field test unit no 001 "model T"

[moodz]

dspic30f4012 UNIPI fitted with 6 channel 24 bit converter + programmable gain amplifiers.

PS dont try to build this ... you dont have the hex yet ... it is still under test ... and the audio output is missing ( need to allocate another pin ).

[hdphilip]

hello Paul

i would have figured you would have used the 4011 chip for sure, i've been using the zero button, also the audio output, which is interesting to hear.

Philip

[moodz]

The code base is the same for 4011 I just need to update schematic for the 4011.
Moodz

[satdaveuk]

Hi
Looking forward to it
I got a couple of IC2s from USA and LCD displays and programmer from hongkong on the way, already got 4066s somewhere here.
Good work

Regards

[hdphilip]

progress

Well,
here's my progress so far on the Rev oo3 front end connected to the 4012 processor.

Remember... This is a developmental thread,
And this is just my findings so far...

build notes:

for the driver IC's i used TC4429 (non inverting) to drive the mosfet and the 4066 and the TC4429 for the bias generators. ( 6 amp drivers instead of the 9 amp). the TC4429/TC4429 are cheaper

C1 i started out using the 10uf but changed it to a 4.7 uf to speed up the damping settling.

1.circuit function:

a.ground balance not used.

b.descrim balance in damping adjust) This circuit works great! using the menu on th 4012, i could fine tune the balance/damping with ease. X1 on the menu made it perfect!

c. bias generator: works great! although i had to replace the 78L05/79L05 when the blocking mosfet shorted.

d. the 4066 and the THAT 1510 worked (didn't blow them up)

e. Series blocking FET's: at first, i figured this area of the circuit would be difficult. which it was, as you can see from my previous posting i made several "plug in" boards so i could experiment with, which turned out to be a good idea.

my findings

the coil i used was made from cat 5 , similar to the coil moodz made.

The coil balanced fine, and before i attempted to try out the blocking fet arrangement i used my 1k resistor/ diode module and the circuit seen here: http://www.geotech1.com/forums/showp...73&postcount=1

which seemed to work well.

Series blocking FET's

When i tried to use the mosfets, at first i tried some (smd)supertex tn2130/tp2635 they got hot..shorted

so i made some different plug in modules for my next sets i ordered. tried some TO-220
400v p and n channel, gate to drain short on Q4, which also blew the 79L05.

OK, next up to be sacrificed would be the bsp230 /bsp130 still no luck Q3 took a hit this time.

i haven't tried the phc2300 yet no point trying them yet if the bsp230 /bsp130 failed.

Thoughts on the blocking FET's.

I would like to do a circuit Sim on the blocking FET's before i try the ones i have left, also i was thinking maybe it would bee a good idea to put a Damping resistor across the coil, I'm thinking the FET's are blowing before the Q5 damping mosfet has settled .

Any thoughts on that?

So i put the 1K/diodes module in and ran some more tests.

sense this board only has 1 sampling window i didn't expect much for discrimination between ferrous/ non ferrous which seemed to hold true.

by experimenting around, i hooked my scope probe on the output (c17) i found by adjusting the damping, i could change the DC level out.

which got me to thinking, maybe a DC balance control on the THAT 1510 might be a good idea, so when we "tune up" this circuit, we would adjust the offset first, then attach the coil, then fine tune the damping, for a null on the output. so that may be a "true balance"

Any thoughts on that idea?

did hook up a speaker, i used a 330 resistor in series with the audio out of the 4012.

at first thought "oh this is a mess" it was screaming all over the place. while it was hooked up to the that 1510. I adjust the audio on the menu, no help. thought about it, OK then i shorted the input of the ADC to ground (after unhooking from the THAT), so the display would be reading zero's all across the screen, then i adjust the "audio" on the menu, well, at about 8180, i heard the "clicking " sound Paul has talk about in the past. like a Geiger counter. so, i reconnected to the That, now the audio was much better.

it did pick up a gold nugget at about 5" and a piece of PCB at 12"

Also i found that hooking a trim pot on the input of the ADC help the digits on the LCD be a little more stable which improves the audio of the device, maybe a threshold adjustment would help.

Any thoughts?

Philip

[moodz]

Good work Phillip ....

Some comments ( may not in order of yours )

The audio code needs an update but good to see you were able to hear the geiger clicks.
I will try and make it a bit smoother and probably auto set.

The blocking FETs I have not extensively tested other than noting the BSP type tend to blow up easily whereas the PHC type were more robust. The main problem is that you dont want to be drawing any significant current ( eg milliamps even ) through these FETS or the bias will be upset and the gates punctures shorting out to the bias supply. It may turn out that the resistor diode circuit is the one to go with ..... I dont think the damping mosfet is implicated because the gate bias keeps it turned on whereas the blocking FETS simply block if the input voltage exceeds the pinch off voltage of the gate ... unless there is a negative swing on the +ve side of the coil and vice versa on the other side. While you are experimenting just use a heavy power fet as the blocking device .. the source / gate capacitance will be a bit excessive but it works fine and quite hard to blow up ... none of mine did.

The DC level out ( of the THAT ) should change as the inputs are adjusted by adjusting the sample start time, sample duration and damping. You should be able to see descriminatinion or target response here or both. ( voltage can swing +ve for copper / -ve for ferrous etc ).

The ADC input range is 0 to 5 volts .... ideally you want around 2.5 volts with no signal .... the output of the THAT is not exactlty in that range so a level shift of some sort is required ...either op amp or capacitor/diodes. Note the input of the ADC wants to be driven by a source impedance of no more than 500 ohms so a buffer amp is good.

moodz

[hdphilip]

audio

hello Paul

i did some testing on the audio, i hooked up a 1k pot between +5 and gnd,i was moving the pot, trying to simulate the detector, it's pretty rough. i was thinking, if the rotary encoder when not in menu mode, could serve as a sensitivity control, it could adjust the dynamic range of the audio, the sat seems to work fine, so that would set the lower end. and the rotary encoder would set the upper end. the max upper range would be another menu item.

just a thought

Philip

i would like the use the 4012 as a back end to another front end idea i have.

[Davor]

How about using the very 4066 as a chopper to stabilise amplification? Should fix the ground reference too. Plus maybe an opportunity for discrimination?

[moodz]

How about using the very 4066 as a chopper to stabilise amplification? Should fix the ground reference too. Plus maybe an opportunity for discrimination?

I take it you mean chopping as in chopper amps where the aim is combatting dc offset and enhancing dc accuracy. The 4066 perform a demod function ... Chopper amps are not particularly low noise for dynamic signals .. The aim is not gain but dynamic range that's why my designs are moving towards 24 bit ADCs. ... The target frequencies band is quite low ...the ground is typically the lowest frequency ... It's the biggest target after all. ( tinkerer is has the right idea )

Moodz

[Davor]

I should have watched your schematic more carefully. It IS a chopper, or actually a SAH. You supply opamp with large C7 and C8 that narrow the bandwidth to a sensibly low region, providing in effect averaging of successive samples and widening the dynamic range. With counter phases in input there is no point in additional stabilisation. This is a way to go

BTW, opamps can be convinced to work as perfectly balanced amplifiers with symmetric but finite input impedance. Your SAH-style demodulator runs on high impedance inputs, so I guess you are not too interested in this solution.

When I saw your limiters my first thought were diode limiters, but they would be noisier with their DC biased resistors. Could prove cheaper though.

I'll follow this topic more closely, as this is the closest thing to the perfect PI that I've seen so far.

[hdphilip]

Davor

This is an insteresting PI setup, one of the reasons why I'm building this, I like the way moodz uses the 4066 in the input stage, to me it makes perfect sense. let the op-amp only be exposed and deal with the information we wish to sample. my next step is to build a level offset/buffer to drive the ADC correctly.
Also I'm in the process of making a "hammerhead" style pi setup with a 4066 in the front end too.

Philip

[Davor]

I think you can even differentiate it to gain motion filter functionality, but heck - what is the use of ADC then? Let the micro crunch it for you in FIR style

[hdphilip]

I think thats what moodz is working on this very moment.

[6666]

Hdphilip
I looked at the " diode module" idea
been wondering what the function of diodes D2,D4, D11, D12 is
and why they are connected to VPP, are they part of the bias generator
are they needed if not useing the bias generator ?

[hdphilip]

hello 6666,
The diodes,
D2,D4, D11, D12 they serve to knock down the fly back spike coming from the coil, those 1K resistors should be 1% and at least 1/2 watt they do get very warm.

The "VPP", even though it's at +12 , it's serving as "gnd" in the analog processing chain, if you look at pin 6 on the THAT 1510 chip, it's also going to VPP which sets the ground reference, So with that being said, when i look at signals with my O-scope i hook the ground on the test probe to VPP as it is my ground reference.
on my o scope, the gnd if floating, if it wasn't, I'd short out my bench power supply.

Philip