Curve Tracer of LockyZ: Better UI and Translation

Hi all,

Since 2014, Lokcy_Z from China released his transistor curve tracer system, and that helps DIYers a lot. The tracer is powerful and inexpensive, and it seems there is no competitor at its price band.

After the production of first batch, the tracer was discontinued for many years. Until 2019, when Locky_Z announced to re-start the production. The new version has smaller formation and some improvements. The ADC and DAC are actually the same, as well as the software.

From early 2020, I got the new tracer and found out that the software have some bugs and poor English translation. Then, I started my work on optimizing the tracer software. Basically the terminology has been re-translated, so that it is much easier for international users. But translate the manual is huge task that I haven't done yet.

Right now, the software version is V3.5.30, in this version there are 2 important changes:
1. The color of curve tracing area has been change to white and blue, etc. This makes UI more friendly and readable.
2. The stamp of lockyZ's logo has been removed, and this helps when copying traces to clipboard and looks cleaner.

You can download the software
here
or go to:
ElecFi Makers: LKZ Tracer ICT V3.5.30i

Any suggestions are welcomed.
 

Attachments

  • 119774013_2639304596332851_5536774781925275812_o.jpg
    119774013_2639304596332851_5536774781925275812_o.jpg
    472.1 KB · Views: 919
  • 119684137_2639305389666105_5658522367848383232_o.jpg
    119684137_2639305389666105_5658522367848383232_o.jpg
    433.2 KB · Views: 926
The 2020 new version of Intelligent tracer, and with full schematic attached.
I am modding the tracer,like upgrading Vref IC and place new low-ESR FP-CAP.

Since I got this tracer very early, it uses some second-handed parts, thus I also have to replace and desolder many relays, transistors and resistors that are crucial.
Some resistors have to be precise and low-temp-coefficient, so that the measurement will be accurate; using 0.1% 25ppm wirewound and even 0.01% 5ppm metal foil make sense.

Now Locky_Z is providing all-new-part version of the tracer, which is a good news to late birds. I would recommend this type, for those who haven't buy a tracer.:)
 

Attachments

  • Full Sch_2020.png
    Full Sch_2020.png
    189.5 KB · Views: 1,033
  • IMG_4321.JPG
    IMG_4321.JPG
    952.3 KB · Views: 958
  • IMG_4324.JPG
    IMG_4324.JPG
    574.1 KB · Views: 928
Hi,

thank You so juch for Your great effort.
having waited for years I also bought the new version with used parts recently and are far from confident with it.
Probabely its still some unfamiliarity with the device, but more often than not it just puts out weird results and wild cards similar to what's shown in Your pics.
I put it aside because of great disapointment and low time.
Maybe Your software changes that.

jauu
Calvin
 
Last edited:
Many many years ago, before he was selling openly, some of our team members bought multiple examples of his CT.
We found out very soon that he was using parts removed from old equipment to cut cost.
Also many of the key components were already obsolete.
So we contacted him through our team in Hong Kong.
We gave him an English translation of his manual for free, made suggestions to use modern active parts, low Tempco reference resistors, etc.
But he decided that this was not the route he wanted to take.

So we left it as that, and built our own curve tracer, with all the things that we want that are different.
It is actually very easy to do, especially if you have a good USB acope with func gen.
Just a bit of programming to control the device, even if you need to use it for quantity matching.
For one-off measurements, fully manual operation is even easier.

But if you just want to see a curve and not bothered about accuracy, then just be happy with it.
You get what you pay for.
A good USB scope (we use Tiepie HS3) is easily 1k€+.
But then you use it for umteen other things than just curve tracing.


Cheers,
Patrick
 
Many many years ago, before he was selling openly, some of our team members bought multiple examples of his CT.
We found out very soon that he was using parts removed from old equipment to cut cost.
Also many of the key components were already obsolete.
So we contacted him through our team in Hong Kong.
We gave him an English translation of his manual for free, made suggestions to use modern active parts, low Tempco reference resistors, etc.
But he decided that this was not the route he wanted to take.

So we left it as that, and built our own curve tracer, with all the things that we want that are different.
It is actually very easy to do, especially if you have a good USB acope with func gen.
Just a bit of programming to control the device, even if you need to use it for quantity matching.
For one-off measurements, fully manual operation is even easier.

But if you just want to see a curve and not bothered about accuracy, then just be happy with it.
You get what you pay for.
A good USB scope (we use Tiepie HS3) is easily 1k€+.
But then you use it for umteen other things than just curve tracing.


Cheers,
Patrick

Hi Patrick,

Sounds good. Do you have more information about what your team has developed? I would be interested too given I already have my modded-IC Tracer of LockyZ.

Indeed, the parts used and commercial strategy that Locky_Z have is somehow weird, I think he is an engineer more than a businessman.

I gave him a lots of advice that may be good for his product's expansion and the suggestion of improvement of both software and hardware; but he is kinda stubborn and unwilling to change.

Take a technical detail for example, the DAC he uses is a obselete old Japanese parts. But we have a bunch of modern, better-performance, and easy-to-find DAC choices here. But he just insist not to change, even till 2020.

Another issue is the circuitry, the op amp was supply with up to single 36~40VDC, while the OP's max is 36V, not to mention op amps are used-parts, and this terrible ideas reflect on high fail and return rate.

About the used-parts, yes, clearly he want to cut down cost to fit the Chinese DIY community, that's understandable, and sometime this create many "black technology" to the western people: How can they make and sell on a such low price?!

Because I am happy to pay more for better performance, I also suggest him, why don't you provide another version that use brand new parts for those customer who cares more about quality and precision?


However, no is no. So, I had to buy new relays, resistors and transistors; desoldered parts and replace them with new parts.

Until now, he finally offers new-parts tracer on the ebay for western customers.

And according to my observation, he is the guy who is sometimes stick his own idea and reluctant to accept better, newer ideas. His using software same as 2014 proves this observation.

That is also the reason I make my effort to optimized the software (and he is for sure OK about this). He do provide English interface for international users, but, you know, that's unreadable Chinese-style English. I have to go through the mechanism of the tracer, learn some electronic engineering knowledge, so that I can translate the UI more accurately.

That takes me several months, and the service manual are mostly still Chinese. So I guess there is still several months more work to do.

The reason I do so is that I hope international users can use a more friendly UI and share the common interface/terminology so that users can discuss and generate more ideas on the same basis without language barrier.

I hope my work can make more people willing to use and discuss about the tracer, like modding the tracer, the software, the circuit topology or even firmware which might give us more function and accuracy. :)
 
Hi,

thank You so juch for Your great effort.
having waited for years I also bought the new version with used parts recently and are far from confident with it.
Probabely its still some unfamiliarity with the device, but more often than not it just puts out weird results and wild cards similar to what's shown in Your pics.
I put it aside because of great disapointment and low time.
Maybe Your software changes that.

jauu
Calvin

HI Calvin,

For normal working of the tracer, I don't want to see unwanted, sudden failure of the tracer.

So replacing all the realys and OP-amps is a must. For precision measurement, the output resistors of BCE channel needs to be replaced with low-tempco, high precision wirewound resistors.

Personally, I also remove the 78L05, and use a precision 5V voltage source (REF02 or something) to supply the DAC.

These are my suggestions, and it can make your tracer far from sudden failure.

Oh, and try to control supply voltage about 36VDC, because that's the upper limit of signal OP-AMPs. You can supply it with 40VDC to get more wide measurement range, but I think that's not a good trade-off.

For +40VDC supply, the OP-Amp is working on the edge. It may or may not fail, but certainly increase the risk. For that 4 more volt range, I don't think it's worthy to take the risk.
 
...But then you use it for umteen other things than just curve tracing.

Patrick,

Do you have more details on this?

I’ve thought that an Analog Discovery 2 with an external attachment might be a good solution for a curve tracer. It’s on the back burner of my to-do list (bad mixed metaphor - sorry), but I’m not proud and would be thrilled to take advantage of someone else’s development efforts. I’m always pleased to give someone else full credit.

Life would be even better if someone develops software to take this and other data and creates a SPICE model...
 
There are many published curve tracer circuits on the net.
This is just one example.
https://www.onsemi.com/pub/Collateral/AN-6610.pdf.pdf

We measure FETs 90% of the time, and then only Id vs Vgs.
But you can easily extend this for BJTs by putting a resistor between driving voltage and base to measure Ib.

And you need a high accuracy 2 channel scope, or you calibrate, or both.


Patrick
 
Hey all,

Why don't make a wish-list that a good amateur tracer must have?
For example, automatic Spice model generating, that would be very powerful and useful.
Auto-ranging like DCA75 pro, the machine knows how to choose the right setting of measurement.
Using power OPA like OPA541 perhaps? It would be convenient and powerful. No discrete power amp is needed, and no worries about small OP's max supply voltage.
I think we can organize a open-source project if the demanding is enough.

Like I am always curious, why are there only Locky_Z's product that can provide such a machine for amateurs who don't nor be need SMUs?
I mean, it's right, Locky_Zs have lots of drawbacks, but a least it is enough for amateurs at its price band. But why there is no such another function-specific, easy-to-use curve tracer?

I know there're some project, but they night need scopes, or have poor resolution, like some arduino-based ones. Locky_Z provides a fully assembled hardware, an almost plug-and-use software, and sells for under 200USD. I think it's fine to accept its drawbacks given its price.

So, I believe there is demanding for slightly better accuracy, better functionality tracer for ~300 USD, designed for DIYers!
 
Hi,

thank You so juch for Your great effort.
having waited for years I also bought the new version with used parts recently and are far from confident with it.
Probabely its still some unfamiliarity with the device, but more often than not it just puts out weird results and wild cards similar to what's shown in Your pics.
I put it aside because of great disapointment and low time.
Maybe Your software changes that.

jauu
Calvin

Patrick,

Do you have more details on this?

I’ve thought that an Analog Discovery 2 with an external attachment might be a good solution for a curve tracer. It’s on the back burner of my to-do list (bad mixed metaphor - sorry), but I’m not proud and would be thrilled to take advantage of someone else’s development efforts. I’m always pleased to give someone else full credit.

Life would be even better if someone develops software to take this and other data and creates a SPICE model...

I found a Japanese who are exactly developing with AD2.

http://daisan-y.private.coocan.jp/
 

Attachments

  • 07140003.jpg
    07140003.jpg
    816.1 KB · Views: 394
  • 2019-7-12-001.png
    2019-7-12-001.png
    25.1 KB · Views: 468
  • 07140004.jpg
    07140004.jpg
    760.8 KB · Views: 357
  • 2019062401_020.png
    2019062401_020.png
    69.3 KB · Views: 404
I found a Japanese who are exactly developing with AD2.

http://daisan-y.private.coocan.jp/

Great find!

There's also this: Analog Discovery 2™ Curve Tracer * knack supply

Neither is exactly right, but getting close. For small signal devices, both will do the job. So will the DCA75 Pro. The gap is for higher power devices, like you'd use at an amplifier output.

The SPICE model extraction part is the puzzler. For me, anyway, since I am not a software expert. Most effective approaches described on the web use some sort of curve analysis method that backs the required information out of the curve. My naive thinking is that if you have a programmable piece of test gear that gives that data directly rather than making a curve that you then back data out of, you'd get better accuracy. And, it would be easier to use, too.

Personally, I just want to design and build various pieces of audio gear. LTSpice and the others are a tool to help me. I want another tool that'll let me sort me devices or generate my own database of models based on measuring a number of samples. That's all.
 
This thread was about a curve tracer project, before COVID...
Curve Tracer or Audio Analyser?

Thanks for reminding me of this thread. I'd seen the first few entries when they were fresh, but hadn't followed it through.

This was primarily about a proposed commercial product, which doesn't seem to be on the market yet. Perhaps due to COVID, as you suggest.

Later posts suggest that other people would like SPICE model generation, too. They were roundly shouted down for some reason...

Anyway, I look at it like this.

SPICE models are usually derived from analyzing curves generated by curve tracers, plus some other data. At one point in time, curve tracers used continuous sweeping of the driving function and synchronous detection of the response. The data was captured by taking a photo of the screen. Later, methods for direct screen capture became available. Fast forward, and you find that the driving functions are generated by a zillion points that effectively make a curve and the response is examined by another zillion sample points that effective make a curve. Then, for the SPICE model, the curve generated from all those sample points is fit by some software to extract data points. To my unskilled thinking, there's a lot of middle steps there that could be skipped if you just generated the points you need and sampled those.

The advantage of an Analog Discovery (just one example) is that a lot of the heavy lifting has already been done, both for software and hardware. Yeah, some adaptation is needed to attain the current and voltage levels we might all like. But, that part seems pretty doable. There's already software available from at least two sources that will do the curve tracing. What is needed is that additional hardware and the software that specifically tests for and extracts the data points needed for SPICE models.

Guys like Bob Cordell and Andy Connor have already shared great work on making SPICE models from the data curves. You'd think that there might be an easier way forward.
 
Another optimization:

the picture in "Diagram" can be enlarged without distortion.
This makes measurement more intuitive, you can looks a the diagram while you are doing measurement.

At the offset test result shown in the picture, the fluctuation is less because I changed the servo OPamps.

The original servo OP that LockyZ used is Chinese-made TPXXX, and I replace them with OPA2177, the result get better obviously.
Actually, you can replace them with OP77, 07, 97, etc. to get stable offset results.

I will share my modding tracer log afterwards, including low-temp-co resistors at the output of each channels, and relays. I replaced them with Finder and some omron and other Japanese-made relays. My tracer is now stable and happily running without problems.
 

Attachments

  • screenshot 2020-09-28 001.png
    screenshot 2020-09-28 001.png
    218 KB · Views: 479
Mine is no.22 handmade by LockyZ, it was produced in early 2020, very first batch of the tracer. At that time, there is only used-parts version. So I have to replace many key parts.

ADC: replace with TL431 1% ref as power supply.
DAC: replace with precision, low drift 5V voltage reference, and low-ESR caps.
Relays: All replaced with new Finder power relay and Takamisawa small signal relay.
Shunt resistor: <1% 25ppm wirewound high power 5W.
OPA: OPA2604 are used, but you can also try TL082 or LM358.
Power: MeanWell 36VDC 200W SMPS.
The test result of Linear supply and SMPS are the same, so use SMPS for better cost-effectiveness and smaller volume.

LDO: New LM317 TO-220 for +36V supply to OPamps.


According to Locky_Z, there are some cases that reports MCU malfunction due to unknown reason. This happened to me, Locky_Z just sent me another MCU board, and solved the problem.
 

Attachments

  • IMG_3739.jpg
    IMG_3739.jpg
    888.4 KB · Views: 474
  • IMG_3871.jpg
    IMG_3871.jpg
    894.7 KB · Views: 475
  • IMG_4171.jpg
    IMG_4171.jpg
    999.3 KB · Views: 466
Member
Joined 2007
Paid Member
I've had one of his earlier models. I inadvertently had power applied on the psu, plugged in the connector, got a large spark, then nothing. I ended up replacing parts with new ones and took the easy route of sourcing a new relay with the same part number from China, but I think the quality of those relays are iffy because it acts up from time to time. Locky_z helped me through it but you would think a primo version would work out for him with less support issues.
 
Hi,

thank You so juch for Your great effort.
having waited for years I also bought the new version with used parts recently and are far from confident with it.
Probabely its still some unfamiliarity with the device, but more often than not it just puts out weird results and wild cards similar to what's shown in Your pics.
I put it aside because of great disapointment and low time.
Maybe Your software changes that.

jauu
Calvin

I second that. Outliers like these simply must not happen:

< microchip_dn2470 | Some Vgs vs. ID curves of DN2470 transist… | Flickr >

And the original Chinese text would have an advantage over the translation,
that the Google translator it is bound to improve over time.

Gerhard
 
I second that. Outliers like these simply must not happen:

< microchip_dn2470 | Some Vgs vs. ID curves of DN2470 transist… | Flickr >

And the original Chinese text would have an advantage over the translation,
that the Google translator it is bound to improve over time.

Gerhard

Any other suggestion about the software? I am doing further change, so please feel free to give me feedbacks.
Any problem on reading technical manuals? English version is under work.