$20 LCR ESR Transistor checker project

[de_light]

I'm trying to compile the 1.40m for BSIDE (as a first small thing to do for others!). I'm getting an avrdude: ERROR: address 0x8010 out of range at line 2049 of ComponentTester.hex error on the files that I am trying to load onto the tester.

Am I right in thinking this error means that the filesize is too large for the Atmel 328p internal memory?


Program size is 33064 bytes (100.9% full)
Data is 241 bytes

EEPROM 1026 bytes (100.2% full)

[madires]

Yep, it exceeds the flash and EEPROM size of the ATmega328. To decrease the size simply disable a few features you don't really need.

[de_light]

For BSIDE ESR 02 tester 1.40m 8MHz and 16MHz

Has autohold enabled, 60s timeout

https://drive.google.com/file/d/1820dat2Fdzn6tY6CqapZ-QGmm6cjZYjO/view?usp=sharing

[indman]

de_light,
there is no need to add ComponentTester, ComponentTester.lss and ComponentTester.map files to the archive contents. They are not needed for the programmer to work. It is more informative if you attach the config.h and config_328.h files to the archive, which contain information about the display settings and other options specific clone. 

[robca]

Hello
search working HEX files and fuse settings for the GM328
Transistor Tester (from Banggood) because I have the Atmega
shot. The original seems to be protected against reading :-(
TX7014DH is on the O-Led, thanks for a feedback

The picture in your post is of what's called an AY-AT around here, not GM328. And the AY-AT doesn't have an oled, but an LCD screen

Assuming you have an AY-AT with an LCD as in the picture you provided, it's identical to mine. My version was not protected, and I attached the files to this message https://www.eevblog.com/forum/testgear/$20-lcr-esr-transistor-checker-project/msg3085915/#msg3085915

[cz939]

Hi! Here are some observations and thoughts regarding a new T7/TC1 V2.12k tester. Have you experienced the same? Are there any good explanations and solutions?
Would like to know whether some problems regard just mine piece, or they are common for this lot/series or the old firmware. Maybe some are due to measuring circuit limitations or probing methods? Possibly some are corrected in new open software versions, but this is poor comfort.

Testing of resistor by means of all 3 leads, with 2 of them shorted gives us resistor array of some R and about 0Ω, obviously. But testing of a diode in the same way, or 3-terminal transistor with internal shortage, shows two almost identical diodes, what is of course wrong. Anybody knows why the shortage is not detected? A bug in the logic?

It's a pity that for resistor array/potentiometer the sum of two resistances is not calculated and displayed.

When measuring of transformer coils the first result can be small, the second and consecutive numbers are much much higher. After exchanging of test leads the situation repeats, similarly when measuring in mean time of another coils of the same transformer, if any. This is effect of magnetic hysteresis and DC magnetizing of core material. But which inductance result is correct? The higher? Maybe the probing sequence should de-magnetize (or slightly magnetize) the core prior the essential measurement(s)? And/or provide both numbers - for AC and DC?

Probably some of us noticed that serial resistance of inductors is sometimes showed e.g. about 140Ω whereas the real value is only a few ohms. This is huge error! But sometimes similar RL elements are measured intact. Strange.
Found a "by-pass" - connecting of small resistor (just 10-20Ω) in series with the measured coil corrects the results! The added resistance must be simply subtracted from displayed R, of course, not sure whether and how to adjust L.

Testers with factory (old) firmware suffer wrong measurement of bipolar transistor hFE and this is known behavior. It is probably caused by big relative error when measuring of small voltages or close to unstable VCC, comparable to ADC accuracy. Have no better explanation as hFE at low Uce (near saturation) is normally much smaller.
I think both measured hFE coefficients along with Ic currents and Uce voltages in common collector and common emitter configurations should be displayed (but there can be no space or memory). Or only for OE, if we know that measurements for OC stage are often worth not much.
Was trying to change transistor work point and measuring conditions by means of additional resistor. It turned out that such slight modification of measuring circuit can dramatically amend the results. By connecting of additional 100Ω in series with BJT collector we can get correct measurement of hFE coefficient!

Did someone manage to change the switch-off time after probing for TC1/T7 family testers with STC15L104W mcu? According to early instructions as well as provided on PCB it can be set to 10-15-20-25s respectively by means of P1 & P2 jumpers combination. How to set them? Just solder short to ground?
Default factory setting is 1-0 = 20s, however both points are left unconnected (open, at supply voltage) 1-1 = 25s. How it works? P1 & P2 pads are provided rather for ISP, and mask around is not scrapped/tinned (dislike EN pad).

The supply mcu works very unreliably - sometimes doesn't react on Start button, sometimes shuts everything off instead of initiating a new test and sometimes doesn't automatically switch the device off after 20-25s. And it resets itself during probing from time to time.
I wonder how the 3V (3.6~2.4V) STC15L104W supply mcu can work up to about 4.8V? Recommended replacement is W series (operating voltage 5.5~2.5V), but the older L chips were installed in T6/T7/TC1.
Guilty can be also noise from converter or poor FW. Blinking red dot may be accidentally observed, signalling false IR waves, even if there is no pilot used or the sensor window is covered.
Mystery transmissions between MCU's has not been probably decoded either.
However if for any reason meter doesn't switch off automatically or manually by long button pressing, it is switched off anyway after 160s from last start/restart.

What do you think (what could be the reasons) of different measurement results for the same part depending of test point/leads order, e.g. EBC=123 and EBC=321. This is for calibrated tester and in case of transistors, whereas for simple 2-leg elements the results are intact and steady. Is it an effect of some internal asymmetry and/or step-up converter noise or supply circuits instability?

Probably not everybody knows that 2.12 FW clones are "SW ready" for control of discharge relay. Unfortunately not HW and there is no place on PCB prepared for transistor, relay and clamping resistors.

Some manufacturers/distributors install/sell clones with poor quality or old accumulators, which are unable to supply testers in the whole 4.2-3.0V range. Low battery warning appears at 3.0V, but for some devices this is over strength - it resets, displays white screen, fades, squeaks - over and over again, until built-in protections switches it temporarily off or someone connects USB charging cable.
It even seems like some accumulators gradually discharge to about 3.3-3.2V, then the voltage suddenly jumps to 3.5V, but soon "dies" after couple tests. The truth is quite different - the battery voltage drops even more and it is unable to supply dc/dc booster and LDO, MCU supply voltage drops below 5V and all measurement results in such conditions are tricky - battery voltage looks higher, silicon transistors get 50uA Ice0 & Ices etc. After recharging of the battery the parts parameters "normalizes".
Isn't it strange that 1S 15C 350-450mAh LiPo battery behaves in such manner? What is such battery life span?
These series clones feature very high supply current and PS structure is inefficient. 3.7V accumulator voltage boost converted to 9.5V (and 35-45V for ZD), then linearly regulated to about 5V and one more time to 3.3V for LCD and LEDs.

The tester seems to be a simple one-button device - "just put-in something and push". Its measuring circuit has been designed to be simple too. Yet its operation is not so straight. Probing conditions can change in many ways and not all parameters are calculated and displayed. Moreover, the conditions often differ from those encountered in popular meters. One has always keep in mind complicated rules and formulas described in manual or at least schematics listed in the short version - ttinfo.
By the way, is there a more current version than the 1.11k KHK '2015 document? And official pdf of English 1.39 m ttester/ctester (I found "only" Latex files and some translations)?

Ordinary multi-meters are of better or less better accuracy (class), but are stable. In contrary, measuring results of "our" testers often scatter in consecutive probes (and still consistent for some simple elements). This is probably due to thermal and voltage instabilities as well as noise. Are other models without converter more stable?

The "instrument" is good for testing of discrete elements "from a drawer" but rather not those soldered on board. This is mainly due to its measuring principles and logic. It seems that almost any additional by-side resistances, capacitances or semiconductors render probing void.
Even a popular DMM with ohmmeter and diode test can be used for in-circuit measurement of resistors by means of low-voltage bridge as well as diodes with current source, but we select appropriate function and polarity. Tester is smart, but not for mixed parts. Maybe newer FW with menu which enables switching for selected-type cyclic measurement in addition to "universal" smart mode behaves better in such situations, but T7 has no menu.

The series is equipped with ZIP socket and cheap small spring hook clip wires - good for small THT parts but not suitable for larger terminals nor SMD. Especially lack of SMD pads (or adapter) sucks.

It seems all clones are equipped with too hard start-button. I think it might be much softer, like typical mouse click or roll. Maybe it is not so important when elements are mounted in socket or clips and you have free hands, but in cases you are using 2 or 3 test leads almost 4-th hand is needed to push such hard button.

[madires]

Please check out https://github.com/madires/Transistortester-Warehouse for the latest firmwares and documentation. See the Clones file in the m-firmware archive for TC-1 modifications to make the TC-x compatible with the OSHW firmwares. I think that the TC-x' boost converter in conjunction with poor filtering and an LM78L05 might be the cause for measurement values jumping around.

[all_repair]

May I know how does one interpret the Capacitor leakage reading?  of course, it is M version .  I am seeing charging current and voltage, and then discharging voltage.  They are happening quite fast.  I got no idea is the cap leakage is low, or not.  No delta T provided.

[de_light]

May I know how does one interpret the Capacitor leakage reading?  of course, it is M version .  I am seeing charging current and voltage, and then discharging voltage.  They are happening quite fast.  I got no idea is the cap leakage is low, or not.  No delta T provided.

https://raw.githubusercontent.com/madires/Transistortester-Warehouse/master/Documentation/English/ctester-1.40m.pdf

Page 30 of the manual talks about interpreting cap leakage.

[madires]

The measurement of the leakage current is basically the same as the k-firmware's Vloss measurement, and instead of the showing the voltage loss in % the voltage difference is converted into a leakage current. At the start of the measurement the cap's voltage is around 1.3V.

Another measurement taken is the self-discharge leakage current for capacitors larger than 4.7µF. It gives a hint about the state of an electrolytic cap. From my tests the typical value for a good electrolytic cap seems to be about:
  - 10-220µF    1-3µA
  - 330-470µF   4-5µA
  - 470-820µF   4-7µA
  - >1000µF     5-7µA per 1000µF

[cz939]

Hope this is proper topic for such discussions. Thank you for reply. The link to the well known content (still evolving) ensures there is no newer docs "hidden" somewhere else.

I've accused the converter ~1MHz PWM modulated signal, present probably almost everywhere in the device, as well as floating supply voltage, for most problems, but is the construction flaw the sole cause of all drawbacks?
Some testers cheep (like SW radio) under some circumstances (higher current, worse coil impregnation). Is it "normal" or worrisome symptom?
Are there any ways to improve screening, decoupling or filtering? Did anybody try to add one more tap (transformer rewind) for 3.3V or LED supply, or to replace some ICs?
Can someone give a positive example of TC1/T7 FW update - both MCUs, without STC15L "replacement"?

[madires]

I doubt that the TC-x can be improved greatly. So I'd recommend to get some other tester clone if you aren't happy with the TC-x' performance. Several users updated the STC15L with the alternative firmware at https://github.com/atar-axis/tc1-u4 to be able to run the OSHW firmwares.

[de_light]

Out of interest, what do you think is the best tester hardware going?

I doubt that the TC-x can be improved greatly. So I'd recommend to get some other tester clone if you aren't happy with the TC-x' performance. Several users updated the STC15L with the alternative firmware at https://github.com/atar-axis/tc1-u4 to be able to run the OSHW firmwares.

[madires]

There isn't any perfect tester unless you build one yourself. If you are looking for a tester clone to buy it depends on what you want. The  AY-AT (AKA GM328A) is still is good choice. And If you want a nice case maybe the BSide ERS02 (DTU-1701). For checking crystals and measuring higher frequencies the Hiland M644 because it comes with the extended frequency counter. Unfortunately it lacks the input buffer stage for direct frequency measurement and a few bypass caps.

Things to know about the TC-x family:
- looks neat
- comes with the dreaded U4 with a high quiescent current
- noisy power circuitry
- not rarely do cheap parts in the power circuitry cause trouble (might have improved meanwhile)
- but they all do their job more or less

[carrascoso]

Hello Madires.
Today I had to check some JFET transistors, it recognizes them as the type of transistor they are but it does not show the pinout well, XGX appears.
I have tried other types of transistors and without problems.

The same happens with the firm of Trunk, although in this case if the names of the pins appear on the figure.
I'll see how to communicate it to see if it can be solved.

In Chinese GM328 and LCR-T4 meters without modifying it does not happen.

Is it possible to give him a solution, or for some reason he has to stay like this?

Thanks greetings.

[madires]

It's not a bug, it's a feature! For many JFETs it's hard to tell drain and source apart since they are nearly symmetrical. In those cases the k-firmware shows a "?" and the m-firmware an "x" for drain and source. Clones with an old k-firmware or modified version of an old k-firmware simply guess the pinout, i.e. it could be reversed.

[indman]

I edited the comparison table for the clones a bit! Fixed information about the display size in inches for some clones.
https://www.eevblog.com/forum/testgear/$20-lcr-esr-transistor-checker-project/msg3088827/#msg3088827
If you find errors in this table, please inform me in your personal mail. 

[Ken333]

On the model pictured below, what is the purpose of the 2 pads next to the power button?  It is the 2 pads above the trash can icon with the X over it.   I have searched for hours trying to find any reference to them.  But I did not find anything.  What are they for?

[de_light]

https://1.bp.blogspot.com/-xE3AIF-ATh0/W90WYaLbjSI/AAAAAAAAAek/r5R8etcv9_MeKlOloMBoaat3NAgicMSTgCLcBGAs/s1600/WP_20181102_20_27_24_Pro.jpg]https://1.bp.blogspot.com/-xE3AIF-ATh0/W90WYaLbjSI/AAAAAAAAAek/r5R8etcv9_MeKlOloMBoaat3NAgicMSTgCLcBGAs/s1600/WP_20181102_20_27_24_Pro.jpg]https://1.bp.blogspot.com/-xE3AIF-ATh0/W90WYaLbjSI/AAAAAAAAAek/r5R8etcv9_MeKlOloMBoaat3NAgicMSTgCLcBGAs/s1600/WP_20181102_20_27_24_Pro.jpg

Looking at this image, it appears that these connect directly to the quartz crystal on the underside. Possibly for measuring the frequency of it for a troubleshooting application?

[madires]

If you want to check the frequency of a quartz oscillator you would do that at the oscillator's output and not at the quartz because the additional load by a scope or whatever would change the frequency and might even cause the oscillator to stop running. But the XTAL1 pin can be also used as input for an external clock signal. Anyhow, those two pads are a little bit strange.

[indman]

On the model pictured below, what is the purpose of the 2 pads next to the power button?

I would think that Chinese friends provided for the installation of a quartz resonator in a different form factor and on the side of the display.

[pinguy]

This is an awesome tester, hats off to all involved in its development through the years!
Seeing all its great functions, I wonder if it could also have some sort of "ring tester" functionality (also known as Blue Ring Q Meter), to quickly find shorts in switching power supplies without desoldering the components, and also test various types of transformers for shorts.
Attached is the "classic" ring tester theory of operation, and schematic.
Can this be "emulated" some way with the ATmega? It would make this tool even more useful!

[indman]

pinguy,
in some cases it is sufficient to measure the inductance of the suspect coil and, if possible, compare the result with a similar one. A transistor tester can easily do this. The photo shows a small experiment in which the inductance was measured with an artificially made short of 2 adjacent turns. The difference in results is obvious.

[jleg]

Hi,

after quite a long time of successful usage i wanted to upgrade my beloved AY-AT tester, giving it a go with v1.40.
Since there were several recommendations for which tool chain to use, i made a little test.

I used good old WinAVR, AY-AT settings from "clones" file (with auto-powerdown, SW_SERVO, 30X32 symbols and the new E-series matching stuff for resistors), and replaced the tool chain binaries with

1) avr8 toolchain 3.4.5 (gcc 4.8.1)
was mentioned here some posts ago

2) avr8 toolchain 3.5.1 (gcc 4.9.2)
last one with gcc 4.x

3) avr8 toolchain 3.6.2 (gcc 5.4.0)
last one on microchip site

4) avr8 toolchain 3.6.2 + Arduino-IDE 1.8.13 avr binaries (gcc 7.3.0)
like 3), but with binaries replaced by IDE ones


TL;DR:
no. 4) did win  - that variant created a binary with less than 100% size (ended up at 97.4% after disabling SW_PWM_SIMPLE, SW_SQUAREWAVE and SW_IR_RECEIVER). See details below. I also attached the binary for the AY-AT, perhaps it's useful for some.

Btw, with a cheap USBASP, using WinAVR is really easy. Just copy over the binaries; and if you add "make upload" (and "make fuses" if needed) in "Tools->Options->Tools", there's even no need to fiddle with avrdude or cli at all... 

And of course - thanks for all the efforts put into this firmware!

avr8 toolchain 3.4.5 (gcc 4.8.1)
AVR Memory Usage
----------------
Device: atmega328
Program:   33140 bytes (101.1% Full)
(.text + .data + .bootloader)
Data:        252 bytes (12.3% Full)
(.data + .bss + .noinit)
EEPROM:      793 bytes (77.4% Full)
(.eeprom)


avr8 toolchain 3.5.1 (gcc 4.9.2)
AVR Memory Usage
----------------
Device: atmega328
Program:   33278 bytes (101.6% Full)
(.text + .data + .bootloader)
Data:        252 bytes (12.3% Full)
(.data + .bss + .noinit)
EEPROM:      793 bytes (77.4% Full)
(.eeprom)


avr8 toolchain 3.6.2 (gcc 5.4.0)
AVR Memory Usage
----------------
Device: atmega328
Program:   33050 bytes (100.9% Full)
(.text + .data + .bootloader)
Data:        252 bytes (12.3% Full)
(.data + .bss + .noinit)
EEPROM:      793 bytes (77.4% Full)
(.eeprom)


avr8 toolchain 3.6.2 + Arduino-IDE 1.8.13 avr binaries (gcc 7.3.0)
AVR Memory Usage
----------------
Device: atmega328
Program:   32794 bytes (100.1% Full)
(.text + .data + .bootloader)
Data:        252 bytes (12.3% Full)
(.data + .bss + .noinit)
EEPROM:      793 bytes (77.4% Full)
(.eeprom)

[pinguy]

@indman that was a good test, thank you, and indeed there's an important difference in the reading of the shorted coils! BTW, is the Q value shown on display the same Q as in the blue ring Q meter? i.e. Q=10 means 10 "rings"? I couldn't find it mentioned anywhere in the manual.