Air tests with Hammerhead ver C

[Tinkerer]

Tinkerer, thanks a lot for sharing all the info about Rx-Tx coils, I can see the advantage of having a separate Rx coil and I will build one as well. If the Rx would have the same diameter as Tx coil, would there still be any advantage (like a faster decay rate)?
Where did you get that recipe from? What's the role of colloidal silica in the mixture?
I am not sure I can obtain that stuff from somewhere.
About the graphite: I know the electrical brushes have high conductivity, but I don't have any. Maybe the thick graphite pencils used by artists could also work (they might not have such a good conductivity, because perhaps they mix graphite with other materials).
I don't have any resin either - buying all that stuff could exceed the price for the 3M shielding tape I actually just searched on ebay and I can buy 5 rolls of Scotch 24 tape, 1 inch x 15 feet for US$20 + US$14 freight. Not that expensive actually.
Also, on ebay there is teflon spiral wrap, for US$75. I should start making some extra money, to buy all these

Regards,
Nicolae

The distance between the TX and RX coils reduces the coupling factor k. this reduces the amplitude of the flyback in the RX coil.
Coloidal silica is used in all paints. Its function is to give the paint "body", so it spreads well but does not run off. All fiberglass applications also make use of it. It is marketed under different trade names like "Aerosil" etc. Fiberglass shops sell it. The recipe is "my own poor man invention"
The same happens with Epoxy resin. It is sold everywhere in small 10cc tubes, but the trade names vary a lot. Try "Araldite" GOOGLE!!!!!!
A roll of PP or Polypropylene string probably costs 1$. It is enough for dozens of coils. Sorry I don't know the trade names used in your country. This is why I use the generic names that these materials are made off.

Pencil graphite has a mixture with clay. It will still work by applying a thicker coat. It is expensive though.

Tinkerer

[nick_f]

The distance between the TX and RX coils reduces the coupling factor k. this reduces the amplitude of the flyback in the RX coil.
The same happens with Epoxy resin. It is sold everywhere in small 10cc tubes, but the trade names vary a lot.
Pencil graphite has a mixture with clay. It will still work by applying a thicker coat. It is expensive though.
Tinkerer

Very interesting about the advantage lowering the k, I haven't thought about that, although it seems now obvious
I even have a few tubes of araldite A+B at home, but I thought you were refering at the more fluid resin, for fibreglass applications.
I will check the price of graphite pencils when I have the chance.
You are quite amaising, seem to know everything about everything

Regards,
Nicolae

[Tinkerer]

Tinkerer,
Your shielded twisted pair was for audio or RF applications?
About the OP37, it sounds good. I can get them from www.futurlec.com (.au) for AU$2.09
About the MOS-FET, I think somebody else found this ones: FQD4N50 (COSS = 70pF), 500V, 2.6A. They can be purchased from Mouser Electronics.

Regards,
Nicolae

The shielded twisted pair is CAROL C1322 FLEXLIFE. I use it because I have 500 feet lying around since 20 years.
RF probably would be best. Only trying will tell.
300 Ohm TV antenna inside a shield would be good too.
FQD4N50 sounds good.

Tinkerer

[nick_f]

The shielded twisted pair is CAROL C1322 FLEXLIFE. I use it because I have 500 feet lying around since 20 years.
RF probably would be best. Only trying will tell.
Tinkerer

I found a link to your type of cable: http://products.isc365.com/AVCat/CTL...&prodid=371459

Type : Low Impedance
Number of Conductor : 2
AWG Size : 22
Conductor Strand : 16/34
Nominal Insulation Thickness : 0.025 inch (0.64 mm)
Nominal Jacket Thickness : 0.025 inch (0.64 mm)
Nominal O.D : 0.239 inch (6.07 mm)
Nominal Capacitance : 31.0 pf/ft (Capacitance between one conductor and other conductor connected to shield)

I will try to find some good cable as well. By the way, at work, we have some shielded cables that go up to 20GHz with very low losses. They have double shielding, very dense. I will measure their internal capacitance, out of curiosity.

Regards,
Nicolae

[hobbes_lives]

If you have access to the 3M shielding material that BBsailor recommends, it is probably the best solution.

I currently use the Scotch 24 tape. I've only worked with monocoils so far, and that's the simplest solution. I figured the nickel spray would be good to spray inside the coil housing when individually shielding coils would be more difficult.

[Tinkerer]

I currently use the Scotch 24 tape. I've only worked with monocoils so far, and that's the simplest solution. I figured the nickel spray would be good to spray inside the coil housing when individually shielding coils would be more difficult.

Good ground connection and lots of space between shield and winding are the most important factors.

Tinkerer

[Tinkerer]

Signal amplitude needed for dsPIC30

A few months ago, I was working on a very similar design using a Microchip dsPIC30 series microcontroller. I could not achieve much sensitivity when sampling right after the preamp stage......

Hobbes_lives,

I quote your comment from another thread. Here is my question:

How much signal amplitude do you need for the dsPIC30 input?
It is my understanding that it is best to A/D the signal as early as possible. The more one amplifies, the more the noise amplifies too, plus other added noise.
Therefore I developed a method to generate a very high signal amplitude right at the front-end. For the 1" square foil I get about 3mV at 30cm from the coil with air test, having a total gain of 1000.
I am now looking at the best way to further process this data.

Do you have any suggestions?

Tinkerer

[hobbes_lives]

Tinkerer,

The A/D converter on the device I was using, a dsPIC30F4011, is a 10-bit A/D. Using a 5 volt reference, the voltage step per bit is 4.88mV. This is the configuration that I was using.

Using a 2.7 volt reference, the voltage step per bit is 2.64mV. I believe 2.7 volts is the lowest voltage differential that can be used for the reference voltage in this device, but the specs are not totally clear on this, or I just haven't been able to find it in the documentation.

Since the 3mV signal that you mentioned would only deflect the digital value by 1 bit while using a 2.7 volt reference and ideal circumstances, I don't think it would be detected reliably in a real-world environment.

I'm curious, what voltages do you get from other targets at varying distances from the coil? What about the aluminum foil at 15cm? ( I could perhaps do the calculation on this, but my brain is already sore today. )

Many of the problems that I was running into with the design were related to noise. I had most of the circuit laid out on breadboard, so I was having all sorts of problems with ground loops, cross talk, interference, etc. The biggest contributor to noise in my layout was coming from the LCD drive signals. I'd get a nice burst of digital into the analog circuit each time the display was updated.

I'm going to get back to it soon, as I have some new ideas that I'd like to try. I'm also going to work on each section, one at a time, and build a permanent module when I'm satisfied with each section. That's the nice thing about conventional PI's, they're very modular.

[Tinkerer]

Tinkerer,

The A/D converter on the device I was using, a dsPIC30F4011, is a 10-bit A/D. Using a 5 volt reference, the voltage step per bit is 4.88mV. This is the configuration that I was using.

Using a 2.7 volt reference, the voltage step per bit is 2.64mV. I believe 2.7 volts is the lowest voltage differential that can be used for the reference voltage in this device, but the specs are not totally clear on this, or I just haven't been able to find it in the documentation.

Since the 3mV signal that you mentioned would only deflect the digital value by 1 bit while using a 2.7 volt reference and ideal circumstances, I don't think it would be detected reliably in a real-world environment.

I'm curious, what voltages do you get from other targets at varying distances from the coil? What about the aluminum foil at 15cm? ( I could perhaps do the calculation on this, but my brain is already sore today. )

Many of the problems that I was running into with the design were related to noise. I had most of the circuit laid out on breadboard, so I was having all sorts of problems with ground loops, cross talk, interference, etc. The biggest contributor to noise in my layout was coming from the LCD drive signals. I'd get a nice burst of digital into the analog circuit each time the display was updated.

I'm going to get back to it soon, as I have some new ideas that I'd like to try. I'm also going to work on each section, one at a time, and build a permanent module when I'm satisfied with each section. That's the nice thing about conventional PI's, they're very modular.

Thanks for the reply. I have a slow internet connection, so to download the datasheet of the dsPIC30takes a long time. So I asked instead.
I have the same problems with my breadboard. Today I had a bad day when I wanted to expose the sensitized copper board to the UV lamp, it did not work. Probably not the right kind of UV lamp. Somehow things just never go right on the first try. Anyway...
If I understand right, at 10 bits, the 3mV would be the lsb. To detect the foil at 30cm would mean a very sensitive detector.
At 15cm the signal amplitude is about 40mV, but to detect this target at 15cm would be very average.
So this means that further amplification is needed.
It is probable that I could get a bit better results at the front-end, I will work on that. In the meantime, I see 2 possibilities:
Use a 12 or 16 bit A/D or a bit more amplification.

Other PIMD's use about 10,000,000 amplification. I am now at 1000, this leaves some scope left.

I found this article, link below, for a different way to A/D, what is your opinion on that?
http://www.robotbuilder.co.uk/Resour...icles/149.aspx

All the best

Tinkerer

[hobbes_lives]

Hi Tinkerer,

Use a 12 or 16 bit A/D or a bit more amplification.

At 2.7 volts, the voltage steps would be 659uV and 41.2uV respectively, which should be more than adequate for your needs. The 10 bit A/D converters in the dsPIC devices can sample at 1 Msps, whereas the 12 bit ones can only sample at 200 ksps. I want the higher speed, due to the type of analysis that I am doing, so the lower resolution one has to suffice for now. There are stand-alone A/D converters that should meet the high sample speed and higher resolution, but I want to do a proof-of-concept with what I have first.

Other PIMD's use about 10,000,000 amplification. I am now at 1000, this leaves some scope left.

But isn't much of that 10M gain after the preamp and in the integrator section? That would be quite allot of gain in the preamp section!

I found this article, link below, for a different way to A/D, what is your opinion on that?
http://www.robotbuilder.co.uk/Resour...icles/149.aspx

It'd be great for slow sampling, as they are doing, but wouldn't work for our application. I think finding a stand-alone A/D converter would be the better option.

You could build a S/H circuit that would hold the voltage to charge the cap, then do the timing, then discharge it, but that would be allot of work for something that can be done much easier and more reliably with a stand-alone A/D converter.

A quick Google search for "a/d 2msps" turned up the AD7266 (2Msps, 12-bit) and AD7641 (2Msps, 18-bit) in the very first links, so I have no doubts that it would be easy to find something that would work well in such an application.

Edit: Here's another one that looks nice and simple: ADS7886

[Tinkerer]

Hi Tinkerer,



At 2.7 volts, the voltage steps would be 659uV and 41.2uV respectively, which should be more than adequate for your needs. The 10 bit A/D converters in the dsPIC devices can sample at 1 Msps, whereas the 12 bit ones can only sample at 200 ksps. I want the higher speed, due to the type of analysis that I am doing, so the lower resolution one has to suffice for now. There are stand-alone A/D converters that should meet the high sample speed and higher resolution, but I want to do a proof-of-concept with what I have first.

At my present 2381 PPS and 4 samples per pulse I don't see the need for a very high sampling rate, when doing a S&H first. I get the FE discrimination at the output of the preamp and the TC of the target calculating the decay between sample 3 and 4.

But isn't much of that 10M gain after the preamp and in the integrator section? That would be quite allot of gain in the preamp section!

Right. I have a gain of 69 *14 for the 3mV at 30cm. So maybe I just try for 69*20 and use the 10 bit A/D.



It'd be great for slow sampling, as they are doing, but wouldn't work for our application. I think finding a stand-alone A/D converter would be the better option.

You could build a S/H circuit that would hold the voltage to charge the cap, then do the timing, then discharge it, but that would be allot of work for something that can be done much easier and more reliably with a stand-alone A/D converter.

A quick Google search for "a/d 2msps" turned up the AD7266 (2Msps, 12-bit) and AD7641 (2Msps, 18-bit) in the very first links, so I have no doubts that it would be easy to find something that would work well in such an application.

Edit: Here's another one that looks nice and simple: ADS7886

I will look at that. thanks.
I need to keep it simple otherwise it will take me years.

All the best

Tinkerer

[taliesin]

Yes, I sample at 2uS. I am keen to see your circuit working to its best, so that I can compare my signal output with yours.
Separate TX and RX windings:
If you make a TX coil of, lets say 28cm diameter and 300uH inductance you have a certain amount of flyback decay time due to the inductance and capacitance of the coil.
Added to that is the decay time caused by the Mosfet Drain capacitance.
Now, if you look at the inductance divided by R 900.
However, the TC for the capacitance is pf times R. So the capacitance of the Mosfet, coax cable and shield added up slow down the decay curve a lot. Specially since the flyback voltage of several hundred V charges all these capacitances up to that voltage.
If you place a fast diode in series with the drain, you can block the drain capacitance and speed up the decay.
If you place most of the damping resistor at the coil, you reduce the flyback voltage on the coax and the Mosfet drain, so again you gain some decay time. Then you add a CERMET POT to do the rest of the damping on the board. This allows you to make fine adjustments and gain more time.
A separate RX coil, lets say with 14cm diameter, 300uH, will be much better for pinpointing. It will show a much reduced flyback voltage. It also needs to be damped but will decay faster. There are quite a few other advantages more.
Now, back to the coil shield:
Aluminum oxidizes instantly when exposed to the air. Aluminum oxide is transparent and characterized by very high insulation capacity. In fact into the thousands of volts. So simply wrapping a copper wire around the alu foil might work for the moment, but depends on constant friction to maintain conductivity.
Alu foil is also a good target, that means that the alu shield on the coil is detected by your circuit. It loads the input.

Make sure you use a spacer between the coil winding and the shield. About 5mm should be OK. You can wrap some of the cheap plastic sting around the coil up to that thickness. The cheapest string is usually made of Polypropylene which is an excellent dielectric. But if you look at the thickness of the string that you build up, you see that it is made up of mostly air. Air is the best dielectric. So then cover the string with painters paper tape and apply the graphite shield to that.

Before applying the shield, add a thin copper drain wire on top of the paper tape. (Later you solder the loose end to the coax shield.) Glue it at a few spots so it stays put while you apply the graphite coating.

Here is my recipe for the graphite shielding:
1cc low viscosity epoxy resin
5cc graphite powder
1cc coloidal or fumed silica
7cc epoxy solvent or thinner.
First mix epoxy, the add graphite and mix again then add the rest and mix well. apply a thin coat. Apply a double coat on the drain wire.
Leave a 1 mm gap in the shield.
After drying, paint pure Epoxy over the shield to keep the moisture out.

If you cant find graphite to buy?

Pick the old graphite brushes out of an old drill or some DC motor like a starter motor. Take some very fine sand paper and sand the graphite brush down until you have enough powder.

Well, this is enough for today, but there is still much more to be done.

Good luck
Tinkerer

just dont get the graphite on the kitchen floor ! i know from bitter experience (avoid the marital)