$20 LCR ESR Transistor checker project

[Fenichel]

Thanks again for your help.

When you press the Test button, does the display backlight come on?

Yes, it does.

You can try to update the ATMega328 firmware located here
https://www.mikrocontroller.net/svnbrowser/transistortester/Software/trunk/mega328_st7565_kit/

I'll try that, but I'm pessimistic.  The fact that the contrast took a few minutes to go away, instead of being lost all at once, makes me believe that it is more likely to be a hardware trace giving up its last thread than a cosmic ray fouling the software.

[indman]

I'll try that, but I'm pessimistic. 

Hope dies last! At the very least, you will know that you have checked all possible causes of the malfunction.
A few more tips - look carefully at how the flat cable that leads to the display is sealed. Very often, simple soldering brings the display to life. Also check if any standard LED blinks if you insert it into test contacts 1-2 (1-3) and press the Test button.

[Fenichel]

look carefully at how the flat cable that leads to the display is sealed.  Very often, simple soldering brings the display to life.

Your remark is apt, but the cable's soldering looks OK, even under magnification.

Also check if any standard LED blinks if you insert it into test contacts 1-2 (1-3) and press the Test button.

Yes, it does.  I think that there must be a hardware failure in the innards of the display.  I'll get a replacement to see if that works.

[Dumidan]

Can't that Zenerboost be used for TRIACS as well?

NOT! Because the test component is connected to and powered from the MCU ports, so with voltages between 0V and 5V. The testing of zenner diodes is done with a voltage of 30-40 V generated by an external MCU boost converter; the voltage is measured by resistive divider, so as not to exceed 5V on the MCU input. They are physically separate circuits.

If you have knowledge of electronics, study the original project documentation - you can find it here: https://www.mikrocontroller.net/svnbrowser/transistortester/Doku/trunk/pdftex/english/

You will need to download all the content > click on Download GNU tarball. For unzip I recommend 7zip - or any program that can handle .tar and .gz archives.

[Maniaxx]

Sorry, I don't have a guide about creating fonts. As a starting point please see bitmaps/font_8x8_hf.h. I think the GLCD Font Creator is able to create the right bitmap format (export as TFT / new GLCD). For linux/unix users I've written a small command line tool to convert bit and byte orders of a character bitmap (hexadecimal input and output).

Is it possible to convert the fonts back from C to bmp to get an overview of all bitmaps?
Where can i find your tool to convert bit/byte order?

[madires]

Is it possible to convert the fonts back from C to bmp to get an overview of all bitmaps?

You could enable SW_FONT_TEST to see the complete font or write a small tool.

Where can i find your tool to convert bit/byte order?

I haven't published it yet. Would you like a copy (no docs, besides the source code)?

[Maniaxx]

Actually i wanted to create a comparison sheet to get a glimpse how all the fonts look like. But i probably have to go for 8x8 fonts/24x24 symbols on GM328A anyway to get enough room to put all the fun stuff into it (8x8 usually looks best anyway). So maybe its not necessary. I don't have a device yet though. Did you ever consider adding RLE compression for the bitmaps?

What do you think about sound feedback? Those little pc speaker tweeters are multifreq and cheap. Static resistor ranges could be mapped to predefined frequencies so a detected 4.7kOhm would always give the same freq beep (say 400Hz), 10kOhm maybe 800Hz so you could identify them (or get a feeling in what range you are) even before looking at the screen. Cap charging could be reflected in realtime (sweeping up sound) so one could actually hear the inner workings of the device. Just a wild thought though. Just reminded me on the old 8Bit computers where we sent the uncrunching data (beside copying it to ram where it belonged) to sound chip to hear the data. You could hear load errors/bad data or if you exceeded ram area. It was a nice debugging feature and cool to hear the true machine workings.

[madires]

I had some thoughts about compressing the bitmap data, but in this case there are more drawbacks than benefits. The compression would even increase the bitmap size. When I get a new toy, the first thing after unboxing is to disable acoustic feedback. Some like it - I hate it. Anyhow, several measurements and functions are time-critical and won't allow any additional fancy beeping. But it can be added to others, and you need a dedicated PWM output.

[jotka13]

Hello everybody,

I am new in this forum, so far I always looked at Mikrocontroller.net for the Transistor Tester AVR. Because of the last posts there, I have now also become aware of this forum here.

It seems to me that it is even more extensive than the forum on Mikroontroller.net. The English language does not make it easier for me.

I have now searched here for layout designs for PCBs to reproduce for the different versions of the transistor tester, but unfortunately I have not found them yet.

Maybe you have one or the other hint for me, if I can find them here at all, which ones are already available here or how I should have searched to find them myself.

Many thanks to you already!

[madires]

You'll find examples for PCBs at https://www.mikrocontroller.net/svnbrowser/transistortester/Hardware/ or https://github.com/madires/Transistortester-Warehouse/tree/master/Hardware.

[Maniaxx]

And there's Indman's comparison table for the clones. Scroll down two posts. Not really "PCB layouts" though in case you want true templates for DIY.

[rhb]

FWIW  A suggested FW enhancement.

I am trying to learn RF amplifier design.  A key requirement for estimating the input impedance is the base spreading resistance and the forward biased base emitter resistance. 

I have means for measuring this using my 34401A, but it would be nice if the LCR tester software would report  rbb and rbe or rbb + rbe in addition to Vbe and Ce, the emitter diffusion capacitance, as currently reported by my BSIDE ESR02 Pro.  For details c.f. "RF Circuit Design" by Chris Bowick p 99 (1st ed) or p 105 (2nd ed).

My Peak Atlas LCR45 also reports complex impedance and other information if you press a button after the initial reading.

Part of my project is to compare the accuracy of HF designs using DC or low frequency measurements to VNA results and actual amplifiers built using the measured transistors.  Though widely used at HF, general purpose transistors such as the 2N3904/6 and 2N2222 don't give S or Y parameters for HF

Have Fun!
Reg

[uski]

what is the advantage of the crystal? Do you get more accurate results?

No, I did not notice any significant difference in the stability and accuracy of the readings between the ceramic and quartz crystal.
The main advantages of a ceramic crystal are small dimensions, there is no need for precise adjustment after installation, and also fewer parts on the board.
But the frequency at which the crystal operates has a noticeable effect on the measurement results.
Previously, my AY-AT had a 20MHz crystal. I noticed that the tester was unstable at this frequency.
Most of the Chinese-made ATMega328 processors are very unstable at 20MHz. This is especially noticeable when measuring the ESR of capacitors.
At 16MHz, these same processors work much more stable and predictable. 

Could this be related to the ADC clock ?
An unstable CPU would create many other issues than just problematics measurements.
However, an overclocked ADC can totally explain unstable measurements but an otherwise stable system

Have you tried to reduce the ADC clock with the 20MHz crystal and see if it solves the problem ?

[indman]

Have you tried to reduce the ADC clock with the 20MHz crystal and see if it solves the problem ?

I already wrote before that all measurement problems are solved by setting the clock frequency to 16MHz.
I am not interested in other experiments with ADC.

[madires]

I have means for measuring this using my 34401A, but it would be nice if the LCR tester software would report  rbb and rbe or rbb + rbe in addition to Vbe and Ce, the emitter diffusion capacitance, as currently reported by my BSIDE ESR02 Pro.  For details c.f. "RF Circuit Design" by Chris Bowick p 99 (1st ed) or p 105 (2nd ed).

I can't promise anything, but I've noted down your suggestion.

[madires]

Could this be related to the ADC clock ?
An unstable CPU would create many other issues than just problematics measurements.
However, an overclocked ADC can totally explain unstable measurements but an otherwise stable system

Have you tried to reduce the ADC clock with the 20MHz crystal and see if it solves the problem ?

The ATmega's datasheet states that the maximum ADC clock is 200 kHz for a resolution of 10 bits. For an MCU clock of 8 or 16 MHz the ADC clock is set to 125 kHz, and for 20 MHz it's set to 156.25 kHz. Since the ADC prescaler can't go higher than /128 it's not possible to lower the ADC clock for an MCU clock of 16 or 20 MHz any further. And it doesn't matter much as long as the ADC clock stays lower than the supported maximum.

Another point to note is the maximum MCU clock. IIRC, some ATmegas max out at 16 MHz, also the new 328 types which don't have the "full swing crystal" option anymore. If you have any problems with 20 MHz move to 16, as indman recommends.

[Maniaxx]

1) When upgrading a GM328A to 16MHz is just the crystal to be swapped? Nothing else (beside software)?
2) Do i have to swap the crystal on the dev board as well?
https://www.aliexpress.com/item/32657285482.html (ATmega8 ATmega48 ATMEGA88 Development Board AVR)
3) Is the ISP/dev board flash method able to rescue the chip in case the fuses were set wrong? I can't remember the name anymore but if i reckon correctly there is another flash method for avr (that is able to flash the chip in all cases). The flasher i want to get is this:
https://www.aliexpress.com/item/32670511994.html (USBASP USBISP AVR Programmer ATMEGA8 ATMEGA128)

[indman]

1) When upgrading a GM328A to 16MHz is just the crystal to be swapped? Nothing else (beside software)?

Yes.

2) Do i have to swap the crystal on the dev board as well?
https://www.aliexpress.com/item/32657285482.html (ATmega8 ATmega48 ATMEGA88 Development Board AVR)

No.

3) Is the ISP/dev board flash method able to rescue the chip in case the fuses were set wrong?

No.

[Maniaxx]

Thanks. Just for reference, the other method is high voltage programming and seems to be a much more complex process (usually not necessary indeed). I will stick with the standard ISP.

1) Why are the GM328A DIY kits more expensive (9.80Eur DIY vs 5.80Eur SMD)? Any specific reason? My pick would be the pre-assembled.

2) I can't see it on the pictures. The pre-assembled version needs 16MHz SMD (not DIP) quartz crystals, right? I mean, the Atmega is still a DIP. They could have put the quartz up on purpose as DIP as well for easier modding.

3) None of the sellers explicitly mention the calibration parts (short/loop cable, resistor, capacitor) or the 9V battery buckle. Did anyone get them along with the GM328A?

[Dumidan]

1) Why are the GM328A DIY kits more expensive (9.80Eur DIY vs 5.80Eur SMD)? Any specific reason? My pick would be the pre-assembled.

The DIY variant has THT components, while the SMD variant has SMD components; THT components are more expensive than SMD ones, hence the price difference.

2) I can't see it on the pictures. The pre-assembled version needs 16MHz SMD (not DIP) quartz crystals, right? I mean, the Atmega is still a DIP. They could have put the quartz up on purpose as DIP as well for easier modding.

The quartz crystal will also be 8 MHz, this is the designed scheme and the Chinese manufacturers respect it exactly. If it is the SMD variant already assembled, I expect the MCU to be also SMD, not DIP. Chinese manufacturers do not worry at all that the user would like to make moddings, they choose the cheapest option of production.

3) None of the sellers explicitly mention the calibration parts (short/loop cable, resistor, capacitor) or the 9V battery buckle. Did anyone get them along with the GM328A?

Those who sell GM328 are not the ones who produce them, most of the time they have no idea what they are selling !! So .... hence the setup instructions!

[Maniaxx]

The quartz crystal will also be 8 MHz,

I want to upgrade it from 8 to 16MHz. Thus i need to know what to buy (THT or SMD).

I expect the MCU to be also SMD, not DIP.

I don't think so. Look at the pictures below. Even though you can just see the backside you can clearly see 2x14 THT pins. That should be the MCU.
https://www.aliexpress.com/item/33036516954.html

1) Is the quartz SMD or THT?
3) Is the MCU socketed already?

Edit: Found this. Looks like both are THT and MCU is socketed already.

[indman]

Maniaxx
Chinese clones GM328A (B) may have several configuration hardware. In the photo that I attached, you can see the SMD MCU.
If this is important for you, then check with the seller which option will be sent to you after purchase.

[madires]

I have means for measuring this using my 34401A, but it would be nice if the LCR tester software would report  rbb and rbe or rbb + rbe in addition to Vbe and Ce, the emitter diffusion capacitance, as currently reported by my BSIDE ESR02 Pro.  For details c.f. "RF Circuit Design" by Chris Bowick p 99 (1st ed) or p 105 (2nd ed).

I can't promise anything, but I've noted down your suggestion.

I've looked into the suggestion. AFAICS, the Transistortester's hardware isn't able to measure those parameters.

[Szybkijanek]

So problem is when put 7-9MHz and 15-17MHz HF crystal, tester show wrong value (300-800kHz) and when put osc. probe directly to crystal during the measurement scope show correctly value. There is something with the wrong reading. Can anyone check it? Max read freq on BNC mode is 1MHz?

[Dumidan]

As far as I know, the maximum frequency that can be measured by the basic scheme is 2MHz. That's why the extension for checking quartz crystals has a prescaler that can be set to a factor of 1:16 or 1:32 with a hardware jumper. But the prescaler option in the software must be activated. Did you activate the option? Or go 1: 1?
The prescaler can be used both for BNC input and for checking LF and HF quartz crystals.