$20 LCR ESR Transistor checker project

[ThinkerSan]

Hey everyone,

I’m trying to source the PCB files for a Mega-4GSL transistor tester 2022 or 2023 build.

Hey.
I created a schematic in KiCad for the Mega-4GSL (Revision A1, 2022) based on a board I purchased. In this version, the software SPI has been replaced with a hardware SPI, a hardware IR receiver has been added, and the encoder connection ports have been changed. The PCB (from my KiCad project) has not been verified for manufacturing, so you’ll need to verify it yourself for your manufacturer, such as JLCPCB.
The archive contains the project for KiCad version 10.

[ThinkerSan]

Hello, everyone.
While assembling the Mega-4GSL tester with a color display based on the ILI9341 controller, I ran into a problem with inverted colors. The ILI9341.h file contains #define CMD_INVERSION_ON and #define CMD_INVERSION_OFF, but these weren’t used anywhere. To fix the color inversion, I added a parameter to config_644.h:

Code: [Select]

#define LCD_INVERT              /* color inversion (needed for some panels, e.g. IPS variants) */ and modified ILI9341.c by adding the following to the initialization section:

Code: [Select]

  #ifdef LCD_INVERT
  LCD_Cmd(CMD_INVERSION_ON);       /* invert colors */
  #else
  LCD_Cmd(CMD_INVERSION_OFF);      /* normal colors */
  #endif
The attached files contain patches for these files. Maybe someone will find them useful.

[GraemeG]

That's a good one! These sources are good but for this project they are clearly redundant.
It is quite sufficient to use the original MCP1702-5002 or LDK320AU50R

After reading hundreds of pages of this thread, often 2 or 3 times, I finally noticed the only reference to the LDK320AU50R. Thank you indman!!

How did I not know about this device??? I am having fun rebuilding and improving 2 old TC1 units and could not find a high precision regulator that is pin equivalent to the 78L05 in a SOT-89 package. I managed to twist the leads of a TO-92 packaged MCP1702-5002 and that works but it is not pretty and I think it is adding to power supply stability problems on the original "transformer" version of the TC1. My search skills must be faulty, I had not discovered the LDK320 before but it is on the list for my next component order! It will be a much more elegant replacement for the low grade 78L05's 

[DisasterNow]

While repairing an electric fence (which I abandoned for safety reasons), I had to deal with a pair of SCRs.
Still wanting to test these two (presumably faulty) components, I also used our LCR T4: in one case, it detected a resistor, while in the other, a typical transistor (pins b, c, e). If an SCR were perfectly functional, the LCR T4 would detect it correctly (indicating SCR?). Do any of you have a working SCR on your bench that you could try and see what the display shows?
Thanks

[madires]

When the test current (max 7 mA) or voltage (max 5 V) doesn't trigger the SCR it can't be detected. Being detected as a resistor is a typical indication of an SCR with a higher trigger current. In this case you could use a lab PSU and a gate resistor to check the SCR manually.

[Yuriy_K]

Do any of you have a working SCR on your bench that you could try and see what the display shows?
Thanks

If your TTester has the K-firmware installed, you should see a picture like this...

For power thyristors, the readings depend on the connection order.

[DisasterNow]

Thanks Madires and Yuriy. I'm attaching two photos of how it detects the two SCRs (S4025L and Tyn606). As already mentioned, they could be faulty.
My model is original; I've never modified it, so I don't know what firmware is installed.
There wouldn't be any problems with the voltage (they require 1.5V), but the current required by the S4025 is at least 30mA, while the other one says 15mA maximum. So it obviously can't detect them correctly.
I'll test these two SCRs manually on the bench, and since I don't have an analog tester, I'll have to create a small circuit to see if the SCR is activated, and stays activated after the gate pulse.
Although the schematics I find online probably refer to SCRs with low IGT, and therefore not functional on my two components.

[Yuriy_K]

I'm attaching two photos of how it detects the two SCRs (S4025L and Tyn606).

S4025L - most likely working.
Tyn606 - check with a multimeter (if available) for possible damage.

[DisasterNow]

S4025L - most likely working.
Tyn606 - check with a multimeter (if available) for possible damage.

The SN4025 is probably not faulty.
I think the Tyn606 is faulty, because during the voltage test, seeing as the circuit's LEDs were indicating an anomaly, I hit it with the handle of the screwdriver, and it made a small bang (and a little smoke...). So I presume it's faulty.
Curiously, LCR T4 still detects its gate, precisely as the base of a transistor.
But that was precisely the opportunity to run the tests to determine the fault.
I have the multimeter, but it seems that the digital one isn't enough, and I also need to test whether the gate activates (and then leaves activated) the A-K channel. I've already tried with some test circuits found online, but they probably draw too little current to the gate, and it doesn't work. I'll have to do more experiments. Thanks 

[stj]

none of these testers are reliable at testing SCR's or TRIACS.

[DisasterNow]

As Madires wrote above, perhaps for SCRs with a max Igt of 6 mA, it should run the test and recognize it if not faulty.

---------------------------

I did, however, at least check the junctions with the digital multimeter:
.The Tyn606 and S4025 do not conduct on the A-K channel, both with forward and indirect bias (and this is fine).
.I then checked the G-K junction in forward bias: both SCRs conduct (and this is fine).
.I then checked the G-K junction in indirect bias: both SCRs conduct! And this shouldn't happen, so both my SCRs should be faulty on the G-K junction.
At this point, I didn't perform the conductivity test on the A-K channel, as they are faulty.
So these two SCRs are faulty, and LCR T4 detects them as in the previous images.

[Yuriy_K]

I managed to add built-in resistance measurements to the k-firmware. However, the problem is with the MEGA328's memory limits; I need to exclude either voltage or frequency measurements. Here's an example of measurements taken on a Hiland M644...

There are no problems with this, but there are problems with germanium transistors...

[indman]

I managed to add built-in resistance measurements to the k-firmware.

This is a useful addition, for the ATmega644-1284, you can even add a resistor symbol to the icon.

[GraemeG]

I'm learning my way around compiling and using 1.57m on my repaired TC1 and it is so much better than the badly mutated code the manufacturers supplied! I am having fun enabling and experimenting with the different options and have a few observations. None of these are critical but seem a little inconsistent with how various software options behave. I have no idea how the code works so it may not even be possible to make changes, or may increase the code size and I fully support keeping the code as small as possible.

I am happy for anyone to comment although these are probably questions for you madires and maybe they are worth considering as enhancement requests 

• I like that a long key press turns the tester off during component probe. But it does not work like this during the zener test, capacitor leakage, Diode/LED test, etc. I have a rotary encoder fitted so this behaviour may relate to navigation with a single button.
• Left turn of rotary encoder in zener test and Diode/LED test does not return to the menu. This is also very useful.
• The battery monitor or display of the battery voltage is not happening during zener, capacitor leakage or Diode/LED tests. The small LiPo batteries in TC1 and similar models can run down very quickly once they get below 3.6V and I have had them go well below the BAT_LOW level before I exited the test.
• The config.h comments say that the automatic power off set by POWER_OFF_TIMEOUT only works during auto-hold mode (I just noticed that 1.51m added it to the main menu). If auto-hold is not active then component probing will turn the tester off after CYCLE_MAX probes. But if a component is left connected, or the tester is left in the Diode/LED test, etc. then it seems the unit will stay on forever. Some tests and some users may want the tester to stay turned on forever but it would be nice to have the option.

These are all fairly trivial, just things I have noticed as I learn my way around the tool.

These testers are so useful. While I was experimenting I found a few VN2222 FETs that I had mixed with my NPN BJTs. I probably thought they were 2N2222s but the tester told me the truth! Once I have both TC1s working reliably I will load 1.13k firmware on one so that I have the best of both worlds 

[stj]

what processor is in your TC1?
they made several versions unfortunatly.

[madires]

• I like that a long key press turns the tester off during component probe. But it does not work like this during the zener test, capacitor leakage, Diode/LED test, etc. I have a rotary encoder fitted so this behaviour may relate to navigation with a single button.

A long key press is already used by several functions/tools to switch between modes or reset a value, besides its single button use.

• The battery monitor or display of the battery voltage is not happening during zener, capacitor leakage or Diode/LED tests. The small LiPo batteries in TC1 and similar models can run down very quickly once they get below 3.6V and I have had them go well below the BAT_LOW level before I exited the test.

The battery check is usually done about every 10 s when running functions/tools. When the voltage drops below BAT_LOW the display will show the battery status and the tester powers down after 3 seconds. Maybe the interval is too long for a small battery like in the TC1. You can change the interval in main.c, in function CheckBattery():

Code: [Select]

  Cfg.BatTimer = 100;              /* reset timer for next battery check (in 100ms) */
                                   /* about 10s */

Please try 20 or 10.

• The config.h comments say that the automatic power off set by POWER_OFF_TIMEOUT only works during auto-hold mode (I just noticed that 1.51m added it to the main menu). If auto-hold is not active then component probing will turn the tester off after CYCLE_MAX probes. But if a component is left connected, or the tester is left in the Diode/LED test, etc. then it seems the unit will stay on forever. Some tests and some users may want the tester to stay turned on forever but it would be nice to have the option.

It should be possible to add an option to power off after n probing cycles, no matter if a component was found or not. But this wouldn't work for functions/tools.

[GraemeG]

what processor is in your TC1?
they made several versions unfortunatly.

Both of my units originally had 324PA chips but I have replaced them with 644PA procs. I killed both 324 chips years ago and put them at the back of the cupboard until recently when I rebuilt them. It's great to have the OSHW firmware and enough flash to try extra functions!

Thanks madires. I'm not surprised about the long press. As I was writing the post I had a feeling it was not a practical suggestion.

I will try changing Cfg.BatTimer but I first noticed a problem when I had an old 1000uF electro in the Cap Leakage function. I was watching the Rl and Rh charge current for several minutes as it dropped, then increased after switching to Rh and then slowly fell again. Then switched to discharge and finally the moment I left the leakage function the Tester showed the battery voltage was 3.2V and turned off.
I have settings:

Code: [Select]

BAT_DIRECT
BAT_OFFSET 0
BAT_WEAK 3600
BAT_LOW 3400 and have changed the input resistor to TP_BAT (PA5) from 1M down to 100K. The measurement of the battery voltage seems quite accurate compared to my multimeter. I am sure this is made worse because my "300mAh" batteries are getting old and fade very quickly when they get low. I will test this further.

With the POWER_OFF_TIMEOUT I was wondering if a universal timeout could be created that will turn the unit off no matter what mode or function it is running. Any button press or encoder turn would restart the timer. I would probably set this to 5 - 10 minutes but this is not so important if I am sure the tester will turn off when the battery is low. Maybe this is a totally impractical suggestion.

[indman]

The battery monitor or display of the battery voltage is not happening during zener, capacitor leakage or Diode/LED tests. The small LiPo batteries in TC1 and similar models can run down very quickly once they get below 3.6V and I have had them go well below the BAT_LOW level before I exited the test.

Nothing prevents you from making the battery symbol display constant in these modes, similar to other modes (this can be easily added to main.c).

The config.h comments say that the automatic power off set by POWER_OFF_TIMEOUT only works during auto-hold mode (I just noticed that 1.51m added it to the main menu). If auto-hold is not active then component probing will turn the tester off after CYCLE_MAX probes. But if a component is left connected, or the tester is left in the Diode/LED test, etc. then it seems the unit will stay on forever. Some tests and some users may want the tester to stay turned on forever but it would be nice to have the option.

Nothing prevents you from enabling the auto‑hold mode and configuring the POWER_OFF_TIMEOUT parameter, for example, to 28 seconds, as done in the k‑firmware.
If you need a cyclic operating mode, simply hold down the Test button for a longer time at startup. Madires made a lot of additional settings in config.h, and they are quite sufficient for everyone to flexibly configure the firmware for themselves. We just need to take our time to figure this out.

and have changed the input resistor to TP_BAT (PA5) from 1M down to 100K

If you don’t have a mechanical power switch in the tester, this replacement will discharge the battery. The discharge current is, of course, insignificant, but it will still be 10 times higher than with a 1MΩ resistor.

[GraemeG]

You are right, I am still learning my way around the code. My programming skills are very rusty and I am more likely to break something because I don't understand it correctly. But I'm learning a lot as I go! And I haven't even started with the k-firmware yet. That will be a whole new experience 

I only discovered that the longer press at startup switches the autohold mode just a few days ago. A very clever and useful trick!

I added a mechanical switch to one of the units a long time ago because it had the problem with the U4 STC15L104 micro draining the battery. I have removed the STC chip from both so now I don't really need the switch. I found the battery voltage readings were not accurate or consistent with the 1M resistor and swapping to a 100K fixed that problem. I am guessing 1M was too high and the input capacitance of the ADC causing the voltage to drop. Maybe a better solution might be to put the 1M resistor back and add a small capacitor (maybe 10 or 100nF?) on the PA5 pin. Just something more to learn about how these microprocessors behave.

Thanks again. I appreciate everyone's help with this project.

[madires]

and have changed the input resistor to TP_BAT (PA5) from 1M down to 100K. The measurement of the battery voltage seems quite accurate compared to my multimeter. I am sure this is made worse because my "300mAh" batteries are getting old and fade very quickly when they get low. I will test this further.

You could replace the battery with a larger one. There's enough space in the case for 600 or 700 mAh.

With the POWER_OFF_TIMEOUT I was wondering if a universal timeout could be created that will turn the unit off no matter what mode or function it is running. Any button press or encoder turn would restart the timer. I would probably set this to 5 - 10 minutes but this is not so important if I am sure the tester will turn off when the battery is low. Maybe this is a totally impractical suggestion.

Should be feasible too. Maybe an option for ATmegas with 64k flash or more.

[stj]

my "300mAh" batteries are getting old and fade very quickly when they get low.

go for a walk, look down.

disposable vapes are the answer.
even the smallest are 350mAH, the big ones are 1000 or even 1800mAH!!!
a "Lost Mary 6000" with the charge display is 1000mAH and very common to find.

[MLXXXp]

disposable vapes are the answer.
even the smallest are 350mAH, the big ones are 1000 or even 1800mAH!!!

Just be aware that all the cells I've found in discarded vapes don't have a protection circuit. However, if you're using a vape cell to replace a failing cell, as @GraemeG might, you could use the protection module from the failing cell, if it has one.

[GraemeG]

disposable vapes are the answer.

Great idea! I am happy to say that I don't see a lot of vapes lying around here but there are some. New batteries are on my list but not urgent. The old batteries are running down quickly because I am spending a lot of time experimenting with the "real" firmware.

I also had a problem with my early style TC1 that has the transformer in the boost converter. When I swapped to the MCP1702 regulator the SMPS started oscillating at about 3-4kHz and would draw peaks of 800mA! I lowered the output of the boost converter from ~9.3V down to about 6V and added a better capacitor between the boost and the 1702. That fixed it. The down side is the zener test voltage dropped from about 47V to 31V but I can live with that.

The transformer version is still not very efficient. With the tester sitting idle at the menu it draws about 150mA from the battery. The other tester is a FNIRSI 4A10 board with two B628 boost converters and it only draws 83mA when at the menu. And both of those measurements are with a freshly charged battery at ~4V.

[stj]

without checking the TC1 schematics i presume the boost circuit isnt switched!

[GraemeG]

Correct! The transformer model uses a centre tapped secondary. The middle tap supplies the raw supply for the MCP1702 and the full winding feeds the zener supply. No easy option to switch the zener supply.

The zener supply on the 4A10 board is not switched either but it could so easily have been done and is on my potential enhancement list. The two B628 PWM chips are fed from the battery in parallel via a FET switch. But the zener supply's Enable pin is pulled high by a connection to the chips V+ when it could so easily have been connected to a pin on the processor (probably needs a resistor and diode to protect the enable pin but still simple). It all works but maybe one day I will "fix" it.