Pulse period/Pulse width

[geoscash1]

A question that may have been answered before or in a book somewhere but I'm unable to come up with a definitive answer to get a better idea of a starting place. I have been experimenting and have been wondering about this and wanted to get others opinions or from an experienced engineer in this particular matter.

See image for clarity of question. And thanks in advance.

[Davor]

That depends on your intended purpose.
The coil charging period should ideally be of equal duration as all the samples in case you are up to ground balanced PI. Otherwise it may be shorter.
If you wish to implement EF sample the period becomes long. But with bipolar pulsing you don't need EF.
etc.
The bottom line is that you get the most out of a PI if you can pack a total period tightly. It improves S/N

[Monolith]

A question that may have been answered before or in a book somewhere but I'm unable to come up with a definitive answer to get a better idea of a starting place. I have been experimenting and have been wondering about this and wanted to get others opinions or from an experienced engineer in this particular matter.

See image for clarity of question. And thanks in advance.

[ATTACH]34688[/ATTACH]

Pulse period for PI:

The shorter the pulse period, the more samples you can integrate for better S/N.

The pulse OFF time should not be much shorter than 1TC of your longest TC target.

[FatBob]

Hmm these answers don't really answer anything. I believe we wish to find a sweet spot for
pulse width and rep rate.

The pulse width will affect the size target you can see. If your looking for 1 pound of gold a long
period would be good (1US?). For small gold a smaller period would be best (50 us?).

A fast rep rate will allow you to see those tiny traces from the small gold (a chain looks like one
link!). 1 khz to 5 khz will see those smaller targets better than lower rep rates.

So if you are looking for small gold try 2.5khz rep rate with 50 us TX width.

For those 2 pound bars try 500 hz rep rate with 250us TX width.

[Teleno]

Here's my take.

1. Use current-mode drive, separate Tx and Rx coils.
2. On-period: 50us, off-period: 950 us, repetition rate: 1 KHz.
3. Sampling after the rising edge sees all targets with taus from 0 to 950 us.
4. Sampling after the falling edge sees taus from 0 to 50us.

[ODM]

It is not a simple answer unfortunately.

Time constant (t) in engineering generally means a time when a variable changes 63% to final value; generally 5 time constants are expected for a full change (rise, or fall). For non-ferrous metal targets this is inductive-resistive time constant, t=L/R .
Highly conductive targets such as copper, aluminium, silver have a longer time constant and poorly conductive targets such as gold have shorter time constants. For ferrous targets the time constant is aided by stronger eddy current formation.
Rings and other "easy" shapes have longer t even if the material is difficult because of relatively large resistance.

For the waveform you presented, the coil TX current excites the target in the "wrong" direction, while the flyback current excites it in the "wanted" direction, the direction that will cause the following amplifiers and integrators to signal a target.

Eddy currents are induced in the target only by the change of current in the TX coil. A constant current would not induce an eddy current in our target. If we wish to have the best ratio of TX current (minimized) to the flyback current (maximized), we want the coil TX current to settle as a constant for 5*t of our target during on-time. Then the TX current induced eddies would have died down, and we would only see the flyback induced eddies.

http://www.geotech1.com/pages/metdet/info/induction.pdf

http://www.geotech1.com/pages/metdet...s/FastCoil.pdf

[FatBob]

Well I think he was looking for "Real World" numbers.

This question has been asked and answered many times here but we still
ask it as the answer is always "it depends".

Maybe a study of specific detectors and their possible settings is in order.

I wanted to make a micro controlled PI and came up with the same question.

There are big objects and small objects, and fast decay and slow decay.

A PI for the beach looking for small gold has to be fast. Older commercial
designs (which many hobby ones are based off of) did not try to optimize
for fast early sampling.

But what is "fast"? On a beach some say if you sample 15us after TX that's
the best you can do as the salt water will be seen below that. Others say
you can go to 10 or even 8us. So that number is a limit you do not need
to surpass.

The TX width almost need to change continuously as it sets the size object
detected. Some of the commercial units use 2 or 3 different pulse widths
to cover more target sizes. Chance uses 2 different widths (pulsetrains) for
this.

I guess a better question is what TX widths will see a gold chain and what
widths will see a big gold ring? That answer will allow you to set a TX width.

The Rep rate will eliminate noise and perhaps see the faint targets better.
Older machines were generally slower in rep rate (100 - 900 hz) while
newer ones tend to go faster (1.5 - 10khz). The problem is high current
draw (which was a problem with older batteries but now we have some
better ones to work with).

[FatBob]

I've done a little reading and found out some info.

For small gold use 50us TX width. For best on all objects use 100us.
For gold the shortest delay you can get. Small gold is in the 5 - 7 us range.
For the beach 15us works good.

The RX pulses are usually 20 - 30us. And delay from 1st to 2nd 100 - 200us.

The Rep rate affects response speed, current consumption and noise. I'm not
sure on this one but faster seems better as long as you have the battery power for it.

[eclipse]

How about the deeper PI designs (i.e Delta Pulse) timings?
Will raising the TX width time beyond 500uS give advantages when you consider it's designed to detect only large items?
Or maybe the advantage will be too little (when considering more heat will be generated) and too little more depth will be gained.

What is the correlation between pulse width and wire specs of the coil? Why wider pulse width designs require higher cross section diameter of the wire?
Is there a formula which can be used for calculating wire specs based on maximum amps going through it?

[FatBob]

Well the TX pulse width chooses the size of the target somewhat. A 1000us pulse is good to find cannon sized targets.

Large object Deep detectors use long and slow. 100pps rep rate 500 - 1000us TX.

A longer pulse will allow the coil current to get to max and stay there for some time so available power is a concern
and wire size to handle the peak power. Most wire companies have a chart of current to wire size. AWG 16 can do 22 amps.
http://www.powerstream.com/Wire_Size.htm

A problem to consider is that a PI can see the wire once it gets to a certain diameter
(though may ignore it if your delay is longer to see large deep stuff).

[geoscash1]

I've done a little reading and found out some info.

For small gold use 50us TX width. For best on all objects use 100us.
For gold the shortest delay you can get. Small gold is in the 5 - 7 us range.
For the beach 15us works good.

The RX pulses are usually 20 - 30us. And delay from 1st to 2nd 100 - 200us.

The Rep rate affects response speed, current consumption and noise. I'm not
sure on this one but faster seems better as long as you have the battery power for it.

First, thanks to everyone that has passed on some input on this. I, as a newbie to this hobby am enjoying the learning of this but
some of the more technical parts (like original post) I am still trying to wrap my head around, its harder than it looks for someone with
no previous exp in electronic design.

Thank you Fatbob for your replies. You were correct in assuming I was looking for real world numbers. I am working on my own design
and it has been a struggle as I am no engineer, but I am having a ton of fun trying to understand and figure some of this out.

To clarify a bit more, I live in gold country and want a "homemade" gold machine in PI form. It must be able to find small gold as that is most
likely what will be dug up if any at all. But and I assume as well, if it can detect small gold it will as well find larger and everything else too whereas
the opposite will not detect the small gold. I would not be using it for the beach as that is just not feasible for me and there are plenty of kits that
cover the beach machines anyhow.

So based on some of the comments and suggestions, it would be very cool if you could show what your suggesting with an edited version of the original
pic I posted so I can get a clearer picture of what and how to understand it if possible. If you could show all the pulse points as a guideline that would even
be better.

[eclipse]

Well the TX pulse width chooses the size of the target somewhat. A 1000us pulse is good to find cannon sized targets.

Large object Deep detectors use long and slow. 100pps rep rate 500 - 1000us TX.

A longer pulse will allow the coil current to get to max and stay there for some time so available power is a concern
and wire size to handle the peak power. Most wire companies have a chart of current to wire size. AWG 16 can do 22 amps.
http://www.powerstream.com/Wire_Size.htm

A problem to consider is that a PI can see the wire once it gets to a certain diameter
(though may ignore it if your delay is longer to see large deep stuff).

FatBob that's very helpful information. According to the table best wire diameter for large pulse width PI (500uS) with coil 40-50cm in diameter
should be around 22 AWG (0.35mm2) wire when considering outer diameter (about 1.5mm in diameter) is not too thick and also wire weight.

[FatBob]

I am mainly interested in beach detecting so I lean tward that way.

For small gold everything should be short and fast. 20 - 50us TX
2000 - 5000 rep rate and 15 - 20us RX pulses. The first to second
spacing should be as much as you can do.

Here's a picture of good settings for small gold;



Here's some target time constants;



A quote from Eric Foster;
"Goldquest SS runs with a 30uS TX pulse, and a 10uS minimum delay. Sample pulses are 10uS wide with 40uS between the first and second; then the next TX pulse starts. This all works out to a pulse repetition rate of 10,000 per sec. This timing is good for medium/small rings and nuggets. Try a copper penny, or a silver quarter and you wont get the best range as the TX pulse is too short for the object time constant, and the sample pulse spacing will also give some signal cancellation."

[kt315]

For small gold everything should be short and fast. 20 - 50us TX
2000 - 5000 rep rate and 15 - 20us RX pulses.

its mean ONLY the high Ohm coil using, ie 1 mH (MILLI-). with some additional inserial resistance in TX driver also (50 Ohm?). NOT means 350-380 uH coils.

Reg calls PI's splitting up them in 2 another types:

1. Typical high power PI's (350uH coils) [Delta Pulse, MPP, Surf PI, Barracuda PI etc]
2. Lower powered PI's (1mH coils) [Goldquest SS is here, C-scope 4PI, C-scope 6PI etc]

http://www.findmall.com/read.php?34,...31#msg-1777531

[godigit1]

Hi Geo,
I live in Tahoe so I trip to gold country allot.
I think as a prospector Ground balance is the most important requirement.
You can do everything perfect to no avail if you cant get past the hot ground. And lower time constants that will not help as the ground becomes too reactive.
Think salt water is bad try to detect around fist sized pieces of hematite.
A 10 us TDI is capable of some very small gold if you sweep very slowly.
You may have seen this but at 3 to 5 minutes is a good example of sweep speed

https://www.youtube.com/watch?v=7rTA2CH_x8Y

The catch 22 is turn the ground balance on and that 1 grain nugget is gone!!1

I think it s great you are making your own design, I hope you share your progress.

Best of luck