Pre Amplifier Design for Geophone GS-20DX

[sanwal209]

Hi Fellaws,

I need help in designing pre amplifier shield of Geophone(GS-20DX By Geospace Technologies) for my main new project. After a lot of search I have found a reference design by Linear Technology but its based on remote 2-wire current loop. Is it possible to convert this design into simple voltage amplifier? I can use this complete design as it is but i am not sure about either its going to effect the accuracy.

If you guys suggests to use the same reference design than please do refer me to the circuit or IC for converting the current variation into voltage which eventually fed to ADC of MCU.

Please help me out here.

I have also attached the reference design.

[Marco]

It's a simple inverting opamp configuration with a lot of extra circuitry to turn it into a current loop device.

Basically nothing in there is useful to you.

[sanwal209]

Can you please suggest some documentation or schematics?

[David Hess]

Keep U3 and voltage divider R6 and R7, remove Q3, and move R9 across C3.  Throw out everything else and take the amplified signal from the output of U3.

Geospace Technologies are jerks about providing documentation so I cannot say if the GS-20DX has any loading requirements.  I suspect not so any inverting or non-inverting amplifier configuration will work.

[sanwal209]

Totally agree they really are jerks. The datasheet of GS-20X is only of two pages . Thanks for the help. What about filters as there are many designs online which involves number filters in series.

[salbayeng]

Typical LT circuit, lots of overpriced parts that don't do anything useful.
U3 inverting input is a virtual earth, so the geophone is loaded with 866ohm (technically it's loaded with 1500ohms when you include it's internal 630ohm resistance). (In this mode the geophone is acting more like an accelerometer).
This means it will behave differently to the case where it is driving a high impedance amplifier. (Where it will behave like a velocity sensor)
If you are looking for low frequency signals, then you will do better with a high impedance amplifier. But the geophone will bang against the end stops with a large stimulus.
Geophones can have really lousy frequency response, at the low end the internal spring limits the low frequency response , that will be the 10Hz in the spec, and higher up there is a pole in the response at L/R where L is the geophone inductance, and R is the load resistance.
Note that geophones designed for vertical use have stiffer springs than those designed for horizontal use.
Given the impedance of ~ 600ohms, you'd probably do well with any dynamic microphone amplifier circuit.

[sanwal209]

Thanks @Salbayeng Target application of the project is to detect the Foot steps(Intrusion).  I have found another preamp schematic and i think this is more better as compared to the last one. All i need to do is to simulate it and check the results. Please check the attachment.

[salbayeng]

There's a datasheet at Sparkfun https://www.sparkfun.com/products/11744 for a similar geophone, that shows the effect of loading resistance. The 10Hz response is for vertical geophones. Horizontal ones are ~ 3Hz from memory.
The guys in our seismic group simply soldered a resistor across the terminals (Around 1000ohms, varies by brand) , then just used ordinary low noise amplifiers, one thing to watch for is triboelectric and piezo electric voltages generated in the cables, low noise mic cable is the best (e.g. the ones with semiconductive screen).  I used them for velocity feedback on some servo hydraulic actuators, and I had to add a pole-zero compensator to push the lower frequency down to 1Hz.

[sanwal209]

Thanks man I really appreciate it. As soon as i will finish the project i will do update here.

[David Hess]

What about filters as there are many designs online which involves number filters in series.

I was keeping it simple.  C3 (1000pF) across R9 (150k) creates a single pole low pass 3dB point of 1kHz.  Additional filtering stages could be added after that.

The impedance of the geophone is low at 630 ohms so bipolar transistor input operational amplifiers can be used for lowest noise which explains the selection of the LT1677.  The rail-to-rail inputs and outputs of the LT1677 allow 3 volt operation but a higher voltage design could use something like an LT1007, OP-27, or maybe LT1028/LT1128.  Modern low noise audio operational amplifiers would work well but may have higher drift.

I agree with salbayeng about using a high input impedance non-inverting amplifier which is a trivial change but without data on the geophone, I assumed the 1500 ohm load was intended for performance reasons because it explains the rather odd value of the 866 ohm input resistor.

I used them for velocity feedback on some servo hydraulic actuators, and I had to add a pole-zero compensator to push the lower frequency down to 1Hz.

Long ago right out of college I did the same thing with some dampened inclinometers to push their frequency response higher and minimize phase error.  Of course doing so raised the high frequency noise unacceptably so following the pole-zero compensation was a 4th order low pass filter.

[salbayeng]

Ok The Johnson schematic is definitely "more better"  , this uses a 1k damping resistor R2,  you may actually find you get better footstep response by leaving off R2.
Note the LMC6462 is a CMOS device, so won't work at more than 5v. If you can't find the esoteric voltage regulator, then you can use a micropower 5v reference chip.
If you are not constrained by power drain, then the OP27 is a reliable low noise workhorse for this sort of circuit.
R5/C5 give the first gain stage a 160Hz rolloff.

The 3rd order 150Hz filter is straight forward: R8/C8 is your simple RC rolloff, typically you would follow this passive filter with something 10x higher in impedance, and there it is 68k.
The second op-amp is a classic sallen-key filter R9/C9 with R10/C10  to which he has added some gain( x 10) by using the voltage divider R11&R13 , c11 just makes it AC coupled.
R6/C6 seem a bit superflous, you would get the same effect by making R7 from two 4M7 resistors splitting the rails.

That's a 4th order filter at 150Hz all up,  so he must be expecting a lot of high frequency noise.
If you are using the geophone on a hard surface (i.e. not pushed into dirt) then it can rattle a bit, best to use plasticine, or blu-tak (poster adhesive), or a cork or rubber gasket to attach to a floor.
Note that all floors have various structural modes, so moving the sensor a foot in any direction can make a big difference in its response to unwanted noise vs desired signal.
You can usually feel the structural modes with your hands.

If you put it outside in the dirt , stay at least 10feet away from trees,  they put a lot of siesmic noise into the ground when the branches hit in a strong wind (shrubs are OK).

[salbayeng]

OK , you might be wondering how the likes of David and myself can sprout out these rolloff frequencies off the top of our heads, here's how I do it, see attached reactance chart (these aren't new, first used in the 1930's by valve jockeys)  I found one in the data section of a 1977 Electronics Australia yearbook, but eventually photocopies of photocopies get a  bit ratty, so I've done a new one:

Given two of resistance , frequency, capacitance or inductance , just go to the intersection to find the other two variables to give you a RLC circuit with a Q of one.
so for example 1nF and 1k gives 160kHz rolloff (just above middle of page) so there's a high pass or low pass rolloff for you.
Say you want to make a Q of 10 tank circuit at 160kHz - that's 1mH and 1nF to get 160kHz ,there are many other combinations, this yields a reactance of 1kohm. So we just need to scale this impedance by 10 to get a Q of 10, two ways : either a 10k resistor across the parallelled L & C  or a series 100R in one leg.

The second page is just zoomed into a couple of squares, so you can get a more accurate value.

[David Hess]

Note the LMC6462 is a CMOS device, so won't work at more than 5v. If you can't find the esoteric voltage regulator, then you can use a micropower 5v reference chip.

The LMC6462 and all of this family of amplifiers are good to 15 volts.  They may however be unsuitable because of high voltage noise.  The LMC6482 is a little bit better as far as noise at the expense of power but it is still horrible compared to a bipolar amplifier.

The claim to fame for the LMC64xx series is that they were one of the early rail-to-rail input designs but they have horrible CMRR even for a CMOS amplifier.  These amplifiers can also have very low input bias current; we used to select LMC6081s (not a rail-to-rail input so it has reasonable CMRR) for 2fA input current or about 6000 electrons per second.

[salbayeng]

@David , oops opened mouth without checking, assumed it was one of the modern 5v series of CMOS (MCP6071??).
Isn't the LMC6462 a Bob Pease / Jim Williams design?, I read about it in one of their books somewhere.
We did a lot of comparisons years back, and a lot of testing at lower frequencies with low source resistance, and found the OP27E was the best all round, (Hard to beat without making a front end out of 8 or 16 parallelled transistor pairs), then we hand selected for lowest noise around 1Hz.

[Kleinstein]

With a 600-1000 Ohms source impedance the OP27 is about the right type of OP.   For just detecting foot-steps this should be good enough. CMOS OPs tend to be rather noise in the low frequency range.

The filtering depends one the way you look at the signal. The upper frequency limit can likely be set in software. So no more than a kind of anti aliasing filter is needed here. The low frequency limit depends on the expected signal from foot-steps.  So it might be that you don't need the very low frequencies and could leave them out.

[David Hess]

Jim Williams had already been at Linear Technology for years when the LMC6462 came out so I do not think he had anything to do with it.  Bob Pease at least worked on testing and applications for the LMC6462 but I do not know if he had much to do with the semiconductor design.

The OP27 is definitely one of the greats and I use them or their equivalents like the LT1007 a lot in low noise regulators and references.  The AD797/LT1028/LT1128 and similar are lower noise yet and suitable for geophone applications where the sensor impedance is so low.

These low noise amplifiers can be pretty expensive with the OP27 probably being the cheapest of the bunch these days but a hybrid discrete input design can achieve low noise at less cost if that is a factor although then, DC coupling can become a problem.

With a 600-1000 Ohms source impedance the OP27 is about the right type of OP.   For just detecting foot-steps this should be good enough. CMOS OPs tend to be rather noise in the low frequency range.

Even the best of the bunch in that series of CMOS amplifiers, the LMC6081, is 5 times noisier than the OP-27/LT1007 and 20 times noisier than the AD797/LT1028/LT1128.  At low frequencies the difference is even more pronounced.  The geophone resistance of 630 ohms is higher than optimum for the AD797/LT1028/LT1128 but not by much and there are modern devices which would be a closer match.  The OP27 is about the least expensive of the good options because it has several sources.

[sanwal209]

Hi,
Sorry for late reply as i was banging my head against ICP20P. Found Broken RF Mimic Transistor. Anyhow I have read all the comments by both of you guys i must say you guys have impressive skill set.

Regarding Design, My Collegue have working DIY geophone at his home and I have discussed with about the design he suggested OP27 just like @Salbyeng and also AD797. Now both doesn't work on single supply 3.3V, guess i need to add dc to dc converter. I have also attached the schematic drawn by the help of my colleague.

Thanks to all of you for helping me out. I really appreciate it

[Kleinstein]

There are not that many BJT based OPs that work well at 3.3 V.

One low noise candidate for a low supply voltage (e.g. starting at 4.5 V) is the OPA209.
An LMP7731 could would from 2 V if needed.

[David Hess]

Regarding Design, My Collegue have working DIY geophone at his home and I have discussed with about the design he suggested OP27 just like @Salbyeng and also AD797. Now both doesn't work on single supply 3.3V, guess i need to add dc to dc converter. I have also attached the schematic drawn by the help of my colleague.

If you do not already have a high voltage supply available, then the original selection of the LT1677 which is basically an OP27 optimized for low supply voltages is more suitable.  I am not that familiar with the low voltage options so let's see ... options which will work down to 3 volts include:

LT1677
OP184
ADA4084-1

There are some others but I think these will be the easiest to use.

Update 1: The LMP7731 is another candidate.
Update 2: With an inverting design, there are other better performing candidate amplifiers which only have a rail-to-rail output.

[sanwal209]

I found this beauty LMH6629. Ultra low noise (Input Voltage Noise 0.69 nV/?Hz|). Next step is to design 2nd order active low pass filter. The target market of this project dont bother regarding price as they need accuracy and reliable system.

Let me know your review about above mentioned amplifier.

[salbayeng]

Whoa there, you've fallen into a newbie trap!.
There are a lot of "low noise, high frequency" op-amps out there, particularly in the 5vss class.
But... One of the critical parameters of any amplifier for geophone/magnetometer applications is the 1/f noise corner.
If you check out figures 28 and 29 in the data sheet you will see the 1/F corner is ~ 3khz. so you need to multiply the noise figure by 20 at 150Hz and 200 at 15Hz.  The noise curves for the OP27 look similar from a distance, but the X axis is in Hz not kHz.

There's a good app note out there somewhere that explains all this and has noise vs source resistance vs frequency plots for a couple of op amps. I'll have a look for it shortly.

[David Hess]

I deliberately avoided suggesting amplifiers which are only available in tiny packages and have excessive bandwidth which will make using them more difficult.

The LMH6629 is optimized for an input resistance even lower than the LT1028/LT1128 is and its low frequency noise is much worse because of a high 1/f corner point.  It is really intended for 50 ohm applications.

As pointed out earlier, the low voltage OP-27 alternatives are a sure thing.

Update: salbayeng, you might have been thinking of this one.  I was looking for the most recent version earlier but got distracted.

[salbayeng]

@David I think that's the condensed version of a much older LT app note? maybe my memory is failing me, but the one I'm thinking of had lines for each op amp rather than just points , (It's not this one LT DN15, 1988) . Curves a bit like the ones here: https://books.google.com.au/books?id=dunqt1rt4sAC&pg=PA75&lpg=PA75#v=onepage&q&f=false
Ok, after an extensive search, I seem to have lost the folder that had all the info for that project, must be on an archive drive somewhere.
 I've found a similar AN to the LT one , with more explanation, at AD http://www.analog.com/media/en/technical-documentation/application-notes/AN-940.pdf.

Another possible pre-amp candidate might be some of the INA architecture pre-amps, the SSM2019 I've used with microphones before (inside an electrically noisy enclosure). It works at 5v. It has a reference pin which can be helpful sometimes. If you didn't require the 150Hz filter, you can get all your gain in one chip. The CMRR is brilliant.

>>> Here's a tiny one: LMP7731  3.3nV/rtHz at 3Hz RRIO 3v <<< This could be the best candidate so far
Anyone mentioned AD822 21nV/rtHz, 3v Vss? (It's a FET input so better for 10kohm input)
LM4562 and LME49860  6nV/rtHz @10Hz 5v

Other app notes I've stumbled on while looking:
http://www.analog.com/media/en/training-seminars/tutorials/MT-047.pdf
http://www.analog.com/media/en/training-seminars/tutorials/MT-048.pdf?doc=CN0345.pdf
http://www.ti.com/lit/an/slva043b/slva043b.pdf

[David Hess]

@David I think that's the condensed version of a much older LT app note? maybe my memory is failing me, but the one I'm thinking of had lines for each op amp rather than just points , (It's not this one LT DN15, 1988).

I actually went looking for the later one because it was updated with newer operational amplifiers.

Another possible pre-amp candidate might be some of the INA architecture pre-amps, the SSM2019 I've used with microphones before (inside an electrically noisy enclosure). It works at 5v. It has a reference pin which can be helpful sometimes. If you didn't require the 150Hz filter, you can get all your gain in one chip. The CMRR is brilliant.

Low noise precision amplifiers and low noise audio amplifiers use very similar input circuits so they all have about the same noise.  Precision amplifiers will have input bias current cancellation which actually increases noise and audio amplifiers have additional refinements to lower distortion.  Neither are needed here but I doubt the difference is significant so I would go with whatever is cheapest and easiest to use.

The ancient LM833 which was one of the first good audio operational amplifiers has an input noise very similar to an LT1007 or OP27 but it has a relatively high offset voltage and without bias current compensation, its input bias current is 50 times greater so it is not a "precision" amplifier in the sense that the LT1007 or OP27 is.  At $0.50 new, it is actually not a bad choice for this circuit but it requires at least 10 volts to operate.