click here for circuit diagram of amplifier
click here for circuit diagram of amplifier
Summary - a laboratory instrumentation amplifier is described that can measure uV DC voltages which will be useful for a range of interesting physics experiments (e.g. Hall effect, small piezoelectric effects in insulators, thermoelectric & Seebeck effects, contact potentials, photo current effects etc. ). The 'flying capacitor' (switched capacitor) input stage provides a true differential input minimising input offsets and earth loops that can cause errors when measuring small signals. etc. The electronics is made from easily obtained components (e.g. e-bay) and can be constructed fairly cheply (ca. 30-40 $ / £).
Introduction
Science is a creative combination of experiment and theory - over time they inform and transform each other. Often science requires one to measure something and electronics - and electronic
amplifiers in particular - have dramatically enabled scientists to measure and explore new phenomena.
To me science is about making and measuring things and electronics provides a wonderfully creative (and economical) way of exploring science through the process of designing and building scientific instruments.
The problem-solving you need to overcome building, testing and calibrating the experiment will also force you to confront your immediate assumptions and preconceived ideas - which is often a good thing. In trying to understand Nature you keep learning.
Building scientific instruments (even simple ones) makes you feel like you are discovering the world all over again :-) ... as if these devices equip you with an extra sense of the world.
Described here is a 'flying capacitor chopper stabilised' (FCCS) amplifier (based on an 'instrumentation switched capacitor building bloc' (more info below)) that will be useful for all sorts of interesting scientific and electronic engineering experiments where you need to measure small DC (or very low frequency AC) voltages (see ideas list at the end of this article).
It's a few years since this circuit was 'state-of-the-art' but I chose this deliberately as it means the integrated circuits are easy to track down and can be brought quite cheaply (e.g. e-bay) so that students can afford to build the device. The single most expensive component in my design is the painted metal box, but of course you don't need to use this! (although a screened metal box of some kind is essential).
Standard op-amp circuit
AC or DC amplifiers with x 10 and x 100 gain (i.e. that will multiply a signal 10 or 100 times) are easily created at low cost using integrated circuit op-amps. These can amplify signals down to a fraction of a mill-volt and
perhaps smaller signals, if you are careful about the design and circuit layout. Improvements can also be made by using a differential input circuit, where common mode noise pick-up can be better cancelled.
It's quite easy to make AC amplifiers from op-amp circuits and a typical hi-fi or radio makes use of many of them (some of them like a phono amplifier for a record player can have quite high gains).
To make a stable DC amplifier is harder, especially if it is to have a high gain or amplify very small voltages. This is because small offsets and drift show up directly, while in an AC amplifier these signals are often decoupled from the next stage of the circuit by capacitors so don't show up in the same way (although they can cause saturation problems including signal distortion etc.).
Here are a few problems you face when trying to make sensitive DC amplifiers:
* output variations due to thermal drifts (in the experiment and the electronics that measure them both slow and fast),
* amplifier input / output offset currents and voltages,
* 50 Hz (and 60Hz) mains pick-up (e.g. 'hum loops'),
* unexpected voltages created across the input leads themselves,
* contact potentials between the components and solder joints etc.
If you want to measure small DC signals (possibly also in electrically noisy environments) then we need to explore other techniques than basic op-amp circuits - hence the reason for this project.
In this project we use three stages:
1) switched capacitor input
2) a chopper stabalised amplifier (x 1000 gain)
3) extra gain block (x1, x10 & x100 gain)
click here for circuit diagram of amplifier
1. Instrumentation Switched Capacitor Building Bloc - using a 'flying' or switched capacitor
One way of removing, or at least reducing some of the problems of rouge signals when we make a measurement, is to make sure we don't directly connect our experiment to the amplifier at all !
You can do this with a circuit that alternately switches a small capacitor across the signal you want to measure, and then shares it (now charged up with a snap-shot of the signal) with a similar capacitor on the input of the DC amplifier.
The 'flying capacitor' or switched capacitor creates an almost ideal differential input stage for a DC amplifier as the amplifier is never directly wired to the signal source.
An integrated circuit called a LTC1043 is a 'Instrumentation Switched Capacitor Building Bloc' (see data sheet link below) that has all the switching internally sorted-out and you just need to add a few other components (the capacitors in particular) to try it out. Check out the data sheet for a better explanation of how it works:
see data sheet here:
click here for LTC1043 data sheet
Note: this technique can't be used at frequencies higher than about half the capacitor switching frequency, otherwise aliasing (signal beating) problems are created. So it's only really useful for relatively low frequency AC or DC only measurements (see data sheet).
2. Chopper stabilisation
In addition to the switched capacitor input stage we will also be adding a chopper stabalise amplifier to get maximum performance. As mentioned above, measuring small DC voltages is also problematic due to the far from 'ideal' performance of a typical amplifier. Input-offset currents and voltages are created by the internal (inside chip) amplifier circuitry that can lead to distortion and changes in amplifier response these can occur due to room temperature fluctuations and internal circuit current heating.
A 'chopper stabilised' amplifier tries to overcome many of these internal problems by using a built-in oscillator arrangement that alternately switches circuit structure to cancel these off-sets and drifts, creating amplifiers with very low drift (typically less than 0.05 uV / C) and low offset (typically less then 5 uV) on the output. From the point of view of the user the chopper stabilised amplifier looks just like a really good amplifier and the internal switching should be invisible from the outside (but see note below and data sheet for more info.)
We will use a LTC1052 chopper stabilised op-amp, see data sheet here:
click here for LTC1052 data sheet
Note: Again the internal switching of this second stage means you can't use the chopper stabalised amplifier at frequencies higher than about half the chopping frequency, otherwise aliasing (signal beating) problems are created. So again its ideal only for relatively low frequency measurements (see data sheet).
There are therefore two different chopping / switching frequencies (from the switched capacitor circuit and the chopper stabilised amplifier) we need to be aware of. In principle with good circuit design, and proper use / application of the amplifier in the experimental, these should be 'invisible' on the amplifier output i.e. to the user.
Bringing the two together
The combination of a 'flying capacitor' input circuit and 'chopper stabilised' amplifier can create a very sensitive, stable and drift-free low frequency amplifier -
and that's what we have here - a flying capacitor chopper stabilised (FCCS) amplifier!
The flying capacitor input stage will not in principle provide any gain or loss (i.e. will have gain of x 1, once the circuit has settled down). The chopper stabilised amplifier in this resistor configuration provides a gain of (100k + 100) / 100 ohm = 1000 so = x 1000 gain. (Note for exactly x 1000 you should really use 99.9k instead of 100k, but see calibration notes below) The two units together, the flying cap IC and the chopper stabilised amplifier will therefore give a gain of x 1000 times.
Here I have used the circuit example on page 10 of the data sheet ('Ultra Precision Instrumentation Amplifier' for the LTC1043 IC) but removed the part of the circuit that generated the small -ve voltage (from the positive) as I use a separate + and - regulated power supply so full (equivalent) + and - input volages can be measured.
3. Extra gain block (x 1, x 10, x 100)
In my prototype I have also added an extra bloc of op-amp circuits, after the FCCS circuits if it should be needed. There are three gain selections (via a front panel rotary switch): x 1 (i.e. no extra gain), x 10 and x 100.
(Note: as the signals are now much larger after stage 1) and 2) a standard op-amp circuit is fine to use, although it will add a bit more noise on the output.)
I used a quad op-amp (LT1014CN) to create this extra gain bloc which provides two inverting amplifiers (so non inverting amplification overall) and provide buffering for the output so that the metering / logging circuitry will not load the amplifier (within reason of course).
As the first (FCCS) stage provides so much gain (x 1000) it may well be that the x 10 and x 100 range of this extra gain bloc will never be needed (but this will of course depend on the particular experiment the amplifier is going to be used with and the type of voltage recorder / logger that might be used).
Layout and circuit design
I built the first stage (FCCS) amplifier and the extra gain block on two different circuit boards and mounted them in a metal (screened) box.
All the 0V (earth) connections go to a single common point toward the center of the box (to reduce hum-loops etc.).
I decided to use low noise regulator power supply boards (e-bay) for both the + and - supply to the electronics and these are used when the amplifier is used on batteries (2 x PP3 9V) or on an external power supply:
positive regulator board based on the TPS7A4700 Low Noise low drop out (LDO) IC (set for +5V, see below)
negative regulator board basec on the TPS7A3301 Low Noise LDO IC (set for -5V, see below)
These small pcbs can only supply a few 100mA but are ideal for our amplifier that does not take much current. The units have solder pads that are used to program the regulation voltage. The defult (i.e. no-pad soldered) is 3V so you cant over volt by accident.
I am not sure we actually need these very low noise regulators in the circuit, but as we are going to measure very small voltages it does make sense to make sure the power supply is as low-noise as possible. The units were quite low priced (e-bay, see below) and were fun to set-up and use (and being a positive experience I will probably use them again in other projects).
Input cable
The input should have a high quality coax cable. In the past I have found coax leads can create small voltages when vibrated or moved - as if they are a microphone (so called 'microphonic' effects) - and as this is a very sensitive amplifier you need to be aware of this. My first prototype used twin screened audio cable just to test things out but better cable is needed - some experimentation needs to be made here to see what works best in practice.
Sockets
The amplifier output is available both on a standard BNC and one of the phono sockets at the right hand side of the enclosure. Another of the phono sockets could also provide access to the output of the chopper stabilised amplifier before of the extra gain block (which might provide a lower noise o/p if no extra gain is needed).
Capacitors
As the capacitor is actually the 'signal measuring device' we need good quality types. I used high voltage types (630V) as I thought that if the insulation was good at high voltage it would be low loss in this application (I used Axial Metallised Polypropylene Film Capacitor 630V 1uF e-bay).
The chopper stabilised amplifier data sheet suggests high quality components should be used for the two oscillator capacitors on the LTC1052, I used 0.1uF 100V Metallised Polyester Film (e-bay).
Front panel meter
I used a small center-zero meter (100uA FSD + range setting resistors) so it can show + and - signals on the output of the amplifier. It is only used as a guide and ideally a digital meter or data logger will be connected to the amplifier output for more accurate measurements / recordings during the experiments. I wired two silicon diodes (e.g. 2 x 1N4148) back to back across the meter which will usually have less than 0.6V across them so will be in a non conducting state, however if the o/p goes too high at any point, the diodes will conduct and protect the meter from large signal full scale deflection 'pinging' (i.e. very large full-scale deflection signals).
Instead of using an 'ON' led to show when the power is turned on, I fitted a couple of white leds into the back of the meter so it lights up brightly when turned on - it looks nice.
Ferrite beads - I added a small ferrite bead (ca. with two or three turns of thin enamelled wire threaded through) on the input to the extra gain bloc and on the two o/ps.
These are not essential but in the past I have found them useful to reduce oscillation and instability on the high gain settings.
Front panel layout of my prototype, from left to right:
ON / OFF switch: center off, up ON PSU, down ON batteries
three way switch: x 1, x 10, x 100 gain select
battery test switch: up + battery, down - battery
meter select switch: up battery, down amplifier output
100uA meter (+ LED light)
meter range switch: up ±1V, middle ±10V, down ±0.1V FSD
right hand side connections:
BNC DC output, four phono sockets: so far only DC o/p
coax cable input
left hand side: 2 x PP3 9V battery and holders
rear: 4 way connector (only 3 cons. used) for external power supply (+, - and 0V)
Notes on using the amplifier
The amplifier chips are powered with +,- 5V so none of the amplifier outputs should go beyond 3-4V (or so) otherwise the amplifier outputs may saturate (become a constant voltage near to the supply rail voltage). In the 4 op-amp chain in the 'extra' gain bloc for example you have to be careful that the first op-amp has not gone out of range (i.e. near to one of the supply rails) as its output will then be saturated and the amp will not respond properly. It won't then pass on a reliable signal to the next stage of the chain. If you can, it's probably best to try and arrange the experiment so that the amplifier output is about 1V. That way variations in signal will keep 'within range' and won't go too high, but you also don't want to have too small a signal so it gets masked in noise. It might therefore be a bit of a balancing act at first to find the best gain setting for a particular experiment.
If you short the input the amplifier output slowly tends to zero. Any residual offset there may be once the amplifier has settled down may be due to small contact potentials from dissimilar metals and solder joints etc. as these will be of the order of 1-10 uV we expect these to appear on the output as 1uV x 1000 = ca. few mV to 10's mV signals.
I found if the input is left open circuit the voltage slowly ramped up. I guess any small input offset currents on the input of the LTC1052 chopper stabilised op-amp (which although internally cancelled but may still have offsets on its input stage) charges up the capacitor(s) causing a voltage ramp on the output. If you are going to use the amplifier in a very high impedence circuit you may need to be aware of this.
Calibration
As stated above the exact gain of the chopper stabilised amplifier implies that you should use a 100R and a 99.9 k ohm for the feedback resistors if you want x 1000 gain. I used standard preferred value 100R and 100k resistors because I will use the simple tests below (e.g. test 1) to calibrate the amplifier. Good quality resistors - i.e. low temperature variation may be more important than exact value but exact resistors could be chosen from a batch / made-up from a number of resistors if you have access to an accurate ohm meter.
Note on e-bay components
Electronic components available on e-bay are usually good quality and in cases may be the only option available for the more rare (out of stock) types. However be aware that second rate components are sold as 'new', so sometimes you just have to take a risk and try them out. As none of the components are very expensive in this project I don't think you can go too far wrong on e-bay.
prices of various key parts (e-bay 2018):
LTC1052 ca. £3-50, LTC1043 ca. £15, LT1014CN ca. £11
TPS7A4700 and TPS7A3301 units ca. £6 each,
Test 1 - mV and uV test signals
In this test a variable power supply was connected across a two resistor potential divider (R1 and R2),
the smaller resistor (R1) of which will create mV, sub mV and uV signals to use as a test signal into the input of the amplifier.
For R2 = 1M and R1 = 100 R pot divider (i.e. divide by 10,000):
10V to the pot. div. gave a 1mV test signal (on the 100R) to the amplifier - this gave (1mv x 1000 = 1V) 1V o/p on x1 as expected
1V to the pot. div. gave a 100uV test signal (on the 100R) to the amplifier - this gave 0.1V o/p on x 1 (1V with x10)
For R2 = 10M and R1 = 100R pot divider (i.e. divide by 100,000):
1V to the pot. div. gave a 10uV test signal (on the 100R) to the amplifier - this gave 0.01V on x1, (0.1 on x10 and 1V on x100 range)
For R2 = 10M and R1 = 10R pot divider (i.e. divide by a million):
1V to the pot. div. gave a 1uV test signal (on the 10R) to the amplifier - this was more noisy and gave ca. 0.001-0.002V on x1, (ca. 0.01-0.02V on x10 and ca. 0.1-0.2V on x100 range).
On the 1uV input the output was noisy and varying around a bit as expected given the 'birds-nest' style croc-clip arrangement of my variable power supply and two resistors. So I am encouraged by these very simple tests.
Test 2 - small loop v = dB/dT
I simply short circuited the input with a short croc-clip lead making a single turn loop of wire about 15 cm diameter. Setting the extra gain unit to x1 I brought a rare earth magnet near to the loop (thereby increasing the magnetic field strength the coil experiences i.e. dB/dT = +ve) and a large pulse was seen on the 1 volt scale (much larger movement when set to the 100mV meter setting of course). When the magnet was removed (dB/dT = -ve) I got an equally strong, but negative going pulse, as expected. So this first simple experiment was pretty impressive considering its only a small single loop coil.
Test 3 - thermocouple
I set-up the input with a k-type thermocouple (T/C) and measured the voltage difference created between room temp (16 C) and when the T/C was in hot water (80 C).
Looking up the voltage created by the k-type T/C for this ca. 64 C temperature change gave a thermocouple voltage of about 2.5 mV. As the preamplifier gain was determined by the first stage (x 1000) and second stage (set to x 1) I should therefore get 2.5 mV x 1000 = 2.5 V output from the amplifier - which I measured - so it works :-)
(Note: I also measured an equivalent negative voltage when the input connections wre reverses, as expected)
click below for higher resolution images:
circuit diagram of amplifier
notebook diagrams of fccs unit
notebook diagrams of extra gain block
low noise power supplies
photos of equipment
Jonathan Hare, July 2019
home | diary
| whats
on | CSC
summary | latest
news
