Trying to build ESR meter

[rqsall]

Hello all,

(yes I know you can buy em for 15 bucks, but trying to learn, and get the hang of hacking things up from the proverbial junk box)

This weekend I tried to make the ESR meter from this page: http://paulorenato.com/joomla/index.php?option=com_content&view=article&id=94:building-an-esr-meter&catid=4:projects&Itemid=4

I have a bunch of small signal (I think!) transformers that I inherited about a year ago, together with all sorts of electronics stuff, including a Tek 465 that broke down a few months ago, but has been repaired with the help of the great people at the yahoo TekScopes group (hi David    )  The transformers I have seem to be 2 : 1 as specified on the schematic.

Now, the problem I was running into is this, as soon as I connect the 12 ohm R4 (10 ohm in my case, but 15 ohm makes no difference) The square wave at the transformer output (top of R4) completely disappears.

First question: Could anyone explain why this happens?

Second question: I have worked around it by replacing the transformer with an emitter-follower (MPS2222A) with roughly 200mv over the 10 ohm emitter resistor as a buffer for the NE555 output. I will post the schematic later. It seems to work on the bread board, but being skeptical of myself I wonder if there's anything inherently wrong with that approach?

Thanks for any help and guidance.

[Seekonk]

Check your DC resistance on your transformer.  It is likely more than a couple ohms and your transformer is likely an audio transformer.  That resistance is killing it.  You want a pulse transformer.

What would work is the input common mode transformer found in the line AC section of most electronics.  Wire it as an autotransformer.  A diagonal wire would connect the two windings and this is where the 15 ohm would connect.  I've used these as transformers and they peak in the 40khz region.

[rqsall]

Check your DC resistance on your transformer.  It is likely more than a couple ohms and your transformer is likely an audio transformer.  That resistance is killing it.  You want a pulse transformer.

Thank you, that would indeed explain it. I did check the transformer for connectivity and remember it was a few dozen ohms, I will check again when I get home. But by the time I had it all wired up I didn't realize this.

What would work is the input common mode transformer found in the line AC section of most electronics.  Wire it as an autotransformer.  A diagonal wire would connect the two windings and this is where the 15 ohm would connect.  I've used these as transformers and they peak in the 40khz region.

I'll give that a shot, would something from an ATX power supply work? (will do the google exercise on "input common mode transformer" jargon  )

Thank you very much.

Still wondering if I am deluding myself with my emitter-follower solution though 

[kripton2035]

here : http://kripton2035.free.fr/esr-repository.html
you will find easier to build yourself esr meters
I like the poptronix one, because it uses easy to find components.
the transformer in the circuit you try is not that easy to find, and surely not easy to begin with.

just my 2 cts.

[dannyf]

First question: Could anyone explain why this happens?

Without knowing if the transformer is really a transformer, it is hard to know.

Second question: I have worked around it by replacing the transformer with an emitter-follower (MPS2222A) with roughly 200mv over the 10 ohm emitter resistor as a buffer for the NE555 output. I will post the schematic later. It seems to work on the bread board, but being skeptical of myself I wonder if there's anything inherently wrong with that approach?

I have used the emitter follower approach and it worked. I think it is a far better and easier solution than a transformer.

The transformer does two things:

1) it impedance matches the R4+dut to the 555;
2) it lowers the output swing on the dut.

You will also find that the whole thing works without the transformer or emitter follower,

[dannyf]

I did an esr build (actually two) on a typical 8-pdip opamp. You can search for it. One version uses a reciprocal scale (ie resistance) and another with a linear read-out.

It uses one opamp for oscillation and another for impedance match / precision rectification.

[rqsall]

I wanted to take a picture of the transformer I used, but my camera's battery was dead and my phone has a lousy cam. Anyway, here's the schematic of the emitter-follower work around I cooked up on the spur of the moment. C2 is the cap under test, VR12 is a variable resistor and VR13 is the meter from an old multimeter that I'm planning on using as enclosure. All other caps are 100nF (from memory)

I have since changed the emitter-follower. Biased the base to 1V with a voltage divider, collector straight to 5V. 200 ohm base resistor to limit current and a 1uF cap to provide the input to the base. It seems to keep the output of the NE555 a bit stiffer when I connect a cap to test.

I like the poptronix solution that kripton2035 mentions a lot. I don't think I have the necessary opamps, but I'll order some and give that one a go also.

[dannyf]

here's the schematic of the emitter-follower work around I cooked up on the spur of the moment.

You may want to think about  a few things:

1) C5 should be in parallel with the two resistors / coil meter;
2) R4 makes the driving mechanism almost a current-drive, as R4>>R5//R6/Rdut.
3) attenuation: However, if you don't use R4, the drive current through the load could be significant. That's why you need a pot in front of the transistor to attenuation the output signal from the oscillator, to control drive current.
4) R9/R8 can be simplified: I would eliminate R8 and put R9 between the base + collector.

[dannyf]

I like the poptronix solution

Some quite observations:

1) the rail spliter, built around a TL0xx, is surely unable to hold the rail steady, and an oxymoron given the current amplifier used Q2.
2) the current amplifier Q2 is not needed, as its load is on the tone of 500ohm. I would get rid of Q2 and put the bridge on the emitter; alternatively, get rid of the IC1-a and IC1b, and use a 555 as the oscillator and go single rail;
3) the differential amplifier IC1-c requires matching resistors. Given the use of a meter as the read-out instrument, I would combine IC1-c and IC1-d into a true difference amplifier (aka the input stage of an instrumentation amplifier) setting gain with 1 resistor.
4) the rail splitter is required in this design as the input signal could potentially go very close to the ground. One way to solve that issue is to use padding on the bridge, or to use an amplifier that is (at least) r2r to the negative rail (ie, an opamp with pnp input stage).

This whole thing can be built around a 555 + a (dual) opamp (or a single instrumentation amp).

I don't think I have the necessary opamps

Many opamps or even comparators will do. The lowly ne5532 would be more than enough here.

[Seekonk]

Good you have something working.  The bad news it is about totally useless.  OK it is almost as good as an old Heathkit C-3 but at least that used a bridge.  That tester had a Power Factor knob that measured ESR, it went up to 250 ohms.  Maybe that was good enough in the old tube days.  Just try telling the difference between .1 and .3 ohms of series resistance.  As said before you have a simple voltage divider and you voltage source isn't stable with load.  You need a bridge circuit or at least measure the voltage across the capacitor to ground.

[rqsall]

1) C5 should be in parallel with the two resistors / coil meter;

This is how it's wired. The mistake is in the schematic I drew.

2) R4 makes the driving mechanism almost a current-drive, as R4>>R5//R6/Rdut.

I have since removed that resistor. There is no collector resistor anymore. See the attachment. I'm also thinking that I can get rid of R6 and change R7 to 15ohm?

3) attenuation: However, if you don't use R4, the drive current through the load could be significant. That's why you need a pot in front of the transistor to attenuation the output signal from the oscillator, to control drive current.

There's a 200ohm (R3) resistor there now. Without it, Q1 got quite warm 

4) R9/R8 can be simplified: I would eliminate R8 and put R9 between the base + collector.

OK. This would limit the current into the base of Q2?

Thank you for your guidance. Really appreciate it.

[dannyf]

R3/R12 provide the attentuation.

Otherwise, I would just use a higher value pot (1k or so for example), and put a serial resistor on Q1's base.

If you don't attenuation, huge current will go through Q1 and that may saturate the amplifier stage later.

[dannyf]

BTW, the relative size of R10 vs. R11 means that the latest stage does no amplification. It may not be what you want.

[rqsall]

Good you have something working.  The bad news it is about totally useless.  OK it is almost as good as an old Heathkit C-3 but at least that used a bridge.  That tester had a Power Factor knob that measured ESR, it went up to 250 ohms.  Maybe that was good enough in the old tube days.  Just try telling the difference between .1 and .3 ohms of series resistance.

I'm not sure if being able to tell the difference between 0.1 and 0.3 ohm is my goal here. Just want to be able to determine if a capacitor is likely to be bad or good.

As said before you have a simple voltage divider and you voltage source isn't stable with load.  You need a bridge circuit or at least measure the voltage across the capacitor to ground.

I see the poptronix one uses a bridge. I also like the LED to indicate a dead short.

BTW, the resistance of the transformer I tried using is 3Kohms! 

Attached is a picture of the transformers I tried.

[rqsall]

R3/R12 provide the attentuation.

Otherwise, I would just use a higher value pot (1k or so for example), and put a serial resistor on Q1's base.

If you don't attenuation, huge current will go through Q1 and that may saturate the amplifier stage later.

I have calculated it at 20mA (yet to confirm with measurements). Also, not on the schematic is a 1uF electrolytic cap between R3 and the voltage divider. I should prolly use a ceramic, but I had this one close. I tried a 1k resistor, but it attenuated too much: the meter would barely move.

[rqsall]

BTW, the relative size of R10 vs. R11 means that the latest stage does no amplification. It may not be what you want.

I think it's mostly a buffer. It is in the same as in the original that I derived this from. I'm not seeing much amplification with the scope, about 2x voltage gain.

[dannyf]

This is what I had in mind + a quick simulation.

V1 is your 555 oscillator (a 555 + 1 resistor + 1 capacitor). Its output is 1.5v Vpp (see that later).

Q1 is the buffer. Its output drives DUT (ESR + C1). The amplifier stage is done by Q2 and the rectifier by D1/D2.

The simulation showed the voltage / current reading on R8. The top line is for 0ohm (0.1mohm in the simulation), followed by 1ohm, 2ohm and 10ohm. The meter is designed to drive a multimeters with center resistance reading of 10ohm. As such, we expect the 10ohm reading to be 50% of the full scale reading.

The simulation shows:

sim % vs. theoretical % of full scale readings:
94% vs. 91%, @ 1ohm
88% vs. 83%, @ 2ohm
52% vs. 50%, @ 10ohm

As to attenuation: you will need more than 0.7vVpp swing to open up Q1. Too much swing will saturate the amplifier. Too little swing will cause the readings to be non-linear (distortion caused by the diodes).

The proper way to attenuation is to use a pot, isolated by a base resistor.

If you want to reduce part count, you can configure a pot so that its wiper goes to the base of Q1, and its terminals go to the 555 + ground, as shown in red in the attached schematic.

You may want to play with the attenuation to provide yourself with the best performance.

[dannyf]

it's mostly a buffer.

Just so you know, you can indeed build an esr meter without amplification, using the 555 to directly drive the dut. I have done that.

The downside is that the choice of meters is limited by the 555's current capability (pretty beefy at 200ma).

[dannyf]

Here is an entirely passive ESR detector - I think I had shared it here some times back.

Its measurements vs. theorectical, of full scale:
91% vs. 93%, @ 1ohm
87% vs. 83%, @ 2ohm
55% vs. 50%, @ 10ohm

It has a significant downside: the current peak is around 5v/(20 + 200/30) = 180ma. So it is not possible to power it via a battery.

It works, only via brutal force.

[rqsall]

This is what I had in mind + a quick simulation.

V1 is your 555 oscillator (a 555 + 1 resistor + 1 capacitor). Its output is 1.5v Vpp (see that later).

Q1 is the buffer. Its output drives DUT (ESR + C1). The amplifier stage is done by Q2 and the rectifier by D1/D2.

The simulation showed the voltage / current reading on R8. The top line is for 0ohm (0.1mohm in the simulation), followed by 1ohm, 2ohm and 10ohm. The meter is designed to drive a multimeters with center resistance reading of 10ohm. As such, we expect the 10ohm reading to be 50% of the full scale reading.

The simulation shows:

sim % vs. theoretical % of full scale readings:
94% vs. 91%, @ 1ohm
88% vs. 83%, @ 2ohm
52% vs. 50%, @ 10ohm

As to attenuation: you will need more than 0.7vVpp swing to open up Q1. Too much swing will saturate the amplifier. Too little swing will cause the readings to be non-linear (distortion caused by the diodes).

The proper way to attenuation is to use a pot, isolated by a base resistor.

If you want to reduce part count, you can configure a pot so that its wiper goes to the base of Q1, and its terminals go to the 555 + ground, as shown in red in the attached schematic.

You may want to play with the attenuation to provide yourself with the best performance.

Built this circuit on the bread board. Seems to work reasonably well, but I had to change R1 to 10k and put a 10k pot in series with it to be able to get the meter in the right range. I might just solder it up and put it in the old multimeter and work on the bridge based poptronix version later. Adapting that to a NE555 should be a nice learning experience.

EDIT: Forgot that I had also changed the gain of Q2. R7 is 500ohm and R5 and R6 are unchanged.

Thanks for the help and suggestions.

[dannyf]

Yeah, you will need to attenuate the output from your oscillator - the oscillator in the simulation has a swing of just 1.5v.

[rqsall]

I got it to 500mV at the test leads (Q1 emitter) to have a fighting chance of using it "in circuit"  . The collector of Q2 goes from 0.6 to 2.5V with the test leads shorted and the meter to full deflect.

[dannyf]

The whole design is a compromise,

1) to minimize the diodes' non-linearity, you want to drive the dut hard;
2) to minimize the follower's non-linearity, you don't want to drive the dut hard.

;(

Using a precision rectifier is the way to go, I think.

[rqsall]

The whole design is a compromise,

1) to minimize the diodes' non-linearity, you want to drive the dut hard;

Just to confirm I'm understanding it: you're referring to D1 and D2 in your 1st simulation right? And the non-linearity is the Voltage/Current curve when forward biased, which has a steep straight slope only when driven significantly beyond Vf with a largish current?

2) to minimize the follower's non-linearity, you don't want to drive the dut hard.

And this is emitter-follower Q1? Whose emitter VI-curve is also not flat in the active region.

Using a precision rectifier is the way to go, I think.

Instead of D1+D2?

Thank you.

[dannyf]

Yeah. Here is a version of ESR meter I built - it is for linear read out not the reciprocal readout but the base precision rectifier section is the same.

A sine wave comes into an attenuator built around R1/DUT (=R2 + C4). The opamp rectifies the signal, based on gain set by R4 (a pot). The current readout is through a coil meter (R10) or in the case of a voltage meter, across R10.

Adjust R4 to achieve the desired full scale setting (20ohm here).

The current readout for this particular meter is

1ohm: 11.3ua vs. 10ua;
2ohm: 22.0ua vs. 20ua;
5ohm: 53.3ua vs. 50ua;
10ohm: 102.9ua vs. 100ua;
20ohm: 200ua vs. 200ua;