What is the best frequency response for the preamp in a PI ?

[Detectorist#1]

Hi pito,
Yes, the HF noise is very easy to be filtered with LP filter. But in PI technic the problem is that useful signals and LF noise are in the same band. In single polar mashines, only way is to use low noise OpAmps.

[Detectorist#1]

Hi kt315,
Slew rate of 22V/us for first Opamp is more than enough for fast regular PI projects, but the noise of Opamp is also important in high end projects for small targets. This is pointed from the practice.

[moodz]

Hi pito,
Yes, the HF noise is very easy to be filtered with LP filter. But in PI technic the problem is that useful signals and LF noise are in the same band. In single polar mashines, only way is to use low noise OpAmps.

For the preamp on the PI detector the useful signals are not in the same band.

Referring back to the noise performance of an opamp below .. The wanted target signals might be from 10 Khz to 1Mhz ( depending on how fast your coil is )

... the USELESS signals are
DC error voltages @ DC
Earth Field Induced @ 0.1 - several Hz
Mains hum @ 50 / 60 Hz

When the target signal is demodulated at the output of the preamp these useless signals will be modulated by the demodulation ( sampling ) process and appear in the demodulation ( sampling ) output and due to non linearity of the demodulation ( sampling ) process will appear as intermodulation / convolution on the demodulated ( sampled ) target signals causing errors.
Most every RF designer knows this principle and it is the same in PI. ( its the physics of modulation / demod theory ). However it seems that not many RF designers also design PI circuits.

[moodz]

Hello folks,

a question... I've been looking at the datasheets of all the opamps suggested here. Is it really worth paying about 20 euros for an AD797, for example, compared to about 1 euro for an NE5434? In my opinion, the performance under real conditions is not much better to make it worthwhile. Now the Analog Devices OpAmps are really very good. But from a commercial point of view they are also completely overpriced. A money-making machine for Analog Devices. For the price of an AD797 you can almost buy the components for about 2..3 complete hobby detector kits the size of a Hammer Head PI detector project. Is it really worth it to have an AD797 in the circuit?

What is your opinion?

winner !

[moodz]

where it matters ... the extra dollars for an AD797 only compensates for a less than optimal preamp design.

Its unlikely that a couple of nanovolts at 1KHz to 10Khz will dominate your other noise sources in a typical PI.

[Detectorist#1]

Hi moodz,
"where it matters ... the extra dollars for an AD797 only compensates for a less than optimal preamp design."
Show us what the optimal design of preamp with NE5534 should be and this design will eliminate using AD797.

[Detectorist#1]

Hi moodz,

Process of sampling (demodulation) is non linear process, and on the output will be: f1,f2,nf1+/-mf2.
But we make filtering of these signals: we use 0-appr.5Hz for non motion mode and 0.1hz-appr.5Hz in motion mode. The 1/f noise appear in this band for unipolar Tx pulses. As Carl mentioned, 1/f noise absent only in bipolar Tx pulses.

[moodz]

Hi moodz,

Process of sampling (demodulation) is non linear process, and on the output will be: f1,f2,nf1+/-mf2.
But we make filtering of these signals: we use 0-appr.5Hz for non motion mode and 0.1hz-appr.5Hz in motion mode. The 1/f noise appear in this band for unipolar Tx pulses. As Carl mentioned, 1/f noise absent only in bipolar Tx pulses.

There are two amplifiers in a PI the preamp ( prior ) to demodulation and the sample amplifier / integrator ( post ) demodulation.

SImply put the preamp is dealing with a signal or harmonic of the TX frequency that is modulated with target signals and other unwanted signals. The post amp is dealing with the demodulated signal as you state within say 0.1 to 5 hz ( personally I use 1 to 10 hertz ) .

So the post amp could use a chopper op amp ( which is why the 1/f disappears in the bipolar case also ).

A chopper amp is also more suited for the post amp than an expensive low noise amp. ( eg OPA2188 )

[moodz]

Hi moodz,
"where it matters ... the extra dollars for an AD797 only compensates for a less than optimal preamp design."
Show us what the optimal design of preamp with NE5534 should be and this design will eliminate using AD797.

You accused me in an earlier post of simulating too much and not building real world circuits .

In reality simulations are just "visual mathematics" and following the old proverb of "measure your wood twice before cutting once to build your house.

There is no optimal amplifier .. depends on many factors .. you must design not read catalogues. If you have deep pockets use the AD797.

[detectormods]

If you want the beez kneez PI front end you can base it on a LM318, connect the dff amp inputs to V- and use a large junction bipolar transistors as a diff pair and connect to pin 1 and 8 comp pins tie together 2 ZTX 851 run 6 ma though each transistor and you will have a NF of just under 0.6 Db and 3 Db FQ point around 30 Mhz.

[Detectorist#1]

Hi moodz,
I not see the answer of my question - how to bild optimal design with NE5534 to avoid using of AD797 in the preamp stage in PI design. Using of OPA2188 after demodulation is not nessesery - the level of the signal is far more than noise at this point.

[Detectorist#1]

Hi detectormods,

This idea to use midle power transistors before Opamp is very old. The problem is that this very low level of the noise is possible only when you use very low values of the resistors as the source of the signal. The PI first stage is not those case. MD SD2000 from Minelab use that idea, but the result is moderate - not fantastiq.

[detectormods]

Hi detectormods,

This idea to use midle power transistors before Opamp is very old. The problem is that this very low level of the noise is possible only when you use very low values of the resistors as the source of the signal. The PI first stage is not those case. MD SD2000 from Minelab use that idea, but the result is moderate - not fantastiq.

Old but works very well, if you want to aim at 0.2 Db NF at 10Hz there are only a few cheap ways to do it.

[moodz]

Old but works very well, if you want to aim at 0.2 Db NF at 10Hz there are only a few cheap ways to do it.

Woody and Dectroist#1 you are both missing the point of this thread ... its not about the design of preamps ... its about the required frequency response of the preamp.

There is no 10 Hz or other frequencies of interest below the primary TX frequency in a PI at the coil preamp level. ( unless your TX frequency was 10 Hz )

The target signals only become low frequency when they are demodulated from the TX carrier.

Below is the spectrum of a PI operating at 10 Khz ... figure it out.

[detectormods]

Woody and Dectroist#1 you are both missing the point of this thread ... its not about the design of preamps ... its about the required frequency response of the preamp.

There is no 10 Hz or other frequencies of interest below the primary TX frequency in a PI at the coil preamp level. ( unless your TX frequency was 10 Hz )

The target signals only become low frequency when they are demodulated from the TX carrier.

Below is the spectrum of a PI operating at 10 Khz ... figure it out.

So you are using a digital based brick wall type filters as to completely stop any signal below 10 Khz?
Interesting.
An AD822 would work well.