Unfortunately, I only have an old analog oscilloscope. I can provide measurements from my multimeter.
When switched on, peak battery current is 144mA, while the response time of the multimeter is 100ms.
When switched on, peak battery current is 3.77A, while the response time of the multimeter is 250μs. (charging capacitors)
The steady battery consumption after switching on is 54mA.
Pictures here:
During initialisation the processor will send increasing pulses to both outputs.
These pulses cause voltage spikes across the current shunt (R30) which are measured by PA0 (pin 40).
If PA0 measures nothing or unexpected values you will see Failure 20.
The probe is connected to PA0:
What you can do with your scope is check after start up for activity on the outputs of U8, U10 and the gates of the mosfets.
Replace resistors R35 and R46 these should be 200k.
Yes. I’m stupid I did not notice the assembly. I took the packaging components and trusted vendors that the values on the packaging match the content.
After replacing resistors everything works as it should.
Thank you very much for all that helped me to clarify this problem.
You’re welcome! I’m glad that it works.
Hi, I’m currently also running into the failure 20 and after two evenings of trying to figure out what’s causing it I’m absolut clueless.
If i mention in the bottom part that something seems to be normal to me it’s just my impression. It’s my first unit so i could be wrong and the behavior is not normal
First the Board (latest 1.3): (24 pin header is the wrong way, but i fixed that with the cables)
Some things i’ve checked:
When I hooked the Scope to the PA0 Pin on the Atmega it looked much different
So next I measure the 520N and the IRF9Z24 fets. All have 8,9V at the drain pin
Some scrope images from the gate pins
Both IRF fets (Q3 and Q6 look normal to me. Gate ~6.1V, Drain and Source 8,9V
520N Q2 and Q1 (for Channel 1) seems to be also normal (?)
But with Q5 and Q4 (for Channel 2) my headache begins
So next im measured the PB Line/Pin on the Atmega and the Gate of the 520N together
Channel A example like Q3 (yellow) and PB2 (blue)
Channel B example of Q5 (yellow) and PB1
That signal is so low and short that i’m not even sure it’s a real output of the Atmega
So now I’m really confused. If Channel A works as intended (which I guess, but don’t know), why don’t i get any signal from PB0 and PB1 to the mosfet gates of Channel B?
Or does the software check one channel after another, so there is a problem in Channel A which i don’t see so it doesn’t even trigger the test on Channel B?
I hope this post didn’t got too long 
Any help or suggestion appreciated.
Hi, maybe a silly idea but do you have right transformers with right secondary impedance? Do you have a correct value of current sensing resistor?
Hi, thanks for the reply. The transformers are the ones that were listed in the BOM list so they should be right. I’ve double checked the resistors and they’re correct too - i’m just not sure how much tolerance the failure check routine allows?
I’m currently building a second unit and just wait tor the second set of transistors. My hope is that the second one won’t get failure 20 so i’ve a working unit to do some A/B measurements - that would help a lot. If i should figure out what the problem was i’ll update this 
My cheap component tester showed a much higher Gate Threshold Voltage on the Fets than it should be. I’m not sure how accurate the readings of those component testers are, but maybe my FETs are the problem.
I was surprised you have right transformers because they are out of stock at Mouser - I had to use 42TU series. I am still waiting for a few components so hopefully it will work.
My component tester measures this on transistor out of the box (ESD safe box ):
IRL520NPBF Vt=1,86V
IRF9Z24PBF Vt=3,58V
I was lucky enough that someone else gave me the first set of transformer. The second set for board two I got from ebay.co.uk. Someone from UK is selling a few there - not the cheapest (paid like 18€ for two shipped to Germany) but better than waiting months.
My second board works now so now i can do some A/B measurements between the working and non working board. My first obversation was that i measure a difference on R34. On Both boards i measure ~5V on the top pad of the resistor, but on the bottom it’s 4.1 on the working board vs 4.3/4.4V on the
non working one. If I remove the LM358N underneath i measure 3.3V on both. I’ve swapped the L358N but they’re both fine so the issue has to be somewhere around it.
I finished my MK-312 today and I also get Failure 20. Incredible… i started my own topic with measurements.
Hello,
I want to share my experience.
I found the source of my failure 20:
2 DAC 1661 was installed, only 1 is necessary !
Now, 1 socket is empty, but the board is ready!
Regards
Julien
how is that even possible? 
I put 2 LTC1661 for U8A and U8B.
Now, only U8B.
The BOM file wasn’t so clear…
(i’m french)
JR
Hello all,
I builded MK312 and flashed f005.bin firmware. Display is working and shows Failure 20. I checked all resistirs and components and Im not able find some problem.
I tryed measure signals on pins 1, 2, 3 and 4 on ATMEGA = signals to gates of mosfets and on these pins aren’t signals. It looks that output to mosfets is not working. The same situation is on PA0 - signal = zerro.
Do you have some idea where can be problem? Is there some action in initialization process before tree testing pulses in startup process?
Thank you for support so much!
Michal
Hi,
Can you post some pictures ?
Best luck
Hi Nestor, pictures are below. My PCB is homemade = no professional production but I checked all connections and parts and looks OK. When I switch it ON, I have error Failure 20 and on PA0 pin is not visible signal with tree test pulses - nothint. I checked output from ATMEGA and LM359 to MOSFETS and nothink - no signal after switcing on. All voltages from regulator OK, only on output of LM2941 in 14 volts and not 12. I tested it with 12V on batery input with the same result.
Display is showing message, but backlight is blinking in interval around 10 seconds = Short backlight ON and than 10 second without backlight. The same situation with blinking is on LED Output A and Output B.
I tryed firmware f005.bin and t002_bootloader.bin
What is the exact type of the P-channel FETs you use?
Hello, P FETs are IRF9Z24N, N are IRL520N. All components from Mouser based on buttshock list of parts.
You use a wrong DAC. With the current firmware you should use the LTC part not the MAXIM.
