$20 LCR ESR Transistor checker project

[madires]

Sorry! I've updated the link.

[Per Hansson]

I would be very grateful if you could give some idea where you think the issue is with the K-firmware that I see.
Am I on the right path with trying to adjust "FAKTOR_ESR" or was there some other much more imprtant thing you had to change?

"FAKTOR_ESR" Only works for 20nF - 200nF capacitors. For electrolytic capacitors, I changed a few places. But I consider myself understanding enough in Assembler to lecture Karl-Heinz. Who does not fit my firmware, do not use. I can't help anymore.

I can understand that you are tired of me Yuriy_K but I really like your dedication!
I used the latest GIT version with only this code replacing line 832:

Code: [Select]

#if (PROCESSOR_TYP == 644) || (PROCESSOR_TYP == 1280)
        #define FULL_PULSE_LENGTH_TICS ((HALF_PULSE_LENGTH_TICS*2) - 1)
#else
#define FULL_PULSE_LENGTH_TICS ((HALF_PULSE_LENGTH_TICS*2)+(MHZ_CPU/14))
#endif
With this change I have the following result:

Code: [Select]

GM328R 20MHz: Hiland644-1*: Hiland644-2*: Hiland644-3*: Hiland644-4*
20mΩ resistor: 0.01Ω 0.00Ω 0.15Ω 0.01Ω 0.03Ω
50mΩ resistor: 0.04Ω 0.03Ω 0.19Ω 0.05Ω 0.06Ω
47nF film cap: 46nF 0.50Ω 46.0nF 0.00Ω 46.1nF 1.10Ω 46.1nF 0.23Ω 47.1nF 12.0Ω
330nF film cap: 319.2nF 0.10Ω 316.7nF 0.00Ω 317.4nF 0.23Ω 316.3nF 0.12Ω 324.7nF 1.74Ω
2200µF lytic (KMG): 2326µF 0.10Ω 2327µF 0.07Ω 2229µF 0.28Ω 2470µF 0.08Ω 2294 0.11Ω
1000µF polymer (PSG): 980µF 0.02Ω 980µF 0.04Ω 942µF 0.13Ω 1044µF 0.05Ω 960µF 0.01Ω
Hiland644-1*= Latest GIT release, no changes (unable to read ESR of small caps).
Hiland644-2*= Latest GIT release, changed GetESR.S according to code above. (reads small caps but with a bit high ESR, same for resistors)
Hiland644-3*= Latest GIT release, same as Hiland644-2 but also changed FAKTOR_ESR on line 578 to: 550000 (reads pretty good!).
Hiland644-4*= M-firmware 1.48m-beta (added by edit later for comparison).

But I don't understand the purpose of the original code:

Code: [Select]

#define FULL_PULSE_LENGTH_TICS ((HALF_PULSE_LENGTH_TICS*2)+(MHZ_CPU/14))I checked the source and "MHZ_CPU" is defined as the frequency in Hz divided by a million, so it is for example "8" with 8MHz CPU.
"HALF_PULSE_LENGTH_TICS" is defined a very similar way.
What I don't get is why it is then divided by the frequency of the CPU by the number 14?
And in your changed version in my example code above this is instead "-1" if the CPU is Atmega644, but why?

[edgard22]

Yes, it is the same to connect only 2 terminals to see the junction capacitance, but sometimes it is inconvenient when there are many transistors. Anyway I copied a few lines of code in the BJT sector and I achieved the task.
In the photos you can see measurements of several 2SC5200 of different qualities. The one that shows Cbe460pF is very bad, measured with another instrument it has a Vbr of 130V (The originals generally measured Vbr above 320V)

[Yuriy_K]

Yes, it is the same to connect only 2 terminals to see the junction capacitance, but sometimes it is inconvenient when there are many transistors. Anyway I copied a few lines of code in the BJT sector and I achieved the task.

Incorrect measurement examples, you cannot compare I_E = 6.25mA readings with I_C = 0.935mA readings. The tester incorrectly detects the emitter and collector pins. From this, the capacity of the transition "eb" is incorrectly determined. Measure on individual wires and see if I'm right.

[Yuriy_K]

I can understand that you are tired of me Yuriy_K but I really like your dedication!

I showed my settings, measurements and firmware with corrections on page 311. I have nothing more to say.

[indman]

on 1.45m firmware, 16M crystal single measurement takes 10..30sec, is it possible to make it faster ?

M-firmware tests crystals only if there is an extended version of the frequency meter! Which version of the clone (device)are you using?

[indman]

Check if you have programmed the MCU fuses correctly?
H:D9 L:F7 E:FD(FC)

[indman]

is it possible to disable TP2 probing ?

There's no need for that.Read the section in the manual for the tester where it is indicated how to start the self-testing procedure and show us these results.

[Yuriy_K]

Sorry! I've updated the link.

I looked, I re-quote the transistor measurements from the previous page. The unconnected lead lies on top of the case. For the bottom row I will explain by numbers:
1. base and collector are connected together,
2. The сollector lead not connected,
3. The emitter is not connected,
4. The emitter is connected to the collector,
5. The base is not connected.

The examples shown show that not connecting free terminals has little effect on the readings of the capacitances of the junctions.

[madires]

on 1.45m firmware, on 16M MCU crystal single measurement takes 10..30sec, is it possible to make it faster ?
or leave only R/C/L mode, disable all others to make faster detection

Are you talking about measuring large capacitors?

[madires]

The examples shown show that not connecting free terminals has little effect on the readings of the capacitances of the junctions.

Thanks for checking! The capacitance between the  terminals is a bit more complex because there a several capacitances adding up. The capacitance of the leads, junction capacitance and diffusion capacitance. The latter two change with bias, frequency, temperature and so on. Anyway, if 'third terminal open' provides a usable value then we can use this method.

[indman]

simply shorted pins PC1,PB2,PB3 (TP2) - now measures much faster and no false detections...

Nothing is clear from your explanation because you don't state your problem correctly from the start and no name your clone model. There are no psychics here to guess at your measurement difficulties.

[madires]

Are you talking about measuring large capacitors?

yes

As larger the capacitance as longer it takes. This is normal for the measurement method used.

[indman]

As larger the capacitance as longer it takes. This is normal for the measurement method used.

Madires,that's right,but not 30 seconds,even if he is it measures 100mF capacitance!
Capacity of 10mF is measured by m-firmware with quartz 16MHz for 5-6 seconds, the capacity of 100mF -15 seconds!

[madires]

Glad you found the cause. I can't remember that I ever heard of that issue, i.e. a broken trace causing a delayed probing cycle.

[edgard22]

Yes, it is the same to connect only 2 terminals to see the junction capacitance, but sometimes it is inconvenient when there are many transistors. Anyway I copied a few lines of code in the BJT sector and I achieved the task.

Incorrect measurement examples, you cannot compare I_E = 6.25mA readings with I_C = 0.935mA readings. The tester incorrectly detects the emitter and collector pins. From this, the capacity of the transition "eb" is incorrectly determined. Measure on individual wires and see if I'm right.

You are right, but the values ​​read with the multimeter, although they differ a little from the AY-AT, maintain a relationship. I believe that an exact value is not necessary since it is used as a reference and it is possible to clearly differentiate them in a primary classification.

[Per Hansson]

is it possible to increase parameters indication time, before next measurement ?

You can change it here in config.h:

Code: [Select]

/*
 *  Maximum time to wait after probing (in ms).
 *  - applies to continuous mode only
 *  - Time between printing the result and starting a new probing cycle.
 */

#define CYCLE_DELAY      5000
But you can also instead enable this and the tester will wait for button or encoder input to measure again:

Code: [Select]

/*
 *  Set the default operation mode to auto-hold.
 *  - instead of continous mode
 *  - uncomment to enable
 */

#define UI_AUTOHOLD

[madires]

... or you can start the tester with a long button press (>300ms) to enter the auto-hold mode.

[Yuriy_K]

madires
I will try to explain the identified problem in main.c. When checking on the "m" firmware of the specified transistor, in my opinion, a very important parameter Vbe was missing. The given corrections allow to show this value. I understand that my solution looks primitive. I hope you find a better solution...

[madires]

In this case it's not a bug, it's a feature. Any base-emitter resistor(s) will screw up some measurements and make related results less meaningful. When no base-emitter diode (or two in series for a Darlington) is found V_BE and h_FE aren't displayed. For measuring the h_FE of a power BJT a higher base current would be helpful. The tester can provide just 7mA which isn't sufficient for this. So a simple trick is used to extrapolate mathematically a higher base current. This doesn't work for a power BJT with base-emitter resistor.

[n6rob]

Greetings;

I have a LCR-T4 (mega328p, 8 MHz crystal, 78L05, TL431, one button, no encoder, no up/down buttons) and I have loaded 1.47m on it successfully (also previously loaded 1.13k, but that's another story).  Anyhow, running 1.47m, and noticing some odd behaviors.

Here's what I did; per instructions I edited Makefile, config.h, and config_328.h with appropriate settings (as best I could).  If I was unsure about a software or hardware feature, then I took it out.  The unit boots, and displays fine (had to flip the display and remove the offset in the x axis of the display), and responds to the button as expected.

Observations:
The unit boots, and runs a test and comes back with an IGBT N-ch (with nothing in the test port).
I put the unit in menu mode by quickly clicking twice, I select test, it asks for a short, I install it, and then it does nothing.  I quick-click, and it returns the menu (if I long-click then it gives an error! and returns to the menu).
If I leave the short in during a normal test cycle, it says "probing..... C", it takes a while, but then shows the short across all three terminals.  (Draws a fair bit of current, about 45 mA, from the power supply, too).
It reads an resistor, LED, or an NPN, or a PNP, as an IGBT.
It reads a resistor as "no component found".

Measuring resistance (power and ISP programmer disconnected) between test pins 1 and 2, I get 39K ohms, from 2 to 3 I get an open, from 1 to 3 I measure about 1.4M ohms.  I'm going to hit Google and look for a schematic for this unit after I post this.

Any suggestions of things I should look for?  Much appreciated.

Rob

[madires]

Have you used the settings listed in the Clones file? There could be something wrong with the pin assignments for the probes. You'll find a schematic at https://disk.yandex.ru/d/yW8xa5NJgUo5z (the T4 comes in two variants).

[n6rob]

Have you used the settings listed in the Clones file? There could be something wrong with the pin assignments for the probes. You'll find a schematic at https://disk.yandex.ru/d/yW8xa5NJgUo5z (the T4 comes in two variants).

Greetings and thank you for your reply.

Yes, I referenced the Clones file when updating the configuration files.
Thanks for the link to the reference files, including schematics.  In reviewing the photos I can say that I have the T4/T3-NoStripGrid version.  That helps.  I just referenced the schematics and the pin designations in the config_328p.h file and I don't see any incorrect assignments.

Since my last post, I have reloaded the original firmware (as it came when I bought it, "MTester", no version indicated).  It exhibits some of the same problems.  I might play with the 1.13k version again (previous attempts have given the same results as the original & 1.47m firmware).

Taking a step back, I think it is likely that I have damaged something at the test ports (probably the chip).  I probably failed to discharge a cap before testing; but that is just a guess.

The next step I'll take is to disassemble the unit from the case and inspect it for damaged traces, solder bridges, or cracked solder joints.  It worked just fine for a few years; this weird behavior is more recent.
 If I learn something new, I will post back here.  Otherwise, I would consider this a case of damaged hardware.

Thanks for the help.

Rob

[madires]

Then it's most likely caused by damaged I/O pins. To verify this you can run the self test and look for strange values (or post them here).

[n6rob]

Then it's most likely caused by damaged I/O pins. To verify this you can run the self test and look for strange values (or post them here).

I de-fluxed the board, cleaned-up a few things, looked for cracks, shorts, etc.  I didn't find a smoking-gun.

I entered menu mode, select "test", it asks for a short, I install the short, and then nothing happens.  I wait a minute, click the button, and it goes back to the main menu.
I entered the menu mode, select "adjustment", and get the same result as "test".

I entered "show values" I find the following:
Ri- 20 ohms
Ri+ 22 ohms
C0 43 pF
R0 0.2 ohms
Vref 1084 mV
Vcc 5049 mV *
Acomp 0 mV