$20 LCR ESR Transistor checker project

[indman]

Feliciano, no, this is not soldering, for the Phil1977 clone, first of all you need to try lowering the BOD threshold to 2.7V. But this will only be possible when installing new firmware, since most likely the Chinese factory firmware will be blocked for reading.

[madires]

The watchdog timer is reset regularly during capacitance measurements (and also other measurements). It will be triggered when the firmware freezes for whatever reason.

[Phil1977]

Thanks for all the support!

As soon as I have a little time for it I´ll connect the scope again to VCC of the microcontroller and upload the result including a video of the restarting device.

I also don't think it´s a bad solder joint because I have two units of the same type and both show exactly the same behaviour. I modded one with better buffer caps, but it didn't change a thing.

Maybe it´s just a bad Atmel-clone inside these devices. If I take the effort to reflash the chip I could also think about getting an original chip from reichelt/mouser and replace the controller.

But that´s an activity for a rainy day...

[Feliciano]

I have a similar unit of yours, bought from one link several pages back on this thread, with the chinese firmware showing 1.12k at the end of the selftest.

It correctly detects 220uF and 470uF, between 1-2 or 2-3, and goes into monitor mode when inserted between 1-3.

[Feliciano]

Regarding the Fish8840TFT (the one with blue PCB and Z14 in parallel with Vcc):

I made some progress: flashing m-firmware allows me to read what's on the screen, but the background supposed to be black it's actually washed-out (gray). Same thing with the foreground colors decolorated (pale). However, when flashing the firmware, the background color really turns black.

The only hardware modification was to convert the "4.xV" of 5V//3.3V into separate 5V and 3.3V lines. Perhaps this ST7735 is supposed to be 5V, and removing Z14 would have been enough fix?

Additionally, the horizontal offset seems not enough in this case. I notice the clones file state 156x128. Is this a non-standard LCD and/or shall I need to play with atypical sizes in order to gain extra shifting?

[madires]

The ST7735 runs with 3.3 V. The Fish8840 display settings in the Clones file seem to be a bit off. LCD_OFFSET_X and LCD_OFFSET_Y are enabled, but without any value (resulting in 0). My guess is that the 156 pixels are simply 160 - 4 (160 for the display, and 4 for the offset). Please try LCD_DOTS_Y 160 and LCD_OFFSET_X 2.

[Feliciano]

The above pictures are with 160 x 128, X offset = 4. 

I also supposed to define a blank offset would mean 0, but I would rather prefer see it dissabled as usual. I also saw the same definitions for some other clones, so may want to clarify that for future releases. And maybe to check the reason behind the atypical 156x128.

Later I will try to see if I can solve the offset issue, but also I'm not sure on why the background is not totally black under normal operation, but only when in flashing mode 

Note: the LCD receives power in the following way: 78L05 -> 100R -> Z14 -> 100R -> LCD. It's similar to the suggested solution except by using a zener instead of an LDO. In any case, it's not getting exactly 3.3V, though.

[indman]

Additionally, the horizontal offset seems not enough in this case. I notice the clones file state 156x128. Is this a non-standard LCD and/or shall I need to play with atypical sizes in order to gain extra shifting?

Resolution of the display - 160х128, but the front frame superimposes a part of the image therefore it is necessary to move and offset the picture a little.

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?

This is a standard ST7735 display and requires 3.3V power. There is no need to come up with any fantastic theories about the special display in your clone. The 662k(XC6206) stabilizer now costs very little money, it can be found on other displays. If this is not the case, then you can use any other one or connect 2 silicon diodes in series as done in the LCR-T4 clone. There are many solution options for you to choose from.

[madires]

I see. Have you any idea why LCD_OFFSET_X and LCD_OFFSET_Y are enabled without any values set?

[indman]

I see. Have you any idea why LCD_OFFSET_X and LCD_OFFSET_Y are enabled without any values set?

I don't have such a clone, but I remember that the offset settings bdk100 made in the ST7735.c file, because the LCD_OFFSET_X and LCD_OFFSET_Y settings were not enough to accurately orient the image. Say "thank you very much" Chinese manufacturers of this clone for this placement of the frame on the front panel.

[madires]

Thanks! I'll add a note to the Fish8840 settings in the Clones file.

[Yuriy_K]

Three years ago I did a firmware update for a friend of mine on such a T-tester. I don't remember if he cut off the upper and lower black frames, but he definitely cut off the side ones. Otherwise, the output text is cut off. The entire image field should be visible in the case window...

Found a photo of the frame removal in my archives, added it. The firmware was created in changes to previous versions for the Russian version. It will take some time to create English firmware with the latest changes.

Added new firmware for 8 and 16 MHz. Deleted old firmware...

Changed display pin configuration.
Try new firmware...

[Feliciano]

An issue here is nobody sells SMD components, so sometimes I depend on scavenging things.

I tried the option of the 2x 1N4148 (apparently more power-efficient, and it could have worked considering the original Q was around 4V for everything). But the backlight is still washing away the black. I'll keep scratching my head about this. Perhaps reducing its bright with a resistor? or I don't know if I would have to insert series resistors on the data lines? As I read, Karl-Heinz implemented setting the LCD lines as open collector (OC) and tweaking the pull-up in order to aproximate to 3.3V. I don't know whether the m-firmware has a similar workaround.

Besides that, maybe I will have to look at the ST7735.c as well. Or even cut the frame as suggested.

Additional comments:
.-Regarding the reference to blank offset settings, please note there's the same suggestion for the BSide
.-I also think this one is a runner for the crown of the sloppy designs

[madires]

Yep, if the backlight is too bright try a larger resistor. The in-series resistors for the data lines (in conjunction with the ST7735's protection diodes) are a quick n' dirty level shifter. The m-firmware doesn't support an open-collector mode (would open a can of worms when driving multiple ICs or when switching to hardware SPI).

[Feliciano]

And I'm curious about another fact: what the OSHW firmware LCD driver does different during the firmware flashing procedure, compared to normal operation? Because in those seconds the black part of the screen works as it should.

[madires]

During flashing the firmware doesn't run. The display will show whatever is in its memory at that time. AFAIK, the Fish8840 drives the ST7735 with 5V directly instead of using in-series resistors. So the ST7735's Vcc would get 5V minus a diode drop via the built-in protection diodes when any signal is high. For example, the TC-1 uses 10k in-series resistors plus a 3.3V LDO. The resistors limit the current and allow the LDO to keep Vcc at 3.3V.

[indman]

 It’s a little strange that most Fish8840TFT owners did not complain at all about the display and its brightness when using OSHW firmware? Experience has shown that many displays do not experience problems with the 5V level on the signal buses(VDD=3.3V), although it is still technically correct to coordinate these levels not using resistors, but using at least 4050 

[madires]

There are different versions of the ST7735:
- ST7735
  - I/O Voltage (VDDI to DGND): 1.65V~VDD (VDDI ≤ VDD)
  - Analog Voltage (VDD to AGND): 2.6V~3.3V
- ST7735R
  - I/O Voltage (VDDI to DGND): 1.65V~3.7V (VDDI ≤ VDD)
  - Analog Voltage (VDD to AGND): 2.3V~4.8V
- ST7735S
  - I/O Voltage (VDDI to DGND): 1.65V~3.7V (VDDI ≤ VDD)
  - Analog Voltage (VDD to AGND): 2.5V~4.8V
And possibly some compatible controllers.

[Feliciano]

Interesting, and thank you all for the feedback.

So for the unit in hand, i.e. m328p+GMT177-05 (ST7735S), if the driver is not refreshing the screen (I/O=0V, I suppose) the colors are OK, but all the colors are "overdriven" when running the out-of-range of Vdd=3.6-100*I and I/O up to 5V-Vsat. (I didn't pay much attention to the screen when running the original firmware with 5V//3.3V=4V, though).

Nevertheless, according to the quoted specs I set Vdd=5-100*I, and that apparently solve this part of the issues, thanks.

Now the unit worth a little more, and I will play around with the settings trying to avoid removing the ugly yellow glue, cutting the box, and putting all back toghether without damage.

[Feliciano]

With the simplest hardware mod and some atypical settings, I think I'm getting there:

Code: [Select]

#define LCD_DOTS_X       128
#define LCD_DOTS_Y       158
#define LCD_OFFSET_X     1 
#define LCD_OFFSET_Y     6
#define LCD_FLIP_X
#define LCD_ROTATEAnd one of my narrow fonts due this odd screen frame size.

My original intention was to run K-firmware (16M) on this one, but I'm not sure on whether I could replicate something similar on those config files, so maybe I'll leave it as is.

Cheers.

[Yuriy_K]

A question for everyone who managed to make a k-firmware for FISH 8840.
What values ​​are in your config.h?

Changed display pin configuration.
Try new firmware at the end of the previous page...

[Feliciano]

To put this in context, the Fish8840TFT seems to have a different LCD pinout than the k-firmware's default. Apparently this was properly considered for the older Rev804 shared on Yandex, but looking at the long and nested if-then-else of config.h pointed out by Yuriy_K, it's different on the current trunk. It could be related with the fact of blue or green variations of this board, I'm not sure, but if that's the case, it should be added a separate branch with the proper note. (I suppose it should be also different from the unboxed monochrome versions).

In any case, the PCB I have is the blue one, works with the binaries shared on Yandex, and/or Yuriy_K on the last page, and also with the one I make now.

Code: [Select]

CFLAGS += -DLCD_ST7565_H_OFFSET=7
CFLAGS += -DLCD_ST7565_V_OFFSET=2 /* garbage seen when looking from below */
CFLAGS += -DBAT_OUT=200
CFLAGS += -DBAT_NUMERATOR=66
CFLAGS += -DBAT_DENOMINATOR=33

I will now play with both firmwares to choose what I like best.

Thanks,

[Yuriy_K]

Measuring small inductances on k and m firmware raises questions. I specially ordered several sets of inductances with inscriptions and SMD 0805. Here are the measurement results. For SMD 0805, I measured at frequencies of 100 and 300 kHz.
The results may surprise...
Comparisons with NJ300A readings.

I found out why the readings of the measured inductances are decreasing. This happens when the quality factor Q decreases - below 10. At higher quality values, the measurements coincide. Look at the values ​​Q -> NJ300A.

Added a confirming example. Approximately equal inductances with different quality factors Q.

[indman]

The results may surprise...
Comparisons with NJ300A readings.

To me, there's nothing surprising here. You keep trying to compare measurements on a simple TESTER with simplistic circuitry for 5-10$ with a serious bridge-type RLC for 400-500$! There are no miracles, the RLC measurement for this project is just a BONUS option.From your photos it is obvious that the tester cannot cope with measuring coils with high active internal resistance. Don't worry so much. 

[Phil1977]

On the contrary, it´s good to see that the results -at least for inductivity- are not so different at all.

If you play with the parameters of the LCR-meter (especially frequency) you probably can provoke a similar deviation.

If you want precise results you really need to specify the test parameters and then you need a LCR-meter.

If you just want to identify an unknown part, the results of the transistor tester are a great start. And that´s -from my point of view- the main application of this great little gadget.