LCR/ESR Impedance Meter UA320

[unitedatoms]

Jan 2 2020: 5 years later. Starting new LCR meter project from scratch. UA320 (meaning 3 digit accurate, year suffux 20).
I will keep this same post for reporting on progress.

General idea: The 10MHz is more interesting challenge than original 100-200KHz meter. There is no much need in new low frequency designs because there is a lot of old designs. I hope that this challenge (to reach 10MHz) will keep me interested. Part of the reason, why I did not finish first model is that there is too many existing designs equally interesting and also winning the cost game (cost game is not very interesting part for me).

Approach to costs: Be reasonable, but somewhat ignore most of the cost related considerations at the phase of proving the concept. Lets define proof to be an accuracy in numbers alone (not in combination with cost).

Design choices: So far the 10MHz range dictates the use of RF method, (so named ?) balanced bridge with null detector.

What killed first project? Lack of knowledge, experience and attention to power supplies. Unnecessary feature to have a standalone display and enclosure before the analog modules were complete. Dead end with use of supercapacitors. It looks that supercapacitors have limited life span in terms of amount of charge / discharge cycles. Dealing with supercaps sucked most of available time for builds. Lack of experience with firmwares in general. Use of Arduino.

What should not be in new design: Skip Arduino story completely, do bare metal firmware with perhaps PSoC5 or similar. Skip display so far. Use either USB/Serial or Ethernet with high level software on Windows PC for concept proving. May be eventually add more of screen and knobs to make instrument a standalone.

What to focus on: High resolution of phase detection. Perhaps 20-22 effective bits at 10 MHz.

What is done so far: The phase detector works on paper (LTSpice) with accuracy of 0.007 degrees. And 0.000 degrees repeatability (because it is LTSpice).

What is next step: Review the phase detector alone. Build it and see how well it runs physically.

Description of project: The voltage signal is pure sinewave applied to HI of component under test. Another current controlled sinewave is applied LO of component to maintain zero. Three resulting quantities (voltage, current and residual zero) are detected by phase detector to determine I and Q quantities. The expected null residual value used to compute new applied values. The corrected new applied values appear at component. Cycle repeats. Some arithmetical aggregation, either sliding average or sliding median is used to recompute impedance quantities.

Measurement cycle: The duration of measurement cycle is defined by dual slope ADC (for example MAX132) to suppress mains AC interference at 60Hz. Expected rate of acquisition is 16 cycles per second. At start of each cycle, the V and I source signals are set to known relative phases and magnitudes. During cycle the V source and V2[Quadrature] source is fed to phase detector. There can be 1 or 2 phase detectors (to speed up measurements, the null detector is just another phase detector with less strict accuracy).

How many sources: So far it looks that 4 identical DDS will do. 1 for V signal. 2 for V2[Q] signal. 3 for I signal. 4 is reference DDS for transfer of high res phase/magnitude value from reference to 3 measured signals at the uncalibrated plane (input of phase detector(s)). Why transfer reference DDS is needed ? It will allow to never need a characterization/calibration of phase detector. Detector has limit of accuracy worse than desired resolution/accuracy, however has high repeatability, it is precise. To rely on good precision and bad accuracy, the arbitrary reference is needed. So instead of traditional complicated analog self tunable loops, which keep null detector at null, the loop is just a software routine based on matching reference to any of the signals.

How high the phase resolution is available: It can be up to resolution of DDS phase accumulator even if direct control of phase in DDS is coarse (16 bit), the 32 bit is achievable through so named sweep capability of some of DDS. The DDS is set instead of single tone to specific frequency plan with exact amount of clocks per step, delta frequency per step, amount of steps etc. to ramp through the plan after triggering and dwell at end frequency as normal single tone with phase continuity maintained. The frequencies in ramp plan can be set with very high resolution. So it gives full digital resolution of phase. The magnitude can also be resolved precisely using DDS analog pinout for current setting with external DC DAC.

===============^{text above is recent}_{text below is historical}================
Nov 18 2018: old 2015 schematics and code attached for history.

I am excited to share with All my Impedance Meter project.
It is named UA315 Open Source Hardware LCR/ESR Impedance Meter. It can be either DIY benchtop kit or completely built.

The thread is being updated in real time as the project progresses

Blog posts, design articles, detailed info, (WARNING: picture of cute dog )

Voting thread
Please add your vote to help me to understand what pricing can be!

Device description
It measures Immitance parameters: Impedance ( Resistance and Reactance ), Admittance (Conductance and Susceptance ). Capacitance, Inductance, Equivalent Series Resistance (ESR), Q value, D factor, Phase angle / Loss angle and Loss Angle Tangent in range of frequencies 10Hz to 100KHz (with possibility to reach 200KHz, but it is yet to be seen).



The project is about 50% done and I will keep posting schematics and code as I go.
The digital board is Arduino(TM) alike:

The analog-to-digital board is in progress. Waiting for first 5 analog-to-digital board prototypes to arrive so I could tune the custom made transformer details.

The Analog to digital board schematics:


There are 2 boards.
Digital board is 75+ parts.
Analog board is ~220 parts.
Also included parts:
- probe
- internal flat cable 34pin
- power connector
- enclosure with fasteners


The brief description of how analog-to-digital board works:
- The AD5933 is single bin Fast Fourier Transformer with Direct Digital Synthesizer used as magnitude and phase sensor.
- The excitation signal is buffered with op amp wired as phase-error reduction amplifier (p.314 of Art of Electronics 3rd ed.) also summing DC offset to provide DC bias to device under test
- Kelvin probe and device under test have current limited by output impedance of a buffer.
- The sense V of D.U.T. and reference resistor is scaled by instrumentation amplifiers
- V and I signals from instrumentation amplifiers are multiplexed and brought to AD5933 input through V-to-I resistor, since AD5933 accepts only current input.
- The data about amplitude and phase both for V and I is processed by Arduino-alike board and displayed to user.

Brief extra details, why the parts count is high:
- Major effort is done to reduce the noise of power supply
- All clocking of every chip is fed from single master clock generator with muxed dividers
- The power supply is supercapacitors banks (graphene based!) which supply power during controlled shutting off of the step-up DC-DC converter
- The power during measurement phase (from few milliseconds up to 1 sec at 10Hz range) is regulated with linear regulators, there is no switching, transitions or power source disturbing software activity is happening during measurement phase
- The external power is isolated to >100-120 Db during measurement phase
- The relays are latchable and are not powered during measurement, so the contacts resistance is not modulated by coil current fluctuations
- The method, algorithm and implementation is done with care to avoid reliance on absolute voltage
values. The design is completely ratiometric, so the major contribution to accuracy is values of reference resistors.
- The DC errors like offsets instabilities, thermals are not affecting the AD5933 converter since it has narrow-band sensitive to only single frequency bin
- Having no need to avoid DC errors, the signal path has no DC blocking capacitors, it adds stability to output buffer and reduces risk of parasitic feedbacks
- Having separate I and V channels, which are measured at separate time, the geometric calculations made possible to move from analog-to-digital part and perform in software, instead of AD5933 original designs, where  measurement relies on assumption that output signal has certain phase.
- Having 2 pairs (I and V) of phase/amplitude data instead of 1 assumed pair and 1 measured, the phase accuracy measurement may exceed the one claimed in AD5933 datasheet
- Having this 2 pairs of data, the design is immune to accuracy of output buffer and does not require the buffer to become a critical part of signal path, like Howland current source, transformers, calibration or highly precise parts etc.
- Replacing I-to-V amplifier with virtual ground with simple reference resistor helps to avoid reliance on phase accuracy of I-to-V converter. Resistor is assumed to have no phase dependency of I-to-V conversion (at least in range of frequencies of interest)
- In low and middle ranges of D.U.T. impedances the probe capacitance is factored in calculations and shows the predicted behaviour.
- For high impedance and low and midrange frequencies the probe capacitance factor is also trivial.
- The only difficult range is high impedances at high frequencies. To resolve this difficulty the prototype #4 (this is 4th iteration since April 2015) has instrumentation amplifiers added. It were simple CMOS followers in prototype #3.

The code is about 25-30% ready:
It is about few hundred lines of Arduino Sketch language. It uses UTFT library. The algorithm is infinite loop repeating the measurement and display. The calculations are all straightforward without magical constants or reference values without physical unit. It is as deterministic as possible, the factored values for parasitics like probe capacitance are explicit and are introduced in understandable, easy to maintain manner.

I will publish my very raw code as soon as will complete the analog board ordered from SeeedStudio and complete the transformer design.

Update Sep 24 2015:

The digital subassemblies progress: 34 out of 50 subassemblies for 50 kits are done
The analog boards are on the way from Elecrow.

[DmitryL]

Hmm... woud you share the device specs first ?
You know, these boring numbers, like: "xxx mOhm- yyy Mohm, zzz pF- www mF, qqq nH- ppp H... with ABC% tolerancy..., test frequencies:bla-blah..."

[unitedatoms]

Yes, absolutely. The specs are preliminary, some are underspecced, some other are too ambitious. I posted it in Crowdfunding section of the forum to get the price poll. Depending on what people think about costs, I will choose most appropriate final specs. For example 0.1% 5ppm resistor is one cost, but 0.01% 2ppm is completely different price range, and so on from component to component.

Release Date:
October 2015
 
Brief Product Description:
UA315 is a reprogrammable benchtop impedance meter for frequencies up to 100KHz. It is Open Hardware and Open Source device designed for electronics enthusiasts, who make their own electronic lab devices.
 
Technical Specs (preliminary design targets):
Model Name: UA315 (revision Beta Aug 2015)
Accuracy: 0.5% or better
Resolution: 3 or more digits
Impedance range: 0.001 ohm .. 1000 MOhm
Frequency range: 10Hz .. 200KHz
Excitation AC voltage: 1V or less
Excitation DC voltage: 2.5V
Protection from input voltage: Up to 500V
Power supply port: 5V / 1A round 5.1mm DC connector with cable for USB style power supplies or any other DC power supply. (Power supply is not included with device, when sold as assembled product or kit)
Dimensions: Approx 4" x 13" x 2"

The L and C specs are yet to be decided: But very preliminary it should be.

0.1pF .. 0.1F
10nH .. 1000H

The lowest values are most difficult.

[DmitryL]

Looks nice.... tough dubious..a bit. :-/
Have you tried to apply to Agilent/whatever as a chief technical architect, just in case ?

[unitedatoms]

Looks nice.... tough dubious..a bit. :-/
Have you tried to apply to Agilent/whatever as a chief technical architect, just in case ?

Ha ha. No. This project is much smaller caliber. May be at the level of "Jr. Design Engineer". If it was the Chief Architect level, I'd use custom hybrid chips and gold plated parts on sapphire ceramic substrate.

[Vgkid]

I know you are this far, but have you looked into the GenRad Digibridge series for inspiration. A million times simpler.

[unitedatoms]

I know you are this far, but have you looked into the GenRad Digibridge series for inspiration. A million times simpler.

They are all simple. I chose AD5933 as 7 chips in one: DDS-DAC, Voltage reference, 2xOpAmps with PGA, ADC, DSP. It makes it super simple. Also most of older devices designs are hard to name completely trivial to reproduce. Their schematics is sure sometimes very simple, but old analog designs are full of decisions, limited but cost/labor situations of the time. Many things changed. Say the steel enclosure with high quality oven may ruin the budget today, but was cheap 30-40 years ago. On the other hand I can show 10 lines of text on single screen.

I think I watched the tear down video. It may have a nostalgia value for some. My inspiration for this device form factor is more from mid90s era of space equipment inside piloted space crafts (USSR or Nasa). I understand that some may adore segmented LEDs which are easy to read. I like larger screen with more text for debugging.

And thank you for feedback. I interpret it as issue with lack of user controlled knobs, switches, buttons. My current plan is to fully automate the ranging functions. But if I find the source for nice generic panel mount buttons, knobs, I will add it to give to user more feel of control. And most of all I hope that users will mod, brand, paint the enclosure the way they like with labels and buttons of all sorts.

Say, someone decides to make a speaker impedance meter with charting. Or audio transformer 4-wire transfer function meter. Or blood impedance meter, oxygen in water, or body fat percentage meter, or use device as chemical frontend of PC based lab, etc. The metal panel mount USB connector which can cover the cutout hole is absolutely impossible to find for sane price. So, for this iteration I gave up on adding USB and removed it from schematics and postponed the idea of making it completely extendable for first version of kit.

[Vgkid]

Didn't mean to shoot down your idea at all. With the .10% resistors if you add in a user available way of calibrating the meter, it will be possible for most with higher accuracy bench meters to calibrate the lcr meter to much higher realms of accuracy. For examplebeing able to calibrate it by range resistors to .05% accuracy using a precision metal film/foil resistor in a low inductance/capicatance shielded fixture. But for the low frequency measurenment(best to cal at a lowish frequency eg. 1Khz) to use the values or the resistor at DC.
for further reading read up at ietlabs.
 http://www.ietlabs.com/genrad_history/historic_appnotes/
genrad 1689 manual
http://bama.edebris.com/download/gr/1689/1689%20Precision%20Digibridge.PDF

[unitedatoms]

Yes, the calibration will be available to users. Thank you for links. The users will have all 100% of Source Code with easy to change values of reference resistors, probe capacitance in Arduino IDE. This is the DIY kit with blank analog-to-digital board and all through hole parts to solder.

For calibrating calculations, when D.U.T is the superior accuracy resistor, say you have 1K 0.1% on board and reading the value at 1KHz of external metal foil 0.01% 1KOhm reference and see 997.075 Ohm on screen. I will provide the Open Source Open Formula procedure how to calculate the more true reference value of on-board resistor to change it in embedded software. WolframAlpha link for example.

And users will always can just change resistors themself and any other part.

The quick comparison of GenRad 1689 design vs reference AD5933 design vs my UA315 design
- I-channel: transconductance vs transconductance vs resistor
- I-channel phase factoring: unknown (software tables may be) vs none (datasheet just has 0.5 degree of angle value to characterise transconductance amp phase error, which perhaps has very good high repeatability) vs cancellation of phase error by using same transconductance amplifier twice (muxing V-vs-I-channel).
- ADC: dual slope DC vs 1MHz ADC vs same 1MHz ADC used twice
- Synchronous detection: Multiplying DAC from Sin ROM table (that is very good advantage - hard to beat, but brings DC into picture, which is costly to implement) vs DSP vs same DSP twice.
- Bandwidth narrowness: Defined by low pass filtering around dual slope vs FFT Hamming window for any frequency vs FFT Hamming windowing for selected frequencies with only integer number of test signal periods to avoid reciprocal mixing on top of Hamming windowing
- Unavoidable reciporocal mixings: unknown vs Spurs of DDS/ADC vs  Spurs of DDS/ADC
- Avoidable reciporocal mixings: unknown, perhaps a lot of care taken to avoid interferences vs no recommendation beyond foorptint example vs switching supercapacitor power supply with all possible shutdown of parts of schematics during measurement phase
- Critical absolute references: R+timebase, R+timebase, R+timebase
- Critical non-absolute but ratiometric references in detecting side: this is no-trivial to pick, since I and V channels are not identical vs Ladder or capacitor inside 1MHz ADC (on the same silicon die as DDS/DAC, which is great if thermal behaviours are taken into account, but there is no claims from datasheet) vs matching-nesness of a pair of instrumentation amps

[MagicSmoker]

...
Brief extra details, why the parts count is high:
- Major effort is done to reduce the noise of power supply
- All clocking of every chip is fed from single master clock generator with muxed dividers
- The power supply is supercapacitors banks (graphene based!) which supply power during controlled shutting off of the step-up DC-DC converter
- The power during measurement phase (from few milliseconds up to 1 sec at 10Hz range) is regulated with linear regulators, there is no switching, transitions or power source disturbing software activity is happening during measurement phase
- The external power is isolated to >100-120 Db during measurement phase
- The relays are latchable and are not powered during measurement, so the contacts resistance is not modulated by coil current fluctuations
...

Turning off the power supply and running from capacitors or batteries to achieve a low noise floor is probably rendered moot by Johnson/thermal noise elsewhere. I would first try a decent toroidal mains transformer with two stages of linear regulators - perhaps in a separate metal sub-enclosure - before going crazy with the graphene supercaps and worrying about contact resistance modulation from relay coil current fluctuations (that has to be a 3rd order effect, anyway!?!).

But it otherwise sounds like you've given the design a lot of thought whereas I am glibly making a quick comment on a forum...

EDIT: this thing is quite close to being a proper (low frequency) vector network analyzer; it just needs to sweep the frequency and display the amplitude and phase response on that lovely graphical LCD panel. That would put it well above the capability of the typical handheld LCR meter (like my B&K 879B).

[unitedatoms]

Yes. The goal is to reach pure Johnson noise floors and thermal/aging drifts and be done with it.

Thank you for pointing out the possible powering from AC. I did it for my personal prototypes, but I decided to remove AC power source from kit for many reasons. There is a reserved 100x100x40mm unoccupied volume inside the box for those who decide to add their own modules, power supplies included. And AC source has great advantage, when zero cross detection is used to sync measurement to 50/60Hz mains for higher repeatability.

I also decided to achieve basic functions first, meter functions before VNA functions, since this is just software, user interface, PC interoperability and so on. These activities can be so involving, that will distract from basic goals. I did not quantify all possible sources of inaccuracies scientifically, but from experimenting and observing similar designs, I sorted the sources and countermeasures in some order by priority (onion shells on the way to accuracy):

1. Multiple clocking generators behave not completely asynchronously - they are plesiosynchronous. I observed the values abruptly bobbing even in carefully made battery setup. The chart showed 2 clusters of values, reminding the strange attractor. So I decidied to limit the clocking sources and make it a single one. The timebase accuracy is easiest to achieve.

2. Linear power regulators in contrast to switching supplies. That is easy to design, so I went with this path.

3. Mechanical relays vs analog switches. This was obvious decision, even if cost went higher. Also the puzzling about analog switches parasitics can make the design inattractive for beginners. However, I have a design in mind, when those can be negated by 140Db CMRR ability of instrumental amps.

4. Instrumentation amps were called to improve the phase resolution at very low signal levels.

Edit: Eh, the order of priorities is not very clearly explained. ..  I simply attacked every noise and uncertainty source I saw, and then resolved before proving the need scientifically. My goal is highest possible repeatability (matching of consequent measurements in short or medium period of time) and basic accuracy is just second to top priority. That is why I dedicate so much to noise cancellation, when, instead, I could focus on drifts, embedded references, self calibration abilities etc. I found that noise is most distracting thing, so to continue improvements I had to spot and resolve the noises first.

TLDR: Repeatability > Accuracy

[TheAmmoniacal]

What about phase angle and dissipation factor?

[unitedatoms]

What about phase angle and dissipation factor?

Yes.
It measures Immitance parameters: Impedance ( Resistance and Reactance ), Admittance (Conductance and Susceptance ). Capacitance, Inductance, Equivalent Series Resistance (ESR), Q factor, D value, Loss angle and Loss Angle Tangent in range of frequencies 10Hz to 100KHz (with possibility to reach 200KHz, but it is yet to be seen).

Phase angle and dissipation factor are always part of measurement. I had to spell the "Q value, D factor" instead of "Q factor, D value". thank you for asking.

[timofonic]

I would like a SMT design. I want to improve my soldering skills and the price reduction could make it a lot more interesting too.

[unitedatoms]

I would like a SMT design. I want to improve my soldering skills and the price reduction could make it a lot more interesting too.

Great to hear this! I did not think that anyone will choose SMT over through hole. The main board has 3 SMD 16 pin chips, so I can leave the soldering to the user completely and stop worrying about solder type mismatch. This can be optional for user to choose with connection to price discount.

I use PbFree Tin Bithmusth Silver solder over (pure ?) tin HASL boards. The solder is (made in Massachusetts) very low temperature one and melts at 138C! It can cause trouble when mixed with lead, that melting temperature can drop too low. It will be great if the choice about environment impact went to users. (Though I have to consider the need to increase the copper from 1 Oz to 2 Oz thickness - so multiple resoldering will not cause too much foil dismembering, and also I have to understand if lead free HASL is acceptable by users, many may prefer lead based or gold plating or something else).

The digital board is TQFP100/0.5mm + TQFP128/0.4mm + 40Pin/0.5mm pitch flat connector. Those are impossible to solder without heat gun or underheat or oven. Plus most people will need a microscope. So I decided to solder the digital board 100% myself.

Another labor/cost reduction idea is to let users drill / cut the enclosure. It is 0.8mm painted steel. Many of builders like to use dremel or have an access to machine shops, it will be great cut for me on labor and good discount for users, if they choose to do so.

[timofonic]

I would like it the hard way, because of didactic and economical reasons. I'm studying at a vocational school, we need to be proficient at soldering and desoldering, even packages using big pins with heat guns and ovens.

Does the project features a good continuity tester feature too?

[unitedatoms]

Continuity tester! What a great use, yes. The instrumentation PGAs are perfect for microohmmetry. I am yet to see the resolution at lower range, but I hope for good resolution / repeatability (ignoring the accuracy of 1% 0.2Ohm resistor - which is an impossible claim of any incomplete datasheet, if you think of typical 50ppm 1% resistor, look only after repeatability).

I saw few mentions of using microohmmeters to find the defects (shorts, tin whiskers ?) location by sliding the probes over copper traces. This would be interesting to see. I am so excited, that you mentioned it, because for most ranges there is no much need to have a high CMRR expensive PGA parts. And I have been mostly worrying about high range with shielding, grounding and parasitics, and forgot about microohms. But for micro ohms it is just a software, and perhaps the little sound speaker - beeper.

Another frontier is supercapacitors and batteries. Those also are in microohm range.

[timofonic]

There were some talks about a specific continuity tester here, maybe it could give you some inspiration in terms of features and such...

https://www.eevblog.com/forum/index.php?topic=15435.0
Continuity tester with glitch capture

[zapta]

OP,  you mention a few optimization that reduce the noise at increased cost (latching relays, powering from supercaps, etc). Are you convinced that they are required and justify the cost?  Have you tried without them?  (premature optimization is a common design sin).

Also, if you intend to power it from a 5V USB charger, why not using a standard USB connector?

Kudos for making it open source.

[unitedatoms]

OP,  you mention a few optimization that reduce the noise at increased cost (latching relays, powering from supercaps, etc). Are you convinced that they are required and justify the cost?  Have you tried without them?  (premature optimization is a common design sin).

Also, if you intend to power it from a 5V USB charger, why not using a standard USB connector?

Kudos for making it open source.

Thank you for asking. Latching relays is byproduct of choosing capacitors based power supply. I chose latching relays (to reduce the consumed power) after choosing capacitors based supply. I tried several ways to power different prototypes: USB power from HUB + LDO as in Analog Devices reference circuit, batteries, AC linear power and AC-CD converter. I did not try to run from supercaps yet, just made a board layout and waiting for boards from Seeedstudio.

I can tell from experience that switching supply is the worst noise wise. The battery based supply gave the best results. I am convinced that supercaps are relatively cheap and reasonable way to build noiseless power supplies, they are not the major cost b.t.w. The major cost comes from precision resistors, relays, instrumentation amplifiers and ADI chip.

The LCR measurement is basically very sensitive I-Q demodulator, similar to synchronous detectors in radio. When one builds sensitive radio receiver, containing in-band transmitter on the same board, which run simultaneously and powered from same power supply, the PSRR becomes a major noise factor. If PSRR is, say 80Db, the microvolt error is caused by 10mV noise on power rail. And if the measured signal is 1mV (in some ranges even less), the error of 1uV is 0.1%.


USB connector: I can not find affordable panel mount USB connector. The requirement is, it has to cover the hole edges completely, and it happened so, that in whole industry all USB connectors are designed for injection mold enclosures. Not for sheet metal panels.  Or may be I was not looking for it well enough.

[timofonic]

Any progress on your project?

Are you considering Sigrok compatibility?

You can put it in GitHub, so we can follow it easily.

Thanks for your project and good luck with it!

[unitedatoms]

Any progress on your project?

Are you considering Sigrok compatibility?

You can put it in GitHub, so we can follow it easily.

Thanks for your project and good luck with it!

Project is still in progress. I have made 42 digital boards out of 50. And now making the first analog board. For last 2 days stuck on 555 timer based power supply part. But resolved it OK. The analog boards will have to be changed. Most of the change is around 555 reset pins.

The Sigrok looks interesting. I will not add it to the first version (batch of 50 kits). Too late to add anything. But will consider it for next batch. The major problem with connectivity to anything is nice looking connector for USB Type B. I can not find any panel mount connector for normal price.

Update: Aug 28th. Received few more parts today and was able to run 2 boards together with half of analog board populated:

[timofonic]

Any more news? Your project looks really interesting!

[free_electron]

uses supercomplex A/D  device , processor and lcd diplay with own graphics processor..
power supply ? 555 driven ... 

[timofonic]

uses supercomplex A/D  device , processor and lcd diplay with own graphics processor..
power supply ? 555 driven ... 

Please....


Would you like to elaborate more about such topics and what's bad with it?

What would be the best approach?

It would be very nice to know a detailed opinion about it

Thanks in advance;


Best regards.

[unitedatoms]

I appreciate your comments ! Thank you

September update: Finished the debugging of layout. Fixed a lot of simple mistakes. Reached the design goal of 0.5% accuracy in 3 of 4 ranges. The lowest impedance range is too sensitive to layout because of copper traces / probes ohmic resistance. Currently reconsidering use of several parts, replacing them with higher cost parts. About 10% of schematics changed. I will publish it ASAP.

To answer the comment about mixing the 555 chip into the middle of high cost parts.

I honestly tried to find the most advanced, contemporary, easy to use up-converter and made a lot of reading about available converter chips. There is a lot of interesting, sometimes simple and sometimes sophisticated ways to generate multiples of voltage rails with integrated DC-DC converters. At the moment with all of my knowledge, there is no single chip converter which complies to my requirements.

DC-DC power supply requirements:
- The conversion must start with control inputs being unpowered
- The conversion must stop when control inputs are triggered
- After certain period, the conversion must restart no matter what control inputs do (for example software errors, poor programming, etc)
- Converter must be "stateless". There should be no latched states, requiring software resets (for example after the overloads, thermal shutdowns)
- Converter must be popular, very well known, understood to let even unexperienced users to gain usable knowledge
- There should be multiple schematics adapted by community to allow easy upgrades, contributions to schematics from community
- Converted must have zero activity (even internal generators must stay completely) shut when required
- Converter schematics must be not tied to any advanced magnetics: transformers with calibrated gaps, ferrites with more than 500 KHz requirements.
- Magnetics for converter must be easy to implement, replace, calculate. So consequently several complex to calculate or tune designs are not acceptable: inductors with high DC bias, autotransformers energy gaps and with pulse width ratios, coupled inductors designed for super high efficiency at 1 MHz, etc.
- Diodes must be a common schotky type, not a specifically rare ones
- The design must be crude, able to start and stop in a worst supply, load conditions
- Efficiency is nice to have but is not critical. Some costly parts can be avoided since users are not interested in super high efficiency.
- Efficient board space is nice to have, but not critical
- Other details: At 3-4 watt, it should feed 6 DC rails (current prototype runs at 4.85..5.15V with current 0.89..1.25A).

- Most important requirement 1: it has to output multiple voltages.
- Most important requirement 2: it must be super easy to understand for starter.
- Most important requirement 3: It must not self destruct with 0 load conditions. Who knows what users will do during the build.

At the end I finalized my choice on 555. (At least I, myself can explain every single stage and part, why it was chosen, how to debug it, how it works).

Other details about what is being changed in power supply and what lessons learned:
- The worst mistake was choosing the particular P-MOSFET. Fixed the mistake by replacing it with "logic gate level" P-MOSFET. The lesson learned is: look at actual curves in datasheets and realize that at worst gate voltages MOSFETs behave as current limiters, not as milliohm resistors. To get to milliohm range at >1A at 4V use the "logic voltage gate" MOSFET. The cost is correlated to gate voltage capability, has to pay for it, but there is good reason why it costs so.

- The PNP current limiter from Art of Electronics. It is useful only when MOSFET control is ideal - high gate to source voltage. In situation with poorly chosen MOSFET the PNP part can be removed completely with no effect on work of the circuit. I still decided to keep PNP after upgrading to ideal MOSFET. It is yet to be seen if it helps. The thought is to protect random unpredictable external supplies from overloading. The absolute limit must be 1.1..1.2A, or else the USB supplies (even ones rated for 2A) will act up. I do not want to force users to start suspecting that their USB wall-warts are at fault (even when they most likely will be at fault if device starts consuming 2A).

- Interesting limitation of 555 chip: The reset input must not exceed 555 VCC rail even few millivolts. So I had to move reset input to blue LED with guaranteed 3V level < 4.25V VCC.

- Another correction (attached schematics has to be changed): power stage with transformer (mid tap) has to be supplied from before 1N4007 diode.

- Another improvements: Add jumpers in several test points (voltage controls, for oscilloscope, multimeter probing etc), add low ohmic resistors as current test points for approximate current measurements.

- Other good find: At low power levels and relaxed efficiency requirements - copper wire of transformer coils works as a free low-ohm high power resistor. It is OK to treat it as resistor for up to 0.25W with no problems.

BTW: 1N4007 may look like another cheap offence to surrounding high cost parts. The reason to choose 1N4007 (not even lower 1N400X) is that I needed indestructible piece of very low doped silicon to have the simple isolating DC switch with leakage current at very low nA level.

[timofonic]

@free_electron

You  may be right, but you can be more constructive too. I understand some things can be an aberration for you, because you are very highly skilled. I really appreciate your works and contributions to the community, I want to buy some of your books when I get enough incomes too.

But it would be nice if you are a bit more friendly and provide better feedback, please not traumatize newbies! People do mistakes, we are here to learn, discuss and help each other

[Towger]

@free_electron

You  may be right, but you can be more constructive too.

I agree. The op has designed and is close to release their project. But there is still no sign of your improved version of the ESR/component tester.

[unitedatoms]

October 7th update:
- Analog board UA315 Version 2. Fixed multiple errors in schematics (missing levels conversion for several digital lines), upgraded output amplifier, upgraded analog switch, fine fixes for DC offsets, more filters for DC rails, star grounding, better pairing of layout for DC rails, shielding for isolation of sensitive inputs from outputs (copper ring), shielding of noisy outputs (thick copper ring), single point connects for grounds from multiple DC rails, added multiple test points with jumpers, better attention to silkscreen labels, upgraded transformer with more taps (12 pins).

- Board size reduced to 102x180mm, was 99x199mm (x99 values is to fit into stupid dropdown choice of some other PCB house).

- Ordered the 5 boards batch from Elecrow. Same company, as for first version.

- This board is expected to be final (placement, routing wise, some resistors choices may still vary). The software is 50% ready and has satisfactory accuracy, readability, speed for 3 of 4 ranges.

I will update how well the V2 build behaved later this month (in October). Couple of parts are still to be ordered from Mouser/Digikey.

The purpose of 1st board is accomplished: to evaluate the noise (bobbing) of measurements. The bobbing is very low, in some ranges the repeatable reading have 0 to 1-2 steps of least significant position (full range is about 10000..20000 counts). The residual noises, interferences are not from any power supply or CPU board, the spikes in bobbing comes only from clocking circuit. This gives me hope that precision of 4-5 digits with minimal bobbing at 4th-5th digit is achievable in final build.

Other note: the joint points count is above 800 per board, which is close to 1000 points of license limit for DipTrace. There are multiple annoyances in Diptrace, but it is too late to change the CAD. BTW, the whole analog board is 100% hand routed. It took about 10 days just to place and hand route it, I really enjoyed may be first 5 days of it

Lessons learned: For any future complex designs, I will choose in favor of more smaller boards, than less amount of large boards. The connectors and cables are costly, but it would make the project so much easier to work with, when obviously independent modules are separate. For example power supplies array can be easily separated from analog frontend + digital-to-analog core.

[unitedatoms]

October 8th update:
While waiting for boards from Elecrow, I am writing notes. The article about ferrite transformer: http://unitedatoms.com/articles/ferrite-transformer-design-using-pc40-tdk-pot2213-core/


And the up converter schematics related to transformer.


The finalized part number for transformer will be RM10 with 12 pin bobbin and +2 pins for clamps B65814B2203X000. So I am getting rid of POT2212 transformer cores on eBay (USA only). Those are good and tested, but are too large / too powerful and occupy way too much space on the board.

[Vgkid]

Looking forward to more progress.

[unitedatoms]

Looking forward to more progress.

Thank you for your interest in this topic. I have added 4 articles to start explaining the design in details, while waiting for PCBs.

Project design articles, in depth:

part-1 "Only R and Time reference needed"

part-2 "Ratiometric use of AD5933"

part-3 "Complete exclusion of DC from measurement"

part-4 "ESR meter vs Network Analyzer"

part-5 "First seven prototypes"

[Vgkid]

That is a very nice write up.

[timofonic]

Great!

Are you considering switching to KiCad? I know it might be hard at first, but you might about these limits and make it true OSHW

[kripton2035]

great explainations.
did you consider including some pictures in your articles ? would be easier to read.

[timofonic]

@free_electron

You  may be right, but you can be more constructive too.

I agree. The op has designed and is close to release their project. But there is still no sign of your improved version of the ESR/component tester.

He would suggest improvements and be more didactic, instead doing a rude reply.

I just hope he had a bad day and apologize about it. I appreciate a lot his skills and knowledge, but not in this way.

People make mistakes, please be constructive and help others to make better designs instead demotivate them

I suspect he will not reply again. I hope he surpasses his proud and meditate about his attitude. Everyone acted wrong, nobody's perfect. But we can always improve, even if you are a guru.

[PointyOintment]

I have for a while wanted a good LCR/ESR meter, but never saw one I liked for a price I liked. I was considering building one, but it would have been nowhere near as well designed as this, so I think I'll build one of these once the design is stable and tested. Do you think it would work well using a Teensy 3 as the controller?

Also, here's a very simple OSHW milliohmmeter for finding shorts, for reference/inspiration: https://hackaday.io/project/3635-shorty-short-circuit-finder

[unitedatoms]

Great!

Are you considering switching to KiCad? I know it might be hard at first, but you might about these limits and make it true OSHW

Thank you, this is good suggestion. First, I will publish all design files in DipTrace format. I tried to use KiCad long time ago, but stuck with very odd UI/User Experience, it was very counterintuitive. As soon as the schematics will be stabilized and solid, I hope someone will redraw it in KiCad, or I will try drawing it myself. Yes, OSHW should be available in KiCad and PDF.

[unitedatoms]

great explainations.
did you consider including some pictures in your articles ? would be easier to read.

Thank you. I will add more pictures, yes. The current pictures are mostly about how this theory of operation is different to common LCR theory. The part 5 of story is just list of my personal mistakes, I will think what pictures to add there. Most of the prototypes were cannibalized for every next prototype, so I thought the pictures of junk boards midway to final boards are not very interesting.

Edit: I keep adding pictures to the article.

[unitedatoms]

Do you think it would work well using a Teensy 3 as the controller?

Also, here's a very simple OSHW milliohmmeter for finding shorts, for reference/inspiration: https://hackaday.io/project/3635-shorty-short-circuit-finder

I think it will work with any CPU at 3-3.3V logic levels. The analog board has 34 pin 0.1 male shrouded connector. It supplies only 5V/200-300mW for digital side.

I was expecting that there will no single digital choice acceptable by everyone. Everybody will like their own favorite CPU, USB IO extender, separate PC, Bluetooth etc.

So I stopped trying to make interface completely universal. For perfectly universal digital IO interface between measurement board and CPU in next versions or close future, I would think of optically / galvanically insulated IO with some kind of slow miltipin extenders on analog side.

[unitedatoms]

October 15 2015. Prototype #8 progress update:


The transformer RM10 with 12 pins is ready and has following winding details. All coils are 0.33 magnet wire.

• Primary pins 1-2-3 is 4+4 turns (4.4V)
• Pins 4-5 is 9 turns (7.5V)
• pins  6-10 is 6.5 turns (5V)
• pins 7-8-9 is 21+21 turns (+15V/-15V)
• pins 11-12 is 33 turns (27V)

 

The DC voltages after rectifiers on supercaps unloaded with input USB voltage 5.02V at the beginning of cable, 5.00..5.01V at the screw terminal.

Results: before the LM317/LM337 regulators were soldered.

5V rail unregulated: 5.19V - little bit (56mV overshoot) more than precalculated range of +4.846..5.134V

7.5V rail before regulator 5V analog: 7.15V is 350mV less than expected range 7.269..7.701V

15V rails before regulators: +17.42, -17.40V perfectly in range of precalculated range 16.961..17.969V

26V rail: + 27.37V perfectly in precalculated range 26.653..28.237V

3V rail: no regulator is soldered yet


October 21st 2015: FedEx is delivering the Mouser's box with parts ... just about now ...
Update: The 1.3K resistor made the rails voltages look better. Also the NDP6020P Logic Level P-FET now does not require 5.25..5.5V supply, 5.000V is enough. Resulting voltages of unloaded rails:

Pulsed USB voltage +4.962V/1.36A at input of DC-DC converter
Filtered output of DC-DC converter:
-17.445V before, -15.082V after -15V regulator
+17.437V before, +14.969V after +15V regulator
+27.45V unregulated LED supply
+5.077V unregulated 5V supply
+7.103V before, +5.035V after +5V regulator, +3.039V after +3V regulator


Wrapping up the power supply part as a blog article:
http://unitedatoms.com/articles/impedance-meter-ua315-part-6/

[timofonic]

Great!


Wow, the board with supercaps is massive!

Please provide updates in the forum, as I'm used to read it more frequently than a blog

Is that transformer difficult to obtain? Why so much voltages?

How did you draw these schematics? They look simple and nice. I only see some pixelation in the cylinder, the result is simple and very clear. I wish to be able to do it at least a quarter as good as you, I'm awful at drawing.

Did you use raster images for compatibility reasons? SVG would be nicer, but I understand it.

I find amusing your use of old school parts with the supercapacitor array, it's a bizarre design. Are they expensive? I still don't understand well why you used it, it seems it's used to avoid noise.

Why did you make your own custom Arduino+GPU? What GPU did you use? I had no idea MCUs would need a GPU for this.

What's the advantage of a custom impedance meter instead a more flexible equipment?

[tridentsx]

Great!

Are you considering switching to KiCad? I know it might be hard at first, but you might about these limits and make it true OSHW

Thank you, this is good suggestion. First, I will publish all design files in DipTrace format. I tried to use KiCad long time ago, but stuck with very odd UI/User Experience, it was very counterintuitive. As soon as the schematics will be stabilized and solid, I hope someone will redraw it in KiCad, or I will try drawing it myself. Yes, OSHW should be available in KiCad and PDF.

I had the exact same experience tried KiCad many years ago and quickly discarded it as a waste of time. Last month I saw some article about Cern investing in KiCad adding push and shove routing, controlled impedance, matched length tracks etc ... and I decided to try it again. This time it didn't crash all the time and I am positively surprised with the later versions of KiCad and have decided to make all my layouts in KiCad going forward.

[tridentsx]

I  don't know were your price point is for usb panel mount, this is the cheapest I have found
http://www.vpi.us/cgi-bin/vpi/pm-usbbf-usbaf.html

[timofonic]

Any news about this project? I really miss to see those frankenboards and articles

[Vgkid]

Any updates.

[Dave Atom]

Watching with a great interest, any news ?

[kandrey89]

Thanks for the project, updates and insights, looking forward to further updates.

[xaled]

I'm also interested in the project and updates.


Sent from my iPhone using Tapatalk

[julian1]

Super interesting project!

I feel it would benefit from,

  - Having the schematics/pcb layout (kicad?) and reference source-code on Github, available for anyone to fork, adapt or modify.

Other OS instruments/tools like the BusPirate succeed in large part because of the plugin software architecture, that allows people to write modules to adapt it for their own specific needs, while adding-value to the whole.

Thinking about the different functional adapations - impedance meter, LCR meter, low-frequency VNA, and I supect this will be too much work for one person to get the software and G/UI right.

  - Slightly more modular design. The power-supply design represents a very specific set of design choices. It might be good to be on a separate board.

  - SMT parts alternative. I suspect a lot of folks interested in assembling a project like this wouldn't be afraid of trying toaster-oven type reflow.

[broderp]

No activity for some time, this was very interesting...

What happened?  Did the project go bust?

[Vgkid]

Looks like UA hasn't updated this in a year and 4months so I guess it died.

[Dwaine]

It's a shame.  I was really looking forward to this project completing.   It would of been nice to have and LCR/ESR bench meter....

[unitedatoms]

Sorry for disappearance, I am back. I rethought everything about this project.
This time will start with enclosure. My current choice is Makersled heatsink slim black aluminum.

http://www.makersled.com/wp-content/uploads/2015/10/MakersSLIM-Get-Mounting-Your-LEDs.png

ordered 2x kits for 30$ each, 12" long. The choice of enclosure is for possible vesa mounting, for incremental build with multiple small boards, multiple front panels like in euro racks, so it will be possible to add small parts of project incrementally.

Dec 08 2018:
Built the power supply stage.
12V 1.5A

[OwO]

I still don't quite get why the design is so complex. It isn't difficult to filter a power supply to >100dB isolation especially at these lower frequencies (last time I needed 100dB isolation at 2.5+ GHz a shield can, careful layout design, and a few stages of ferrite-capacitor filtering was all that's required). Passive filtering takes care of the higher frequencies and for the lower frequencies you just need a linear regulator with high PSRR like the SGM2028. I also don't quite get your description of why supercapacitors are needed; are there very sensitive components that need a specific power off sequence?

[unitedatoms]

Agree. It is too complicated.

One day I saw that supercaps are cheap. And it was the original cause.

I chose supercapacitors because I had inertia of thinking over some radio circuits phase noise and low noise DC. It looked for me that noise wise that capacitors are better than batteries, batteries are better than photoelements, photovoltaics are may be better that references inside the linear supplies, etc.

The hypothesis was DSP wise, that DC source noise is a phase noise of Zero's Hz generator.

So by not looking at role of noise in ratiometric AC I overbuilt the noise defenses.
It is "volt nuts" in bad way. Or design rabbit hole.

I had so many mistakes. But I am glad I that I tried. I still have better ideas to continue all over.
One idea is to overcome few limitations of this DSP chip and use semi-analog approach in phase detector to reach super high resolution on phase.

Instead of digital multiplier, use MDAC four quadrant kind, and go to low MHz. May be up to 10MHz with no FPGA or even DSP.

[OwO]

I chose supercapacitors because I had inertia of thinking over some radio circuits phase noise and low noise DC. It looked for me that noise wise that capacitors are better than batteries, batteries are better than photoelements, photovoltaics are may be better that references inside the linear supplies, etc.

Actually this is a good point, a supercap is probably a easy way to get a lower noise floor at the <100Hz region than the best low noise LDO you can get (provided the cap is big enough that the voltage rolloff doesn't start becoming noticeable).

One idea is to overcome few limitations of this DSP chip and use semi-analog approach in phase detector to reach super high resolution on phase.

Instead of digital multiplier, use MDAC four quadrant kind, and go to low MHz. May be up to 10MHz with no FPGA or even DSP.

I would still go with a DSP based solution, maybe just two ADCs feeding a FPGA like all the open source VNA designs out there these days...

If you are only dealing with <10MHz measurements you can directly digitize the AC signals, then do everything in the FPGA. The full system could just consist of the FPGA, a DAC, 2 ADCs (one for voltage and one for current), and some amplifiers.

[Chris56000]

Hi!

Could anyone point me in the direction of where to purchase the PCBs and download the code and documentation for the through–hole version?

I'm interested in building the through–hole version as I'd want to make an instrument that's easy to maimtain & repair, so fancy surface–mount boards are a NO–NO for me!

Chris Williams

[coromonadalix]

i would love to make it portable,  i have nice lithium ions battery packs to use

Is there any project update(s) ??

[asdf]

Hi
I'm writing to let you know about my project "Precision impedance meter asdf_lcr1".
I don't want to hijack the thread (please don't discuss it here) but maybe some of you, who are interested in oshw lcr meter, may want to take a look:
https://hackaday.io/project/165409-precision-impedance-meter-asdflcr1

[unitedatoms]

Restarting the project. This post has attachment with concept of phase detector. I am hoping to hear any comments, opinions if possible.