$20 LCR ESR Transistor checker project

[rfink07]

Thanks - the main reason I am confused is I measured ~10 components and they are remarkably consistent, all between 360-380, which differs from the datasheet

[indman]

I bought a variant of this product recently, mostly to measure transistor gain. I've measured several 2n3094 and 2n3096 components, and they all measure hFE ~ 370. If I'm reading the datasheets right, it shouldn't exceed 300 (and usually be much lower).
https://www.amazon.com/Mega328-Digital-Transistor-Resistance-Capacitance/dp/B07WT9VVZB#customerReviews
Am I missing something? I saw that there can be measurement errors for leaky parts (germanium), but haven't seen complaints about common Si parts.

You must clear understand that ONLY the original autors firmware installed on the device can to a greater extent guarantee the reliability of the results when measuring transistor parameters. And this is provided that the hardware of the tester is in good working order, the supply voltage of the microcontroller is well stabilized and clean. If your tester has original Chinese firmware installed, no one will give you any guarantees regarding the correctness of the results of testing various parts.

[YouCanDoIt]

If you buy a DIP version, you have better chance of getting a real 328.

but what is the difference between DIP and non DIP versions?

(Besides one being rectangular and larger whereas the other square and smaller), as we have been discussing on this thread, it seems the knock-off MCUs are being manufactured on small square encapsulations (cheaper), so that's why some of us think if you look for a DIP28 ATmega328P-PU, and if the vendor photo it's accurate, you have chances of obtaining a read one (with 32KB of flash, of course).

There is still the problem that an ATMega328 can't run a full feature firmware and its low accuracy. Given the low price of the controller, I am wondering if I could just replace the bundled ATMega328 with an ATMega1284 and flash it with the related firmware.

If you have a real ATmega, you can evaluate what changes you could make. Or, if you have an LGT8F328P-based unit, and you have a real ATmega, you could adjust the pinout and make the replacement.

Many thanks for your help. I will ensure to follow these guidelines if I will end up with a clone. I will post the parts I found and I would appreciate if you could ensure I will not do mistakes.

If someone can post some documentation source I would really appreciate. The PCB photo shows a non existent link to https://www.mikrocontroller.net/svnbrowser/transistortester/ and I'm wondering if this is a dead project. The firmware is heavily customized and I have read somewhere it is private. I am concerned about updates given the fact that the official ones regularly published will not work for sure.

kubi48, madires, and indman have posted documentation on different sites. madires has gathered many of that  (including an overview I composed) at github.

Thanks, the link helped me a lot. I have read all the stuff included the faqs and now I have a better idea on how things work. However, even after having followed all the links available in github and its pdf files, unfortunately I didn't found anything about Mega 4gsl.

[rfink07]

Update: I made a simple voltage measurement circuit for the transistors and it read... exactly the same as the meter.

[Feliciano]

Many thanks for your help. I will ensure to follow these guidelines if I will end up with a clone.

Thanks, the link helped me a lot. I have read all the stuff included the faqs and now I have a better idea on how things work.

Welcome, and please note there's a "Say thanks" button on the forum 

[Feliciano]

I've updated my FAQs. Attached.

[YouCanDoIt]

Many thanks for the updated Faq!
In the spirit of sharing I post a link where is possible to buy the parts for the MEGA 4GSL. Prices are good and should be available as DIY kit or professionally assembled. Revision should be the latest (2022) while the one I saw posted in the topic is 2015.

I am in doubt if I should go with a Mega 4GSL or an Hiland M644 (which I will customize with all the possible mods I can find here in the forum including the MCU replacement). Do you think they are at the same level or a Mega 4Gsl is still far better even than a modded M644?
What you would buy if you were me?

[Feliciano]

Continuing with my WIP clone, I found a major issue with one of the latest m324pa transformerless TC1 clones, limiting the OSHW mod posibilities: the manufacturer modified the push button behaviour so is no longer directly connected to the STC, but between two(three) transistors, where the power_lock is no longer triggered by the STC but from the ATmega itself (pin16 high on my m324pa), and the push button sends a low on pin42 of my m324pa. Therefore, the known "two vanilla transistors mod" to flash OSHW firmware is no longer valid for this model.

I attach a [WIP] diagram looking similar to the unit on my benchtop.

So what do we do now: create a new circuit for U8(STC) substitution, and/or create a new variant of the m-firmware config for this case? (I can test either).

In any case, a friendly warning for other Users until this is solved: please check what ATmega TC1 clone variant you have before attempting to flash the OSHW firmware.

[indman]

Feliciano, you have become the “lucky” owner of another FNIRSI product, congratulations!
The engineers of this company seem to have set the task of making a product that does not lend itself to such a simple upgrade to the original transistor tester design.
However, this remark applies to all FNIRSI products - for the most part, hastily put together low-quality products with the goal of getting the maximum benefit from the sale as quickly as possible and then forgetting about support and eliminating errors for the consumer.

My thoughts on finalizing this miracle of Chinese thought:
1. Remove STC and forget about it!
2. Try to improve the circuit using this option, which I showed below.
3. Check the assignment of the display contacts to the microcontroller, as they may also differ from the standard ones.
4. Of course, you will need to adapt the autors' firmware for this clone, since cutting the tracks and laying new conductors will be ugly.
But, if desired, this option can be implemented. Although, to be honest, I wouldn’t waste my time on this.

[carl1961]

GM328A up and running on 1.13k   Non removable MCU
Files I changed attached
mega328_color_kit Makefile and lcd_defines.h  This display is mounted the opposite of other older GM328A I had to reverse the settings in lcd_defines.h

How I did it I put on GitHub
https://github.com/carl1961/New-GM328A

[indman]

carl1961, What frequency is the quartz currently set to?
1.13k is set to 8 MHz, and 1.51m is set to 16 MHz. Does this tester work stably at 16MHz?
And one more note - during measurements, tests and work with the device, no conductors should be connected to the ICSP bus! This can greatly ruin calibration and test results.

[Feliciano]

Feliciano, you have become the “lucky” owner of another FNIRSI product, congratulations!

to be honest, I wouldn’t waste my time on this.

indman, I respect your opinion, but you should agree that it's not only FNIRSI, as most of "the chinese comrades" cut costs/corners wherever they can. So for most of us, unless we are lucky and get a very old stock, the chances for flashing the OSHW and enjoying all its benefits are more limited as time passes. At least, when rolling the dice, my luck got me an apparently genuine 44pin MCU.

So, I rather invest some time on this unit than putting it aside or thowing it away. But to convert this schematic to the original one it's not feasible, as the changes are too many. Therefore if @madires and you and other readers can please support this new mod, I will appreciate it, and I think it will be of benefit for other "lucky" owners too.

My thoughts on finalizing this miracle of Chinese thought:
1. Remove STC and forget about it!
2. Try to improve the circuit

As I don't want to cut/lay many traces: I haven't spent much time trying to understand the m-firmware code, I don't know whether the communication between PD1<->PD2 is needed (which could be acheived by a resistor). And about pulling PD1 low for reset or menu access, and PD2 high for triggering the boost converter, I don't know whether this can be achieved by another STC substitution circuit, and/or by connecting it differently, and/or by some tweaks on the code. In fact, by reading the power supply part of the similar schematic (which I'm adapting from an m328p TC1 clone), on my unit the press of the button is detected by a low on pin 42, answered with a high on pin 16 to trigger the boosters. Could it be to use the traditional auxiliary PCB, but connecting P1 to Q2c and P3 to Q3b (by an additional resistor) the way to go, ar at least part of it? As I said, I don't know the code, so I can't tell.

3. Check the assignment of the display contacts to the microcontroller, as they may also differ from the standard ones.

I will. In the meantime, I can tell the ISP is the standard one, and gets recognized by my USBasp as an m324pa.

[madires]

Continuing with my WIP clone, I found a major issue with one of the latest m324pa transformerless TC1 clones, limiting the OSHW mod posibilities: the manufacturer modified the push button behaviour so is no longer directly connected to the STC, but between two(three) transistors, where the power_lock is no longer triggered by the STC but from the ATmega itself (pin16 high on my m324pa), and the push button sends a low on pin42 of my m324pa. Therefore, the known "two vanilla transistors mod" to flash OSHW firmware is no longer valid for this model.

I think you can simply ignore the STC of your TC1 variant, i.e. no need for the mod. Or remove it, as indman suggested.

[madires]

So, I rather invest some time on this unit than putting it aside or thowing it away. But to convert this schematic to the original one it's not feasible, as the changes are too many. Therefore if @madires and you and other readers can please support this new mod, I will appreciate it, and I think it will be of benefit for other "lucky" owners too.

All what we need might be just the pin assignment of the ATmega324 to configure the firmware accordingly.

[Feliciano]

I think you can simply ignore the STC of your TC1 variant, i.e. no need for the mod. Or remove it, as indman suggested.

Thanks for replying. I had that gut feeling: the mod could be simpler. However, do the m-firmware?:

the press of the button is detected by a low on pin 42, answered with a high on pin 16 to trigger the boosters.

(I meant at the boot time. Of course pin 42 should be also read for menu access, as pin 16 should be high during operation).

On the other hand, I'm checking this soldered display, and uses a different pin mapping (from right to left bottom-view):
1=GND, 2=Vcc, 3=PB4 (44), 4=PB3 (43), 5=PB5 (1), 6=PB7 (3)
Is that what you meant by:

All what we need might be just the pin assignment of the ATmega324 to configure the firmware accordingly.

So what do you think? Shall we forget the auxiliary PCB, remove the STC (for power consumption reduction), adjust the config, make, and flash? Or something else needs to be adjusted?

[carl1961]

It is running with the 8 hz crystal , I removed the 16 hz when I was having screen issues after I read in Karl-Heinz manual where he said something about , Logic Gate Chips on some of these boards and 16 hz not working on them, as this board has a no name china 20 pin Logic Gate Chip.

[indman]

carl1961, This is exactly the nuance I wanted to clarify with you.
The use of an additional microcontroller, which acts as a command translator for BGR displays as well as for the 8-bit paralell display that is installed in your clone version, does not allow full operation at a frequency of 16 MHz.
For whom this frequency factor is important, you should refrain from purchasing such GM328A models or find and connect a standard classic RGB SPI ST7735 display directly to the ATMega328 ports without an extra chip.
Somewhat earlier, one of the participants in the Russian-language forum identified the marking of this 20-pin chip - HC89S003A.
A similar controller 24-pin HC89F0531 can also be installed.

[carl1961]

carl1961, This is exactly the nuance I wanted to clarify with you.
The use of an additional microcontroller, which acts as a command translator for BGR displays as well as for the 8-bit paralell display that is installed in your clone version, does not allow full operation at a frequency of 16 MHz.
For whom this frequency factor is important, you should refrain from purchasing such GM328A models or find and connect a standard classic RGB SPI ST7735 display directly to the ATMega328 ports without an extra chip.

I bought this thing before I even knew about the firmware.
Thanks for the advice too, I will remove the ISP jig when I am done playing with this. I still trying to get the 1.51m to work, seems all the reason for not working is the 7735 driver.
Yes ,  now that you gave that chip number, it is more readable LOL in this photo of another similar model as mine.

[jdev99]

Just received this one, with original ATmega328P. They are still out there...

Many thanks for the link. It seems to be a version of the GM328A different from the ones showed in the comparison chart. Isn't "M328Kit+TFT(AY-AT)" a better choice?

I don't know if it is better, it does have firmware version 1.12k

[NF6X]

I am new to the transistor tester world, and I'm missing some context and history. Why is that additional STC microcontroller in there? Is that a feature of any of the OSHW designs, or something that the clone manufacturers added? I only see it connected to two ATMega pins in Feliciano's draft schematic. If it's not driving any other hardware, I don't understand why they spent money putting it on the board instead of implementing whatever it does in ATMega firmware. I saw indman's comment about it translating LCD commands, but I don't grok its reason for being there yet.

[indman]

NF6X, STC is an 8-pin power management controller that Chinese engineers have implemented in some clone models.
And the 20-pin(24-pin) microcontroller is another direction of Chinese engineering. They use crappy cheap BGR displays and connect them to ATMega that way. It's clear? In both cases, these Chinese tricks only worsen the design of the original author's project, which is optimally designed with a minimum number of parts used.

[NF6X]

I'm having a hard time wrapping my brain around the idea that they spent money on a whole microcontroller to manage the power button instead of implementing it in the main microcontroller firmware for free, given how profit-motivated I expect the clone manufacturers to be. Does it normally have additional connections not shown in Feliciano's draft schematic?

[indman]

NF6X, using STC involved adjusting the time (from 10 seconds to 25 seconds) until the device was completely powered off.
But this time can be easily configured in advance and in the author’s project. So I don’t see any justification for using an additional controller. Maybe someone else can explain it to you better?

[Feliciano]

The short STC story:
The STC15L104 on the older TC1 and variants is used as a timing/power MCU. On the older factory firmware there was a communication between it and the MCU in order to hold the 5V, and for how long. It had high standby consuption (IIRC it was the 3.3V unit overdriven @VBatt-Vd). Later the manufacturers updated it with a better model. Later the manufacturers removed as by the schematic indman shared recently (and also replaced the transformer with a second booster). You can refer to the pages around May 2018 for the whole STC&TC1 discussion/details. Here I attach the original schematic as reference.

And that's what's puzzling me with my different STC-based TC1:
At the first press of the button, 5V is generated and the MCU will power up, with a low on pin 42. It shall reply with an steady high on pin 16 for keeping the 5V. If another low comes on pin 42, it shall be processed as menu enter (if held low, IIRC), or reset (if quickly returned to high). As I said, I don't know what normally m-firware does with these pins, so I don't know whether it can be adjusted on the config, or it had to be a different subrutine, or it can be accomplished by an external circuit taking Vbatt and 5V (from different places), reading the low from the push button from Q2 transmitting it to PD1, and reading the high on PD2 and transmitting it to Q3. I'm waiting for confirmation on whether the additional workaround is not needed, else, which shall be the better approach.

[NF6X]

The best hypothesis I've come up with so far is "firmware engineer's uncle owns the local counterfeit STC factory, and snuck the part into the tester design for a kickback".