delay, auto tune, and threshold. This design does not have a digital display or control pad. It is all knobs and switches.

I show the analog output going directly to the speaker driver. But this chip might not be fast enough to do VCO, so the 555 might have to go back in. I have not tried doing VCO yet so I don't know.

Robert

[Pawel]

Re: Digital PI schematic

Hi Robert,
According to my own experience it is much better just to use a timer (auto reload mode) to generate the acoustic signal.

Regards,

Pawel

[Carl]

Further simplification...

Replace all those pots, with a push-button selection switch, a rotary encoder, and an LCD display. Use the selection switch to cycle through the settings, and the rotary encoder to vary the setting. The LCD tells you which setting you are adjusting, and it's value.

That would be a nice design for an underwater detector, where you want to drill as few holes as possible.

The Tesoro SandShark uses a PIC micro, and can do both VCO and non-VCO.

- Carl

[Robert Hoolko]

Re: Digital PI schematic

Pawel

In this chip the timer auto reload is only 8 bits so the frequency would not be low enough for VCO. I would have to add external parts.

The chip would be fast enough for VCO if it did not have anything else to do, but I am afraid that it might take too many instruction cycles and not leave enough time for the main task. I will look at this later. I am already chopping the output at a fixed frequency to produce audio, but that frequency is a divisor of the sampling rate and does not take much time.

Robert

[Robert Hoolko]

Re: Further simplification...

Carl

I have been using digital interfaces for over 30 years and I still don't like most of them. It generally takes much longer to make adjustments and even takes time just to see what the current settings are. Also writing the display software can be more effort than the parts that do the real work.

I do acknowledge the importance of minimizing holes underwater.

If my digital PI works at all I can see a digital display in the future. At this point I still don't even know if it is going to work on a PI signal.

Do you know if the SandShark PIC processes the signal or just does the timing? The PIC designs I have seen only do timing.

Robert

[Pawel]

Re: Digital PI schematic

Robert,

As far as I know the counter/timer 2 can be used in 16-bit auto reload mode in all 8051 clones having 3 counters. ADuC is not exceptional.

Regards,

Pawel

[Pawel]

Re: Further simplification...

Carl,

Some time ago I was playing with a similar interface. I used a double-function rotary switch (having a push-button function) and implemented "click" and "doubleclick". It worked fine, but I still prefer old analog potentiometers. They occupy a lot of space, do not look modern, but in my opinion they are just the most comfortable control devices.

Regards,

Pawel

[Pawel]

Re: Further simplification...

Robert,

Once upon a time ;-) when I connected the controlling potentiometers just as you did I found an unpleasant instability of the instrument (that was not a metal detector). It was due to the subtle (one bit) jitter of the potentiometer readout. I had to add some hysteresis to the readout procedure...

Regards,

Pawel

[Robert Hoolko]

Re: Digital PI schematic

Pawel

I have not been using T2 because it is also the baud rate generator. With the AVR board I had to use the serial port for debugging. The ADuC has a built in debugging mode and I have not had to write any of my own routines yet to support debugging. But I have been assuming that T2 is tied up while the debugger is running.

Robert

[Pawel]

Re: Digital PI schematic

Robert,

Just do not store the sound data in rcaps during all debugging procedures (perhaps you could send them to the PC instead) and implement the audio function when all other processes are OK.

Regards,

Pawel

[Carl]

Re: Further simplification...

Having the multiple knobs makes for an easier interface. On my DFX, I get tired really fast, having to scroll through endless menus. A few knobs would help that machine.

However, for underwater use, one knob & one switch would be nice. The current all-discrete design does not allow this at all, whereas the microcontroller approach would.

My recollection on the SS is that the PIC does timimg & audio only. I believe it uses opamps for the differential integrator, and a final gain amp. The gain amp drives the PIC, a 16C74.

- Carl

[Dartañan]

Re: Digital PI schematic PIC16f877

I am desing is PIC16F877-20
-Generate a power for IC (15V) non ICL7660, for 2 diode and 2 electrolitic
-Timig
-Rs232 and bootloader for fast reprograming
-Programing in C PICC

[simeon webber]

Re: Further simplification...

see my design at the top of the page
it offers on the fly waveform per coil display
so adjustment and preset scans are easy too do

simple solution

use more than one micro
and to adjust the power use a variable resistor r2r chip like the ad8403

a fet driver solves the problem of speed and some pll lock from a small calibrated pick up in the coil assembley assures the correct freq feedback

so out of lock = no power on that coil


why stop at one coil you guys
havent you ever heard of having more than one beer or two eggs at breakfast


interigate your find
without diging it

then walk away if your scanning detector shows the outline of an old tin can or ringpull

you can even tell it to display a ringpull x marks the spot

depth becomes a matter of how hard you interigate and stay still on a spot

auto interigation is an easy option
and finds tally is also easy

all the pic designs i see offer no memoory of finds only of the data
{the settings uses no room at all in memory
nanao have i see ones with an external eeprom or as another above had the idea of rs232 connection }

yes all possible

multiple coils is the only new advance in years to pi......

so try this
if you want a challange
post above.....


i agree about timing

so simple run one core on a nop interupted schedule to give a very low max speed then just int update it with new settings using another pic to load but only when you load a new set
this way auto load at rest can be done
saves transitions to a new freq each time you dial a digit

all these things increase the use

cpu done right is the best way
for actualy finding things at depth and in
water esp salt water....

as for speed the above is true a 20 mhz pic has a nop time of 200us so
200us time frame is possible control of power level is simpl to achieve by using a passive transisotor and variing the voltage from the cpu drive pin via a digital pot also loaded from the same core as the pwm driver runs

also on an interupt
all controlled by the main cpu

however an interesting descussion

[Nicko]

Re: Digital PI schematic

Hi Robert,

I think that you'll find that the ADuC812 needs low impedance ( 600 ohms and you lose at least one bit from the ADC. In the diagram you have drawn, you will need to buffer (voltage follow) all the pots - this is a pain, and seems to be a weakness in this chip - the AD development board for the ADuC8xx series has buffers for all the analog pins.

Also, this chip is really not very powerful - it is only a 20MHz 8051 which is not really enough for DSP, especially as it has no DSP instructions (i.e. bit-reversal indexing, composite MAC etc.)

Ideally, you should be looking at something stonking, like the AD21990 - it uses the ADSP-2191 core which gives 160MIPS - I use the same core (ADSP-2191M in LQFP144 package) in other projects with great success. Only problem is that it uses approx 200mA, but look what you get for that...

Nicko

[Robert Hoolko]

Re: Digital PI schematic

Nicko

Thanks for the tips. In this application the input bias is not as important as it is in other cases. For one thing the signal is obtained by taking two successive samples from the same ADC channel and subtracting one from the other. This tends to cancel out the bias error. Besides, in a metal detector we are more concerned about changes in the signal level than we are about the exact level. In any case the bias would be removed when the signal goes through the high pass filter.

For the pots I don't think a 5 mv offset will be any problem. I was only planning on using the upper 8 bits from the ADC.

I agree that the ADuC812 is under powered for this application. I have to constantly worry about how many cycles I am using, and there are some things I cannot do at all. But I don't think I would want to go to a 200 mA chip either. Analog Devices is coming out with an ADuC841 upgrade to this device. It is much faster than the 812. It is still not in the DSP class, but it is fast enough for a metal detector.

Robert