SandShark details

[Anonymous]

OK, I re-opened my SandShark, measured the following:
Transmit pulse width: 71.5us
Main sample delay: 22.8us
Main sample pulse width: 21.2us
The SS uses an 80-volt PNP for the coil switch, and a PIC16c73B-04 microcontroller run with a 3.58MHz ceramic resonator (Matsushita).
To decrease the sample delay to around 15us, the main clock needs to increase to around 5MHz. However, the microcontroller is a 4MHz-max type, so this will overclock the part by 25%. I personally don't have a problem trying this, but I don't recommend it to anyone else. Matsushita does not make a 5MHz resonator, so I'll have to find something else that will work, or just use an external pulse source.
I checked the TX pulse, and it was solidly clipping at 80v. So I swapped the PNP for a 200-volt IRF9630 PMOS, now I get a 200-v pulse. No change in detection depth, which could be due to the high coil resistance of the printed spiral coil. So I'd like to try a wire-wound coil, and re-optimize the damping. It's also possible that the base/gate drive circuitry is limiting the turn-on of the PMOS, as it was designed for the PNP.
So, I suspect that, like the SurfMaster PI, the performance of the SandShark can be improved upon, but it might require getting rid of the printed spiral coils.
- Carl

 

[Anonymous]

Keeping in mind as the clock frequency is increased, not only will the pulse delay be shorter, so will the transmit on time, the sample gate times, and the delay between samples assuming a differential integrator is used, and an increase in the pulse frequency. It will be interesting to see how this affects things. You may find that shorting the pulse delay is a plus, decreasing the pulse duration is a minus, increasing the pulse frequency is a plus and everything is a wash in the end. What do you think? Let us know.
Thanks,
FJ

 

[Anonymous]

I expect that decreasing the transmit pulse width is a minus, but that can be compensated for by different coil parameters. The secondary sample delay is not a critical factor, and the sample pulse width seems to already be plenty, so reducing it should not hurt.
The biggest potential problem I see is the audio frequency will increase, and end up sounding like a pig sqealing. Could then rename it, "Tesoro SandPig".
- Carl

 

[Anonymous]

Carl,
The transmit pulse width does not have to be very long in order to detect the low conductive targets that the shorter sample delay time will bring into range. The TC of a nickel or thin ring is very short so a short transmit pulse is OK. Eric is using a 30uS transmit pulse in his newer designs. It has been my experience that one can easily run most PIC microcontrollers with a 50% increase over their stated maximum speed. It is also possible to replace the PIC in the Sandshark with a PIC16F84 (flash version) and write ones own code. The newer chip runs at 10MHz @ 5 Volts.

 

[Anonymous]

I sure don't want to get into replacing the PIC, as it is a SMD. Plus, I'm looking at what very simple changes can be done, to improve performance. Swapping the resonator, and maybe a few resistors.
The transmit pulse width affects whether the coil is fully "charged" or not (max I), which then affects depth (dI/dt). My concern is that, by shortening the TX pulse by 50%, the coil might not be at the peak I at the end of the TX pulse.
- Carl