New tool to find short capacitors on motherboards : shorty with display

[megaraider]

Originally Posted by megaraider

[And above posts...]
My thoughts FWIW:

e)
Changing supply voltage from 5V to 3.3V will degrade performance and accuracy for both MCP3421 (INL vs. Supply Voltage) and AD8628 (Signal Voltage Gain).

f)
Even for DIY project it's taking too long to see the daylight!
IMHO it's well time to move the hands from keyboard to a bread board or a prototype board...

[clearchris]

Originally Posted by megaraider

e)
Changing supply voltage from 5V to 3.3V will degrade performance and accuracy for both MCP3421 (INL vs. Supply Voltage) and AD8628 (Signal Voltage Gain).

I checked the datasheets, and didn't see anything that would indicate less performance or accuracy at 3v3 vs 5v. Maybe I'm missing it?

I'm asking because it would be really nice to sub out a pi pico for the nano. Nanos had their time, and that time is passing. Unfortunately, it's a 3v3 board (can take power from 1.8 to 5.5v) and the pins aren't 5v tolerant. I have an pi pico in the mail, should be here on mondy, I'm going to see if I can get the software running on it. Would make an interesting option, and not very expensive at 6 usd.

https://ww1.microchip.com/downloads/...ets/22003e.pdf
https://www.analog.com/media/en/tech...-8629-8630.pdf

[stj]

the original design was 3.3v - so i presumed it could work.
and the oled runs at 3.3v no issues

[redwire]

I think the original circuit does not push much current through the probes?
BC557 switching around 56mA but most of that goes through R2 2Ω which is in parallel with the tested resistance, and by current-division gives around 1.4mA only to the probes. Test voltage aound 113mV which is good.
If this is right, then the source needs an update to pass more current through the resistor under test. Not a new vreg.

[clearchris]

Originally Posted by kripton2035

I added a voltage regulator 1755S to bypass the one inside the arduino. just thinking there would be a better regulation than the 1117. I also added a constant current source with the pnp transistor, to have better accuracy and better reproductibility of the measures.

Originally Posted by redwire

I think the original circuit does not push much current through the probes?
BC557 switching around 56mA but most of that goes through R2 2Ω which is in parallel with the tested resistance, and by current-division gives around 1.4mA only to the probes. Test voltage aound 113mV which is good.
If this is right, then the source needs an update to pass more current through the resistor under test. Not a new vreg.

I read through the thread and found this post by krypton. Sounds like he added a voltage regulator and constant current source to the pnp transistor. I believe he later said he removed the transistor entirely and just used the constant current power source.

I'm not really sure of the severity of the issue he was trying to hunt down, the videos sure look like he has a functional device. Could be an issue of the perfect being the enemy of the good. I'm pretty sure I'd be fine with an 80% solution when what I have right now is a 0% solution.

I took a look at the code, he bit bangs the registers, so I'm going to have to get the datasheet out, but should be able to figure out what he's doing and get it going on rp2040.

[redwire]

I'm wrong about some of the test currents. I checked math again and the circuit with 5V is putting a max of around 58mA through the resistor under test, with 114mV open-circuit. So this looks quite reasonable, assuming the (Arduino Nano V3) and 9V battery power supply holds up.
With a test resistor of 1Ω that's 38.6mV and 38.6mA
With a test resistor of 0.1Ω that's 5.6mV and 55.6mA
With a test resistor of 1mΩ that's 58.4uV and 58.4mA The op-amp is a bit noisy then, after it is 1.578mV to the ADC input, a little small.

I don't see any deal breakers, other than 5V '328 with 3.3V OLED and no cap across R8 and R7 for filtering incoming noise, you could try average it out in firmware.
In the mΩ measuring it's in the noise floor and battery limit. You guys want 1A?! Any inductance and that would wallop what you are testing...

[megaraider]

Originally Posted by clearchris

I checked the datasheets, and didn't see anything that would indicate less performance or accuracy at 3v3 vs 5v. Maybe I'm missing it? (...)

JFR see attached images taken from datasheets.
Off course we're talking about peanuts... but when it comes to detecting µΩ

[megaraider]

Originally Posted by clearchris

(...) I have an pi pico in the mail, should be here on monday, I'm going to see if I can get the software running on it. Would make an interesting option, and not very expensive at 6 usd.

Go for it... much more interesting indeed

[clearchris]

Originally Posted by redwire

I'm wrong about some of the test currents. I checked math again and the circuit with 5V is putting a max of around 58mA through the resistor under test, with 114mV open-circuit. So this looks quite reasonable, assuming the (Arduino Nano V3) and 9V battery power supply holds up.
With a test resistor of 1Ω that's 38.6mV and 38.6mA
With a test resistor of 0.1Ω that's 5.6mV and 55.6mA
With a test resistor of 1mΩ that's 58.4uV and 58.4mA The op-amp is a bit noisy then, after it is 1.578mV to the ADC input, a little small.

I don't see any deal breakers, other than 5V '328 with 3.3V OLED and no cap across R8 and R7 for filtering incoming noise, you could try average it out in firmware.
In the mΩ measuring it's in the noise floor and battery limit. You guys want 1A?! Any inductance and that would wallop what you are testing...

That sounds like why he added the constant current source for the probes to limit current through the item being tested. IIRC, he said it was limited to 30ma. Probably a really good idea.

Megaraider:

Yeah, you got me, good catch. There's definitely value in feeding the AD8628 5v. There's likely also a lot of value in adding a constant current source the the probes. Reading the code, krypton uses a gain of 1 for the MCP3421. Not sure if there's any value in changing the gain on the ADC, if we want more gain, we should probably change the amplifier to get a bit more range. Upping the gain on the MCP3421 adds a bit of noise, though I'm not sure how much a difference of 5uV matters. For this application dealing with uOhms, maybe a lot.

If we are dealing with 1.578mV to the ADC input we could run the ADC on 3v3 (to get 3v3 communication to the MCU) and it should be fine. Everything else can run 5v including the OLED.

I'm just throwing out some ideas, I'm not necessarily advocating (ok, maybe for the rp2040, but I'm not married to it) you guys know this stuff far better than I do. I suspect krypton was putting some effort into keeping the BOM down, but I wouldn't have any issues bumping it up (suggestions so far, not all by me, amp, mcu, transistor, constant current source, more caps) if it increases quality.

BTW, I noticed that the VoutAOP line to the MCU in the schematic isn't used in the code. Is there any reason you would need this line? It's probably just a source of noise if it isn't used.

[stj]

someone on eevblog came up with this mod

[sam_sam_sam]

Is the person who started this post going to make a kit available for purchase just asking the question

[clearchris]

Originally Posted by sam_sam_sam

Is the person who started this post going to make a kit available for purchase just asking the question

The latest I have heard is that he was going to put the pcb design files up for sale so people could order their own, but he has been too busy.

[clearchris]

Originally Posted by stj

someone on eevblog came up with this mod

Nice. I reread this thread yesterday, I'll go through and reread the eevblog post today. Did you find that in the original krypton post or was it another post?

[clearchris]

I went through the Badcaps and EEVBlog thread and pulled out the relevant posts before our most recent discussion. Here it is for reference:


## Link to Krypton's Shorty Page
http://kripton2035.free.fr/Projects/shorty-display.html

## EEVBLOG POSTS OF INTEREST

"this tool should not cost more than $20"

###Clearchris note: BOM pressure
https://www.eevblog.com/forum/testge...28/#msg3461128

"with the integrated 10b adc you can have sufficient resolution to find shorts, you need to be below 1mΩ to find the short precisely. you can play with the adc amplification factor to get this right. but you loose the highest value you can measure at 400mΩ max. with the 18b adc I use, I can go from 10µΩ to around 3.5Ω that's a far better for the use of finding shorts.
and by the way I added a new digit to the display, to read the 10µΩ and it is still quite stable reading !"

###Clearchris note: The code uses only 16b, suspect it's a noise or speed issue to use the full 18b
https://www.eevblog.com/forum/testge...98/#msg3515698

zoltahn:

"I opted for fast conversion and oversampling and I can go up to 1.8 ohms and down to 10 microohms accurately and stable and it seems to me it is perfect for locating the short in efficient and effective and quick manner.
After first three versions I gave up to five digits after comma as you gain little bit more speed and stability and make no difference in comparison with four digits in finding the short circuit.
I have seen at somebody from Russia a schematics of milliohmeter just with MCP 3421 without opamp and it seems also quite ok. On other hand I believe Scullcom used also oversampling with external ADC and opamp on his milliohmeters, I do not remember exactly as at the time I have seen his work I was not interested in milliohmeter subject."

https://www.eevblog.com/forum/testge...94/#msg3515994

"I left the possibility to use the internat 10b adc in my schematic, the aop output goes to the arduino and the mcp3421.
if someone don't have or does not want the 18b adc he can still use the same pcb."

###Clearchris note: AOP line explained
https://www.eevblog.com/forum/testge...70/#msg3516070

nikifena:

"Just a few ideas. What about adding an input protection circuit - two back-to-back 5V Zener diodes or unidirectional TVS? In order to reduce the HF noise what about a 100pF capacitor across R7 and R8 and for better HF DC filtering 100nF across C3."

https://www.eevblog.com/forum/testge...38/#msg3539638

nikifena:

"1. Switching to tactile switches for powering and all other functions. Using 12mm tactile switches with caps has a few benefits - they sit well next to the display and the height is good to fit everything into a box. Also, having an electronic power on, there could be integrated auto power off. Two switches can be used for four different things:
SW1 short press - on/off
SW1 long-press - mute
SW2 short press - relative mode
SW2 long-press - calibration

Auto power off after some time 30-60seconds when the tester is not used.

2. Adding a lithium battery with a built-in charger and battery protection - usually, there are cheap PCBs from Aliexpress and they could be used here in order to make the project simpler. I realized that there is not necessary to have a 5V boost circuit. The voltage from a single battery can be used directly to power the Arduino and the display board. Only the current source left, so I suggest using a precision reference voltage with an enable pin. I found LT4120 as a good solution. There is a version with 2.5V output. Just an extra transistor will increase the current capabilities. A precision voltage source will improve the stability and the readings too.

3. Adding relative mode. There are two ideas:
-A classic relative mode - Just connect the probes to the circuit and press the relative mode to zero the reading or
-First select relative mode, then connect the leads to the circuit. When the value reading is stable, a beep indicates that the relative mode is used. In this mode, when a low value is measured, two fast beeps could indicate that for faster short finding.

4. Increasing the range of the unit. This could be done for the next revision of the unit. Switching some resistors could do that. There could be two more ranges - micro ohm range and ohm range (up to 100 ohms for example) The ranges could be automatically selected.

I've been working as a PCB designer for the last few years, so I'm happy to do a good-looking PCB. I can't modify the code, so I will be happy if kripton2035 will agree to improve the whole project.

I will put a schematics idea soon."

https://www.eevblog.com/forum/testge...94/#msg3540794

"I HATE auto power off systems because they NEVER really switch off. and when you need the device that sat in a drawers for some time, the battery is flat
this never happens with a good old mechanical switch.
I'm already thinking at REL mode and greater range, but as you said it will be for the next release."

https://www.eevblog.com/forum/testge...04/#msg3541104

Nikifena:

"Dear kripton2035, the auto power-off circuit is not leaky and won't suck the battery. Both MOSFETs are off when the circuit is off and there is no leaky path.

For the current path resistors, I choose 2512 footprint for R2 and 1206 for the 1R resistors. The bigger resistors have higher thermal mass to keep the resistor temperature more stable over time keeping the readings accurate. I added some diode protection for protecting the frontend circuit.

The ADC is pretty good and it has a build-in PGA (programmable gain amplifier) of up to 8x gain which could be used for increasing the range of measured resistance. In this case what about reducing slightly the gain of the differential amp to 10?"

###Clearchris note: New shorty schematic attached to this post. Also, does increasing the gain increase the range of measured resistance? I would think that increasing the gain *reduces* the range.
https://www.eevblog.com/forum/testge...28/#msg3541928


nikifena:

"What about removing the current precision opamp gain stage and using only an ADC with built-in PGA with a variable higher gain. There are a couple of up to x128 gain ADC's with 24-bit resolution. The marked in blue MAX11270 and CS1237 are my favorites for now. What is necessary: fast readings with more digits"

##Clearchris note: attached updated schematic and list of ADCs with high gain and sample rate. I don't know if deleting the opamp and substituting with a high gain ADC is a good idea, but higher sample rate could be desirable. I can handle the software end of things if people choose to go this way. It may be desirable to have a higher sample rate with the existing opamp. I'm not convinced more bits is able to be used (though desirable), in my experience, past 16/18 bits tends to be all noise.
https://www.eevblog.com/forum/testge...20/#msg3544020

kripton:

I've implemented 6 digits to my shorty prototype. 10µΩ resolution. with 16 bits and pgax8.
but it's not stable. only the 5th digit is ok. this is not a precision instrument. not sure if 24 bits will add a lot of usability.
you may have to change a lot of other parts of the design to achieve it.

https://www.eevblog.com/forum/testge...22/#msg3544622

kripton:

I've received the pcb's.
had to deal with a voltage regulator mcp1755s not working at all when in parallel with the arduino regulator. this took quite some time to figure out.
rebuild the code to have a simple calibration procedure: you measure 0 ohm and 1 ohm and the calibration is done.
now I only have some instabilities that are not on the prototype ( ! ) that makes the measure instable at "high" ohms
you can measure in theory up to 7 ohms
I also changed the constant current (that was not one) with a zener.
we are almost there. will make a demo movie soon.

https://www.eevblog.com/forum/testge...40/#msg3578140

kripton:

I would like the tool to be able to measure up to 1-300Ω to handle any kind of situation
but it would require a more capable adc.

###clearchris note: seems like scope creep, is this needed for finding shorts?
https://www.eevblog.com/forum/testge...56/#msg3578156

BADCAPS POSTS OF INTEREST:

the microcontroller is an arduino nano
the ADC is an mcp3421 in 16 bits mode, the 18bits being too slow at 3 samples/sec

###clearchris note: speed is an issue with the mcp3421
https://www.badcaps.net/forum/showpo...07&postcount=5

###clearchris note: pic of probe tip connection, twisted two wires to probe?
https://www.badcaps.net/forum/showpo...1&postcount=12

" I'm thinking of adding one pad with a 1Ω resistor to calibrate easily. but it will be defined soon."

###clearchris note: Probably not really useful as a resistor on the board doesn't include the probes
https://www.badcaps.net/forum/showpo...8&postcount=44

I received the pcb some times ago. but there is a failure I did not find until now.
I made some changes in the schematic, using a constant current source for better repetability,
well I'm still investing. the voltage regulator is giving 2.4V instead of 5V
... and it does not seems to be a short circuit !

https://www.badcaps.net/forum/showpo...1&postcount=78

I added a voltage regulator 1755S to bypass the one inside the arduino. just thinking there would be a better regulation than the 1117. I also added a constant current source with the pnp transistor, to have better accuracy and better reproductibility of the measures.
now I'm debugging this new schematic, it's almost working, but as a reminder it is a hobby project, and I have been very busy these last months.

https://www.badcaps.net/forum/showpo...5&postcount=83

I checked on a lab power supply, the whole device is now using 50-60mA.
the current constant source now is using 30mA.

https://www.badcaps.net/forum/showpo...0&postcount=85

transistor is on an analog pin 3 A3
anyway, the transistor is always on, and has been removed from the source code when I use the constant current source.

### clearchris notes: transistor delete?
https://www.badcaps.net/forum/showpo...4&postcount=87

this is the actual state of the schematic of the shorty-with-display input stage.

### clearchris notes: schematic pic of constant current circuit and input stage attached
https://www.badcaps.net/forum/showpo...6&postcount=88

[clearchris]

Ok, I was able to get the software compiled and on a Pi Pico (rp2040). There were some register bit banging, I didn't translate that yet, but I don't see why that wouldn't work.

Again, I'm offering this as an option, I don't have PCB skills, so I'm not driving the bus should anyone decide to make a PCB.