<krzk>
I tried few relays, including ones from AliExpress which all look the same... but the AliExpress ones barely work or do not work at all. It seems they sell sometimes relays which do not pass quality check :)
<krzk>
wwilly: so better to buy from regular store
<krzk>
With the voltage drops I had issues multiple times. I have pretty long cables, also with a cube-board in the middle (bottom-left on th epicture)
<wwilly>
krzk, I have the same thing with a cube-board
<wwilly>
I guess the relays died...
<wwilly>
it's benching like almost all the time, cpu at high speed...
<krzk>
All my problems of voltage drop was the cable connectors. All. You have to be sure that the two cables are really connected. When I was using cube or relay itself and just plugged the cable - I had huge drops, 5.2 V -> 3V with high load
<krzk>
I measured all the drops and when do they happen with oscilloscope and the solution - proper connecting
<krzk>
Let me prepare a picture...
<wwilly>
I use a tester, and have 0V sometime directly from the relay screw
<krzk>
V?
<krzk>
0V? That's crazy :)
<wwilly>
yes, I guess the relay is dead
<wwilly>
because isn't 0V all the time, and other relay isn't affected
<wwilly>
I don't know the physics of the relay, maybe heat of the relay or a shock move something inside... ?
<krzk>
So I moved the connections of cables from relay to another cube-board and joined the cables with proper tool
<wwilly>
ok nice
<krzk>
So check your connections. Also depends what cable diameter you have - there might be significant drop for high current and long cables
<krzk>
In my case these are I think 1mm cables so very wide, but necessary for 5V and 2A
<krzk>
The relay itself (assuming you have similar) has a coil and moving parts which last for 10k of switches.
<krzk>
Depends... cheap ones might last 1k... good ones maybe more
<wwilly>
ok, I haven't switch on/off the relay so often
<krzk>
Anyway, the first is to check the cable connections, how they are connected together and how they fit into the erlay sockets (screwed enough or not)
<wwilly>
my wires are UL1015 18 AWG, less than a meter I guess
<wwilly>
yeah but I tested voltage from the screw head of the relay, this screw shouldn't be affected by the connected wire right?
<krzk>
18 AWG is quite wide (1 mm), so good enough. Then the connections.
<krzk>
What do you mean by "affected"?
<krzk>
If it is not screwed properly, it will be affected
<wwilly>
uhm yes
<krzk>
18 AWG by itself on 1 meter cable will have a voltage drop of 100mV, so should not affect too much itself
<wwilly>
I will rewire my stuff
<wwilly>
what is your wire gauge?
<krzk>
My power supply is adjusted to provide 5.25 V, so to the boards gets ~5.1V which fits in the tolerance (usually tolerance is 5%)
<krzk>
1 mm, so 18 AWG as well
<krzk>
but in few places I was using 0.5 mm because they did not fit in to the cube (two cables into one socket)
<wwilly>
ok I haven't adjusted the power supply, but other board are working properly without voltage drop
<wwilly>
ok
<wwilly>
the tape is to limit electrocution when you have your hand on it?
<krzk>
Yeah, only to cover it from accidental touches
<wwilly>
the "thing" is that I soldered myself the plug for the board, maybe I did "shit" for this one plug...
<wwilly>
because the other are working perfectly fine
<wwilly>
arf, back to solder again...
<krzk>
Could be. All the joints/connections between elements are candidates for voltage drop. The same if some things do not fit (too big or too small).
<krzk>
The actual plugs I bought from Alixpress for cheap - I have plenty of them for different configurations and it will be still better quality than my soldering :)
<krzk>
So the only soldering I had to do, is the voltage divider for one board 3.3V -> 1.8V for serial/UART
<krzk>
The relay you have is good, similar model as mine. The PSU is different - I don't know the brands that much to judge. It is quite cheap for that power :)
<krzk>
At the end the quality of PSU output also is important, if you run for example multiple boards all the time. You know, that 350W is in theory :)
<krzk>
If you have some oscilloscope or good meter unit you can check whether there are drop outs on the PSU exit - then it would be a hint it is not enough.
<krzk>
but before buying anything new - check the cable connections :)
<wwilly>
yeah, if by any chance I've bad soldering skill, and the connection shortcut, do this kill the relay and or relay board?
snawrocki has joined #linux-exynos
<krzk>
wwilly: You can kill the board or the power supply
<krzk>
The relay actually only in case of short during the switch... so yes
<wwilly>
ok :(
<krzk>
if during the switch, huge current goes via relay, it can create an arc there and melt the relay connectors of this coil
<krzk>
But relay is cheap :) Boards are more important
<wwilly>
the board runs fine with another cable
<wwilly>
ok, time to change relay then, and rewiring too
<krzk>
I mean, not connectors of coil but connectors of the wires operated by coil. I don't know the english terminology here.
<wwilly>
pfff, will never reach the end of the PhD with this wasting time arf....
<wwilly>
the fine wire inside the relay to make the coil you mean?
<wwilly>
also, do you think that I could relay a 220VAC by this module? the relay say so, but the board?
<wwilly>
like give a 220VAC by this, and use the normal power block for the board
<wwilly>
I'm asking this because I also have a jetson tx1 and tx2 that use 19V
<wwilly>
so my idea was to use this relay board to control a 220VAC, and use stock power block for the board
<krzk>
I mean, that the relay has a switch and these switch connectors/wires/ends/pins are either disconnecter or connected. In the transition period there is always some smaller or bigger arc. If huge current goes, the arc could be bigger.
<krzk>
the relay is as well for 220 V AC so you could do it. You should be just careful about isolation but except of this, no problem.
<krzk>
and of course different topic is to mix up cables so you won't connect 220 to Odroid expecting 5 V DC :)
<wwilly>
krzk, yeah, I would use the stock power block to attack the board then
<wwilly>
just this relay board deliver the 220VAC to the stock power block of the board
forkbomb has quit [Ping timeout: 256 seconds]
<krzk>
wwilly: this should work. I never did something like this with my setup but it is the same as physical swithc on some power/socket bar.
<wwilly>
krzk, the fun fact also is that the stock power block outputs 5VDC 4A, written on the sticker. however, when you have a nice cooler on the board itself, and run a stress workload that correctly use the superscalar cpu capacity, the power sensor on the board (on the XU3), the cpu could draw like 8A just for the big cluster
<wwilly>
I use a microbenchmark that stress only the superscalar integer part of the cpu
<wwilly>
when all 4 cores of the big cluster run my microbenchmark, the board could draw this much of power. but quickly reduce the cpu frequency because of heat and draw far less
<wwilly>
so, I guess the stock power block rate 4A but for a short time, knowing that the board will not drawn much after "stationary" load with a steady non critical temperature
<snawrocki>
hey, CPU operates on much lower voltage than the main 5V supply and there is a switching power regulator on the board which could deliver much higher current at 1V than what is drawn from the external power adapter
<krzk>
wwilly: 8A? i guess it is not on 5V, like snawrocki is saying. it's too much for the board.
<krzk>
The highest values I saw were around 2.5 A (although with GPU and few USB devices this could go higher.. but not 8)
<wwilly>
yea, cpu at 1.3ishV
<wwilly>
arf, my bad, 8A wasn't the right metric, but W .... my bad
forkbomb has joined #linux-exynos
fourkbomb has joined #linux-exynos
forkbomb has quit [Ping timeout: 272 seconds]
_whitelogger has joined #linux-exynos
chewitt_ has joined #linux-exynos
chewitt has quit [Ping timeout: 246 seconds]
dlan has joined #linux-exynos
dlan has joined #linux-exynos
dlan has quit [Changing host]
chewitt_ is now known as chewitt
genii has joined #linux-exynos
mszyprow|home has quit [Ping timeout: 240 seconds]
wwilly_ has joined #linux-exynos
wwilly has quit [Ping timeout: 265 seconds]
mszyprow|home has joined #linux-exynos
mszyprow|home has quit [Ping timeout: 265 seconds]
wwilly_ has quit [Quit: Leaving]
mszyprow|home has joined #linux-exynos
mszyprow|home has quit [Ping timeout: 258 seconds]
snawrocki has quit [Ping timeout: 240 seconds]
paulk-leonov has quit [Ping timeout: 260 seconds]
paulk-leonov has joined #linux-exynos
steev has quit [Read error: Connection reset by peer]