$20 LCR ESR Transistor checker project

[TechieTX]

Odd observation...
I've been using an AY-AT with the accuratizing parts and 1.13K for a couple of weeks while I tried to figure out how to build the 1.32m firmware.  Building 1.32m was WAY more painful than it needed to be!  Anyhow, the 1.13K has worked fine on everything I've tried, and it identified some bad 20-year-old MOSFETs I'd been keeping for future projects.  Finally this evening I worked out the last of the problems with building 1.32m and started testing, after calibrating it.  Mostly OK, although I noticed it's more variable on repetitive readings on many of my ceramic caps.  A quality film cap might read 960-963pf, but a ceramic would read anywhere from 9616pf to 10.7nf (about a 10% total range), and it never stabilized.  Aluminum electrolytics and tantalums?  Still within < 1% range on successive reads, but the ceramics I have are all over the map.  1.13K with that same 10nf ceramic only hit a range of 10.06 to 10.16, or around 1% variance with multiple measurements.

Looking at the results after adjustment, that VCC is wrong.  1.13K is reporting 5.00V for VCC and my Fluke and Wavetek DMMs both read 5.00V; that 5.025 reading may have something to do with the inaccuraty.  I haven't tried recompiling to use the LM4040AIZ reference yet; that may cure the issues with the ceramics and/or varying results.  I'd left it remarked due to comments about the external ref being useless if you have a precision LDO regulator.  Maybe not so much.

[madires]

Didn't the settings listed in the Clones file work? If so, please PM your settings for updating the file. The asterisk after Vcc indicates that an external reference is used. Since Vcc is exactly 5V please disable HW_REF25 and try the ceramic caps again.

[VEGETA]

So what is the latest\best version of this thing? I know the banggood one is famous, is it the one?

thanks

[JonnieCache]

So what is the latest\best version of this thing? I know the banggood one is famous, is it the one?

It's hard to tell with some of them. Look for ones that specifically say AY-AT on the board. However, one of the ones that I bought showed AY-AT in the photo, but not on the actual board when it arrived. It looks the same to me, but I wonder if it is somehow inferior. But, you are looking for the one with the 3 wire terminals and the rotary encoder. Those are the latest boards.

[TechieTX]

Well, I've beat it to death by now, and got everything in 1.32m to compile without any horseplay.  Something had been enabled that was sending out a 5us HIGH pulse on PD3, which wiped out reading the encoder (see the photo of the PD3 pin).  I worked back and forth between the (heavily modified) config.h and config_328.h I got working compared to the distribution file and got it going without much modification.  The ZIP file has the Makefile, config.h and config_328.h for the AY-AT with the precision LDO and resistors and the original 8MHz crystal.  I've only tested the compile with WinAVR as that's all I have installed.

I still get erratic readings (10% variance) with cheap ceramic caps, stable with mica, film, aluminum electrolytics and tantalums, so I'm OK with it having 10% variation on cheezy caps that have over 110 ohms ESR.  The other features that I tested work fine, so it's a GO for me.  I did disable the 2.5V reference chip, and it didn't make any difference.  I'd added some extra grounding to my board since the layout for ground is absolutely pitiful, but removing my extra grounds didn't change anything.

I modified the Makefile to use my USB ASP programmer under WinAVR; your configuration may vary.  I used the -F switch (ignore ID bytes) with avrdude 'cos most of my testing was with an ATMEGA328 (no P), but the code works with 328 or 328P, no difference as far as I could tell.  The last two pics are after an adjustment cycle with both 328 and 328P.

One suggestion: change the configuration in your standard config_328.h file to match the clone settings for the AY-AT.  The default settings won't run the ST7735, which causes trouble for anyone with an AY-AT.  Since it's the most popular board with that display, it makes sense to have the defaults match the AY-AT.  Fewer complaints of "I did the build and my display doesn't work!" is a Good Thing.    I'd also recommend the #define ENCODER_A        PD2 to be #define ENCODER_A        PD1 for the same reason.  If > 70% of the people here are using an AY-AT, then the defaults in the code should match the most common configuration. (I'm guessing on that 70% from the posts over the last 8 months).

EDIT:  I recompiled for 16MHz and inserted a 16 meg crystal, and the inaccuracy with the ceramics doesn't happen, for whatever reason.  The readings are stable, and roughly the same as what I remember Karl-Heinz' version reading.  I'll do 1.13K next with the 16M crystal and update the results.
Update twice: I had to gut the menu out of 1.13K 'cos it was compiling to 112% of flash, and I couldn't find the wasted space where the bootloader was being added in.  It was also unstable with the cheap ceramic caps at 16MHz, although Karl-Heinz did say that measurement accuracy would be effected in his code with a 16MHz crystal.

Thanks for all your work, everyone!

[VEGETA]

I modified the Makefile to use my USB ASP programmer under WinAVR; your configuration may vary.

Where are the files hosted? I didn't think it is true open source though.

It's hard to tell with some of them. Look for ones that specifically say AY-AT on the board. However, one of the ones that I bought showed AY-AT in the photo, but not on the actual board when it arrived. It looks the same to me, but I wonder if it is somehow inferior. But, you are looking for the one with the 3 wire terminals and the rotary encoder. Those are the latest boards.

I was speaking about this one:
https://www.banggood.com/DANIU-3_5inch-Colorful-Display-Multi-functional-TFT-Backlight-Transistor-Tester-p-1083042.html

I cannot tell if it is latest edition but it has nice enclosure which is important for me. ZIF socket is there as well.

There is this one too:
https://www.aliexpress.com/item/Multi-purpose-Transistor-Tester-128-160-Diode-Thyristor-Capacitance-Resistor-Inductance-MOSFET-ESR-LCR-Meter-TFT/32812264664.html

But I don't know which one is better as there is no comparison sheet between them. I wish someone makes a shootout for these.

It is worth mentioning that there is this one too for SMD components:

https://www.banggood.com/BSIDE-ESR02PRO-Digital-Transistor-SMD-Components-Tester-Diode-Triode-Capacitance-Inductance-Meter-p-1218985.html

Eventually, we are lucky to have such very cheap good and working devices. Otherwise, we'll be paying $$$.

[JonnieCache]

I modified the Makefile to use my USB ASP programmer under WinAVR; your configuration may vary.

Where are the files hosted? I didn't think it is true open source though.

Files are here and are open source: https://www.mikrocontroller.net/svnbrowser/transistortester/

I was speaking about this one:
https://www.banggood.com/DANIU-3_5inch-Colorful-Display-Multi-functional-TFT-Backlight-Transistor-Tester-p-1083042.html

I cannot tell if it is latest edition but it has nice enclosure which is important for me. ZIF socket is there as well.

There is this one too:
https://www.aliexpress.com/item/Multi-purpose-Transistor-Tester-128-160-Diode-Thyristor-Capacitance-Resistor-Inductance-MOSFET-ESR-LCR-Meter-TFT/32812264664.html

But I don't know which one is better as there is no comparison sheet between them. I wish someone makes a shootout for these.

It is worth mentioning that there is this one too for SMD components:

https://www.banggood.com/BSIDE-ESR02PRO-Digital-Transistor-SMD-Components-Tester-Diode-Triode-Capacitance-Inductance-Meter-p-1218985.html

Eventually, we are lucky to have such very cheap good and working devices. Otherwise, we'll be paying $$$.

I didn't realize you were asking about a completed product to purchase rather than a DIY version of the tester. I don't know if anyone can give you the information you are seeking for pre-built testers. To me, it appears that many companies are making them so you can't be guaranteed even which one you will get other than that they will look the same. However, the quality of the parts inside could be very different.

I'm not saying that those testers are bad, but you may not know what you are getting until you receive it. My conclusion when it came time for me to buy one was to build my own because that would be the only way I could make sure I was using quality parts. The resistors, voltage reference, and a couple other parts are crucial to having a good tester. It took me a long time, but I finally have one that appears to work very well.

The testers you have linked to do not have any rotary encoder so they would not be the latest iteration of the tester. However, if you really want to buy a pre-built tester, you should look for one that looks like the one attached below.

This layout has a rotary encoder, ZIF socket, 3 wire terminals, and SMD tester. I have not seen one like this that comes in a more polished, more commercial-style enclosure, but you can find them on eBay and other sites pre-built with an acrylic enclosure. For example: https://www.ebay.com/itm/Transistor-Tester-LCR-Diode-Capacitance-ESR-meter-Signal-Generator-MOS-Triode-/172845809087

The downside to many of the enclosures I have seen and actually tried is that some of them don't leave enough room to use the ZIF socket lever easily and cover up the SMD tester if that is important to you.

As for the shootout idea, while it would be great, I don't think it is possible because you never know which company made the tester just from the web site. I would be wary of buying one pre-made. You may find that to get good testing results, you have to tear it apart and replace various parts anyway.

I hope this helps.

[Twenty4Pack]

Quick comment on the ZIF:  I put a short length of heatshrink tubing on the ZIF bar - left the end on - to get around the acrylic cover's interference issue.

[lypse]

Hi guys,

I am wondering, is it possible to buy an official tester with good parts and decent case recent firmware ect..?

I don't mind soldering if its sold as a kit.

[madires]

There isn't any official model, but the AY-AT type clones are nice.

Quick update on the m-firmware: The framework for a very simple command parser is done.

[VEGETA]

Hi guys,

I am wondering, is it possible to buy an official tester with good parts and decent case recent firmware ect..?

I don't mind soldering if its sold as a kit.

I want that too, since the ones in Banggood and Aliexpress have nice enclosure.

[bitseeker]

Having occasionally seen posts requesting a high quality, tested, and/or fully featured version of this tester, I'm curious what kind of price would be acceptable to those who want them. A kit or even fully assembled tester with nice case can certainly be made. It's not a technical hurdle. But making them in small quantities will cost much more than the typical ones that can be ordered from China.

I haven't done a cost analysis to know what such a tester would actually cost to make. Hypothetically, though, instead of being $10-20, if a "good" tester cost $50, $75, or even $100, would that still be desirable or would you just purchase an existing commercial tester (many of which don't have as much functionality in a single device)?

[lypse]

I guess at 100$ it should feel like a similar priced multimeter, case, button, screen, test leads and so on.. maybe styled as one of the cute pocket meters?

at 50$ it should be like the current clones with a "decent" case, but good components

Firmware updates should not require much coding skills if any.. (unless the user wanted to)

Where would I see the AT-AY testers mentioned? is it the version with output terminals and rotary encoder?

Having occasionally seen posts requesting a high quality, tested, and/or fully featured version of this tester, I'm curious what kind of price would be acceptable to those who want them. A kit or even fully assembled tester with nice case can certainly be made. It's not a technical hurdle. But making them in small quantities will cost much more than the typical ones that can be ordered from China.

I haven't done a cost analysis to know what such a tester would actually cost to make. Hypothetically, though, instead of being $10-20, if a "good" tester cost $50, $75, or even $100, would that still be desirable or would you just purchase an existing commercial tester (many of which don't have as much functionality in a single device)?

[bitseeker]

I guess at 100$ it should feel like a similar priced multimeter, case, button, screen, test leads and so on.. maybe styled as one of the cute pocket meters?

That makes sense from the customer's perspective. I wonder, though, what the required manufacturing volume would be to be able to accomplish it. Likely quite a financial commitment.

at 50$ it should be like the current clones with a "decent" case, but good components

That sounds like a feasible middle ground that might be able to be accomplished without major upfront capital investment.

Firmware updates should not require much coding skills if any.. (unless the user wanted to)

I'm not sure how much additional hardware and/or software would be required to be able to update it without installing dev tools, hooking up an ICSP cable, etc. It would be a worthy enhancement from a product perspective, though.

Where would I see the AT-AY testers mentioned? is it the version with output terminals and rotary encoder?

By "mentioned," do you mean how do you make a positive identification? The "AY-AT" lettering is generally silkscreened in small font near one of the edges of the PCB with some other identifiers. The ones I usually see use a red PCB with three separate two-pole terminal blocks and rotary encoder.

[texaspyro]

I'm not sure how much additional hardware and/or software would be required to be able to update it without installing dev tools, hooking up an ICSP cable, etc. It would be a worthy enhancement from a product perspective, though.

Fairly trivial if it includes a serial port and bootloader...

[bitseeker]

Yeah, it'd need space for a bootloader. The 328 is full with all that's possible in the m firmware.

[robca]

Yeah, it'd need space for a bootloader. The 328 is full with all that's possible in the m firmware.

Methinks that the biggest issue for a newbie is figuring out how to build and which binary to flash.

If someone were to make a ready-to-go version and provide a single repository of binaries, ready to be flashed with a USBASP, the issues that most new users face would almost completely disappear. Provide a simple set of flashing instructions for a specific board, and a repository with the latest, tested binary for that

IMHO, a bootloader would not be the most critical element. The main challenge for someone new is figuring out what binary is safe to flash, given the immense variety of boards and firmware combinations

[timelessbeing]

If someone were to make a ready-to-go version and provide a single repository of binaries

You mean, like on the very first page?

Message #4

Flashing instructions should be all over the internet.

[VEGETA]

current "clones" are ok but you here have latest firmware so that is the addition, not to mention better parts. So the enclosure is the same while the firmware is updated.

If it was 30-40$ only then it would be nice.

[madires]

Yeah, it'd need space for a bootloader. The 328 is full with all that's possible in the m firmware.

Karl-Heinz has modified optiboot to program the EEPROM too (the original version does only the flash memory): https://github.com/svn2github/transistortester/tree/master/bootloaders. And the ATmega644/1284 based tester with more flash and I/O pins allows us to have a dedicated hardware USART. Possibly will see a tester with bootloader in the future

[madires]

IMHO, a bootloader would not be the most critical element. The main challenge for someone new is figuring out what binary is safe to flash, given the immense variety of boards and firmware combinations

Karl-Heinz already offers compiled firmware (k-firmware) for a bunch of different tester models. For the m-firmware there's the Clones file which lists the settings for a few models. If you got a tester not listed please send me your working settings and I'll update the file. We try to provide support for most models, but we're also just humans with limited resources. Maybe I've given you an idea on how to help

[JonnieCache]

Having built a couple of these already, I think it would be a challenge for someone to build an affordable device, with an enclosure and appropriately spec’d components. For example, the high-quality resistors for the testing portion of the AY-AT circuit total $8 all by themselves (full retail, of course). The reference voltage transistor is $1.72. That’s already $10 in parts.

It seems to me that unless these things are built in seriously high volume, it would not be possible to build a quality kit or completed product including a nice enclosure for even $50 with a decent profit margin.

[robca]

IMHO, a bootloader would not be the most critical element. The main challenge for someone new is figuring out what binary is safe to flash, given the immense variety of boards and firmware combinations

Karl-Heinz already offers compiled firmware (k-firmware) for a bunch of different tester models. For the m-firmware there's the Clones file which lists the settings for a few models. If you got a tester not listed please send me your working settings and I'll update the file. We try to provide support for most models, but we're also just humans with limited resources. Maybe I've given you an idea on how to help

I know, and I found everything I need thanks to all the folks who did the work. But, based on the questions in the long thread, there are many people who have problems finding "the right" device to buy and then struggle figuring out which binary to flash. Unless you are 100% sure that you have the right build for a device, there's always a risk that flashing it would make it inoperable. I bought a blank 328p to flash, just to be on the safe side and be able to go back to factory image if needed, even if I was pretty confident

My point was that, if anyone were to offer a pre-assembled device with a simple repository of builds (actually, just one or two), it would go a long way towards solving the problem. Going ever further and offering a bootloader (which is what I was replying to) won't offer that much more value

The real problem, as others are pointing out, is that any decent quality kit, once the cost of the better components, assembly and shipping are taken into account, would be too expensive for most users (or has to be sold at cost, making it a bad proposition for whoever assembles it)

[bitseeker]

Yeah, it'd need space for a bootloader. The 328 is full with all that's possible in the m firmware.

Karl-Heinz has modified optiboot to program the EEPROM too (the original version does only the flash memory): https://github.com/svn2github/transistortester/tree/master/bootloaders. And the ATmega644/1284 based tester with more flash and I/O pins allows us to have a dedicated hardware USART. Possibly will see a tester with bootloader in the future

Ah, I didn't know about the optiboot. I was thinking more about the larger ATmegas as you mentioned (644, 1284, 2560, etc.).

[bitseeker]

Having built a couple of these already, I think it would be a challenge for someone to build an affordable device, with an enclosure and appropriately spec’d components. For example, the high-quality resistors for the testing portion of the AY-AT circuit total $8 all by themselves (full retail, of course). The reference voltage transistor is $1.72. That’s already $10 in parts.

It seems to me that unless these things are built in seriously high volume, it would not be possible to build a quality kit or completed product including a nice enclosure for even $50 with a decent profit margin.

Exactly. Just guesstimating, even at modest volumes for parts, yields a pretty high overall cost. That's why I was just wondering what people's expectations were as far as price. Such an endeavor is only feasible if the market can bear it.