10 Transistor ESR Meter Adapter - Now 6 Transistors!!

[Jay_Diddy_B]

Hi Group,

Given the interest level in both of my earlier ESR meter designs:

The 5 Transistor ESR Meter

https://www.eevblog.com/forum/projects/5-transistor-esr-meter-design/

The ESR Meter Adapter

https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/

I have decided that it was time to do another one.

Features of the new design


No integrated circuits, including voltage regulators
100kHz sine wave measurement
AC coupled to protect the ESR meter adapter from charged capacitors
No special parts
Potentiometers for zero and gain adjustment.

This was challenge in analog circuit design.

In the other designs I started with modelling the circuit using LTspice. I developed this circuit using LTspice, but I am going to start off by sharing some pictures of the prototype.

The PCB is 2 inches x 3 inches. It built the prototype using an LPKF c60 circuit board plotter.



This is the solder side of the board. There are lots of parts:

10 transistors
32 resistors
2 pots
11 capacitors ...

This is the component side of the board




Preliminary tests

The adapter is designed to work from a single 9V battery. The output is 10mV / Ohm.



If the battery voltage is changed by 1V the reading changes by about 1.5% which is not too bad.

Test current

This shows the test current measured with a Tektronix P6022 current probe, 2mA / div.





These are preliminary results, the testing was started today.

Regards,

Jay_Diddy_B

[dardosordi]

Really nice, looking forward to se the rest of it.

[kripton2035]

I answered here to get notifications on this interesting design !

[2N3055]

Another one of your little gems... 

[BravoV]

Subbed. 

Preparing pop corn and watching one of the analog and LTSpice kungfu master around (at least for me) .. in action. 

[Jay_Diddy_B]

Hi group,

I am going to show the design and evaluation of each of the functional blocks that were used in this circuit.
Most of the blocks are currently working very well. One block has some challenges, but we can work through the challenges together.

I should also add this meter adapter, like many other, measures the impedance of the capacitor at 100kHz, not the ESR. It does not measure the resistive part.


Oscillator - AC sinewave reference

This circuit comes from a design proposed by 'X.G."

I have analyzed and engineered the circuit to make it more robust.

LTspice Model


This is the LC oscillator implemented with NPN transistors.

A .meas LTspice directive is used to measure the output amplitude of the oscillator.
This model can be used to explore sensitivity to various change in operating conditions.



And the output waveform:





Note the output is with respect to the positive rail.
The amplitude is  2x Vbe p-p

Variation with Inductor Resistance

If resistance of the inductor resistance is stepped and the amplitude of the oscillation is measured:




Results



The magnitude of the oscillation is largely independent of the resistance of the inductor. The resistance should be around 1


Variation with supply voltage

The supply is changed from 8V to 10V



The results



A change in supply voltage of 1V results in a change in amplitude of 1.25%

Junction Temperature


The model was modified to sweep the junction temperature from 15 to 40C:



The results of sweeping the junction temperature:



The result is - 1mV per degree C

or 10mV / 550mV x 100 = 1.8% change in amplitude for a 10C change in junction temperature.


Observation

This relatively simple circuit is suitable for a reference in the ESR meter adapter.


Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

In the last message the design of the 100kHz reference was presented. This message will show how the 100kHz voltage is converted to a current.

PNP versus NPN

It is sometimes easier to develop circuits using NPN transistors with the ground at the bottom of the circuit and then 'flip' the circuit to use PNP devices. The current source was developed in this way.

LTspice model



Q3 is a modulated current sink.
Q4 is a constant current source.

The modulated AC current flows through C6 and is the test current for measuring the ESR.

The circuit will find its own dc operating point.



Variation with Supply Voltage




The output amplitude is almost independent of the supply voltage.

Variation in current with ESR



The current source is close to ideal. It is equivalent to a current source in parallel with around 20k

Hardware Testing

The complete circuit, the oscillator and the voltage to current converter was built:



Test Results



testing at 9V



testing at 7V


The current is relatively insensitive to battery voltage.

Regards,

Jay_Diddy_B

[Jay_Diddy_B]

Hi,
The first part of this development went well the ac current source is working nicely.

This is the concept:



The right hand side of the circuit is an ac voltmeter. It has to rectify and amplify very small 100kHz signals.

To allow the ESR meter to be used in-circuit, the signal levels have to be kept small at the maximum ESR.

The current source 1.6mA RMS so 1.6mV with a 1 ESR.

This part of the circuit didn't go as smoothly. I will share the experience.

Precision Rectifier


If you Google 'precision rectifier' it won't take very long before this circuit shows up:



The idea behind this circuit is the opamp turns the diode D1 into an ideal diode. That is a diode with no fwd voltage drop.

Implementation

The opamp circuit was implemented using an opamp built with BJTs:



The circuit was tested and didn't work as well as expected, especially for small signals.

Analysis



This is a model of the precision rectifier.



Even its very idealized form the circuit doesn't perform well with small signals.

If the diodes are replaced with ideal diodes, the performance improves to:



The trouble is that you can't buy ideal diodes.

Offset

If input offset is added to the model:



The performance is further degraded:




It seems that the precision rectifier isn't all that precise, especially at small signals, when the circuit is implemented with real components.

(I have a solution that seems to be working well enough. This will be presented in the next installment).

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,

I started to think about improving the non-linearity with small by amplifying the signal before the rectification.

x100 Amplifier

This circuit that can be built with two common BJT has remarkable performance:



The small signal bandwidth (-3dB) of this circuit is 13MHz.



That is a gain bandwidth product of 900MHz

Not bad for two transistors!!

Note:
C1 and C2 have been made large to improve the LF performance. They can be smaller in the ESR meter application.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi,
I took the amplifier and converted it to a precision rectifier:



I then measured the linearity from 100uV to 20mV RMS



or on a LOG - LOG scale



This circuit performance is 'tremendous' given the simplicity.

Note: there is an offset. The offset is the emitter current of Q1 x emitter resistor R2. This is about 40mV

 Offset and gain are easily dealt with. Non-linearity is very difficult to correct.

The next section will be hardware tests of this rectifier.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

Here are the hardware tests of the rectifier circuit.

Construction




The circuit was built on a single-sided circuit board using SMD components. The SMA cable is for the signal input.
The two pots are for zero and gain.

Test configuration



A 40dB attenuator and 50 Ohm termination were placed between the HP3325B function generator and the rectifier circuit. The output was measured with a Fluke 289 DMM.

Measured Results

Data



and the graph



looking at the small signals 100uV to 1mV




I am happy with these results.

Regards,
Jay_Diddy_B

[Jay_Diddy_B]

Hi group,

the assembly of a prototype was finished and the testing has started.

Modification

The bandwidth of the rectifier had to be reduced to stop it from oscillating.



A 15pf capacitor was added in the location shown. It reduces the GBW product of the amplifier to around 190MHz.

Complete schematic

Here is the complete schematic:



Note that the capacitor being tested is ac coupled to the circuit and the circuit is protected from applied voltage, that is charged capacitors. This is the scheme that was used on my original ESR Meter Adapter.

Assembly

The PCB is 100m x 100mm. The schematic and artwork were made using KICAD.
The prototype was made using a single sided board on an LPKF Protomat c60.



Other side



Testing

The scaling factor is 100mV per 

0 Ohm Short



With the short applied to the input RV2 was adjusted for 0V on the DMM

5 Ohm




10 Ohm




At this stage the prototype seems to be working well.

Regards,
Jay_Diddy_B

[wasyoungonce]

Suscribed 

[tomasis]

Maybe should I try to build this one instead of the 5 transistors design?

 

[Jay_Diddy_B]

Maybe should I try to build this one instead of the 5 transistors design?

 

Hi,

As far as I know, I am the only person who has built this design. I am not sure how sensitive it is to the layout. The 5 transistor design is not sensitive to the layout.

The 6 transistor uses a DMM for the display
It is very well protected against applied voltage and charged capacitors


There is this one too:

Link:https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/

Regards,
Jay_Diddy_B

[ali_asadzadeh]

Nice topic, Can it be ground referenced, so a Normal MCU like STM32 can be used for measurement, instead of the Usual DMM for reading!
Thanks for sharing

[bd139]

Sorry for necroposting on this one but I decided to build this. I built your original ESR meter design a few years back. Unfortunately due to a cupboard avalanche the meter was destroyed  . I don't enjoy using analogue panel meters anyway so scouted around and found this as an alternative. It also fits the mantra I'm trying to work with at the moment which is discrete transistor designs and no expensive or weird opamps

So far I put the oscillator side together and all seems good  .

Next half tomorrow...

[bd139]

Ok built it fugly style and it doesn't work (yet).

So outcome so far is:

1. can zero it with RV2 fine.
2. RV1 - what does it do? assuming diode balance?
3. Getting 10mV on 1 ohm which is weird.

Checking my work for boo boos...


Edit: traced it down to the AC amplifier / rectifier. I've shorted the terminals and injected 2mV p-p into the left side of C6 and can see no signal worth mentioning at C7


Edit 2: fixed that. Boo boo on my part. I am now shorting the DUT terminals and injecting signal and it approximately matches the linearity specs on earlier post. Still not working quite as intended however. Further debugging to take place on Sunday...

Current outputs:

20 ohm -> 0.493
10 ohm -> 0.239
5 ohm -> 0.112
1 ohm -> 0.015
0.5 ohm -> 0.005

Edit 3: established RV1 is gain after reading the entire thread properly. Issue with oscillator...


Edit 4: think I found the problem. The oscillator output is low, producing half of the expected signal amplitude into the DUT. I ran the simulations in LTspice and compared to reality.

This is RMS mV across the DUT



I am now shorting the DUT and injecting calculated RMS values into the amplifier to check that it can be calibrated into range. That will entirely isolate the oscillator as the cause if that is the case...


Injecting signal after R13 was fruitful. Clearly shows the amplifier works with expected values from the simulation.

The issue is the oscillator level by the looks. Validating my construction versus simulation with scope next...


Ok so the oscillator amplitude is way off. After some experimentation, this is probably the Q of the LC circuit. I used a polybox (EPCOS B32529C1223K) and a 100uH RFC (CPC Multicomp MCAL0410A1-101KU). I will try some different parts to see if I can knock the Q up.


Ok after various modifications I can't get this to perform anywhere near the original prototype. Shelved for now. Further analysis at some other point in time