Impedance Analyzer Build and Experiments

[jaxbird]

I've been playing around with designing and building an impedance analyzer based on the AD5933.

It's a nice little chip. Capable of generating a sinewave from low Hz (some clock scaling needed) up to about 500kHz output sweep. And on the input, using a 12 bit ADC with up to 1 MSPS, it will do a 1024 point DFT and provide the complex imaginary and real values for each step in a sweep.

It is a bit complicated to get the best performance out of it.

And I'm mostly interested in low range impedance (below 100 Ohm). The chip is not really designed for that. But adding a buffer on the output making it capable of sourcing some more current should help.

My current setup is something like this:




And the current prototype:




(Not ideal to breadboard, but it works, only sacrificing some resolution really)

I will need to have some kind of reference measurement to compensate for phase and impedance changes caused by the buffer, I think the best approach will be a few calibration resistors that can be invoked on the input.

Impedance can be calculated as:

Z = SQRT(I * I + R * R)

And phase:

Phase = ATAN(I/R)

Should be possible to calculate capacitance and inductance as well once the resonance frequency is found.

Not sure if I can calculate ESR and any other interesting parameters, suggestions much welcome

[dannyf]

The chip is limited for both lower end and high end Z measurements.

Using buffer is one way. Another is to use an outboard amplifier.

the typical approach to counter phase shift is to use a calibration resistor for each measurement.

Yes, it can be used to measure ESR, just not at very low-end (<10mohm would be hard).

There is a new chip introduced by Analog a couple months back that has a CM3 chip built in, with higher resolution as well.

[jaxbird]

The chip is limited for both lower end and high end Z measurements.

Using buffer is one way. Another is to use an outboard amplifier.

the typical approach to counter phase shift is to use a calibration resistor for each measurement.

Yes, it can be used to measure ESR, just not at very low-end (<10mohm would be hard).

There is a new chip introduced by Analog a couple months back that has a CM3 chip built in, with higher resolution as well.

Yeah, it's not really designed to do below 10kHz using the internal oscillator, but with a configurable external clock I get good results down to as low as 10Hz. (takes some time for it to sample and calculate)

I will try adding a few relays with resistors to run a calibration sweep before each measurement, that should allow compensation of both phase and buffer bandwidth limitations. Probably some resistors with minimal inductance for best results.

I'm not sure how to calculate the ESR with just the frequency and real + imaginary values from the DFT available.

It does generate a decent 500kHz+ sine wave, but clearly at the edge, not sure the measurements will be very accurate though with only 1 MSPS:




Any idea if AD's new design is available at this time? I wouldn't mind including it in my experiments

[dannyf]

I'm not sure how to calculate the ESR with just the frequency and real + imaginary values from the DFT available.

Depending on the set-up, but generally by performing a delta. Measure the dut, obtain real + imaginary values; Short the dut, obtain real + imaginary values.

The delta of the two is your dut.

It does generate a decent 500kHz+ sine wave, but clearly at the edge,

A better approach, from a signal + drive point of view, is to use a DDS + CFB opamp. But you have to perform the DFFT after that.

There are quite a few "extensions", some on Analog's own website, to push the performance envelope for 5933/32.

[Conrad Hoffman]

Grab the first item here for checking your conversions- http://www.conradhoffman.com/chsw.htm
From here use the handy formulas and capacitor loss info to do it yourself- http://www.conradhoffman.com/electronics01.htm

Given any two parameters of value and loss, you can convert to everything else.

[jaxbird]

A better approach, from a signal + drive point of view, is to use a DDS + CFB opamp. But you have to perform the DFFT after that.

Maybe some day .. I like that the AD5933/34 does all that for me, but, of course with limitations. Some can be worked around but it will never go above ~500kHz. I think I can live with that limitation for now.

Grab the first item here for checking your conversions- http://www.conradhoffman.com/chsw.htm
From here use the handy formulas and capacitor loss info to do it yourself- http://www.conradhoffman.com/electronics01.htm

Given any two parameters of value and loss, you can convert to everything else.

Very nice collection, thanks  I'll have to study it in more detail.

[jaxbird]

Spent a bit more time on this project writing something to actually plot the output, it's cleaner than what I expected.

Here measuring a 10R resistor, so primarily showing phase and impedance error from the device plus added buffer, although the resistor probably got some inductance as well:




It appears to work well all the way up to 500kHz, so 10Hz-500kHz bandwidth should be possible.

The tricky part will be to keep it within usable range. The output swings around VCC/2 and can be scaled from 200mV to 2000mVpp in 4 steps. Input got a PGA before the ADC, but unfortunately only two levels of gain x1 or x5.

[Zad]

I did some work on using the AD5933 a few years ago. At first, I couldn't understand why the chip wasn't more commonly used in impedance measurement, but the more you look into it, the more annoying the drawbacks become. I wrote up some of my findings in my blog

(See here http://electronicsdesigner.blogspot.co.uk/2011/04/design-for-precision-lcr-meter-part-2.html )

In order to get an acceptable frequency range, you really need an external oscillator with some method of selecting different clock frequencies. I went with a low-end DDS chip. Cheap to buy and easy to program with the same embedded micro you are using to talk to the 5933. I wouldn't run the drive oscillator faster than the design limit of 100kHz. As you have seen, at that point, you start to get components from other frequencies showing up, so you really need to filter the output, which of course adds its own phase+magnitude curve to be calibrated out.

Impedance range limitations: In order to measure the sort of inductors and capacitors a typical hobbyist may use, the 5933 is going to be outside it's normal measurement limits. Fine if you want to measure a 10K resistor, less so if you are measuring a 10pF capacitor. As you would expect, Analog Devices has some excellent application notes, and it is well worth digging around in them to see how they suggest increasing drive current, and improving sensitivity dynamic range etc. I ended up using an analogue multiplexer to select drive and receive gains via feedback resistors in the op-amps. I also included a small TrimDAC both to zero the offset, and induce a DC drive offset to investigate how capacitance changes with respect to voltage (3 years before Dave's video!)

As you can imagine, this changed a neat little solution into a bit of a Frankenstein's monster - although I must admit I thoroughly enjoyed doing it. After some thought, I realised I could probably just replace the AD5933 with an ADC and DAC driven straight from nice cheap fast ARM DMA ports. I wouldn't need the DDS then, and I would have much more control about how the measurement is made, and the processing done on it. Of course, you could just drive it directly from the DDS, way up into the RF regions, and simultaneously measure the current and voltage with much higher precision than the 5933 does.

I really should write up how it progressed, and the tests I did with it, but for the limited audience it doesn't really seem worth it, and it has mostly been cannibalised for other projects now. If I prototype a "new improved" direct measurement version, then I will definitely publish it with photos etc. I just need the enthusiasm! Like Dave and his uPower power supply design, you get to a point where you have done all the interesting stuff and it becomes hard work for little return.

[jaxbird]

Zad, Very good info thanks for sharing.

I wanted to have a go at the AD5933 it because it initially seemed like a very simple "all-in-one" solution, but based on experiments and supported by your experience a lot of tweaks are indeed needed.

It is lacking features to keep it within usable range / level. The data sheets and app notes are good, I've been studying most of them in detail.

Personally I would certainly be interested in how far you have taken your design, but I fully understand that it probably is a very narrow audience and the lack of interest and motivation once you have the problems solved .. many of my projects end up like that as well.

I don't have a specific goal for this project, but it would be a nice tool to have in the collection if it ends up in a usable state.

[jaxbird]

Here a couple of plots with calibrated phase and impedance, first a capacitor:




And a small inductor:

[dannyf]

I could probably just replace the AD5933 with an ADC and DAC driven straight from nice cheap fast ARM DMA ports.

You will find that to be fairly involved, due in large to

1) limited current capabilities from the dac - you will likely need additional drivers for <1kohm dut;
2) low speed of the adc / adcs - this can be addressed by fancier external adc (Analog makes quite a few parallel adc, with identical specs as AD5933/34), mcus with super fast adc (LPC has a 70msps part for example), or by layering the samples: you would take 1 sample at one point in a cycle, and take another sample at a later point in the next cycle, .... and reconstitute the data afterwards.

Not to mention that math becomes a little bit difficult for duts with mostly reactance (phase angle approaching 90 degrees).

However, I agree with your basic notion that the part is quite limiting. They made the assumption that the output signal = signal on the dut. Had they implemented another adc (to measure the voltage on the dut), this could have been such a fantastic part.

[dannyf]

first a capacitor

You can clearly see the inductance in that thing.

ESR would be the trough on the plot.

[jaxbird]

You can clearly see the inductance in that thing.

ESR would be the trough on the plot.

I can't guarantee that it's correct as it's just a quick calibration and a measurement with everything still on breadboard.

but at least it resembles what it's supposed to look like



And my plot is using a linear scale for Ohm.

[dannyf]

Your chart looks spot on.

[The Electrician]

Your plot of a capacitor doesn't look quite right.  The phase angle at frequencies below the SRF should be negative, but you show it as positive.  Maybe you've plotted the phase upside down.  The phase of the inductor plot also has the wrong sign.

Also, your plot shows the capacitor phase tending toward zero degrees at the very lowest frequency; it should tend toward -90 degrees at the lowest frequency, and toward +90 degrees at the highest frequency.

Furthermore, the shape of the impedance curve doesn't look right for a linear impedance scale.

Here's a measurement of a motor run capacitor over the same frequency range as your plots, but with impedance shown on a logarithmic scale:



Here's the same plot, but with a linear scale for impedance:

[jaxbird]

Maybe you've plotted the phase upside down.  The phase of the inductor plot also has the wrong sign.

Thanks, very much appreciate you posted a comparable curve 

Always a problem when trying to build something without having a reference to compare against. (A bit of a chicken and egg problem I guess)

It is highly likely I have messed something up in the phase calculations causing it to be inverted. The calculation is fairly simple:

Phase(radians) = tan^-1(I/R)

Then convert to degrees and correct for each of the 4 quadrants.

Subtract system phase.

And plot it with screen Y address coordinates being flipped. (likely where I messed up)

Furthermore, the shape of the impedance curve doesn't look right for a linear impedance scale.
...

Yeah, I believe there is a much bigger problem than just the inverted phase, clearly still some work needed. But at least now I have a pretty good idea of what it is supposed to look like 

[dannyf]

This is what the impedance curve looks like, in linear scales.

[jaxbird]

This is what the impedance curve looks like, in linear scales.

Thanks, that is a nice simulation 

I think I better switch to a log scale on both axis.

Actually the chart I posted earlier was neither linear nor logarithmic, I didn't take into account that the amplifier configuration I was using wouldn't provide a linear output. 

Anyway, these are some of the configurations I'm experimenting with:

This one does not work well.



This is better:



It will not give a linear output, but that's fine, I can compensate.

I'm currently using AD8532 for the opamps and a 3.3V supply.

[dannyf]

This one does not work well.

How so? To me, they look the same.

[jaxbird]

This one does not work well.

How so? To me, they look the same.

I think it's the opamp's output impedance that causes poor results with the first configuration.

I will try look for something other than the AD8532 I use now.

[dannyf]

Since the DC performance doesn't matter here, have you tried those CFB opamps? AD815, tpa6120 (and its ths equivalent)? Pretty much any xDSL drivers will do here.

[jaxbird]

I found a few BUF634, I will have a go with them and see how well they perform.

http://www.ti.com/lit/ds/symlink/buf634.pdf

[jaxbird]

Haven't rebuilt it yet, but been playing around with LTSpice:



Green is OPA2134 + BUF634
Red is AD8532

If the simulation is correct there should be a significant gain in performance using BUF634.

But with OPA2134, BUF634 and AD5933 it's a bit like using a BMW opamp with a Ferrari buffer to boost a Toyota chip 

[dannyf]

BUF634, in a composite amp, should be good. But I thought they are fairly difficult to find and expensive.

[jaxbird]

BUF634, in a composite amp, should be good. But I thought they are fairly difficult to find and expensive.

Don't know about general availability, but I believe they are in production. I found a bag of 4 in my box of slightly exotic components, don't remember if I bought them for a project or just found them nice to have .. They are quite expensive though, something like $5-10/each.

I can probably get away with using just 1 of them as I will need a dual supply anyway, so shifting the signal to swing around ground and then back to around 1.65V on the output should be doable.

[jaxbird]

Not really related, but I do find it odd the way TI decided to market the BUF634:

Among their main use cases are motor and Solenoid driver. I cannot think of a case where anyone would need a 150MHz bandwidth and 2000V/us slew rate buffer to drive a motor or a solenoid 

Surely a much cheaper solution would do just fine for those applications.

But as a generic opamp booster it's very potent.

[dannyf]

But as a generic opamp booster it's very potent.

More so as a hi-fi headphone driver.

[Zad]

In the AD app notes they suggest using an AD8606 which has 1 Ohm output impedance. See their circuit Lab reference CN-0217. As a suggestion, try taking the feedback for the drive amplifier from the DUT side of the drive current limiting resistor.

[theatrus]

There is also the LME49600 opamp buffers from National (TI). They're intended for audio use as well but have similar redonkulous slew rates.

[jaxbird]

In the AD app notes they suggest using an AD8606 which has 1 Ohm output impedance. See their circuit Lab reference CN-0217. As a suggestion, try taking the feedback for the drive amplifier from the DUT side of the drive current limiting resistor.

Thanks, that's good advise, I must have missed that application note. In the main datasheet they only suggest AD8531,
AD820, AD8641 and AD8627. Of which AD8531 looked the best.

I'll try get my hands on a few AD8606.

In the meantime I'll try out this configuration:



I will also need to do some level shifting.

[jaxbird]

More so as a hi-fi headphone driver.

Yeah, that was probably the purpose I had in mind when I bought the BUF634.

There is also the LME49600 opamp buffers from National (TI). They're intended for audio use as well but have similar redonkulous slew rates.

Very similar specs, and cheaper  good suggestion.

[jaxbird]

Got the large breadboard out to build up a Buf634 based circuit with associated components:




It's a good improvement in performance:




But it's not a good idea to attempt accurate low impedance measurements using a breadboard circuit where each connection point got something like a 50 milliohm+ resistance 

Next I'll make a PCB for more accurate evaluation.

[dannyf]

using a breadboard circuit where each connection point got something like a 50 milliohm+ resistance

Depending on the math utilized. You can, for example, decompose the amplitude + phase angle into real + imaginary parts and null the impact of the wires if you perform two measurements at each frequency. The resulting real + imaginary parts can be re-converted back to amplitude + phase angle of the DUT.

[Odysseus]

using a breadboard circuit where each connection point got something like a 50 milliohm+ resistance

Depending on the math utilized. You can, for example, decompose the amplitude + phase angle into real + imaginary parts and null the impact of the wires if you perform two measurements at each frequency. The resulting real + imaginary parts can be re-converted back to amplitude + phase angle of the DUT.

Another option is to use a kelvin (4 wire) connection to the DUT.  Kind of a must at low impedances.

[jaxbird]

using a breadboard circuit where each connection point got something like a 50 milliohm+ resistance

Depending on the math utilized. You can, for example, decompose the amplitude + phase angle into real + imaginary parts and null the impact of the wires if you perform two measurements at each frequency. The resulting real + imaginary parts can be re-converted back to amplitude + phase angle of the DUT.

Probably could, but with breadboard, you just poke a wire or component and the resistance changes

My current calibration/compensation is fairly basic, it will be more involved to get full calibration as the frequency response changes slightly with different ohm loads.

I haven't thought of doing the calibration with the real and imaginary numbers, currently I use the magnitude and phase. I'll try have a go with that method once I get the PCB together. I wouldn't even need to decompose as I have the complex real and imaginary available directly from the DFT.

[jaxbird]

Quick proto PCB:



Single sided, not ideal, but many times better than breadboard

Just need to get it soldered together and hope I haven't made too many mistakes.

[jaxbird]

Another option is to use a kelvin (4 wire) connection to the DUT.  Kind of a must at low impedances.

Yeah, Kelvin connections would be ideal, but I think I'll try get away with really short wires for now, as it will at best only resolve down into the lower portion of the milliohm range.

[jaxbird]

A bit old school looking



Might have gone a bit overboard with the decoupling.

[dannyf]

Nicely done.

[jaxbird]

Nicely done.

Thanks, it does provide much cleaner (and hopefully more accurate) results:



Note: the degree markers are not placed correctly.

[Zad]

Boards fabricated during development often do look oldschool! What matters is that you can get into them to measure and scope signals, and add/remove/swap components as needed. Unless you are working >GHz then track distances and compact boards aren't massively important. I'd be more than happy working with a board like that. If you do another one, maybe bring out the test probe connections to the edge, so the wires aren't trailing over the board?

[jaxbird]

Boards fabricated during development often do look oldschool! What matters is that you can get into them to measure and scope signals, and add/remove/swap components as needed. Unless you are working >GHz then track distances and compact boards aren't massively important. I'd be more than happy working with a board like that. If you do another one, maybe bring out the test probe connections to the edge, so the wires aren't trailing over the board?

I think you misunderstood my old school statement, it's not meant that old school isn't perfectly fine in most cases. It's more that well, it does look old school Probably 90% of the boards I make are old school, much easier to solder together without soldermask than smd, and much easier to bodge, swap components etc.

I do all my prototype and one off boards on CNC and I can do fine details down to 0.5mm pitch/10 mil tracks, but it can be time consuming and tricky to solder without mask and you don't want to rework a board like that too much. I try to stick with through hole unless it's e.g. a switchmode or something like this CPLD board where everything except the connectors is surface mount:

[jaxbird]

One of the limitations of the AD5933 is that the ADC sample rate for the DFT depends on the clock, so to capture a full wave for the DFT below ~10kHz an external clock is needed. The optimal is to capture just a single wave in the 1024 points sampled by the ADC. So it does require some slightly complex clock control.

Here a capture with the range down to 100Hz:




It is very slow doing the DFT at the low range though, not sure if the internal DSP runs off the same clock as the ADC.

[jaxbird]

Current setup, getting a bit large:



It works pretty well, but I've been investigating some noise issues.

First major noise source was the CPLD, I had the outputs in fast mode added a some better decoupling and changed to slow mode plus a small resistor on the output.

But still some significant noise present:




Turns out it's from the USB/UART converter, probably primarily from the USB data lines:




So I'll need to build a small isolation board with some optocouplers next and hope that gets rid of most of it.

[jaxbird]

After a bit more probing around, it's not the USB/UART, that was just from bad probing 

It appears to be when I use the external clock, here with internal clock (all measured directly on the AD5933, on input between SGND and feedback using very short ground connection, the output is clean)



Reasonably nice and clean.

But when I use external clock, it looks horrible with small signals:



And zooming, it's surely the external clock:





Edit: added details on measurement

[jaxbird]

I tried separating the DGND from SGND and AGND, although the datasheet says:

IT IS RECOMMENDED TO TIE ALL SUPPLY
CONNECTIONS (PIN 9, PIN 10, AND PIN 11)
AND RUN FROM A SINGLE SUPPLY BETWEEN
2.7V AND 5.5V. IT IS ALSO RECOMMENDED TO
CONNECT ALL GROUND SIGNALS TOGETHER
(PIN 12, PIN 13, AND PIN 14).

As according to their own demo board schematic they do keep the DGND seperate, but no real improvement. 

GNDs should probably be joined close to the AD5933, but I believe I have a bigger problem that that, with the clock noise I'm observing.

[jaxbird]

And just to confirm it's not because the external clock signal got wild over/under shoot:



Not sure yet what is really causing the noise on the AD5933 feedback or how to deal with it.

[Zad]

Have you tried reducing the clock drive level? That might drop the feed-through.

[jaxbird]

Have you tried reducing the clock drive level? That might drop the feed-through.

It's a good suggestion, thanks, but I have tried to reduce it, I have a 100R resistor on the output and then I pull it down with 1k at the AD5933, but I haven't observed any improvements. It only swings 0-3V with this configuration. I can try reduce it further, but I doubt it will improve on the noise.

I'm starting to wonder maybe the AD5933 I'm using have suffered from the rough environment of prototyping, but I don't really think so as it does work fine in all other aspects.

[dannyf]

The chip is a current drive. So the issue you are seeing maybe decoupling or earthing?

[jaxbird]

The noise is only really a problem with measurements in the very low range, in something like this inductor measurement it's got no effect:




But the noise really annoys me, it should not be there. 


Edit: No indicators, and the top log scale is not correct, but this chart is 1 Ohm to 500 Ohm scale.

[jaxbird]

The chip is a current drive. So the issue you are seeing maybe decoupling or earthing?

Good suggestion, I was thinking decoupling as well, but I already have 10uF tantalum + 100nF ceramic on it (as close as possible to the chip, soldered onto the pins), plus I don't think there shouldn't be a difference from using internal and external clock if it is caused by lack of decoupling. Also in the configuration I'm using, the chip is only driving a very benign load.

On the earthing, I am investigating, but I've not been able to measure any significant bounces anywhere, it is a fairly low consumption circuit after all, max something like 80mA.

[dannyf]

That's really nice. So it is connected to and controlled by a PC? Maybe after this, you can have Analog publish your work on their website.

[jaxbird]

That's really nice. So it is connected to and controlled by a PC? Maybe after this, you can have Analog publish your work on their website.

Thanks .. Yeah I've found it easiest to explore the functionality from a PC. Just simple USB <=> UART <=> mcu <=> I2C <=> AD5933.

Also I don't really need the CPLD anymore, by using a pic24 I can use a PWM to generate a clock up to ~6MHz, so one less device to program.

It's a fun project, but I doubt Analog would have any interest, I'm just playing around really Trying to push the AD5933 to the limits.

[jaxbird]

Actually the clock noise signal is present on the output (didn't notice it because I was dealing with a 2Vpp signal vs 10+mVpp noise), so at least that gives me some options to reduce it before it enters the current amp.

With some filtering applied (not visible, all smd stuff soldered on the copper side) measuring a pretty low impedance film capacitor:






Near 10 milliohm without any significant noise is pretty good I think, the capacitor resonance peak might be lower than that, but I don't think it would be realistic to aim for anything much below that with the setup I have

[jaxbird]

Been studying and playing around with the math involved to be able to plot an ESR curve. The key is really calculating the loss angle.

Here a capacitor measurement with ESR included: (it's a bit noisy, will have to clean that up)



I believe it's correct, but I cannot be 100% sure.

[jaxbird]

And a different capacitor (electrolytic):

[jaxbird]

And just for fun, the impedance of a small 2 way speaker: (Linear scale 20 Ohm)

[The Electrician]

Wow! All the way down to 10 Hz! 

Is 500 kHz the upper limit for the AD5933?

[jaxbird]

Wow! All the way down to 10 Hz! 

Is 500 kHz the upper limit for the AD5933?

With internal 16.776MHz clock it will do down to about 10kHz, below that it will no longer capture a full wave. But with a variable external clock signal it's no problem getting down to 10Hz .

It does take a slightly more elaborate algorithm to scan the lower frequencies, something like: set the clock to desired frequency (I use mcu PWM as clock for the lower frequencies) to capture at least 1 full wave for the 1024 point DFT, then scan 1 octave and repeat.

Downside is that it's pretty slow at low frequencies, as it needs to lock on a few waves first. But speed is not really high priority, beyond faster is preferred

The official, linear until, upper frequency is 100kHz as far as I remember, but it will do up to 500kHz with a roll off that starts around 200kHz. I can compensate for the roll off above 200kHz, with some loss of resolution.

I have considered trying e.g. a 20MHz external clock just to see if it will still run reliably 

[jaxbird]

Build up a simplified 2nd AD5933 to make sure the noise from external clock I'm experiencing, isn't just due to a faulty chip from all the prototyping short circuits etc.

But no, unfortunately it behaves exactly the same, very strange 




Tested with the same "gentle" clock signal I posted a capture of earlier, no over/under shoot, and lower in level than the AD5933 supply.

[jaxbird]

Also had a quick go at it with a 20MHz external clock, but, while the DDS parts works fine, the internal ADC does not want to play along, unfortunately.

Well, it was worth a try, and my clock signal wasn't the cleanest, but I don't think it will go much faster than the internal 16MHz.

Quick and dirty clock source:



Not too bad for 20MHz on a breadboard:

[sorin]

Which program did you use to visualize the output ?
Matlab, Labview ...

[jaxbird]

Which program did you use to visualize the output ?
Matlab, Labview ...

I use C# and do all the drawing for now, but it shouldn't be too hard to either export or write a direct interface for Labview etc.

[fonak]

I've been playing around with designing and building an impedance analyzer based on the AD5933.

It's a nice little chip. Capable of generating a sinewave from low Hz (some clock scaling needed) up to about 500kHz output sweep. And on the input, using a 12 bit ADC with up to 1 MSPS, it will do a 1024 point DFT and provide the complex imaginary and real values for each step in a sweep.

It is a bit complicated to get the best performance out of it.

And I'm mostly interested in low range impedance (below 100 Ohm). The chip is not really designed for that. But adding a buffer on the output making it capable of sourcing some more current should help.

My current setup is something like this:




And the current prototype:




(Not ideal to breadboard, but it works, only sacrificing some resolution really)

I will need to have some kind of reference measurement to compensate for phase and impedance changes caused by the buffer, I think the best approach will be a few calibration resistors that can be invoked on the input.

Impedance can be calculated as:

Z = SQRT(I * I + R * R)

And phase:

Phase = ATAN(I/R)

Should be possible to calculate capacitance and inductance as well once the resonance frequency is found.

Not sure if I can calculate ESR and any other interesting parameters, suggestions much welcome

Hi all

This is a very interesting project. In atachment please find formulas to calculate other parameters.
sorry for my bad english

[jaxbird]

Hi all

This is a very interesting project. In atachment please find formulas to calculate other parameters.
sorry for my bad english

Thanks, that is a very nice description and collection of key calculations 

I have been doing a bit of work on a 2nd iteration prototype, mainly added some filtering to compensate for the roll off above 100kHz plus the ability to measure voltage across DUT using MCU ADC and some programmable lowpass filters and gain.

Sill waiting for a few parts, so not built it up yet.

[jaxbird]

Another detail I have found is that the AD5933 does not behave so nice when switching frequency range:




It does not have much effect when measuring a capacitor, but with an inductor it can cause a jump in the impedance.

[jaxbird]

Designed a new prototype. It includes kelvin connections, some frequency response compensation, and a few more features.

Ready to engrave on CNC:

[jaxbird]

And the 2nd proto pcb after 20 minutes of engraving, drilling and milling:



Happy I now have my CNC being 99%+ wysiwyg with my custom software.

[jaxbird]

And testing the nearly fully assembled 2nd proto pcb:

[jaxbird]

Testing went well until I had a minor setback oops moment, heard a pop and sensed that smell you just don't want to smell when working on electronics, I'm sure everyone here is intimately familiar with it 

This is where the smell came from:



Of course it had to be the most inconvenient component to replace 

Not sure exactly what happened, but I was flipping over the pcb to get some measurements, I think some wires on my messy table shorted it out, I did have my psu limited to 200mA though, so not entirely sure.

Anyway, it seemed to be working fine before the "incident" so I'll just replace whatever blew and go from there.

[tonyarkles]

And the 2nd proto pcb after 20 minutes of engraving, drilling and milling:

As a curiosity, what are you using for bits? Those look like very nice isolation routes.

[jaxbird]

As a curiosity, what are you using for bits? Those look like very nice isolation routes.

Hi Tony,

I use cheap Chinese 1/8" carbide 30-40 degree v-bits. With a new sharp bit I can do really fine details down to 0.5mm pitch/ 8-10 mil traces or better, but it's a bitch to solder, and I tend to use my bits beyond dull

I find the most important factor is a very solid CNC and minimal runout of the spindle for best results.

[jaxbird]

Got it up and running again after the unfortunate oops moment 

The damage went beyond the blown 74hc4053 and included the whole negative rail (2 x 7660) plus the main AD5933 

Not sure how that happened, but I guess the large amount of decoupling did it during an unluck short.

Looking on the bright side, the opamps and buffer are fine

So back to testing.

[Odysseus]

I spent a semester working on a similar project for an independent research project.  I took a very different approach, preferring to produce the IQ values as DC  voltages using mixers, subsequently read by a precision delta-sigma ADC.  I found a lot of fixable issues with my design once the dust settled, but I haven't gotten around to addressing them in a new design.  It also relied on the use of bench-top lab equipment, so it wasn't a stand alone device, either.

Anyway, here's a link to all the documentation I produced for that project.  Maybe it will be of use to you or others. Good bedtime reading at least.

https://drive.google.com/folderview?id=0BzbfuLG7ZVrvcDhmajVXejZkN1k

[jaxbird]

Been a while, but I think I got as much as possible out of the AD5933, nice chip for impedance analysis up to 100kHz+

New toy should allow me to expand on that range. Using the digilent Analog func gen and scope shoudl allow up to near 10MHz, however not quite what I need just out of the box.

Here a nice capture of a 220nF capacitor using nothing but a 50R resistor and various cables:

 

[jaxbird]

So while pretty nice, it does show a number of issues, like in dB, not Ohm, and very small area being actually reliable (400k to ~ 1MHz) and of course missing many of the interesting traces like ESR etc.

[jaxbird]

Picture of 220nF film capacitor for reference:

[jaxbird]

And using a very low impedance 25uF capacitor, things get a bit noisy:

[jaxbird]

Picture of large 25uF film capacitor for reference:



For comparison, you can still see the "smallish" 220nF film at the edge. This is a very low impedance capacitor, never mind the "audio" label and 1% precision.

But this is a capacitor designed for precision speaker crossover design. While I'm aware of the general ignorance here on EEVBlog forum of the importance of using quality components for audio. I have to point out that passive crossovers is one of the areas where you cannot get away with shitty +/-20% electrolytic capacitors as you need to precisely align the crossover frequency with the mechanical phase alignment and phase errors of drivers to achieve a repeatably successful design, and we are working in the preferably <1% area for good results here.


Edit: Rant over, Triggered by a capacitor, sorry, I just read too much "shit" from "engineers" believing they always know what they are talking about.

[jaxbird]

Anyway, the question got to be, should I add a pre-amp or an amp to increase the sensitivity of the scope.

Hard choice, I have experience of building a high frequency constant current Howland based drive for the driver, but my experience with the Howland is that you cannot control the offset and it's very difficult to balance, as in you need some serious precision resistors for reliable results.

I think I could amplify the differential output using some e.g. AD8051 in stages of x2, maybe 3 stages in total for optimal results, then using some 74hc405x to switch various loads and gains for extended ranges, all controllable by the device using the available digital IO.


 

[jaxbird]

Interesting old school 100nF versus multilayer 100nF:





Edit: 100nF not uF ( )

[jaxbird]

Pictures of above measurements:

Oldschool:



Multilayer:

[jaxbird]

Next target, not sure if I can capture it, don't have kelvin clips that small, but it's 0805 100nF for comparison to the previous measurements:



edit: not uF, but nF, sorry

[jaxbird]

Not easy, but managed to capture the tiny 0805 100nF cap:

[miguelvp]

Friend of mine in Spain that is a ham as I was calling him to wish him a happy new year pointed me to this:

http://miniradiosolutions.com/

And this:

https://greatscottgadgets.com/hackrf/

last not related but awesome too

[c4757p]

Ooh, I hadn't seen this before, this project looks nice. I'll have to bookmark it. I'm working on something quite similar but over a 1 kHz to 150 MHz range as my BSEE senior project - perhaps I'll make a post about that too.

[jaxbird]

Friend of mine in Spain that is a ham as I was calling him to wish him a happy new year pointed me to this:

http://miniradiosolutions.com/

And this:

https://greatscottgadgets.com/hackrf/

last not related but awesome too

Thanks, but I think those are slightly higher in frequency and lower in power than what I am trying to achieve.  My goal is more in the 10Hz to 10MHz range, but being able to provide plenty of current to measure low impedance at these frequencies.

[jaxbird]

Ooh, I hadn't seen this before, this project looks nice. I'll have to bookmark it. I'm working on something quite similar but over a 1 kHz to 150 MHz range as my BSEE senior project - perhaps I'll make a post about that too.

Thanks, but keep in mind this project was originally intended to push the AD5933 to it's limits, which I pretty much did earlier with reasonable results, but now I have switched to using a pre made Analog Discovery Unit from Digilent to aim even higher, as in beyond the 100-500kHz limit of the AD5933. Hoping to achieve close to 1Hz - 10MHz instead. And hopefully being able to probe down to near 10mR.

My goal now is primarily to write some software and create a module for the digilent unit for precision impedance analysis across a wide range, including plotting charts for impedance, capacitance, inductance, ESR, etc.

[jaxbird]

Interesting old school 100nF versus multilayer 100nF and 0805 100nF x7r (50V of questionable origin, but claimed Samsung)

Clear winner is the smd 0805 x7 50V, but closely followed by the multilayer 100uF through hole and as a sloppy last is the old school single layer 100uF ceramic.

[dannyf]

Pictures of above measurements:

Those kelvin clips of yours are just nice,

[dannyf]

Those chips have a big limitation: they assumed that the signal applied to the dut is the same as from the chips. AD could have implemented another ADC channel and to get a separate voltage measurement and do FFT from there. That would have made those chips infinitely more versatile.

On that front, have you experimented with a discrete buffer on the excitation pin? Essentially, run a pair of current amps (pnp+npn, with a resistor in between the bases and emitters).

The gain loss should be small. and can be corrected in software if you allow calibration.

[robrenz]

Pictures of above measurements:

Those kelvin clips of yours are just nice,

Yes they are awesome! Anyone have a link where I could get a set? 

[dannyf]

My kelvin clips - practically free.

I could have squeezed in a 1/8w 100ohm resistor there.

[jaxbird]

Those chips have a big limitation: they assumed that the signal applied to the dut is the same as from the chips. AD could have implemented another ADC channel and to get a separate voltage measurement and do FFT from there. That would have made those chips infinitely more versatile.

On that front, have you experimented with a discrete buffer on the excitation pin? Essentially, run a pair of current amps (pnp+npn, with a resistor in between the bases and emitters).

The gain loss should be small. and can be corrected in software if you allow calibration.

The AD5933 is somewhat limited, but works well within 1kHz to 200kHz. I used a Howland current source on the output with reasonably good results. I haven't attempted a discrete buffer.

I have now switched to experiment with an Analog Discovery unit from Digilent, basically 2 x 100Msps 14 bit ADC, plus 2x100Msps 14 bit DAC

(http://www.digilentinc.com/Data/Products/ANALOG-DISCOVERY/Discovery_TRM_RevB_1.pdf)

It is of course 10x the cost of AD5933 (http://www.microchipdirect.com/ProductSearch.aspx?Keywords=TDGL023), but much better suited and useful for many other purposes. With the dual ADC I can do a DFT and get absolute values.

But I probably cannot get away with a purely passive solution (simple shunt resistor) as it isn't happy driving more than 50R load at 2Vpp amplitude (already rolling off a bit early at that load) and the ADCs LSB is something like ~300uV at highest sensitivity.

[jaxbird]

Pictures of above measurements:

Those kelvin clips of yours are just nice,

Yes they are awesome! Anyone have a link where I could get a set? 

Hehe, yeah well, it works

I have added some actual kelvin clips now, the other solution was a bit fiddly.

[jaxbird]

So I've been playing around with the software, rewriting it a few times before getting to good results. My first attempt was to simply use 0 crossing and RMS on voltage on the sampled data to calculate phase and amplitude, but, while it works, it isn't easy to get accurate phase at higher frequencies even with some interpolation to improve the accuracy.

The solution where I get best results is using DFT. And a few tricks of averaging multiple sample buffers depending on amplitude. That does give reliable results even when the signal is down to 1mVpp. But still it does get too noisy when attempting to measure down near 10mR.

Here a capture of the new software with a film capacitor:

[jaxbird]

And here an electrolytic capacitor:



Still need to add ESR and other interesting values to make it more useful. Great thing is that this approach does not need any calibration as the Digilent/Analog unit got good accuracy.

[jaxbird]

With the Kelvin clips shorted and 50R shunt it measures ~13mR, so that is the lower limit in this configuration, the 13mR probably come from leakage between the channels.




So to get below 10mR I'll need either more current or higher sensitivity or maybe both.

[jaxbird]

Did some work on cleaning up the software while waiting for parts:

[codeboy2k]

I've been following this thread.. excellent work!

I want one:)

[fonak]

Great project. Is there a chance to public full schematic, software and firmware?

[kripton2035]

as I already have sample of ad5933 for some years, I want one too !

[dannyf]

Nicely done.

[jaxbird]

as I already have sample of ad5933 for some years, I want one too !

He is using the Analog Discovery for hardware, not the ad5933.

AcHmed99 is correct I have switched to using an Analog Discovery unit (14 bits vs 12 bits, 100 Msps vs 1 Msps etc) plus you get do do your own DFT (reference against DUT) with the Analog unit.

Sure it's 10x the cost (~$220) of a single AD5933 (~$25), but IMO well worth the extra money.

Currently I'm working on an amplifier allowing me to measure down to near 1m?, as the current minimum of ~13m? just isn't low enough in many cases. I've got some AD8038 on order that will hopefully allow this with the full 10+MHz bandwidth and some gain.

With the aim of 1m? to ~500k? within 1Hz to 10MHz, I might have to compromise a bit on accuracy in the extremes, but it should be possible.

[jaxbird]

Nicely done.

Thanks, encouragement is a good motivator

[moffy]

Your posts have been very informative! I have to agree that the Analog Discovery appears well worth the money.

[fpliuzzi]

I remember the excellent thread in 2013 by The Electrician about capacitor measurements on an Impedance Analyzer and thinking to myself how really useful a device like this would be on the bench.

Little did I know at the time that I would later be witnessing a capable Analog Discovery/Impedance Analyzer blossom into reality. This is a fascinating thread indeed. 

[jaxbird]

I've discovered that too long (200mm) measurement wires do introduce a relatively "large" amount of inductance (250nH+) and does have some effect on e.g. the resonance frequency when measuring caps and inductors. I can only estimate the error but measuring a 100nF capacitor should probably resonate around 5-7MHz but I only measure half that, of course dependent on the capacitor, and not a common published spec, so a bit of guesswork. Anyway, I find very short Kelvin connections (~30mm) give significantly better results at higher frequencies.

I'm thinking about methods to compensate for longer leads as really short leads are more difficult to work with, I'd like to be able to stick probes in a PCB and measure impedance where interesting, decoupling etc.

Here a test of a somewhat common (I guess) 10uF, 100nF, 10nF ceramic capacitor combination, picture of the setup:




Apologies for messy soldering, did it on the table with a 3.5mm tip I had in the iron already, really need more than 2 hands to hold tweezers and iron when doing this. (so many excuses)

[jaxbird]

Measured at the point of the 10nF capacitor:

[jaxbird]

And measured at the point of the 10uF capacitor:




Interesting difference, just moving the probes a few mm, anyway, still got some ~30-50nH and ~20pF error. And results above 10 MHz are questionable, plus currently unable to measure below 13mR, so some error on the low.

[jaxbird]

Less fortunate combination of 1000uF electrolytic with 220nF film in parallel:

[The Electrician]

Pictures of above measurements:

Those kelvin clips of yours are just nice,

Yes they are awesome! Anyone have a link where I could get a set? 

I see several on eBay:

http://www.ebay.com/itm/1-Pair-of-Large-Kelvin-Clip-Fine-Tip-Gold-Plated-for-LCR-Meter-981-002-USA-/191227969924?pt=LH_DefaultDomain_0&hash=item2c86134184

[The Electrician]

Jaxbird, the old HP/Agilent/Keysight fixtures for impedance analyzers are available on eBay:

http://www.ebay.com/itm/HP-Agilent-16047A-/400834797655?pt=LH_DefaultDomain_0&hash=item5d539da057

http://www.ebay.com/itm/HP-Agilent-16034E-LCR-Meter-surface-mount-Test-Fixture-with-case-HP-4274A-4275A-/141416235256?pt=LH_DefaultDomain_0&hash=item20ed1080f8

They are somewhat pricey, but I've noticed that there are Chinese look-alikes available for much less:

http://www.ebay.com/itm/TH26001A-4-terminal-test-fixture-for-LCR-meter-TH2811D-/261419837445?pt=LH_DefaultDomain_0&hash=item3cddd62805

There is a surface mount version that shows up occasionally on eBay, too.

The spacing of the BNC connectors on other-than-Agilent impedance analyzers, such as Wayne-Kerr, Hioki, Tonghui, etc., is the same as the Agilent, so all these various fixtures will fit all the various impedance analyzers (there may be the occasional exception).

I'm sure you would get better results at the higher frequencies if you would install your project in a metal box with 4 BNCs that would accept a fixture on the outside of the box.

[jaxbird]

I remember the excellent thread in 2013 by The Electrician about capacitor measurements on an Impedance Analyzer and thinking to myself how really useful a device like this would be on the bench.

Little did I know at the time that I would later be witnessing a capable Analog Discovery/Impedance Analyzer blossom into reality. This is a fascinating thread indeed. 

Yeah, the many measurements by The Electrician with his precision Hioki analyzer and explanations have been a great inspiration.

Of course this solution will not provide anywhere near the same level of precision, but still very useful.

[jaxbird]

Pictures of above measurements:

Those kelvin clips of yours are just nice,

Yes they are awesome! Anyone have a link where I could get a set? 

I see several on eBay:

http://www.ebay.com/itm/1-Pair-of-Large-Kelvin-Clip-Fine-Tip-Gold-Plated-for-LCR-Meter-981-002-USA-/191227969924?pt=LH_DefaultDomain_0&hash=item2c86134184

I believe they were just having a laugh at my early "Kelvin clips"

[jaxbird]

Jaxbird, the old HP/Agilent/Keysight fixtures for impedance analyzers are available on eBay:

http://www.ebay.com/itm/HP-Agilent-16047A-/400834797655?pt=LH_DefaultDomain_0&hash=item5d539da057

http://www.ebay.com/itm/HP-Agilent-16034E-LCR-Meter-surface-mount-Test-Fixture-with-case-HP-4274A-4275A-/141416235256?pt=LH_DefaultDomain_0&hash=item20ed1080f8

They are somewhat pricey, but I've noticed that there are Chinese look-alikes available for much less:

http://www.ebay.com/itm/TH26001A-4-terminal-test-fixture-for-LCR-meter-TH2811D-/261419837445?pt=LH_DefaultDomain_0&hash=item3cddd62805

There is a surface mount version that shows up occasionally on eBay, too.

The spacing of the BNC connectors on other-than-Agilent impedance analyzers, such as Wayne-Kerr, Hioki, Tonghui, etc., is the same as the Agilent, so all these various fixtures will fit all the various impedance analyzers (there may be the occasional exception).

I'm sure you would get better results at the higher frequencies if you would install your project in a metal box with 4 BNCs that would accept a fixture on the outside of the box.

Thanks, it would be absolutely nice to have some proper test fixtures for both leaded and SMD , but I think they might be a bit overkill for this project, I'm not convinced that the somewhat questionable precision I will be able to achieve would justify the nice fixtures. But might change if I find a reliable way to compensate for the added inductance and capacitance in software .. still thinking about that one. I think I will have to read some more of this document :

http://cp.literature.agilent.com/litweb/pdf/5950-3000.pdf

it's been very helpful so far on a lot of practical considerations and the basic complex number math for many of the calculations.

But before that I'll attempt to build an amplifier circuit that will allow measurements down to near 1mR with full (10+MHz) bandwidth with minimal phase change.

[jaxbird]

Since it's weekend, I've been working on a reliable way to calibrate the little instrument.

The Agilent Impedance Measurement Handbook has been a great help. (thanks Agilent, or rather Keysight).

It does describe in some detail the method used to perform an Open/Closed calibration to compensate for parasitic inductance etc.




There is a bit more to it, but that formula is a good start.

[jaxbird]

Example of difference in measurement of a 1uF capacitor, with and without calibration:

Without calibration:




With calibration:




Quite a difference in the resonance frequency.

[jaxbird]

Did a quick accuracy test following the open/short calibration. This calibration doesn't require any references, it simply requires a scan with disconnected Kelvin clips followed by a scan with shorted clips.

Anyway, using a few resistors from 10k to 10mR, mostly standard 1% and 5%, but very low values (10mR and 20mR) are just short pieces of wire made using a conventional milliohm/microohm meter.

All measured at 1kHz using 50R shunt:

Nominal value:	Measured value:
10k	9.466k
1k	0.987k
100R	98.8R
10R	9.95R
1R	0.986R
100mR	98.7mR
50mR	50.8mR
20mR	19.8mR
10mR	10.5mR

At 10k and 10mR is pushing it a bit, but besides that it's pretty good accuracy within a few %.

It will easily do higher than 10k, with e.g. a 5k shunt. Input impedance should be in the mega ohm range, so good accuracy should be possible up to a few 100k. Will test that at some point.

[jaxbird]

Picture of test resistors for reference:

[jaxbird]

While I have figured out how to perform and apply calibration, I am still fumbling with a simple problem.

The problem is, when applying calibration, I need to use complex numbers, meaning I build tables of these, indexed by frequency.

But as the measurement frequency is likely to be between 2 frequencies in the lookup table, interpolation is needed.

Is there an easy standard way to interpolate between 2 complex numbers? (just linear interpolation, no fancy spline needed)

Must be a somewhat common problem, but quick google just seem to suggest everyone use matcad 

I'm sure there is a simple way of doing it, and I'll figure it out eventually, but any hints much appreciated.

[jaxbird]

Regarding to doing the accuracy measurements at 1kHz, why not higher with 10MHz bandwidth available? Well, it's simply to avoid any errors introduced by the resistors.

Nicely explained by these charts: (again credits go to Agilent)

[jaxbird]

Here demonstrated with real measurements: (the 10k resistor is a bit too low in value to show much stray capacitance, but trend visible at 10MHz)

[jaxbird]

Using 100mR it's much easier to see the effect of lead inductance:

[jaxbird]

I have very recently received a nice bag of goodies, it was generously sent by The Electrician:



5 capacitors and 2 inductors, including captures of measurements on his high end Hioki analyzer.

Very nice references for this project, huge thanks to The Electrician.

[jaxbird]

This is great because it means no more fumbling around in the dark on whether this device will perform usable measurements or not.

Here first using the small low impedance electrolytic Sanyo 22uF/16V OS CON.

Reference measurements:





And my measurements:

[jaxbird]

The measurements are much more noisy at low impedances, but accuracy is reasonably good.

Comparison based on the cursors at 120Hz:

Unit	Reference	Measured
|Z|	57.589R	57.347R
Rs	923.101mR	887.992mR
D	0.014433	0.015486

[jaxbird]

and at ~100kHz:

Unit	Reference	Measured
|Z|	79.990mR	73.992mR
Rs	21.821mR	19.488mR
D	0.282733	0.284133

[The Electrician]

Hi, Jaxbird!

I'm glad they came through OK.  Did they make you pay any customs duty? 

First thing I notice is the SRF of your measurement appears to be at about 400 kHz, whereas the sweeps I included indicate about 750 kHz.  I made sure the parts were inserted into the fixture with the absolute shortest leads possible, which typically is right up against the body of the part in the case of the capacitors.  I suspect you have just a little more lead length involved in your measurement, or perhaps your compensation for the parasitics of your clips is a little off.

But, besides that, this is a really great project!  Wouldn't it be nice to be able to get reasonably accurate impedance sweeps for only a few hundred dollars instead of 10's of thousands?   

[jaxbird]

Hi, Jaxbird!

I'm glad they came through OK.  Did they make you pay any customs duty? 

First thing I notice is the SRF of your measurement appears to be at about 400 kHz, whereas the sweeps I included indicate about 750 kHz.  I made sure the parts were inserted into the fixture with the absolute shortest leads possible, which typically is right up against the body of the part in the case of the capacitors.  I suspect you have just a little more lead length involved in your measurement, or perhaps your compensation for the parasitics of your clips is a little off.

But, besides that, this is a really great project!  Wouldn't it be nice to be able to get reasonably accurate impedance sweeps for only a few hundred dollars instead of 10's of thousands?   

Thanks for sending the references, I very much appreciate it  No duty luckily I hardly every have to pay extra duty (only when I receive items via FedEx or DHL)

Yeah, I have tried to study the comparison of the measurements, I believe I got some residual inductance I need to deal with (probably around 10-20nH or there about), here an overlay of the 2  measurements on same scale:



I hope to be able to construct and share such a device, sure, it will not be a reference device suitable for publishing datasheets, but something generally useful to get a good idea of behavior/performance.

[The Electrician]

I hope to be able to construct and share such a device, sure, it will not be a reference device suitable for publishing datasheets, but something generally useful to get a good idea of behavior/performance.

I think it's doing quite fantastic!  Don't forget, by the way, especially for electrolytics, the ambient temperature effects.  It's instructive to run the analyzer in continuous sweep mode and heat up a component with a "warm" air gun, or even just grasp it, and watch the ESR of an electrolytic change; of course, you can do that without needing a sweep.

People buy a DE-5000 for $100.  I'd easily pay a few hundred for sweep capability.  A person learns after playing with an impedance analyzer that you can tell the health of an electrolytic in an instant with a single sweep.

[AlfBaz]

I hope to be able to construct and share such a device, sure, it will not be a reference device suitable for publishing datasheets, but something generally useful to get a good idea of behavior/performance.

I think it's doing quite fantastic!  Don't forget, by the way, especially for electrolytics, the ambient temperature effects.  It's instructive to run the analyzer in continuous sweep mode and heat up a component with a "warm" air gun, or even just grasp it, and watch the ESR of an electrolytic change; of course, you can do that without needing a sweep.

People buy a DE-5000 for $100.  I'd easily pay a few hundred for sweep capability.  A person learns after playing with an impedance analyzer that you can tell the health of an electrolytic in an instant with a single sweep.

+1
I've been messing around trying to use FG and scope under automation to do bode plots and impedance sweeps.
Some success with bode plots on two port systems. Nothing noteworthy doing impedance sweeps on single port DUT's. Many, many sources of errors/limitations and assumptions in my methods

[jaxbird]

I hope to be able to construct and share such a device, sure, it will not be a reference device suitable for publishing datasheets, but something generally useful to get a good idea of behavior/performance.

I think it's doing quite fantastic!  Don't forget, by the way, especially for electrolytics, the ambient temperature effects.  It's instructive to run the analyzer in continuous sweep mode and heat up a component with a "warm" air gun, or even just grasp it, and watch the ESR of an electrolytic change; of course, you can do that without needing a sweep.

People buy a DE-5000 for $100.  I'd easily pay a few hundred for sweep capability.  A person learns after playing with an impedance analyzer that you can tell the health of an electrolytic in an instant with a single sweep.

Thanks, this is good advice  Also as you mentioned earlier, not being sloppy with the calibration makes a huge difference .. I've made a few changes, better Kelvin Clips, shielded sense wires and a few more details. All contributing to better results.

Current setup looks like this: (still need to add the low level amplification opamps etc)

[jaxbird]

Here a measurement of the 6.8uF film capacitor:



Reference measurements:





My measurement:



Overlaying the measurements:




Edit: adding images

[The Electrician]

This is most excellent!

See that bump in the ESR and DF at about 3 MHz?  The first time I saw that on a large film cap, I thought something was wrong with the analyzer or the fixture.  But it's real; it looks just like that on an Agilent 4294 and on the Wayne-Kerr.  I think it's an internal resonance due to the nature of the metallization sprayed on the ends of the extended foil.

Your analyzer shows it perfectly!

[jaxbird]

This is most excellent!

See that bump in the ESR and DF at about 3 MHz?  The first time I saw that on a large film cap, I thought something was wrong with the analyzer or the fixture.  But it's real; it looks just like that on an Agilent 4294 and on the Wayne-Kerr.  I think it's an internal resonance due to the nature of the metallization sprayed on the ends of the extended foil.

Your analyzer shows it perfectly!

Thanks, I'm pretty happy with this performance, but it's still got some areas where the device does not perform this well. I'll try post a few examples. E.g. when doing inductor measurements the Rs does not quite follow. I'm experimenting with some dynamic test current adjustments (voltage/shunt resistance) based on impedance. That should also extend the range up to ~500k Ohm.

I was wondering about that bump, seems like nearly all film capacitors 1uF+ got this behavior. I was thinking they had maybe a smaller 2nd capacitor included either intentionally or due to mechanical construction. Thanks for providing the actual explanation 

 

[jaxbird]

Got the amplifier part of the board populated, it's not going to win any "pretty soldering award", but it works:



It should allow measurements down below 1mR, here with a large film capacitor:



And zoomed in on the resonance frequency:

[jaxbird]

Still needs some work on the calibration to get the amplification completely working, I get some phase offset although the amp circuit should be very close to 0 degrees from 1 to 20+MHz.

Anyway, here with a piece of ~1mR wire:



And a 100kR 1% resistor:

[jaxbird]

Spent a few evenings getting the amplification to behave, according to simulation it should be flat to 20MHz with almost no phase shift. But after measuring it, in reality there is an ~8 degree phase shift already at 10MHz, probably should have used more isolation between traces on the PCB.

Anyway, I can compensate for the phase shift during calibration, as the instrument is capable of measuring itself.

The great thing is that with 16x amplification it is now capable of measuring very low impedance, down to below 1 milliohm.

Here a test using the large 20uF reference capacitor from The Electrician:

[jaxbird]

Reference measurements for the 20uF:

[jaxbird]

And my measurement:




I don't think it will get much better using the current hardware, but there are still a few tweaks I want to play with.

[The Electrician]

That is astounding! 

It matches the reference curves so well it's hard to believe you can get such good performance with so little hardware.

This shows that a low priced USB based box could be sold for a few hundred dollars--certainly well under $1000!

This has got to be one of the neatest projects on the forum.

[kripton2035]

and the $250 of the dev board he uses could easyly be replaced with a small mcu + good adc board self made for some $50 IMHO.
the dev board helped to develop the thing faster, but is overpriced for how he uses it ?

[Kevin.D]

Been following this interesting thread/project. Jaxbird .
What's your plans when completed project. ? 

[kripton2035]

and the $250 of the dev board he uses could easyly be replaced with a small mcu + good adc board self made for some $50 IMHO.
the dev board helped to develop the thing faster, but is overpriced for how he uses it ?

He already has it, so its adding addtional functionality to the tool. The discovery can be had for 99 bucks, that all I paid for mine. I would be interested to see your 50 buck implementation though.

the $99 is for an US student only ... otherwise it's $279

To qualify for student pricing, the customer must be a student at a US academic institution and be required to purchase the board for a class. The quantity is limited to one board per student.

[jaxbird]

That is astounding! 

It matches the reference curves so well it's hard to believe you can get such good performance with so little hardware.

This shows that a low priced USB based box could be sold for a few hundred dollars--certainly well under $1000!

This has got to be one of the neatest projects on the forum.

Thanks, I'm happy the amplification was quite successful. It's been a long, interesting, journey (I actually find it relaxing to work on ), while the instrument is quite usable, there are a few things I want to change, so I think it might be time for a new prototype soon.

The credit for the good performance has to go to the Analog Discovery unit, it is a very nice little device that packs a lot of power in a very small footprint. The dual 14 bit 125Msps ADC (AD9648) is pretty amazing.

But I might have painted a little too rosy picture of the performance in my eager to show what it can do. Under certain circumstances the results look significantly worse. e.g. measuring Rs of a small capacitor at low frequency. And Rs on inductor measurements is lower that what it should be.

Also you can see that the Dissipation factor is slightly below what it should be, but as long as it tracks nicely, I can easily live with that.

It is a bit time consuming to get a measurement like the one I posted earlier: I need to first let everything warm up, then do an open/short calibration, and then I can do the measurement.

And as you advised, being careful with calibration does indeed pay of, I have found that even the few nH in the tips of the Kelvin clamps can make a difference, so to make sure I get open/short at the same point I do it like this:

Open:


Short:


Do you only use open/short calibration, or do you use a reference as well?

Another issue I would like to have a look at is compensating for longer cables. Currently I have to use very short cables to get good results.

[jaxbird]

and the $250 of the dev board he uses could easyly be replaced with a small mcu + good adc board self made for some $50 IMHO.
the dev board helped to develop the thing faster, but is overpriced for how he uses it ?

You would not be able to replicate the performance of the discovery module for $50, the ADC alone is something like $80. However it could be possible to do something clever and use the same technique as sampling scopes, as what we want to sample is a repeating wave (using scan and not sweep) and we already know the frequency, so maybe a dual DAC, one channel to generate the output and one channel combined with a fast comparator to sample the signal. Should be possible and a lot cheaper than high speed ADC and FPGA, mem etc.

Might also be possible to use a completely different approach with MLS sequence to sample everything in one go.

While fun and interesting, I don't think I'm prepared to go that far

[jaxbird]

Designed a new version, slightly simplified but should have equally good performance.

Here ready for engraving on some double sided FR4:

[jaxbird]

Been following this interesting thread/project. Jaxbird .
What's your plans when completed project. ?

Thanks, well, I don't really have any plans, it's just something I found interesting to pursue. A very useful instrument to have in the collection was my motivation . Surely not something anyone would use on a daily basis, but occasionally you just want to know the characteristics of e.g. a capacitor or inductor, plus all the other areas where impedance analyzers are useful. I'd like to make a probe I can stick in a PCB, not so much for fault finding, more to check designs and get better insight into various behaviors.

[jaxbird]

...
the $99 is for an US student only ... otherwise it's $279
...

You can get them from MicrochipDirect for much less than that without being a student, plus I believe they run a campaign with free shipping this month. I'm sure I got an email somewhere, and that reminds me, I should order some stuff

Just don't get it because of this project, as I have no idea when I'm confident enough in the design and software to share.

Edit: to clarify, Microchip sells these for ~$220/each no matter who you are

[jaxbird]

New board ready:

[jaxbird]

Testing the extremes of the instrument. I've found this little 22nF Wima is very good to test the extremes in both frequency, sensitivity and linearity.

[jaxbird]

Measurement of 22nF from 1Hz to 20MHz, 10M to 10m range:



Much better than I had expected, I was expecting upper limit to be about 200-500k, but 5M+ is indeed possible.

[AlfBaz]

I'm curious, how long does it typically take to do a complete frequency sweep and analysis from say, 1Hz to 10Mhz and what is the minimum number of samples you would need for that range?

[jaxbird]

I'm curious, how long does it typically take to do a complete frequency sweep and analysis from say, 1Hz to 10Mhz and what is the minimum number of samples you would need for that range?

I haven't actually timed a full scan, the time depends on how low in frequency you want to go, anything below 10Hz is pretty slow. I always use the full 2x8k sample memory, but I do have some optimization, below 10 Hz I only sample 2 full waves, so that is 2 seconds+ per sample @ 1Hz, between 10 and 100Hz I sample 4 waves, 100 to 1kHz is 8 waves and above that 16 waves for the DFT. Retrieving the sample data and computing DFT plus various calculations for each sample is pretty quick,

For e.g. a capacitor measurement I rarely use anything below 100Hz, unless it's a large capacitor used in rectification.

In most cases 200 samples across the frequency range is enough to give a nice detailed measurement. It's what I use by default.

[jaxbird]

I've received a large number of private messages on this project while i've been away. It's always been my intention to release something at some point, whatever that means, but sometimes life happens, however I will post a video very soon explaining the functionality and I will make the design and software available free of charge, I hope my many hundreds hours of original work put into this project will be appreciated and not just copied into some shitty open source kickstarter like many before me or sold as original work on some shitty chinese site.

TLDR; I will post instruction video, schematics and software within the next week or so.

[dannyf]

great work, jaxbird.

not just copied into some shitty open source kickstarter

Likely that's where it is going.

Providing some hack-proof and visible signature in the software will help deter that.

[Wolfram]

I've received a large number of private messages on this project while i've been away. It's always been my intention to release something at some point, whatever that means, but sometimes life happens, however I will post a video very soon explaining the functionality and I will make the design and software available free of charge, I hope my many hundreds hours of original work put into this project will be appreciated and not just copied into some shitty open source kickstarter like many before me or sold as original work on some shitty chinese site.

TLDR; I will post instruction video, schematics and software within the next week or so.

Amazing work, looking forward to finally being able to make a decent impedance analyzer.

[jaxbird]

Would everyone prefer a kit available on eBay? (including pcb, essential high speed opamps, caps, shunts, cables and clips)

or just making a pcb themselves being ignorant of the essential knowledge how to achieve the best performance?

[jaxbird]

[fpliuzzi]

Your ebay kit idea sounds like an optimum way for a user, such as myself, to more easily mirror the quality measurement results that you have produced using your unit. I would certainly be interested in a kit containing the items you've listed above.

[jaxbird]

Small teaser:




I think I do need to add at least a little bit of input protection, perhaps a handful of BAV99 and some resistors.

[LaserSteve]

Considering your software looks an order of magnitude or two  better then the AD Demo board's included code, I'm highly interested. I work at a university and this chip would be the cat's meow for one of our small projects.  I have the demo board, and while the board is excellent. I have my doubts that the contract programmer knew what he was working on.

Steve

[fpliuzzi]

That's some teaser. Your Impedance analyzer software is pretty remarkable. You've really included quite a bit of flexibility into its operation.

[jaxbird]

Considering your software looks an order of magnitude or two  better then the AD Demo board's included code, I'm highly interested. I work at a university and this chip would be the cat's meow for one of our small projects.  I have the demo board, and while the board is excellent. I have my doubts that the contract programmer knew what he was working on.

Steve

Well, I'll release it for free once I have everything ready, just need to find the time to get everything together. I do contract work, but I doubt you'd find my rates reasonable .. especially if looking for a bargain. I mostly deal with very large corps where I really do need to factor in a lot of overhead, insurance, travel, meetings etc.

That's some teaser. Your Impedance analyzer software is pretty remarkable. You've really included quite a bit of flexibility into its operation.

Thanks, I did put a lot of effort into this project. I do think it's got a lot of potential, but sure, only a somewhat narrow audience.

[voltsandjolts]

This is fantastic jaxbird. Well done. 
Dave needs to do a review on this puppy when its complete, "young players" could learn a lot from this tool. And, ahem, some older players like me.

[rx8pilot]

This is a cool project. Wish I could have this right now.

[quarks]

 great project

[jaxbird]

This is fantastic jaxbird. Well done. 
Dave needs to do a review on this puppy when its complete, "young players" could learn a lot from this tool. And, ahem, some older players like me.

Thanks, I really find this tool has been very helpful in getting a better understanding of the relationship of many impedance related parameters, especially how to determine the quality of a capacitor, inductor etc.

I just wanted to make a usable ESR meter, but I kinda got caught up it in, and then The Electrician showed off his beautiful impedance analyzer, then that became my new benchmark, so I guess it's an unhealthy obsession coming to an end really. I'm proud of the end result though, I achieved or exceeded all the parameters I had for the project.

I have posted a video of the plug in board for those intersted.

[jaxbird]

New video, describing the plugin board and the design decisions involved:

[fpliuzzi]

I like the addition of the option to install SMA connectors on the input side of the PCB. It's always nice to be able to add an assortment of test jigs to a measurement device if needed. Nice touch.

[ruairi]

Fascinating project, thanks for the great video.

[jaxbird]

Fascinating project, thanks for the great video.

I like the addition of the option to install SMA connectors on the input side of the PCB. It's always nice to be able to add an assortment of test jigs to a measurement device if needed. Nice touch.

I will probably make a kit available, I think it will be around $50 including all parts (except the Analog discovery itself), it will work with both version 1 and 2 of the Analog Discovery and include everything including reasonable kelvin clips. I hope that is reasonable, I think it will cover the cost of all the parts with shipping from various vendors etc. It should not be a problem to solder the kit together for anyone interested, it's all 0805 or larger parts and so8/16 parts.

Shipping will be an additional $10 world wide. I will only be able to handle small quantities, but I think not a problem as the interest in this should be quite narrow. Anyway, the software if free for anyone who'd rather build their own, just keep in mind this is a hobby project so while I'll do my best to fix any bugs it might take a bit of time for me to respond to any bug reports.

I'll post another video on general usage and calibration of the device for best performance.

I'm sending one to Dave for evaluation, will be interesting to hear his opinion on the project.

[ruairi]

That sounds great, I'm in for a kit whenever they are ready. 

Thank you,
Ruairi

[steverino]

just a thought, but why not approach digilent about stocking your designs?  anybody interested in these will have to purchase an analog discovery.  maybe you could earn some $$ from your work.

[fpliuzzi]

Thanks for all of the effort you have put into this project. If a kit is made available, please count me in for one.

[jaxbird]

That sounds great, I'm in for a kit whenever they are ready. 

Thank you,
Ruairi

Thanks for all of the effort you have put into this project. If a kit is made available, please count me in for one.

Thanks, any money made on the kit will go towards getting a Red Pitaya and implementing the same functionality for this device. I saw from one of Dave's recent videos they already have some very basic impedance measurement software, but it didn't look all that great and it looked like it didn't do any calculations to include parameters beyond basic impedance, I believe I can do much better than that, also I noticed they sell their switching board at a whopping $300+, that seems very expensive.

[jaxbird]

just a thought, but why not approach digilent about stocking your designs?  anybody interested in these will have to purchase an analog discovery.  maybe you could earn some $$ from your work.

Thanks, I might try contact them to see whether they would be interested in at least featuring the project on their website, I believe the impedance analyzer solution does provide enough value to purchase the Analog Discovery purely for this purpose.

[AlfBaz]

I believe the impedance analyzer solution does provide enough value to purchase the Analog Discovery purely for this purpose.

+1 That is precisely what I have done, solely on the basis of what you have demonstrated. Put me down for a kit and as a bonus there's your excellent Audio Analyser Suite

[kridri]

Are you planning to release the software? In a kind of OSHW license?

[jaxbird]

Are you planning to release the software? In a kind of OSHW license?

Unlikely, I don't have good experience releasing open source software.

But in case I lose all interest in maintaining the project I will make the source code available and hope someone else will take over and keep it alive (unlikely to be honest, most people just butcher such projects for their own benefit)

[jaxbird]

Some kind of soft count of who is interested in a board for this project would be very helpful, if you are interested, please reply to this message, it's not binding, just to give me some idea how much I should order.

Cheers

Edit: just to be clear, I give no promises on performance.

Edit again: Don't be shy, a PM will work as well.

Open for 2 weeks, then it will probably take me a week to pack everything and 1-3 weeks shipping

[Carrington]

Some kind of soft count of who is interested in a board for this project would be very helpful, if you are interested, please reply to this message, it's not binding, just to give me some idea how much I should order.

I'm one of the lucky ones that already have one. And quite happy with the first tests (take a look below). 
Get the correct value of any of these inductors using a DE-5000 is simply impossible.

[Selectech]

Some kind of soft count of who is interested in a board for this project would be very helpful, if you are interested, please reply to this message, it's not binding, just to give me some idea how much I should order.

Cheers

Edit: just to be clear, I give no promises on performance.

Edit again: Don't be shy, a PM will work as well.

Open for 2 weeks, then it will probably take me a week to pack everything and 1-3 weeks shipping

I'd take a couple.

[kridri]

Are you planning to release the software? In a kind of OSHW license?

Unlikely, I don't have good experience releasing open source software.

But in case I lose all interest in maintaining the project I will make the source code available and hope someone else will take over and keep it alive (unlikely to be honest, most people just butcher such projects for their own benefit)

The only reason why I asked it. Is because I don't have an analog discovery but I do have an FPGA board with some adc/dac on. So I would implement an impedance analyzer on the FPGA and you're software is pretty amazing as pc software for showing the results.
But no hard feelings I will do my best and write my own GUI, but he won't be as nice as yours. 

[jaxbird]

The only reason why I asked it. Is because I don't have an analog discovery but I do have an FPGA board with some adc/dac on. So I would implement an impedance analyzer on the FPGA and you're software is pretty amazing as pc software for showing the results.
But no hard feelings I will do my best and write my own GUI, but he won't be as nice as yours. 

Like everything, if you understand, it's really not so complicated, but it does take a lot of work.

I have found this HP/Agilent/Keysight document a very helpful reference: http://cp.literature.agilent.com/litweb/pdf/5950-3000.pdf

Cheers,
Jake

[jaxbird]

Some kind of soft count of who is interested in a board for this project would be very helpful, if you are interested, please reply to this message, it's not binding, just to give me some idea how much I should order.

I'm one of the lucky ones that already have one. And quite happy with the first tests (take a look below). 
Get the correct value of any of these inductors using a DE-5000 is simply impossible.

Thanks, it's great to get some feedback on the performance at these somewhat extremely low inductance values. My testing is primarily based on the set of reference components I got from The Electrician with plots from his ~$10k Hioki analyzer.

[RandallMcRee]

Count me in for a kit.

Already have an Analog Discovery--this seems like an excellent addition to its functionality.

Question about calibration--do we need to come up with precision L,C values in order to do that?
I do have a good bench meter and frequency counter.

Thanks,
Randy

[Carrington]

Thanks, it's great to get some feedback on the performance at these somewhat extremely low inductance values. My testing is primarily based on the set of reference components I got from The Electrician with plots from his ~$10k Hioki analyzer.

Thanks to you, thanks for sharing your work.

[Carrington]

Already have an Analog Discovery--this seems like an excellent addition to its functionality.

No doubt!

Question about calibration--do we need to come up with precision L,C values in order to do that?

Nope, just a simple open/short calibration.

[kridri]

The only reason why I asked it. Is because I don't have an analog discovery but I do have an FPGA board with some adc/dac on. So I would implement an impedance analyzer on the FPGA and you're software is pretty amazing as pc software for showing the results.
But no hard feelings I will do my best and write my own GUI, but he won't be as nice as yours. 

Like everything, if you understand, it's really not so complicated, but it does take a lot of work.

I have found this HP/Agilent/Keysight document a very helpful reference: http://cp.literature.agilent.com/litweb/pdf/5950-3000.pdf

Cheers,
Jake

Hi Jake, I'm aware of the Agilent document and I have read it multiple times during the last years. I also read some service manuals from HP lcr/impedance  analyzers. There is also allot of great information written. I will keep you guys posted of my adventure with the fpga based impedance analyzer.

[casinada]

+1 for Kit 

[pmcouto]

I'm in for a kit.

[rx8pilot]

I have a pick and place machine/line and could possibly assemble a small-med batch if it is helpful.

Sent from my horrible mobile....

[ruairi]

I'm definitely in for one.

Thanks,
Ruairi

[fcb]

+1 for a kit.

[azer]

Count me in for a kit

[Wolfram]

I'm in for a kit as well.

[Fraser]

Just found this thread. I believe I had an original AD impedance analyzer evaluation board that uses this chip.

I shall have to dig it out 

Thanks for a very interesting thread 

Fraser

[Fraser]

TL;DR


Just found my Analogue Devices AD5934 development board. Pictures attached.

Just in case they are of any interest to readers.

I bought it to build a low frequency impedance analyzer but it never happened  I do not see me needing one now either.

If anyone has a need for this brand new unit, make me an offer on it 

Fraser
UK

[Fraser]

AD5934 data sheet.

It has a lower frequency range of 1kHz to 100kHz excitation frequency so I was going to use it on speaker networks and transducers.

Fraser

[timofonic]

What's the advantage over the Component Tester? I'm "confuccio"

[jaxbird]

Just a quick thanks to everyone expressing interest in this project.

I have a number of PM's I haven't responded to yet, sorry, it's been another busy week at work. I'll respond this weekend.

Hope everyone will have a nice weekend.

Cheers
Jake

[jaxbird]

Count me in for a kit.

Already have an Analog Discovery--this seems like an excellent addition to its functionality.

Question about calibration--do we need to come up with precision L,C values in order to do that?
I do have a good bench meter and frequency counter.

Thanks,
Randy

Hi Randy,

Short answer: nope, not needed. Just do the Open and Shorted calibration.

Long answer: Simple Open/Close calibration will work just fine. I have been working on a Open/Close/Reference in case it makes sense for anyone (you'd need some reference resistor with very low inductance/capacitance), but my experience is that the simpel Shorted/Open calibration is most accurate.

Cheers,
Jake

Edit: just saw Carrington already answered the same So I guess double confirmation.

[jaxbird]

What's the advantage over the Component Tester? I'm "confuccio"

Not 100% sure what the Component Tester is, but I assume you are referring to the excellent cheap Atmel based component testers doing resistors, diodes, transistors, capacitors, inductors etc. This is an impedance analyzer, as in it will tell you the impedance of any DUT at any frequency (well within 1Hz to 10MHz+) this means you can plot e.g. the ESR and dissipation factor for a capacitor at any frequency within range whereas a normal LCR meter (or the component tester) is locked into a specific frequency or in large steps usually maxing out at 100kHz. Viewing a plot for a component's parameters vs. frequency is a good way to get a better understanding of how this component will perform in a specific circuit.

Or you could use it to measure the impedance of a loudspeaker, or find the resonance frequency of a piezo, or even test the impedance of your blood.

[jaxbird]

Just found this thread. I believe I had an original AD impedance analyzer evaluation board that uses this chip.

I shall have to dig it out 

Thanks for a very interesting thread 

Fraser

Hello Fraser,

I originally tried to make some kind of analyzer using the Analog Devices impedance IC, but at some point I switched to the full blown Analog Discovery unit as I found I needed better performance than what I could ever achieve with the Analog IC alone, that was kinda where things got serious

Current project is 100% based on the Analog Discovery and it's dual 100Msps adc and dac.

Cheers,
Jake

[Fraser]

Ah I see.

Many thanks for bringing me up to speed 

Fraser

[jaxbird]

+1 for Kit 

I'm in for a kit as well.

Count me in for a kit

+1 for a kit.

I'm in for a kit.

Thanks, I'll add you to the count.

I have a pick and place machine/line and could possibly assemble a small-med batch if it is helpful.
Sent from my horrible mobile....

Thanks for the offer, I think for now the volume is much too low to consider any manufacturing, it will probably remain that way, it's a bit of a niche project, I expect it will be below 50 boards in total. But if things change, like e.g. Dave or Digilent/National Instruments promotes the project and hundreds of boards are needed I'll make sure to ask you if you are still interested.

Cheers,
Jake

[Neganur]

Fun project, I'd take one.

[rx8pilot]

ably remain that way, it's a bit of a niche project, I expect it will be below 50 boards in total. But if things change, like e.g. Dave or Digilent/National Instruments promotes the project and hundreds of boards are needed I'll make sure to ask you if you are still interested.

Just for clarification, I was only intending to help the forum members as a favor. I do not have an assembly business, but rather a business that is capable of assembly. The goal was to push/support this project if needed. If you get a deal with distribution - I would be the worst possible option for assembly. 

[jaxbird]

Fun project, I'd take one.

Thanks, I promise hours of joy and maybe some frustration Impedance measurement is serious business actually it's very educational to spend time measuring everything you can get your hand on and observing all the parameters, like how many uH does this wire have per mm or if I twist those wires how much capacitance etc.

[jaxbird]

ably remain that way, it's a bit of a niche project, I expect it will be below 50 boards in total. But if things change, like e.g. Dave or Digilent/National Instruments promotes the project and hundreds of boards are needed I'll make sure to ask you if you are still interested.

Just for clarification, I was only intending to help the forum members as a favor. I do not have an assembly business, but rather a business that is capable of assembly. The goal was to push/support this project if needed. If you get a deal with distribution - I would be the worst possible option for assembly. 

Understand, I'd be happy to send you a batch for assembly for anyone who'd rather not do the assembly themselves, but it's really not hard to manually assemble, just SO8/16 parts and 0805 components, only some protection diodes are SOT23 and might be a bit challenging for those wielding a 100W solder gun from Frys.

The great thing about a project like this is that anyone who'd find this even remotely interesting and fully understand how to use it, is likely to be an expert at SMD soldering already

[jaxbird]

I'm still working on improving the instrument, it will be an ongoing effort, don't expect perfect performance at all impedances right out of the box, but it will be improving as I get better at understanding calibration and compensation.

Anyway, this week I've done some work on the open/short/reference calibration method, originally I never used this method because my implementation provided worse results than the simple open/close calibration. However, spending some time understanding the complexities involved, I have come up with a better implementation that provides superior results to the simle open/close calibration. Important is that the reference resistor must have as little inductance and capacitance as possible, I have been using 0805 resistors for my experiments with good results, these are tiny, but near idea for this purpose.

This is the general idea for this kind of calibration:



Edit: keep in mind these are all complex values.

[ruairi]

Excellent stuff.  Do you think this will allow the use of longer test leads with acceptable accuracy?  I am considering building the whole thing into a dedicated instrument of sorts with a tablet.

Cheers,
Ruairi

[jaxbird]

Excellent stuff.  Do you think this will allow the use of longer test leads with acceptable accuracy?  I am considering building the whole thing into a dedicated instrument of sorts with a tablet.

Cheers,
Ruairi

Thanks, unfortunately longer test leads will never work well with this solution, only so much signal to work with and added inductance will have a significant effect at higher frequencies/lower impedances.

I have tested configurations like these nice shielded leads: (but results were not as good as short leads)



I have consulted experts as evident in this picture:

[jaxbird]

Quick update I've ordered enough for about 25 kits, that should cover everyone, I'll post here when I have received all the parts, I think Paypal is probably the best payment even they have fairly high fees per transaction, but they also have a lot of buyer protection.

Anyway, make sure you understand what I will supply is a kit, you have to solder it together, if you are not familiar with surface mount components, don't order this kit. Also if you do not own an Analog Discovery unit (1 or 2) you should not buy this kit.

Also you are only paying for the hardware kit, the software, it is free for any non commercial use in the current version.

[Floyo]

I might be a bit late to the party, but if another kit is available I wouldn't mind receiving one.
There are lots of useful things I could use it for. Soldering should not pose any problems

[aardvarkexpress]

Hi jaxbird,

I just found this thread.  It looks like an awesome project!  I'd love to be able to play around with this sort of instrument.

Do you have enough parts for another kit?

Cheers,

Tom

[Rerouter]

Any idea what a ballpark price would be if you bundled it with a discovery, everywhere i am looking they are about $500 vs the $100 mentioned earlier,

[Efig]

Hi jaxbird,

Excellent stuff. I am interested in playing with it. This will add great value to discovery.

Do you have more kits left available?

[ruairi]

Any idea what a ballpark price would be if you bundled it with a discovery, everywhere i am looking they are about $500 vs the $100 mentioned earlier,

I can't see how it would be feasible for Jaxbird to by and resell Analog Discoverys with his kit.  It would require a huge investment, not to mention tax/duty issues for international sale.

Digilent list the Analog Discovery for $259 USD on their own store.  The founder of Digilent did a fascinating interview recently on The Amp Hour and gave a discount code too, check out the interview, it may still be valid.

[steverino]

Any idea what a ballpark price would be if you bundled it with a discovery, everywhere i am looking they are about $500 vs the $100 mentioned earlier,

I can't see how it would be feasible for Jaxbird to by and resell Analog Discoverys with his kit.  It would require a huge investment, not to mention tax/duty issues for international sale.

Digilent list the Analog Discovery for $259 USD on their own store.  The founder of Digilent did a fascinating interview recently on The Amp Hour and gave a discount code too, check out the interview, it may still be valid.

Digilent had a discount code of AMPHOUR recently, but not sure if it's still valid.  It's worth a try if you're purchasing a Discovery.

[Rerouter]

Sorry if I didn't make it clear but yes in such a case I would pay him before he ordered the thing, I thought he was still a student so the discount would apply for him.

[jaxbird]

Sorry if I didn't make it clear but yes in such a case I would pay him before he ordered the thing, I thought he was still a student so the discount would apply for him.

Nah, sorry, I can't offer any specials on the Analog Discovery units. I don't think I'd ever qualify for the student discount, by age 17 I was laying down natural gas pipelines instead of studying, university never really was an option for me, but I've caught up later in life through reading and general interest.

I believe I got mine for around $200 from Microchip direct, but I don't think that deal is available anymore, another option is, since it's an instrument required for many electronic engineering courses, to try find one on ebay etc. surely there must be a lot of students who needs money more than their Analog Discovery. My solution is designed to work with both version 1 and 2.

[Wolfram]

I've received my Analog Discovery, and I'm very eager to construct the impedance analyzer and start playing with it. Has the schematic for the plug-in board been published anywhere? The video on the plug-in board should contain enough information to build it, but constructing it from a known correct schematic is less prone to errors.

[jaxbird]

I've received my Analog Discovery, and I'm very eager to construct the impedance analyzer and start playing with it. Has the schematic for the plug-in board been published anywhere? The video on the plug-in board should contain enough information to build it, but constructing it from a known correct schematic is less prone to errors.

I recommend you wait for the kit, likely it will be cheaper than what you'd pay gathering the parts by yourself, even if it's slightly more, consider that you'd be contributing to the project.

I've already ordered parts for the first batch, first come first serve once I have everything ready.

[Wolfram]

Excellent, looking forward to it. I would of course also still buy your kit in any case, that's the least I can do.

On a more technical note, if more supply current was available for the plug in board, would it be possible to improve the performance when using longer test leads? If so, would it be difficult to add an option in the software for arbitrary input and output scaling, to allow the seamless use of output buffers and more stages of input amplification?

[ruairi]

Hi Jaxbird,

Any update on this project?  I am very interested in building one.

Thanks!
Ruairi

[3roomlab]

this project is so awesome
isnt it possible then to measure the Er of circuit board?

[ruairi]

Agreed, I'm hoping Jax still has this on the back burner.

[steverino]

Jaxbird's profile indicates he hasn't been active on the forum since July.

[ruairi]

No updates to his website either.  Life gets in the way.  I hope he gets back to it, it seems like a fun project.

[ruairi]

Hi all,

Jaxbird posted a new Impedance Analyzer video on Youtube!

Fingers crossed that the project will be released soon.



Cheers,
Ruairi


 

[groggen]

Hi,
I just found this thread.
This project looks amazing and I'm very excited.
I would definitely buy one kit if/when it becomes available.

Cheers,
groggen
 

[ruairi]

Unfortunately this has been untended for a while.  I tried PM Jax to no avail.  I hope he gets back to it!

[bitseeker]

I hope so too.

Jake, if you're still out there, I just got an AD2 and look forward to trying one of the kits. Meanwhile, I'll check out your audio analyzer suite. Cool stuff.

[casinada]

He has a website:
http://www.thestuffmade.com/
May be he'll reply overthere 
He seems to be replying to people, at least for the audio analyzer project. 

Here is something similar??
http://designideas.cocolog-nifty.com/blog/2016/12/high-impedance.html

[modmix]

At diyAudio, Frex reports here

Just some pictures of my small test fixture to use the Analog Discovery as Impedance Analyzer with the cool software of TheStuffMade.
It's only connectors and one resistor..Difficult to do simpler than that !

He's using Analog Discover Legacy (AD).

While my AD2 works nicely with the WaveForms network analyzer, AD2 isn't recognised using Impedance Analyzer with the same elements attached (100R/10R).
Audio Analyser sees the AD2 (Not so sure whether it works as it is supposed to do - no time to check, yet).

Anyone being able to use Impedance Analyzer with AD2 instead of AD?

look forward to trying one of the kits.

Me too ,-)
Don't know how to apply for a kit or a board, though.

Ulli

[alocam]

I hope so too.

Jake, if you're still out there, I just got an AD2 and look forward to trying one of the kits. Meanwhile, I'll check out your audio analyzer suite. Cool stuff.

Hello, I would also be very interested in getting the PCB and components as a kit. However, I have just found out about this thread, and it looks like the project is having no activity since beginning of 2017. I look forward to any news, as it looks quite an interesting development!.

[R_G_B_]

I'm not sure if there are people who are aware that digilent have released an update to their analogue discovery waveforms 215 that now includes an impedance analysis.

[steverino]

I'm not sure if there are people who are aware that digilent have released an update to their analogue discovery waveforms 215 that now includes an impedance analysis.

Here's a digilent blog entry regarding the updated waveform software, including the impedance stuff.

https://blog.digilentinc.com/software-update-waveforms-3-6-8/

[bitseeker]

Yep, they added that and the ability to use scripting with the spectrum analyzer. It's a nifty little device.

[rx8pilot]

The AD2 is quite a bang for the buck....I want one now. The impedance functionality is an excellent upgrade since the project that started this thread seems to be abandoned.

[jaxbird]

Just to nekro my old thread and add an ending.

The whole project with all details can be downloaded here:

https://www.dropbox.com/sh/ylf9i87dac65q2k/AABA_Rqgm1r5Ql0MsNyM7ajta?dl=0

And some basic details here:

[jaxbird]

I understand a lot of people say: hey, the Waveforms software does this now?, why would I even need this anymore?

True, the official software did add impedance measurements after I showed my work, however they didn't implement any calibration, this makes their software absolutely useless. Not kidding, it cannot be relied on without calibration.

[voltsandjolts]

Welcome back jaxbird, I thought we had lost you! 

[jaxbird]

Thanks 

You did lose me for a while, I had to do some regrouping and make some really hard choices on what really matters in life.

But yes, I'm both alive and back.. thanks..

[bitseeker]

I second that. Great to have you back!

[peteb2]

Of interest, has anyone who is following jaxbird's fantastic work on measuring impedance, have you managed to have his greatly appreciated S/W for download from via his Dropbox link run with an AD2? I'm finding with the program running under Win7 OS displaying the GUi & plugging in the USB to my AD2 the audible announcement occurs (the driver is running) but then instantly there's a flash of status text for USB voltages etc on the GUi and the message "Device not available" returns. The AD2 onboard LED continues to flash.

If i run the AudioAnalyzer however things are absolutely fine!

I guess in a round about way, what i'm asking is has anyone managed to make 1.3.0.43 work with their AD2?

[peteb2]

...OK i placed comments on issues with the AD2 not working with this great S/W on the YouTube page relating to this current problem on this really neat project. It seems however my comments may have been flagged by their algorithm as SPAM or abusive.

TheStuffMade has been replying as much as possible but just in case i'll repeat my YT comments explaining my findings:

The Impedance Analyzer software boots up 100% OK and without an AD2 connected via USB i can interact with it as example: changing options in "settings" or playing with pre-assignment of plot lines colors for variables.

The instant i plug in my AD2 (using a Diligent supplied USB cable) the standard audible sounds, a green LED begins flashing on its PCB & less than a second later the software splashes a change for an instant in the lower right from "device not available" to the text for USB parameters all reading zero and instantly back to "device not available". The lower LH "Status: Analog Discovery not available" message never changes.
 
It would indicate an attempted connection is being rejected. If i unplug/replug, nothing occurs not even the momentary flash of the USB parameters as before. It's as if the software now rejects the AD2 & has gone deaf.
 
In comparison the Audio Analyser software will not boot up unless the AD2 is plugged in and then it works perfectly as well as does Digilent's Waveforms2015

I've played around with FTDI's FT EEPROM Programming S/W just to have a look at the architecture of the FT232HQ device used in the AD2 for USB. The driver option is set for D2XX Direct as opposed the option Virtual Com Port and Driver Version is listed as 2.12.28.0 (i mention this as a FWIW).
 
I must say thanks for replying to my question. I truly appreciate all the work you have put into this project. This subject of measuring impedance with at least some accuracy has always been a challenge for me in my role as a technician and as a electronics hobbyist. I really hope we're able to have the AD2 work with your software...

​ TheStuffMade  Thanks! I've also reached out on Dave Jone's forum (https://www.eevblog.com/forum/profile/?area=showposts;u=116817) . I have only the newer version AD2 to hand, (i'd need to buy a AD (1st model version) but they appear to becoming a rare item now not to mention prices are 'interesting' but will keep my eyes pealed for a bargain). I've tried a restart of my PC & even another completely different machine both running the most up to date release Win 7 & always left your designed active plugin board out of the AD2. I've even used a Freeware "Simple USB Logger" to see if there's a pattern. I can see multiple bulk URB data transfers have originated from the PC side (the Impedance S/W running on the PC & AD2 connected) but the AD2 pretty much acts dumb i.e. nothing much being returned by it other than the odd byte). I hope there's a solution. This is such an interesting project i'd hate for it to fall by the wayside. Thanks again. 

[modmix]

Just to nekro my old thread and add an ending.
The whole project with all details can be downloaded

What a nice surprise, jaxbird!
A short while after I got my AD1 to do some impedance measurements with your fine software this software was gone.
Glad to have it back - THANKS.

The AD1 in your material has a cable going into it - what is it for? where is it connected internally?
TIA
Ulli

[bitseeker]

The AD1 in your material has a cable going into it - what is it for? where is it connected internally?

See:

[modmix]

I built a plugin board v2.2 (thanks for the Gerber files !).
Works quite fine 
One question:

A 500R shunt gives quite more "noise" at low frquencies - see lower green curve
The upper green curve let's me wonder why 500R is selected...
Is there a way to change that to select the lower value shunt a bit earlier?
Thanks
Ulli

[xaled]

Hi modmix,

do you have a BOM for the plugin board or parts supplier project to share?

Thanks.

[modmix]

Most of the R and C I already had (size 0805).
Parts I got from DigiKey

• AD8039ARZ-ND
• BAV99-E3-08CT-ND

I got from Reichelt

• 74HC 4053D NXP (SO-16, not 16SOIC)
• TAJ 3516 10/16 SMD-Tantal Kondensator, 10 μF, 16V
• BL 2X17G 2,54 2x17pol. Buchsenleiste, gerade - didn't find a 2x15

hth
Ulli

[xaled]

Thanks Ulli!

[aandrew]

Thank you for this, Jaxbird, it's very much appreciated.

On the software side: is there anything interesting/unusual/tricky you're doing to drive the AD2? Is it just basic setup sine sweep and measure or do you have more magic in there, particularly around cal?

I ask because I'd like to get this working under Linux and/or OSX without running a virtual machine or WINE. I think I can drive the AD2 reasonably well by monitoring the USB interaction on Windows and replicating things with libusb.

[Elmar]

Hi modmix,

do you have a BOM for the plugin board or parts supplier project to share?

Thanks.

this is what I got so far:

qty   MPN                           designator
4   ?                           ?
2   ?                           H1,2
1   PPPC152LFBN-RC   ?
2   RMCF0603FT68R0   R15,16
10 RMCF0603FT1K00   R2,R3,R4,R5,R8,R9,R10,R11,R19,R20
4   RMCF0603FT3K00   R1,R6,R7,R12
2   RMCF0603FT10K0   R17,R18
2   ?                           C15,C16
4   CL10B103KB8NNNC   C1,C3,C9,C11
9   CL10B104KA8NNNC   C2,C5,C6,C10,C12,C19,C20,C21,C22
2   TAJA106K016RNJ      C4,C7
8   BAV99-E3-08            D1,D2,D3,D4,D5,D9,D10,D11
3   74HC4053D,653       IC3,IC5,IC6
2   AD8039ARZ              IC1,IC2

-I can't figure out the value of C15, C16
-not sure what coaxes to take
-also not sure about the function of the screws

I was wondering if the coaxes would make any sense because of the added inductance

[croma641]

Elmar,

go to this Youtube Video/page:




(and watch the old  related videos/pages)

best regards

Andrea

[justanothername]

I want to order a PCB for the impedance analyzer but the board outline seems to be missing in the gerber files.
Where did you all order your PCBs?

Edit: I've generated an outline from the silk layer. Attached.

[justanothername]

So, the Parts and the PCBs came. But I can't find a placement plan.
This is a puzzle for very dedicated people I guess....

Edit: I made an assembly plan. Parts on bottom are just bav99s and one 10µF

[justanothername]

OK i built this but I'm inconclusive.

You can see my build in the picture with accessories:
A: Adapter for small parts
B: short for calibrating
C: DUT (Wuerth 885012106018)
D: Kelvin clips adapter (not used yet)

I calibrated the software with open/short calibration before measurements.
Maybe it is the socket header I used, but the response has a weird spike at 1MHz. On the other side the point of lowest impedance matches better to wuerths data than the measurments with WaveForms.
Anyway, the measurements from the Digilent adapter board have no spike at 1MHz, they look more or less like the data as given from the manufacturer, resonance point is off though.

Could as well be that I did something wrong by trying to figure out the assembly plan...

[faraday]

does it work with latest waveforms?

[croma641]

 It works perfectly !