$20 LCR ESR Transistor checker project

[oitar]

Brag time

1. Original Atmega324(slightly defective) replaced with Atmega644.
2. The power management ic replaced with the 2 transistor hack(see attached, yeah looks like a dog's breakfast, but it works nicely; will put a bit of hot glue over).
3. Button replaced with rotary encoder- helps prevent arthritis from multiple button presses
4. Don't think I'll even replace the probe resistors- I'm happy with the precision already.

Changing the zener testing to separate terminals- away from the zif socket should be done as a matter of necessity, sooner or later one will forget either a charged cap, or  the probes in the wrong zif pins, and toast the cpu.

Cheers to all, thanks @madires for still keeping the upgrades going, and answering even the dumbest questions!

[BennehBoy]

I've just come to use my T7 after quite a period of none use and found that the screen repeatedly cycles between full white and off whilst trying to charge the unit.

I'm assuming that it's been killed at some point - maybe even the last time I used it...

I've not read all 380 pages (eeek!), and a quick google did not turn up a similar symptom, but any ideas?

I'm hoping it's just a case of reflashing something/doing the transistor mod on the STC15L104W.

Thanks

[Calambres]

I also have an AY-AT with genuine DIL ATMEGA328 chip that suddenly and randomly started to write some crap characters on the screen.
Quite annoying if you ask 

[madires]

I've just come to use my T7 after quite a period of none use and found that the screen repeatedly cycles between full white and off whilst trying to charge the unit.

Please check the voltage of the LiPo battery. It's possibly very low (deeply discharged). If charging doesn't help try a new battery.

[BennehBoy]

I've just come to use my T7 after quite a period of none use and found that the screen repeatedly cycles between full white and off whilst trying to charge the unit.

Please check the voltage of the LiPo battery. It's possibly very low (deeply discharged). If charging doesn't help try a new battery.

It was very low - 0.1v when I checked it just before the time of my last post.  It's been sat trying to charge since then, will check it shortly.

UPDATE - no change to the voltage, I guess it's toast?

[Yuriy_K]

For those who see similar readings on silicon transistors on the k-firmware, which is incorrect (upper left corner). This is not a program error, but an incorrect measurement scheme. If we add inductance and capacitance to the scheme, highlighted in red SMD, we get correct readings for silicon transistors. For germanium transistors, measurements are shown - lower right corner.

[madires]

UPDATE - no change to the voltage, I guess it's toast?

Yep, the LiPo battery needs to be replaced. I'd recommend also to add a small switch to the battery (to prevent draining it by the high quiescent current).

[madires]

For those who see similar readings on silicon transistors on the k-firmware, which is incorrect (upper left corner). This is not a program error, but an incorrect measurement scheme. If we add inductance and capacitance to the scheme, highlighted in red SMD, we get correct readings for silicon transistors. For germanium transistors, measurements are shown - lower right corner.

Also recommended by the ATmega's datasheet, i.e. a small inductor plus cap for the ADC's power supply.

[indman]

For those who see similar readings on silicon transistors on the k-firmware, which is incorrect (upper left corner). This is not a program error, but an incorrect measurement scheme. If we add inductance and capacitance to the scheme, highlighted in red SMD, we get correct readings for silicon transistors. For germanium transistors, measurements are shown - lower right corner.

Also recommended by the ATmega's datasheet, i.e. a small inductor plus cap for the ADC's power supply.

Here is a reply from Karl-Heinz Kübbeler on the advisability of using a inductance(throttle-coil) in the AVCC power supply circuit:
"There is no need to use a throttle (coil) for the AVCC supply.
During the ADC period no digital activity disturb the VCC power supply.
Any inductance in the AVCC supply also have a resistance.
The influence of the resistance of the coil probably interfere the
measurement result more then a remaining noise at the VCC power.
The enhancement of ADC resolution requires a noisy or unstable signal.
The mean value of several input values is always the input value itself,
if there is no change of the value. Only if the input signal changes a
little bit, the mean value can interpolate the steps of the ADC by
a statistically effect. For example the mean value of
10, 11, 11, 12, 11, 10, 11, 10, 11, 11 is 10.8 ! "

[madires]

There's no simple answer. It depends on how well the power supply circuitry is designed. Some Vcc ripple helps (is necessary) with the ADC (averaging multiple readings). But a sudden larger load change will cause a short drop/rise of Vcc. With a marginal power supply and no inductor for AVCC the ADC's 5V reference voltage would change too. The LC filter reduces that change of AVCC. For a proper power supply the LC filter isn't really needed (might even be counterproductive). But for a poor one it can be a good idea. And the values of the LC filter should be choosen to suppress ripple to some extend while keeping enough noise for the ADC. This is also one of the reasons why some clones deliver results which are a bit off.

[indman]

But for a poor one it can be a good idea.

It is a good idea for poor and rich people to make a quality power supply for this device that is not afraid that a sudden change in load will cause Vcc to drop/rise, thus eliminating unnecessary parts in the form of external REFs, coils, i.e. crutches for the lame.

[BennehBoy]

Yep, the LiPo battery needs to be replaced. I'd recommend also to add a small switch to the battery (to prevent draining it by the high quiescent current).

I've ordered a larger capacity lipo (that will fit the case) and will certainly be adding a switch, thanks.

[Feliciano]

Or you could adapt any old cell phone battery you may have handy.

[Mecii]

I just came across this thread when I realized some of the manufacturers of these testers were producing fakes. I just received this one today and wonder if this is a fake or worth keeping. I have not yet had a chance to test anything since I received this at work and I will have to wait until this evening to test. Sorry, for the quick ask but this is a large thread and I will read more tonight but I was hoping for some quick feedback.

Thank you.

[madires]

Seems to be a genuine ATmega328.

[Phil1977]

May anyone tell me about this device?




It´s a great little device but it has one flaw: It reboots itself when *middle*-sized caps, around 100uF, are measured. I´ve read somewhere in the manual of the original device that there are a some brownout-protection registers in the Atmel that can cause this behaviour.

Any chance to fix this for such a clone device?

[madires]

Looks like a variant of the AY-AT/GM328A. To change the fuse bits you need an ISP programmer (e.g. USBasp), a programming software (e.g. avrdude), and also have to solder a few wires for an ISP header.

[Phil1977]

Thank you!

Is there any tutorial for the ISP programmer you can recommend?

Can I set the fuses without deleting the code? Of course, it´d be best to program the original code - but how are the chances it directly runs on this hardware?

[madires]

There are tons of tutorials about programming an ATmega/AVR. Just run a search and include a few keywords of your specific setup (programmer, OS, ...). Usually the clone manufacturers set the lock bits, i.e. you would have to erase the ATmega. But this allows you to switch to one of the two OSHW firmwares.

[Phil1977]

I just realized YOU are the more or less main owner of this project.... So thanks for this wonderful code and I´m very glad to look for an ATmega&&non(Arduino)-tutorial on my own.

[Feliciano]

It´s a great little device but it has one flaw: It reboots itself when *middle*-sized caps, around 100uF, are measured. I´ve read somewhere in the manual of the original device that there are a some brownout-protection registers in the Atmel that can cause this behaviour.

Any chance to fix this for such a clone device?

It doesn't happens to me with the clone firmware. Have you tried with another USB power source, or using a battery? Else, you could use the OS firmware, overwriting the it came with.

[Feliciano]

Regarding the Fish8840TFT (the one with blue PCB and Z14 in parallel with Vcc):
Not having an SOT-23 regulator handy, I inserted a Resistor between Vcc and the 3.3V Zener (didn't modify the LCD data lines, thinking on the open collector option). The Voltages look reasonable and no longer overcurrent/overheating, but the contrast went too low. So I flashed indman's k_based-firmware (8MHz), but the LCD only shows different shades of white. One additional thing I notice is: while re-flashing, the background turns black, and then I can read white text on the screen (15V testing..., that would be needed to be adjusted on the makefile for this 33k/33k variant, as well as inserting the "$(OBJDIR)" compiler options on the older makefile).
 
What else would I need to modify to set the proper foreground/background colors? Or maybe this is a new LCD variant supposed to work on 5V and not 3.3V?

[Phil1977]

It´s a great little device but it has one flaw: It reboots itself when *middle*-sized caps, around 100uF, are measured. I´ve read somewhere in the manual of the original device that there are a some brownout-protection registers in the Atmel that can cause this behaviour.

Any chance to fix this for such a clone device?

It doesn't happens to me with the clone firmware. Have you tried with another USB power source, or using a battery? Else, you could use the OS firmware, overwriting the it came with.

Yep, I´ve tried different power sources, I added additional decoupling caps at the CPU, I also checked the supply voltage with a scope. The 5V at the CPU are rock stable, all other tests are running perfectly but 47uF or 100uF caps can not be tested.

Caps <10uF and >1000uF seem to be okay.

[indman]

Yep, I´ve tried different power sources, I added additional decoupling caps at the CPU, I also checked the supply voltage with a scope. The 5V at the CPU are rock stable, all other tests are running perfectly but 47uF or 100uF caps can not be tested.

There are a lot of questions about this (and not only) problem with measuring capacitors on AVR ATMEL (purely Chinese MKs are a different story), so let's try to summarize what we know at the moment. There will be a lot of text, so whoever is not interested or too lazy to read, you can skip it.
So, description of the problem:
The tester turns off, "you (are) cut off", the display goes off during the measurement of capacitors, mostly of large capacity. When measuring through the menu, there is usually no such effect.

This problem is typical for software from Karl-Heinz k-firmware (1.12-1.13k).The software from Markus Reschke (m-firmware) also has this effect.
What do you mean it turns off? That is, for some reason the MK resets or the control program fails. The most likely reasons:
1. The power supply voltage of the MK and/or display disappears for a short time;
2. Power supply of the tester (MK) from a pulsed source with large ripples, interference and noise;
3. Triggering of WDT (watchdog timer) - enable/disable is set by the appropriate fuse before compilation or in the software;
4. BOD activation (supply voltage level monitoring system) - enable/disable, the threshold is set by the corresponding fuses before compilation.
5. MK defective (quite possible, given the strong rise in price of original chips and the slag that the Chinese industry is now trying to “sniff”).

The solution:
1. In 99% of cases there is a contact problem (bad or “cold” soldering, “straight-handedness” of the assembler).
2. The effect is eliminated by supplying power to the MK from a high-quality LBP or battery.
3. The authors suggest setting WDTON to the disabled state. In the software of both authors, this timer is reset programmatically before each measurement cycle and is activated if the operation (measurement) is not completed after a certain time. Usually in such cases the message TimeOut appears on the display. For more details, see the manual from Karl-Heinz.
4. The authors recommend turning on BOD and setting the response threshold to 4.3V. For the clock frequency (quartz 8 MHz), high-quality +5V stabilizer and filter capacitors, this threshold does its job perfectly and no problems were noticed with any measurements.
Factory Chinese components in most cases cannot digest this threshold, since under increased load (measuring a capacitor is one of the longest processes in terms of time and load), short-term voltage dips appear, which are sensitively detected by the BOD. Therefore, advice such as set additional fuses and lowering the BOD threshold to 2.7V sometimes really helps solve (mask) the problem, but not always.
With an increase in the clock frequency of the MK to 16-20 MHz, the current consumption and, accordingly, the load on the power bus and stabilizer also increases. The safe operation zone of the MK narrows.The picture that explains this is below.
You can also lower the BOD threshold or turn it off completely; the MC will not reset, but then there is a very high probability that measurements during such voltage dips will be incorrect or distorted. The optimal solution would be to install a QUALITY +5V STABILIZER (for example, MCP1702) and similar quality filter capacitors at its input and output!
5. In most cases, a faulty MK is determined quickly when running self-testing and calibration using hardware tests. If possible, you should try the m-firmware software or replace the MK with another one.

[Feliciano]

As indman suggest, if you're still using the chinese firmware, check carefully for a bad solder. Sometimes they're not easy to spot. If you have another tester, you can measure the power-related components for clues.