BB AI-64 + KiwiSDR 2 NO HARDWARE DETECTED
Hi,
assembled a BB AI-64 + KiwiSDR 2 board and get NO KIWI HARDWARE DETECTED.
Steps done so far:
* Flashed + booted http://kiwisdr.com/files/KiwiSDR_v1.665_BBAI_64_Debian_11.9.img.xz -> shutdown -> power off
* Attached the board according to https://forum.kiwisdr.com/index.php?p=/discussion/comment/14651/#Comment_14651
* Attached the PSU again to BB AI-64 (not to the kiwi board)
* Booted -> connected via ssh -> run via root (via multiple steps) "apt-get update && apt-get upgrade && reboot && cdp && make install_kiwi_device_tree && reboot && cdp && up && reboot"
journalctl -u kiwisd.service tells (see full log attached):
* EEPROM check: open /sys/bus/i2c/devices/5-0054/eeprom No such file or directory
* NO KIWI HARDWARE DETECTED
Looked like a hardware issue, so started to measure according to http://www.kiwisdr.com/ks/troubleshooting.pdf:
* 3.3A: 0.52V
* 1V: 0.43V
* 1.8V: 0.57V
* 3.3G: 0.6V
* 5V in: 0.5V
* 3.3V GPS: 0.53V
* 3.3V: 4.04V
* 5VE: 0.9V
* 5V: 0.86V
All too low BUT 3.3V being too high (=4.04V)?
The PSU (MDR-60-5 = 5V and 10A) delivers exactly 5.03V to the BB AI-64 and the BB AI-64 runs stable under load. But something seems fundamental wrong.
Is the BB AI-64 broken? Or is the kiwi board broken? What else can I do? Is it worth getting another BB AI-64 or giving the kiwi board also direct power supply (from the same PSU)? Any idea welcome.
Many thanks
Comments
Disconnect the beagle board from the Kiwi. Measure the resistance to ground at all the power control points listed on the first page of this guide and display the results.
The resistance measure result against ground:
Thanks!
Did the Kiwi board ever work before being used in the AI64? Like with a BBG?
Did you follow the AI64 installation instructions avoiding the extra P9 connector pins? Because if you didn't the Kiwi board is likely destroyed:
https://forum.kiwisdr.com/index.php?p=/discussion/comment/14651/#Comment_14651
Also: DON'T EVER use a multimeter to measure the resistance of FPGA low voltage power inputs like the Kiwi 1.0 and 1.8V regulator output test points. The multimeter has a source voltage in resistance mode that will likely exceed the maximum allowed voltage on those FPGA pins.
the kiwi board was never used before ... ordered it as an extra board end of 2023 and got it beginning of 2024 together with a completely assembled BBG KiwiSDR 2. Use the assembled kiwi so far and now found the time to get the extra board attached to an BBAI64.
Yikes, my multimeter measures resistance with 2.8V. I hope this didn't brick it.
Try the board using the BBG of the Kiwi-2. You're very unlikely to hurt the BBG from doing so.
ok. I'll test with another BB ... gave the BBG Kiwi-2 to a friend... either I'll get that back for testing or I'll get another BBG or BBAI64. This will take some days. I'll come back.
Thanks so far!
@jks , Although the multimeter is in resistance measurement mode and has a high voltage on the probes, it does not provide enough current to cause components to fail. I have never seen anything like this!
@dbast , Your measurements show very low resistances! How could this happen? A thunderstorm?
@studentkra the board was never used before... ordered it as extra board and didn't use it before... it was stored disassembled and safely packed inside the house. No thunderstorm or anything else should have touched it so far.
low resistance as too low or is there still a chance that this will work with another BB?
@dbast, are you realy measure resitanse, or use testing diode metod? If it realy resistanse, so.. its too bad. I dont know, how its heppend..
@studentkra its a cheap 27 years old multimeter ... not sure what it actually does
@studentkra Yes the current is low. But the cost of devices with low (1.2, 1.0, 0.8) Vcc supplies is typically high (advanced FPGAs, CPUs). I wouldn't risk possible problems with the breakdown voltage of ESD diodes on low voltage pins. See eevblog forum for more info. Multimeters specifically designed with low burden voltages (< 1V) are a thing.
@dbast , The lack of voltage on pin 5VE looks very, very strange. After all, this voltage comes directly from the BB-64 AI power connector. Perhaps the KiwiSDR board is connected incorrectly? I cann not attach a picture with schmatic of BB-64 AI. I got a "network error" message.
@jks Hi, John! I've been browsing the eevblog forum, but I haven't found a suitable post about multimeters. Could you provide me with an accurate link to this topic? I've been repairing electronics for many years. This is the first time I've heard that a multimeter can damage electronic components. Thank you!
[Warning: AI slop summary below links. YMMV]
Warns that DMM test voltages (3–10V) can exceed breakdown voltages of low-voltage components, causing hot carrier effects, noise, or failure. Risks include base-emitter breakdown in bipolar transistors or gate damage in MOSFETs/ICs. Advice: Avoid on delicate, unpowered circuits; use low-voltage modes if available.Warns that DMM test voltages (3–10V) can exceed breakdown voltages of low-voltage components, causing hot carrier effects, noise, or failure. Risks include base-emitter breakdown in bipolar transistors or gate damage in MOSFETs/ICs. Advice: Avoid on delicate, unpowered circuits; use low-voltage modes if available. Warns that DMM test voltages (3–10V) can exceed breakdown voltages of low-voltage components, causing hot carrier effects, noise, or failure. Risks include base-emitter breakdown in bipolar transistors or gate damage in MOSFETs/ICs. Advice: Avoid on delicate, unpowered circuits; use low-voltage modes if available.
Focuses on continuity (a form of low-resistance measurement) voltages (up to 7.9V) damaging high-impedance inputs in low-voltage ICs/microcontrollers (e.g., 3.6V max). Can forward-bias ESD diodes or cause false readings/shorts. Recommends meters with <1V output for safety.
Users share FPGA failure stories, including overvoltage from probes/tools. While not multimeter-specific, one post mentions accidental voltage injection during unpowered testing (e.g., via probe slips), relevant to resistance probing.
Discusses safe probing of ICs, including avoiding resistance measurements on powered or low-voltage pins due to test current/voltage risks.
----
Re "network error" message: I think this is just this forum's poor way of saying it wants a .jpg or .pdf file. Or it could be a database insert error. There have been many times when I post something, or do an edit, and it fails with some error, but succeeds on a second try. The forum software is terrible.
Thanks to John and AI for this links )) I will study it!
I now tried everything.. measured the beagleboard without kiwi board etc and then realised after a while ... well, cough ... I managed to spare the right pins, but all I can say 180° ... /me hides. Now everything works. Sorry for the confusion and thanks for the great support here!
180° ?! You mean to say... Damn! But how?! )))))
Do I understand correctly? The Kiwi board survived being plugged into the AI64 rotated 180°? That's amazing. Perhaps this worked because you also used the opposite end of the P8/P9 connector where power would not have been applied to incorrect pins potentially causing damage. The power pins are in the green area of the "left" end of P9 as shown in this image: https://forum.kiwisdr.com/index.php?p=/discussion/comment/14651/#Comment_14651
Ah! That's the problem. On the BB 64-AI, the ethrenet connector is located on the other side. Now I understand "how" ))
Yes, orientation of the cape on an AI64 is completely ambiguous. I don't think there are even silkscreen markings for P8/P9 on the PCB (unless they fixed that on builds later than what I have). Worse, they added those 4 non-standard power pins to the "left" end of P9-1. That's why I spent some time creating that diagram.
yes, I spared the marked red "Do not use!" pins and it survived full 180° connected to the green area... so p8 was connected to p9 (no shift outside of the green area)... "survived" in a sense that the board is now detected, showing a complete spectrum with some poor indoor antenna. Next: Doing a Self-test.