chance pi build

[6666]

Hi Dan,
Your programmer is a clone of USBASP. So, you can use any software you want :
AVRDUDESS
EXTREME BURNER
ARDUINO (you can programm standalone chips with arduino software by Ctrl+Shift+u)
You can also use avrdude by command lines such as ; avrdude -c usbasp -p m328p -u -U flash:w:yourfirmware.hex

Regards

(ARDUINO (you can programm standalone chips with arduino software by Ctrl+Shift+u) would you mind explaining that a bit more thanks

[6666]

(ARDUINO (you can programm standalone chips with arduino software by Ctrl+Shift+u) would you mind explaining that a bit more thanks

Never mind its the same as clicking the yellow arrow to program.

[dantech]

http://chomikuj.pl/dantech/Chance+PI
v172

Rebuilt and increased twice speed of entire timing. Decreased power consumption. A lot of code simplifications and improvements.
For instance main pulse has 200Hz and 60us width (previously 100Hz and 120 us)

[baum7154]

Thanks Dantech,
This should be interesting to try out. Is the sample window now reduced from 12.5us to 6.25us? There could be a down side to the reduced main pulse of 60us as it may not fully energize the coil, thereby reducing detection depth. This will be dependent upon the properties of the coil used. I think the original Version 1.2.1 had a multipulse burst rate of 227 per second with an 11.0592 mHz clock. The multipulse burst has 16 short pulses and one long 120us main burst.

Regards,

Dan

[dantech]

I can try whatever you want Even better to get to know how it works.

In the v121 the internal timer generates interrupt with frequency 200Hz but... inside the timer interrupt support we can see two different subsupports. Shall we say odd cycle and cycle. It means that frequency of each odd and even cycle is 100Hz.
In the odd one there is 8 prepulses and one main pulse 120us, the damper is off, only one sample is taken to get signal strenght.
In even cycle there is 16 PWM pulses. PWN series is almost 600us wide, the dumper is on, 3 samples are taken (samples gap is 12us) to get information about signal slope.


In the v172 the internal timer generates interrupts with frequency 400Hz, so odd and even cycle have 200Hz. There is no 8 prepulses in the odd cycle and main pulse has 60 us. In the even cycle PWM pulses are shotened 4x. PWN series is almost 140us wide. Pulses gap is still 12us.
I did not noticed any difference with detection depth and VDI behaviour
It is still under observing and any suggestion is appreciated.
The main advantage at the moment is reduced power consumption

Any suggestions?

[baum7154]

dantech,

In the present version 1.2.1 my detector draws about 90ma with no backlight on. What is the current draw with using V172? I believe that the V172 will fully charge my 3DSS coils in the range of 335uh and I will have to test the field performance with the standardized targets I have been using for the past few years. I still must program a chip with the firmware you have provided and make a few mods to my pub. As soon as this is done I may have some suggestions based upon test results.

Thanks,

Dan

[dantech]

It looks that the power consumption depends on coild build. Dedicated coil with v121 takes 160-190mA. At the moment I am testing different timings and detection depth and power consumption are changing. AT the end of tests I shall try to select optimal timings. Generally I am trying to get good results with 2x faster timings than v121.

Has anyone any experience with optimal timing for PI?
I mean pulse width, frequency, pulse sequences, etc.

It seems that short prepulses serie before main pulse makes better sensitivity for smaller objects.
I can change everything in the software for this hardware project , so each idea is appreciated

[baum7154]

[QUOTE=dantech;221348]It looks that the power consumption depends on coild build. Dedicated coil with v121 takes 160-190mA. At the moment I am testing different timings and detection depth and power consumption are changing. AT the end of tests I shall try to select optimal timings. Generally I am trying to get good results with 2x faster timings than v121.

Has anyone any experience with optimal timing for PI?
I mean pulse width, frequency, pulse sequences, etc.
---------------------------------------------------------------

Dantech,

I think the answer to your questions about the 'optimum timing for PI' is that there is not a universal timing for all target sizes and and materials. An analog detector optimized to detect small gold will not optimally energize large targets composed of 'slower' metals and its short Sample Delay may well be over before a slower targets response happens. Likewise the same detector optimized to detect large slow targets will completely miss the response of fast small gold because the Sample Delay is set too far out in time.

For example a tiny .3 gold nugget would be best detected at a high pulse rate (perhaps 3000pps) and a short sample window (perhaps 7us or less). Your comment "It seems that short prepulses serie before main pulse makes better sensitivity for smaller objects." to me is correct. See post 298 and others in the Minipulse Forum Thread 'Getting ready to build MPP' as it is geared to the detection of small gold with the Minipulse Plus analog detector and discusses timings for those targets. GREEN has posted target decays in that and other forums and these decays give an idea of minimum required Sample Delay to see small fast targets.

I think part of the reason for Chance PI having a complex multipulse design is to allow it to detect a wider range of targets both large and small composed of a wide range of materials from iron to gold. Its complex pulse pattern also allows sensing of ground and discrimination processing. My experience with the V1.2.1 software is that the discrimination needs 5 to 6 units of signal strength in order to give credible results. This phenomenon is typical of several other 'discriminating' detectors. Certain target geometries such as common 'crown' bottle caps present themselves as gold or aluminum until signal strengths increase perhaps well above 6 units. Discrimination of gold from aluminum of the same size is practically impossible as they respond similarly in time.

A large gold nugget of 5 grams requires a significantly different optimal detector setting than a 0.5 gram gold nugget. I think this is part of the reason for the 120us long pulse of the V1.2.1 firmware, to strongly excite larger targets. On small targets much of that pulse is probably wasted.

To me CHANCE PI was initially released as a coin detector using a coil that cannot operate effectively below about 18us because it is not shielded and is not optimally damped. Damping must be optimized in order to allow it to operate with Sample Delays in the range of 5 to 10us necessary to detect the smallest gold targets. For this reason I abandoned the original coil design in favor of the 3DSS design of the Chance PI coil thread and these coils, properly damped, will allow sampling at 5 to 6us. Also a 2 stage preamplifier is a required modification for operation below 10us because the original design preamp goes into saturation, preventing sample delays below about 10us from being used.

I invite others to join in this discussion as there is so much I am not covering here.

Regards,

Dan

[dantech]

OK. Let's make great PI

8 prepulses in the odd cycle are made of 20us pulse and 60us gap, so period is 80us. It gaves us 12,5kHz (1/80us). It seems too high frequency for anything. Main pulse 120us is quite universal. Also it seems better to use 400Hz (200 for odd cycle and 200 for even cycle). In the v121 was 200 Hz only (100 for odd and 100 for even cycle)

What about it?

At the moment it seems also that PWM series in the even cycle is made to get higher coil current without increased power consumption. It look that current grows through entire PWM width and reaches high value. With the same width of single pulse we could consume a few times more power.

Am I right?

[dantech]

v174
http://chomikuj.pl/dantech/Chance+PI

Real timing 400Hz. In the odd and even cycle the main pulse has the same width 120us, so this is real 400Hz. 8 prepulses and main pulse are the same in each cycle. There is no PWM and damper with diode in the even cycle

Samples gap in the even cycle is 7us

[Spyke]

One question, if I want to modify my chance Pi (latest original PCB version) with latest code from dantech, which hardware changes are necessary?

Thanks in advance.

[dantech]

No changes. My release is backward compatible with hardware for v121

[baum7154]

OK. Let's make great PI

8 prepulses in the odd cycle are made of 20us pulse and 60us gap, so period is 80us. It gaves us 12,5kHz (1/80us). It seems too high frequency for anything. Main pulse 120us is quite universal. Also it seems better to use 400Hz (200 for odd cycle and 200 for even cycle). In the v121 was 200 Hz only (100 for odd and 100 for even cycle)

What about it?

At the moment it seems also that PWM series in the even cycle is made to get higher coil current without increased power consumption. It look that current grows through entire PWM width and reaches high value. With the same width of single pulse we could consume a few times more power.

Am I right?

------------------------------------------------------------------

Dantech,

Have you read this entire thread? There are many things covered in this thread but without the source code from FAndy some speculation on the theory of operation was necessary.

There is a description from Tinkerer in post 45 of this thread regarding the 16 pulses of V1.2.1 as follows:

"The sample taken during the short pulses is probably for the Ground Balance. Minelab uses a similar method. Several closely spaced short pulses also act like a demagnetizer for the magnetic properties of minerals in the soil.

With the original coil (is it 400uH?) and 390R? for damping we get a long coil decay slope. Somewhere along this slope, the decay slope from magnetic targets crosses the coil decay slope. Using this crossing point as a pivot, we can discriminate magnetic metals."

I have also posted pictures on page 3 of this thread and one of them shows that the pulses are gradually increasing in their pulse width and are not fixed pulse width.

I do agree with your statement of the pulses pumping the coil to higher energy as FAndy's Chance PI site pictures and description also explain. I believe that Mosfet Q9 is not for damping or dissipation of coil energy but provides a short across the coil between pumping pulses. This short does not dissipate flyback energy but causes most of it to be retained in the coil until final flyback is allowed. At that point Q9 is not allowed to short the coil and only the damping resistances of R16 & R13 are in play. If these resistances are not optimum for a specific coil the coil fly back time will not be the minimum it can be. Too high a resistance and the energy won't dissipate fast enough, too low a resistance looks more like a short and the coil will not dissipate fast enough to see small gold.

Forum member TEPCO also had some good insights into the operation of the Chance PI and his input is worth reading in this thread.

Regards,

Dan

[dantech]

I shall read this thread more carefully again.
Facts from code below:
1. 16 pulses acts only in the even cycle. During entire even cycle Q9 is closed and coil does not transmit entire power. So I do not thing that Q9 circuit + PWM makes something useful.
At the moment I switched off PWM and use one pulse only. I do not see any difference in the field. Fandy told me in the past that Q9 circuit does not work in fact.

2. Sample taken during "short" pulse (it means in the odd cycle in the v121) is for signal strenght only.
3. Samples taken during "long" pulse (it means in the even cycle in the v121) is taken for VDI purposes

Now, every cycle has 8 prepulses and one main pulse. It works quite stable in my garden

v176
http://chomikuj.pl/dantech/Chance+PI

- a lot of changes eg. there is possibility in the menu to tune additional VDI sample shift and VDI samples gap, default setting restoring, backlight, new timing. More in changelog.txt

By the way. At the moment default U min of battery is 11V proper for my 12V gel battery so if someone need to decrease battery U min and message about low battery is displayed, one should start detector with Menu button pressed

[Spyke]

No changes. My release is backward compatible with hardware for v121

Thank you!