$20 LCR ESR Transistor checker project

[madires]

ive purchased one of these on ebay and want to add m328 but have trouble finding latest hex. the mikro site is a hodge podge of confusing tex and hex files with line numbers and dont resemble any file type that avrdude recognizes. can someone tell me how to get a normal hex file there or link to another location? im mostly interested in inductance and milliohms, your help here would be greatly appreciated.

Unfortunately the website doesn't support direct download of individual files. If your tester matches the standard design please go to  http://www.mikrocontroller.net/svnbrowser/transistortester/Software/trunk/mega328/ and click "Download GNU tarball" for the tar.gz archive of the selected directory.

[paulie]

thank you. that is what i needed to know. it wasnt obvious to me what tarball meant.  i hope the cheap version i got on ebay ($9 bid) will allow me to do inductance and low ohms measurements i need. specially nice are the smd pads on that board.

my plan is, once its tested with 9v battery, will try to convert to 5v dc benchtop supply. hopefully just tying the battery check adc input high will disable any low voltage warning. thanks again for helping.

[madires]

my plan is, once its tested with 9v battery, will try to convert to 5v dc benchtop supply. hopefully just tying the battery check adc input high will disable any low voltage warning. thanks again for helping.

You're welcome! If you're going for a direct 5V power supply I recommend to consider to add a 2.5V voltage reference. The low voltage warning/power-off can be disabled in the firmware (adjust values in config.h and recompile the source). Or a 7.5-12V supply would also be fine (based on the LDO regulator).

[paulie]

i guess you want 2.5v ref because dc supply is noisy and not stable? specially usb cable from pc in my case. i do have lm336 if that seems to be a problem. high accuracy is not important to me but i would like to measure lower inductance and lower capacitance. i guess soldering a 27pf and subtracting would work for caps but what about lower inductance?

would 20mhz xtl instead of 8mhz improve ability to measure lower inductance?

can you describe the alorithm used for measuring inductance?

minimum hardware is two 680r and two 470k?

compiling is not convenient atm but i am set up for atmel asm and would like to experiment.

edit: nevermind... i found all info needed on inductance algorithm in section 5.4 of document "ttester_eng106k.pdf". thanks for an excellent design.

[madires]

i guess you want 2.5v ref because dc supply is noisy and not stable? specially usb cable from pc in my case. i do have lm336 if that seems to be a problem. high accuracy is not important to me but i would like to measure lower inductance and lower capacitance. i guess soldering a 27pf and subtracting would work for caps but what about lower inductance?

The reason is the accuracy of the supply voltage. With a MCP1702 as LDO regulator you get 5V +/- 0.4%. Your PSU might deliver 4.7V or 5.3V and the voltage reference helps to determine the offset of the supply voltage. The tester will then compensate that for the measurements. A LM4040 would be a nice voltage reference. The LM366 is a hex sense amp ;-) The k-firmware supports capacitance measurements starting at about 30-35pF and the m-firmware at about 5pF. For measuring low inductance you'll need a proper LCR meter.

would 20mhz xtl instead of 8mhz improve ability to measure lower inductance?

The maximum clock rate currently supported by the firmware is 16MHz. The higher clock rate increases the timer resolution but that doesn't help much with low inductances (just a little bit). To measure lower inductances a higher test current would be needed. The inductance measurement is actually a nice by-product because the circuit wasn't designed for that. Other measurement methods are more suitable for supporting a larger range of values.

minimum hardware is two 680r and two 470k?

Three each. At the moment Karl-Heinz is testing also lower resistor values to gain higher test currents which improve some measurements.

[neslekkim]

Do anyone have any link to one of these that have the 328 microcontroller?, and possibly not one that is sanded down, but maybe all still are ?

[paulie]

The LM366 is a hex sense amp ;-)

you misread. its lm336 which is 2.5v reference. ive been using them for years. not particualrly accurate but thats not important atm. i do have a couple 4040 but they are expensive and i got hundreds of lm336 in the junk box. in fact because accuracy is not needed at this stage i will probably not use any reference. im just thinking of throwing something quick together while waiting for the ebay one to arrive.

The maximum clock rate currently supported by the firmware is 16MHz.

right. my intent was to experiment with a basic asm version but now i see that inductance feature will never get down to the nanohenry range so will skip that idea. ill maybe build an oscillator type but even that may not be practical for the 433mhz and 2.4ghz indictors im dealing with. idk.

Three each.

if i only wire up 2 ea will it still perform as a two terminal tester? resistors, diodes, caps, coils? or does it check for all 6? right now im trying to throw together some minimum hardware till the one with full circuit and  0.1% resistors arrives.

i have one last question. does the hex file you gave me support serial? its mentioned in the diagram. i really want to build a usb/ftdi powered version like in the to-do list of that pdf. it would avoid the lcd and complicated power ckt. im hoping one of these can be built with 4-5 wires and couple components instead of dozens. ie minutes to build instead of hours/days. the fancy unit may not arrive for weeks.

[paulie]

Do anyone have any link to one of these that have the 328 microcontroller?, and possibly not one that is sanded down, but maybe all still are ?

a quick search on ebay for "transistor tester" will come up with hundreds of links. many clearly show mega328. generally it looks like the $16-$20 ones are m328. the $8-$12 ones may be m8 or m328. seller says the one i got for $9 is m328 but ive learned not to believe a word. ive got large supply of m328 anyway so its not important in my case.

[neslekkim]

ok, I thought most of the used the 168, but it's just a chipswap if so happens, then it should be fine.

[paulie]

i think you are right, a lot of the middle priced ones seem to be m168. im not sure theres any practical difference at this stage of firmware except m8 definitely dont work with latest version. according to sellers the surface mount units all use the bigger chips but i would avoid them in case of bricking. the one i bought has no isp pads so thats another reason for socket. most ebay ads have photos showing the board so that helps.

[neslekkim]

Earlier in this thread I think someone published links to ones that was sure to be the newer one with the 328, but they are old links, so i was hoping that someone had some that was checked recently.
I see there are three of four types:
1) one with ZIF socket, and avr in dip: http://www.ebay.com/itm/271366797178

2) one that is boxed: http://www.ebay.com/itm/181109931348

3)  this one seems to have an 8L http://www.ebay.com/itm/271367115286

4) smd version, 168: http://www.ebay.com/itm/181161606431

And probably lots of other variations too..

[paulie]

the 8$ one in your 3rd link is exactly what ive got coming. he swears its m328 but thats doubtful. he also said it was backlit but that is probably not true either. personally i dont care either way because i got tubes of m328 and backlit displays are cheap.

what i would really like to do is, while waiting, build a stripped down version with nothing more than four 5% resistors with power and txd going to a pc serial adapter. im wondering if that will work with the hex file posted above. it would also be nice if internal pull-ups were active so i dont have to wire external ones.

edit: i connected only 5v and pc4 and every time i reset get the following characters at 19200bps:

12V
 0V 1.32V

or

13v
5.02 VBF C

or

16v
F C

probably what it thinks is battery and some "ghost" component or self test warning. lots of garbage characters too. so thats a good sign. pretty good for NO external components at all. very encouraging so next im going to hook up some resistors and a cap across tp1-2 then see what comes out.

[neslekkim]

Based on description, it also seems like this is using same code?

but, support for big screen, that is something the chinese have added?

http://www.ebay.com/itm/181164691633

[paulie]

yes, i saw those. also the ones with signal generator built in. $50 is too much though imo.

so i wired the six resistors, tied pc4 & pc5 hi, and pullup on the button. it seems to be working after a fashion but not quite behaving as described in ttester_eng106k.pdf. on reset or button push with .1uf inserted i sometimes get a message "not calibrated" followed by confusing instructions to short terminals and/or plug in 100nf cap.  a couple times i got "100nf" reading but only one or two times at the beginning and not since. usually just a few cryptic letters.

this never happens when ports are shorted together as described in manual. sometimes i do get "0ohms 0ohms" which makes sense but, again, not reliable. no hint of self-test or calibration.

it might be wiring error or bad contacts or who knows what else. it does look like there are issues with the serial routines. lots of blank spaces and odd characters. im using a crystal and spacing is very consistent so unlikely a baud problem. maybe set for some oddball terminal protocol. it would be nice if it just printed a line or two with crlf in industry standard vt100/hyperterminal format.

i wont be able to play with code til vacation is over in couple weeks and it will be interesting to see how the ebay unit behaves when that arrives.

[madires]

you misread. its lm336 which is 2.5v reference.

I see.

if i only wire up 2 ea will it still perform as a two terminal tester? resistors, diodes, caps, coils? or does it check for all 6? right now im trying to throw together some minimum hardware till the one with full circuit and  0.1% resistors arrives.

For self-adjustment all three are needed. And I think it might break some tests.

i have one last question. does the hex file you gave me support serial? its mentioned in the diagram. i really want to build a usb/ftdi powered version like in the to-do list of that pdf. it would avoid the lcd and complicated power ckt. im hoping one of these can be built with 4-5 wires and couple components instead of dozens. ie minutes to build instead of hours/days. the fancy unit may not arrive for weeks.

Please see the Makefile:
CFLAGS += -DWITH_UART

The software UART is enabled for that firmware image.

[paulie]

yes, since that post i wired the six resistors, tied pc4 & pc5 hi, and pullup on the button. it seems to be working after a fashion but not quite behaving as described in ttester_eng106k.pdf. in fact i bult two units like that. whats strange is different m328 work differently.

is it necessary to program the eeprom file?

i also see lots of signals that are not serial appearing on pc3. it is consistent though so dont look like noise. is this for standard vt100/hyperterminal type terminal? is there an example of serial output available anywhere?

[kubi48]

You have always to program the flash memory with the .hex file and the EEprom with the .eep file for correct operation.
The serial output should be nearly the same as displayed on the LCD.
It is output with 9600 baud (1 Stop bit) with a 5V level.
The capacitor symbol is shown as -||- and the resistor symbol as -RR- .
The diode is shown as  ->|-  or  -|<-  .
The omega sign is replaced by Ohm and the µ sign is replaced by u .


 

[han]

i build with atmega 328, relay protection, USB supply,and internal voltage reference

i find the accuracy is depend on stability of the USB supply voltage (different USB power source different result)

i plan to use DC-DC converter from digikey.

whats the beter solution: ?

1. use 5V to 5V DC-DC converter
2. Use 5V to 9V DC-DC converter + 7805


 is there anyone that using dc-dc converter to supply the device?

[paulie]

You have always to program the flash memory with the .hex file and the EEprom with the .eep file for correct operation.

thank you for that info. that may explain why same type chips operate differently. i will burn the ee file too next time. this is odd considering flashing new chips that have blank ee will not work.

im disappointed serial output is so complicated and dont work with standard terminal.

i will try to fix both problems when vacation is over and i can play with code and recompile.

[Macbeth]

2) one that is boxed: http://www.ebay.com/itm/181109931348

I've got a couple of these. The display is backlit, and the AVR is a 168 (in a socket) and comes with 1.07k firmware. The ZIF is a nice touch!

[madires]

whats the beter solution: ?

1. use 5V to 5V DC-DC converter
2. Use 5V to 9V DC-DC converter + 7805

7805 or 78L05 aren't recommended. Please use a MCP1702 or something similar. Also add a proper ripple filter to the DC-DC converter.

[madires]

thank you for that info. that may explain why same type chips operate differently. i will burn the ee file too next time. this is odd considering flashing new chips that have blank ee will not work.

Several strings, display characters, tables and adjustment data are stored in the EEPROM. No wonder you get strange display outputs ;-)

im disappointed serial output is so complicated and dont work with standard terminal.

There's an arduino version (http://www.pighixxx.com/lavori/ardutester/) supporting serial output and including a nice GUI tool.

[paulie]

i may try that ardunio version someday but im not a big fan of the environment. i prefer winavr by itself because its more compact, flexible, and powerful. uses 1/10th of pc resources too. admittedly ardunio is more popular with better support for beginners but not for me.

anyway thanks to that last hint from you and kubi eveythings working great now. obviously all the extra characters and signals were simply 0xff bytes from blank eeprom.  i have two of these working now at less than $2 in parts including avr. accuracy and stability are excellent.

this is an incredible project and already dozens of local fellow students and friends are as excited as me. thanks markus for a wonderful design and generous help.

[madires]

this is an incredible project and already dozens of local fellow students and friends are as excited as me. thanks markus for a wonderful design and generous help.

Markus Frejek designed the tester. Karl-Heinz (kubi) has carried on with the developement and I (another Markus ;-) have joined in with an alternative firmware. And we're happy to hear that you like the project.

[paulie]

aha... i didnt know kubi was karl. what a team. thanks to both of you for bringing me up to speed on importance of ee in that build. im guessing putting those strings in ee is to save flash for the 8k version?

ive already sorted out half the "mystery" components in my junk box. i see what my final requirements are now and adjusting the makefile should do it. a stripped down version for mega8 with no powerdown seems to be the best fit for a group of local engineering students that im setting this up for. unfortunately i wont have access to my main computer for a couple weeks.

do you know if there is a hex file for mega8 anywhere on the net? also can you tell me what a "tex" file is? im not linux and a lot of this is new to me.

edit: i know what "tex' is now. nothing to do with hex or largest state in usa. i was using wrong search terms before.

i could still use a link to hex file for mega8. nothing useful on google so far. if you or anybody could help it would be appreciated.