In circuit capacitance meter part 2-modified with inclusion of inductivity

Measuring capcaitors from 1pF to 80mF and/or inductors from 1microH to 40H using a low voltage method that is robust against parasitic parallel or seriess resistance such that reliable measurements can be made withou removing the items from a circuit.

As part II of the previous Capmeter project
here a variant and extension of the in-circuit capacitance meter is presented.
The present version allows the measurement of cpacitors from 1pf to 80mF
AND inductors from 1uH to 40H.

The measured values are robust against parallel (for C) or seies (for L) resistance
over a wide range. Additional indicative information on C series resistnce is provided.

Voltages applied to Cx are confined between +-200mV and the maximum effective voltage
across Lx would be 10mV. Thus most semiconductor elements connected to them within a circuit
would still not conduct. Parallel resistors to Cx or (series) resistivity of Lx are
simulatneously measured and do not influence the measured capacitance or inductance
as long as overrange ssituations can be avoided.

The version 2 setup can be considered as an autonomous separate analog part that
yields Cx, Lx and Rparallel .. values in terms of DC-voltages.
The second part consist of a MIcroocontroller (Arduino) with a (4 channel) ADC.
In my realisation the ADS1115 (4 channel, 16bit) is used in order to supply
good resolution. Still a simplification using A2-A4 analog Arduino inputs should work
(with somewhat less accuracy/resolution).

Range switching is done manual, which gave better control on the measurements and
explorations.
The Microcontroller is ssupplied with the analog output voltages together with
voltages the encode the setting of the range switches. Using all of these inputs it
computes the Cx, Lx and Rp, Rs values and supplies the in an 20x4 lcd display.

The basic principle of the measurement has alrready been describe in the description
of the first version. But a very short of the essentials is reiterated here:

Starting with a triangular drive voltage U(t) with a given slope we:

a.) apply the (attenuated) U(t) to a test capacitor and measure the resulting current Ic(t)

b.) use a resistor R to convert the drive voltage U(t) to a drive current I(t) through an inductor Lx and measure the voltage Ul across the inductor, the maximum Ul of 10mV is small enough to not
significangtly change the drive current (using U=+-2.5V).

Effective sampling of the signal voltage (R_shunt*Ic) or (Ul) allows to eliminate the effect of
parallel respectively series resistances. Difference to off-center samploing windows
yields information on the latter resistances.

Using "zero-offset" high bandwidth op-amps (OPA2182) allows to work with current measuring shunt
voltages inte range of 1mV. But the small residual effect of the shunt voltage is further reduced (compensated)
by a (positive) feedback of this voltatge to the excitation U(t). In particular this helps to
compensate to the shunt contribution to Rs.

Detailed comments on the sschematics and e:xamples on real world measurements are in the detailed description.

Projektelemente

LC_meter_schematics1.pdf (PDF)

LC_meterV1_3.zip (ZIP)

Arduino program for the LC-Meter

CapMeter_description_part2.pdf (PDF)

Updates vom Autor

monk99 vor 1 Jahr

Finally here is an update to the LC-meter comprising:

1.) a partial autoranging (for the current shunts in C-mode) or the current sourcing in L-mode
2.) algorithms to improve estimates of the effective R-parallel and R-series resistances

For the new version uase the new schematics LC_meter_AR_schematics.pdf and LC_meter_V3.ino program.

Some hints for realisation of the circuit:

OpAmps: for most of the OpAmps any reasonably fast ( > 3MHz) type may be used. EXEPTIONS are:
the intgegrator U4A should have a low bias current (FET type) and most important the
(pre)amplifiers U8A and U8B should have negigible offset voltages. The comparators U2, U3 and U5
may also be any reasonably fast and accurate comparator or OpAmp.
My actual selection was mainly from what I had available in stock and seemed viable.

For best accuracy it may be the best idea to use the slightly more expensive
OPAx182's throughout and possibly dedicated comparators for U2, U3 and U5.
However, the present choice yields decent results.

Autorange switching:
for switching the shunt resistors R24,25,31,33,34,35 in C-mode a low power MosFet BS170 is used,
which can directly be driven by the Arduino outputs. The body backward diode of these MosFets
does not hurt since the voltages across them are within +-10mV only.
For the lowes shunt resistor values the intrinsic on-resistance of the BS170 (abou 2 Ohms) is rather
high. Presently I used (with limited accuracy) just the measure on-resistance (here 1.8 Ohm) for the
lowest range. A simple improvement would be to replace the low range Fets Q3 qmd Q4 by high current Fets
with just some milli-Ohms on (TBD).

For the current resistor switching in L-mode I tried a newer switching IC: DG413 with an on resistance of
about 20 Ohms (which at that stage is superior to the several 100 Ohma of CD4016) it also can be
driven directly by the Arduino outputs.

Note: the reference for the logic voltage VL of this IC, if tied
to the Ardunio 5Volts should be connected only via a current limiting resistor in case that the 9V (batteries in
my tests) come later than the Arduino 5V! The same holds for the inputs (2.2k).

In that respect also the ADC inputs have to have current limiting resistors to protect them in case
of OpAmp output voltages above 5V.

C-Measuring voltage:
The maximum voltage for C-measurements is given by Uref(=2.5V)*R9/(R8+R9) (i.e. about 250mV).
In most cases this is ok, however, it is marginal if Shottky or Germanium diodes are present.
The attenuation factor can easily increased by decreasing R9 e.g. to 50 Ohms.
As a consequence the sensitivity for very low C values goes down and the upper range limit for
large C's goes up.
In order to still get the proper Cx values the attenuation factor variable in the Arduino program
has to be set to the new value.
Note: the full factor has also to take account for the feedback resistor R32 parallel to R9,
the corresponding correction should accounted for by adding the value assigned to correction-factor
in the ARDUINO program.

Feedback:
The feedback adjustent may be critical, since it is a positive feedback at some point the circuit will
develop oscillations, the proper setting is somewhere close to the onset of oscillation if adjusted with
a high quality capacitor with negligible series resistance. The feedback basically compensates for the
effective series resistance of the current measuring shunt resistor.
As long as the time constant of Resr * Cx is less than say 4 times the delay between triangle edge
and triangle voltage zero crossing (where the C-current is sampled) the feedback and/or the series resistance
will not influence the resulting Cx value.
The optimal feedback is rather important to get the best Resr estimate.

Measuring strategy:
From the above remark and the fact that there may be other transient phenomena
(like damped oscillations due to stray capacitance or inductances and or feedback+OpAmp response)
upon slope change of the triangular drive voltage/current it would increase accuracy if the
central sampling happens after a sufficient time delay. The most transients are
constant in their duration, so setting slower voltage slopes yields "cleaner" measurements.
The latter measure, however, is limited by the resulting signal voltage after amplification.
If this voltage is too low other (offset, digitisation) errors are introduced.
A potential remedy could be cranking up the amplification factor.

Typically one should start with the 1 sec ramp time wait for the autoranging. If e.g. C is too large (> 1mF)
one may stick with overrange and switch to 10sec (needs patience).
If the values stays too low (see the mV display at the end of the first line) one may decrease the rampm time.

The best insight yields monitoring of the siganl ICx(t) at the output of the preamp.

Modifications:
One may increase the range of allowed parallel resistance (C-mode) by narrowing the central sampling window,
which can easily be achieved by reducing R21 and R22. Note, however, the fraction of sampling time
to repetition (ramp) time scales the time constant of the final averaging by R41 and C5, a narrow sampling
may ask for a reduction of C5 in order to keep the reaction time reasonably fast.

So still some room for experimantation ..., have fun !

LC_meterV3.ino (25kb)
LC_meter_AR_schematics.pdf (414kb)

Antworten

Alain Dupont vor 1 Jahr

Hi,

Its written about 'LC_neterNeuV0_2.ino' Arduino program... I think it's missing.

Thanks for this great project!
Alain.

capture-decran-2024-03-18-a-183153.png (110kb)

Antworten

2 Attachment(s) 1 Kommentar(e)

Diskussion (2 Kommentare)

Wilhelm Krug vor 3 Monaten

A very interesting Project.
I often had to repair older Circuits.
To check the electrolytic Capacitators without desoldering them is often not possible with a normal Cap Meter.
Your Project can be the solution for this Problem.
I think a few parts are missing for a practical measuring Instrument.
1. Power supply:
To get the +/- 5V ( +/-9V ) Supply for the C-MOS IC's ( 10V should work ) You can use 1 or 2 isolated DC/DC Converters 5V/5V.
So You only need a single 5V Supply Voltage.
2. Battery Power:
I guess the best solution will be an 2S Lipo Battery ( nom 7,4V ) followed by an 5V Voltage Regulator.
So the complete Circuit will be a stand allone Instrument.
3. Controller:
I know - Arduino is an very versatile Platform.
IMHO a single Microcontroller would be the better Solution.
You can insert the same USB Chip that was used on the Arduino Plattform.
Then take the same Ports of the Controller and give him the Arduino Bootloader.
So the Arduino Platform should play with.
As alternative an complete shield with Your Circuit will be fine too.
I played around with the newer AVR32DB32 Chips. They offer new very interesting features ( integrated OP-Amps, zero crossing detectors, 12Bit A/D Convertes, D/A Converter ) possibly some IC's could be replaced with. Memory is availible up to 128k and 48 Pins ( AVR128DB48 ).
A negative point is they only can be programmed and debugged via UPDI, and they are not Arduino compatible.
4. Display:
I prefer OLED Displays, because the viewing angle increases significant and the current draw goes down.
But sometimes they are a little bit noisy.
Graphical OLED Displays can be an Option too.

If my suggestions are practical must be tested ( noise, stability, s.o ) but with an PCB it will be an great measuring Instrument for everybody who had to repair Circuits.

Antworten

0 Kommentar(e)

Michael Mahoney vor 1 Jahr

This is a fascinating project I think I will build one as I already have most of the components, Could you post a link to the rotary encoders used in this project?

Antworten

monk99 vor 1 Jahr

Thank you for your encouraging comment. Concerning rotary encoders: I used an extra set of contacts on
the mechanical switches for range selection with resistors to create a volatge that encodes the swiitch setting.
The switches are from Pollin:

Drehschalter CKA3X04, 26 mm, 3x4 Kontakte

or the 2x6 Variant for the slope control... mostly because I had them, the pertaining circuitry is in the schematics.
Meanwhile I also tried a further variation, using MosFets and MosfetSwitches for the shunt resistors in order to
establish a partial autoranging. I will post this as an update to this project in the next days, also
with an improved Arduino program.

I would like to learn about your expereince with it.

Antworten

1 Kommentar(e)