Designing a PI.

[Bill]

Hi all,
I have been working on the front end of the Hammerhead PI trying to produce better results with different types of coil dampning and active pull down. I spent a few years developing linear accelrators and quick collapse is whats needed there too. I have decided on using a diagnal half-bridge topology to shorten collapse times. So far results are good. The target metal I am after is gold. My goal is to build a unit whoose characteristics lend itself to finding gold., ie short early sample, fast collapse, etc.. small size at a decent depth. My question is this, what benifits does a DD coil have compared to a mono, if any, in PI use? I noticed alot of the Minelab Gold machines used a DD. How does the DD work in a PI. One coil is Tx and the other Rx? I am writting up a nice article on the improvements I am making and will post it here when its done. Thanks for the help!
Will

[Carl-NC]

Glad to hear your working on improvements. I did a little bit with active pulldown and found only a modest improvement in turn-off time, as reflected in the peak flyback voltage. Probably a reduction in load capacitance (coil/cable/MOSFET, etc) would bring better results.

DD PI can be set up two ways. One is by pulsing & differentially sampling both coils simultaneously, see Gary's project or the Twinloop article. This has the benefit of canceling noise. Or, have one coil as TX and one as RX. For fast sampling, this is the way to go.

Look forward to your article.

- Carl

[Bill]

I will try the 1 for Tx and 1 for Rx. Anything that will allow a earlier sampling time. The diagnal half bridge does make a difference, not big, but every little gain here and there will add up. Has anyone tried using germainium diodes for the clapms on the Rx end? Any benefits? I am attacking the Hammerhead one system at a time. My overall goal is to make a version thats better suited for smaller Au tragets. I bought a PCB from Carl and will assemble it to stock specs. This way I have a baseline and can tell if the changes I make help. Thanks for the help! Any ideas?

William

[bbsailor]

Designing a PI.

DD coils can use one coil for TX and one coil for RX. The Input resistor R12 (1K) needs to be disconnected from the coil and connected to the center of the RX coax. Use twin coax (SVHS wire) and keep the shields separate until you tie them the together at the circuit board. Consider using a 4-pin mic connector for the DD coil.

The one benefit of the DD coi is that the RX coil can be a little faster than the same size mono coil because the RX coil does not see the MOSFET capacitance which can be from 50 to 300 pF depending on the MOSFET you use. In this case, less capacitance is better and results in a higher value higher resonant frequency and higher damping resistor value.

Use a shield that cannot be deteted by a coil at a fast delay setting. I use 3M 1190 copper fabric tape. It comes in 1" and 3/4" " widths and is self adhesive. Use a plastic spiral wrap spacer between the coil wire and the shield to reduce capacitance and keep the coil self-resonance as high as possible. Coil, shield and coax should be resonant around 600Khz plus or minus 50 Khz for a 300 to 320 uH coil.

Use the coil housings from Hays Electronics. They make homemade coils look professional.

bbsailor

[Bill]

I got my coil housings in from them on Saturday. Very nice and fast shipping! I bought two diff diameters. I am using 1200V 60A IGBTs for my switches as well as a Highspeed driver IC to drive them. Thanks for the info!

[Bill]

BTW, what switch capacitance value makes the most difference? When I build the stock hammerhead as a bench mark I want to use the best parts possible. Is it: input(Ciss) capacitance, output capacitance(Coss), or reverse transfer capacitance(Crss)
Thanks for the help guys!
William

[Unregistered]

THANX WILLIAM,
I THINK YOUR PRODUCT AND WAY OF REASONING FOR A REALLY SIMPLE,GOLD-ONLY DETECTOR IS LONG TIME OVERDUE AND I WISH YOU GOOD LUCK!!!!
HOPE YOU COMPLETE THE PROJECT VERY SOON
CHRIS

[treasurediver]

Have tried Germanium diodes. Seem to get more noise level, but tests were flawen at another level, as I found out later.
Treasurediver

[Bill]

I got my PCB from Carl on Friday and have completed assembly of my base model. (Thanks Carl!) The PCB is an excellent make, who builds them for you BTW, great work! Per Carls advice I used polystyrene caps for all non polarized caps. Basic IRF740 is the switch since thats what most people use. The housing is a nice cast aluminum unit with gaskets and removable front and back endplates. Batts are a shrinkwraped unit consisting of 10 2300mAh AA NimH=12VDC. I built a few of these. I also purchased about 5 small gold nuggets, ranging from .5 grams to 2.5 grams. The whole purpose for this unit is for a baseline. I will get it tuned the best I can for the gold nuggets and track my results. Then I can tell if the mods I made on the HH/Au unit help or not. I have a feeling that a proper coil is going to make a ton of difference. I plan on using a minelab DD. Heres my question du jur, to make coil removal easier, can I mount a jack on the grounded enclousre, and plug the coil leads into it? Or would this cause some noise issues? I wouldnt think so but I figured I would ask the experts. This would make comparisons accurate since I can use the same coil for both rigs. Thanks guys!

William

[Carl-NC]

"Has anyone tried using germainium diodes for the clapms on the Rx end?"

I've tried Schottkys, they did not work as well as the 4148s.

"The PCB is an excellent make, who builds them for you BTW, great work!"

PCBs are made by Imagineering (www.pcbnet.com).

Not sure about grounding the enclosure, it might cause noise problems. Just make it easy to disconnect and try it both ways. If you ground the case, I would not connect the "ground" lead of the coil connector to the case, keep the coil ground isolated all the way to the PCB.

- Carl

[FJ-FJ]

PI for small gold

WilliamN,
I'm curious why you are using such heavy switches? For what it sounds like you are building, you don't need anything near that heavy. I'm wondering what you are expecting to gain. Obviously the most important design criteria for your goal is going to be going to the shortest possible pulse delay you can for the small nuggets. If this is the case you should be choosing switches with low capacitance. The 740 FETs can be bought in low capacitance versions. There are similar FETs now with much lower on resistance and low capacitance as well. Maybe the IGBTs you chose have a much lower capacitance and on resistance and you just haven't mentioned it. It's just that the specs you do mention really don't seem to have any consequence for what you are trying to accomplish. Just curious.
Good luck,
FJ

[Bill]

Yes they have very low, Ciss, Crss, and Coss. Besides, they are also being used in a different topology. Most people use a single fet or flyback topology. I am using a completely different method (two switches, a kind of quicker active pulldown). I stand to gain a little(I mean little) faster collapse which means earlier sampling times. Carl- I have found in my research that it is best to have the coil wires only connected at the PCB as you have stated, which I will do. As for the grounded case, doesnt everyone ground their cases? I would think it help keep out RFI.

William