SD2000 easy 10% greater depth mod.

[Woody.au]

I have been experimenting with different configurations for the SD2000 front end and after doing a couple of trials with different components an easy 10% increase in depth was noticed. The front end switching fet VN2410L was replaced with Zetex Fet ZVN4424A and the depth and sensitivity both increased. The specifications on both parts are near identical except that low RDS on is achieved with half the gate threshold voltage with the ZVN4424.
So why did ML not use this fet to start with? The fet was only released in late 1994 after ML finished the original design.

[Reg]

Hi Woody,

What are they doing with the FET? Do you have a schematic showing the main power FET and associated circuitry to the front end? I am curious about the extra FET's and how they are connected.

BTW, does anyone have the front end schematic of the later SD's or GP's? I am curious about them.

Reg

[Woody.au]

Hi Reg,

You changed your email as it keeps bouncing.....

The N channel fet on the front end is switched off during flyback and decay then switched on to allow the receive signal to pass. This is then connected to a P channel fet that feeds the input of the first ampifier...SD2000 -2100 is a LM394 and later versions to a AD797. The P channel also feeds a timing pulse to the timing electronics.

Minelab make most of their gains via getting rid of the series 1K ohm dropping resistor that is used on other P.I front end designs. The noise from a 1K resistor would be 4nv/hertz/2 but a bipolar input stage needs to take into account current noise that can be up to 100nv/hz/2. Even the AD797 is not optimum because when the device is open circuited due to turn off via the front end fet the AD797 goes into unstable oscillation and noise generation that is not rapidly brought under control by sourcing a low impedance path.

There is plenty to do on a P.I front end but the Candy approach is in the right direction.

Just by shunting the input of a AD797 with 5K ohms the noise decreases somewhat. There is plenty of scope for more experimentation on P.I front ends.

[Reg]

Hi Woody,

Thanks for the info,

I haven't changed my email address, so I am not sure what is going on. I should be getting all emails sent to me. I will double check my settings but nothing should be stopped that I know of.

Thanks again,

Reg

[Woody.au]

I have a R..S.... at comcast.com is that correct?

[Reg]

Hi Woody,

No, it is at comcast.net. You can also reach me by simply sending me a private message on this site. Comcast.com will bounce back.

Reg

[JC1]

Current Noise

Hi Reg,

You changed your email as it keeps bouncing.....

The N channel fet on the front end is switched off during flyback and decay then switched on to allow the receive signal to pass. This is then connected to a P channel fet that feeds the input of the first ampifier...SD2000 -2100 is a LM394 and later versions to a AD797. The P channel also feeds a timing pulse to the timing electronics.

Minelab make most of their gains via getting rid of the series 1K ohm dropping resistor that is used on other P.I front end designs. The noise from a 1K resistor would be 4nv/hertz/2 but a bipolar input stage needs to take into account current noise that can be up to 100nv/hz/2. Even the AD797 is not optimum because when the device is open circuited due to turn off via the front end fet the AD797 goes into unstable oscillation and noise generation that is not rapidly brought under control by sourcing a low impedance path.

There is plenty to do on a P.I front end but the Candy approach is in the right direction.

Just by shunting the input of a AD797 with 5K ohms the noise decreases somewhat. There is plenty of scope for more experimentation on P.I front ends.

The noise from a 1K resistor would be 4nv/hertz/2 but a bipolar input stage needs to take into account current noise that can be up to 100nv/hz/2.

Uh current noise should be in amps and not volts but sure

glad to see someone mention current noise, as all the noise

talk has be voltage.

Did you mean nA?

Of course you did.

What is the current noise in a 1 kohm resistor?

[JC1]

answer

depends on the bandwidth.

ok 100 kHz bandwidth.

what is the current noise in a 1Kohm resistor?

[Woody.au]

http://www.nikhef.nl/~jds/vlsi/noise/basicnoise.PDF

[Carl-NC]

what is the current noise in a 1Kohm resistor?

4nV/rtHz divided by 1k. And it's all the same noise.

[Mr. Bean]

Input

HI Woody,

Thanks for input, I will have to give that a read sometime.

The discrete Fet input has been discussed before and I like

the way you think. Or maybe the way Dr. Candy thinks.

Those LSK389s according to the data sheet can still get a

bit noisey max spec and perhaps can find a few a bit

better. But is still a good way to go, the fet that is.

[JC1]

Carl wins the Cigar Again !!!

4nV/rtHz divided by 1k. And it's all the same noise.

Give that man a Cigar.

He deserves it.

And the square root of 100 kHz is what?

316

and that times 4nV/1K is what?

1.2 nA

And that current may go around a 1 Meg feedback

in a 5534 set up.

and that voltage is what?

1.2 millivolts.

Yuk !!!

[Mr. Bean]

Not on the op amp

Hey Carl,

On the op amp inputs its not the same thing

and you have to calculate both.

And then add them.

Right?

[Mr. Bean]

Before I forget

Here is a link about keeping the old supply lines clean

might be important after low noising the front end, maybe

not, maybe there is enough PSRR, could be dangerous on

the discrete stuff, who knows?

http://www.wenzel.com/documents/finesse.html

[Mr. Bean]

Hmmm,,,,,,,,,,

Hi Woody,

Well I still am not sure what you meant with the current

noise in a 1 Kohm resistor 4 pA/rtHz and a bipolar input

op amp. The NE5534 is common bipolar op amp and it is

1.5 pA/rtHz.

Anyway so 100nv was not 100nA or 100pA?

Not sure what you meant.