Ghost in the system?

[Czauto]

Hi guys,

I noticed something very strange this morning. I managed to get a very short pike monitored by ICC. Very short and very High. 98,000Watts to be precise. Inverter percentage spiked to 400% at the same time but nothing else spiked. Any ideas? I suspect it was the Axpert's idea of a joke. Haven't noticed anything funny after that. Dashboard shows Max load day = 97885Watts.

I'm going to check for bad connections after work. All I can think of is a loose Neutral connection somewhere.........

[Czauto]

that would be a spike, not a pike

[Guest]

I would not worry. We have seen same data coming in.

[Czauto]

Almost got a heart attack.

I inverted the axperts cooling fans last week. Cannot believe the temperature difference. Average of of 20degrees drop. Why would they not fit the fans with upwards airflow in the first place??

Sent from my S60 using Tapatalk

[___]

False reading. The kind of thing that happens when you do a subtraction sum with the wrong data type.

0 bottles of beer on the wall! 0 bottles of beer! Take one down, pass it around, 18446744073709551615 bottles of beer on the wall!

[Guest]

35 minutes ago, plonkster said:

0 bottles of beer on the wall

Thanks Plonkster ... now that tune is stuck in my head! (cwl)

[___]

True story. I once owned a 1.4 liter fuel injected Citigolf. The particular ECU in that car was known as the MP9. Now the MP9 had an interesting little bug. It had a TPS (throttle position sensor) on the throttle body, as well as a stepper motor that controller the idle speed. It would use the TPS to measure the resting position of the throttle. This value was also stored in nonvolatile memory the moment you turned off the ignition. For the most part it worked well. As time went by, the resting position of the throttle would gradually increase as dirt collected in the throttle body. But then, usually when the car was around 5 years old, some of that dirt would occasionally get dislodged and the throttle body would register a lower resting position than yesterday. All fine and well, except the ECU had a bug where it would do a subtraction between this value and the stored value. This should lead to a negative number, but in this case it wrapped and became a very large positive value. The ECU had safeguards though: When that happens, it logs an error (throttle position out of bounds) and defaults to safe values. This in turn causes erratic idle and messes with low speed driveability.

The firmware apparently cannot be fixed. The only fix is to clean/replace the throttle body and reset the stored value to the factory value, so it can start at the bottom, buying you another few years. I was in luck, I knew the guys who made a reset tool. The trouble as they explained it to me is that while they can write to the storage, they cannot modify the value in memory. You needed to have the ignition on to program the ECU, and that would load the current stored value to RAM. Even if you overwrote the value in storage, it would simply reset to the wrong value when you turn off the ignition. The trick was therefore to overwrite the stored value, and then rip out the ECU connector so it could not overwrite it.

In the end it only cost me a new throttle body. I heard horror stories of people who also replaced the ECU, which as you might imagine, is not cheap.

[Czauto]

Wait...what? I'm completely confused now. I don't drink beer, only ciders[emoji23][emoji39]

Sent from my S60 using Tapatalk

[Sidewinder]

I saw more or less the same happen to me on Sunday afternoon. I trying to completely setup my Infini, I thought that the settings from the LCD is only a few of the basic setting that is required to get it working. I loaded the factory CD software and connected it to the Inverter with a RS232 cable. I forgot to unplug the USB, so for the next day, I would get 0V input from Grid reported on ICC Tread Info, Emoncms etc.The main thing was that the Load Watts was hovering around -29699W! Reboot of Pi brought things under control. I'll try again this weekend without USB cable!

[___]

14 hours ago, The Terrible Triplett said:

Thanks Plonkster ... now that tune is stuck in my head! (cwl)

Fwiw, I did not just make up that number. If you know where it comes from... well then (like me) you probably didn't spend much time in the sun when you were younger. Other possibilities are 4294967295, 65535 and 255 :-)

There is this other story, when my first child was born and the nurses strapped on that contraction monitoring device (made by Philips), I asked them what the numbers mean and they said they don't know, all they know is it stops at 127. I sat there grinning like an idiot after that...

[Czauto]

43 minutes ago, plonkster said:

Fwiw, I did not just make up that number. If you know where it comes from... well then (like me) you probably didn't spend much time in the sun when you were younger. Other possibilities are 4294967295, 65535 and 255 :-)

There is this other story, when my first child was born and the nurses strapped on that contraction monitoring device (made by Philips), I asked them what the numbers mean and they said they don't know, all they know is it stops at 127. I sat there grinning like an idiot after that...

I would guess something to do with binary code, which I know very little about so you lost me at the 0 bottles of beer on the wall!!

[___]

30 minutes ago, Czauto said:

I would guess something to do with binary code, which I know very little about so you lost me at the 0 bottles of beer on the wall!!

Okay, so this has to do with how numbers are represented in a computer. Such numbers can either be signed or unsigned. Unsigned numbers are used when we never expect the results to be negative. The upside is that we can store twice as large a number if that assumption holds true.

Very basic math. You may recall from grade 1 how you can do addition by splitting numbers into ones, tens, hundreds, thousands, etc... where you have ten possible digits to employ in each slot. Now with binary we only have two, so instead our columns represents ones, twos, fours, eights, sixteens, etc... as you can see, it doubles each time. We say that the decimal system has a base of ten, while the binary system has a base of 2, so:

dec(1) = bin(1)
dec(2) = bin(10)
dec(3) = bin(11)
dec(4) = bin(100)

As you can see, just like the decimal system, when you reach the highest available digit (a one), it rolls over to a zero and it carries a one.

With that basic concept out of the way, lets assume we have a CPU with registers that are 8 bits wide. So the largest binary number we can write is 11111111, and working from the right using the grade-1 column approach, 1+2+4+8+16+32+64+128 = 255. The largest number we can represent with 8 bits is 255. But we cannot represent negative numbers.

What computers do to represent negative numbers is sacrifice one bit, usually the one on the left. If that bit is a 1, then the number is negative. If it is a 0, the number is positive. This however means you only have 7 bits left to represent the actual number, and once again, bin(1111111) = dec(127).

To summarise then, with unsigned integers we can represent from 0 to +255, and with unsigned integers we can represent from -127 to +127. Note however that for the first 127 numbers the representation in binary (whether signed or unsigned) is exactly the same, in other words, as you count up from the bottom everything goes fine until you hit 127 (01111111 in binary). Adding one to that will cause a one to be carried into position 8, ie you will end up with (10000000). Now, if you're using unsigned types, that is equal to 128 BUT.... if it is signed, that is -128 (exact reason why it works that way is the remaining 7 bits is the Two's Compliment representation, not important here).

Now imagine you were counting backwards from 255 to zero. The same thing happens, especially in smaller CPUs that don't detect overflow. As you hit 00000000, and you subtract one from that, it "clocks over" to 11111111, which SHOULD be -1, but if you incorrectly interpret it as unsigned, it will be 255 instead.

The maximum number you can represent with 8 bits is 255 as I said, but for 16 bits that number is 65535, for 32 bits it is 4294967295, and for 64 bits it is 18446744073709551615.

The reason why I found the 127 limit on the contraction meter amusing, is because it leaked a tiny bit of info about the internal implementation: They are using a 8-bit signed integer and have properly added safeguards to limit overflow. It's interesting that they used signed math, because one would assume contractions can't be negative? But then, it is also common for a programmer to use signed math even when not necessary precisely to avoid the bugs that creep in when things accidentally go negative.

Now... that large spike in the consumption? What probably happened is there was a large spike in consumption, large enough that the left-most binary digit turned into a one. With sufficient chains of interpretation, some which assume the number to be signed and some which don't, that can turn into impressively large numbers. Imagine if it started as a 16 bit number, went to 32768 (which is -1 if you assume signed), and you accidentally re-encoded -1 as a 32-bit number and made the mistake a second time. Well, now 32768 turned into 4294967295.

It's probably something like that...

Edit: And this is why programmers confuse Christmas and Halloween. Because Dec(25) = Oct(31). Octal being a numbering system using 8 as a base, ie you count up to 7, then it clocks over. And then there is Hex that counts up to 15 (0 to 9 and A to F)... it took me a decade to learn to read hex and octal but once I did... so much simpler than converting back to decimal :-)

[Czauto]

En daar donner my tor van sy bootjie af

[___]

48 minutes ago, Czauto said:

En daar donner my tor van sy bootjie af

Let me dumb it down :-)

Computers have limited numbers and they work like those old car odo meters that "wrap around". When you spin them past 999999, they go back to 000000. But sometimes we number them from -500000 to +499999 because we also need to display negative numbers (ie zero is in the middle of the range now). Sometimes we don't know what kind we have in our hands and when we attempt to turn to 500 000km we end up at -500 000 instead, and at other times we turn it back aiming for -1 and we end up at 999999 instead.

:-)

[Czauto]

No I get it. nothing more interesting than binary and hexadecimal, I've just never spent the time to properly introduce myself to it.........maar my tor het amper verdink!!

[SilverNodashi]

98 is a very specific number. I would imagine that one of the sensors in the inverter did this. a DS18B20, for example, has a startup temperature of 85 degrees. My guess is that whatever caused it, just "restarted" and that's what you saw.  

[Chris Hobson]

On 22/05/2017 at 5:56 PM, Czauto said:

 

Why would they not fit the fans with upwards airflow in the first place??

Sent from my S60 using Tapatalk
 

I have flipped mine too. It is a matter of conjecture but an explanation I have heard is they expect the Inverter to be mounted above batteries and the downward airflow was an effort to prevent battery vapours entering the Inverter. Is this true? Maybe

[Czauto]

That's a stupid assumption if you ask me. You what they say about assumption.......
I never even hear my fans anymore. Temp avg just over 40°. Heck it's so quiet now I'll even install it in my bedroom[emoji3]🙄

Sent from my S60 using Tapatalk

[Guest]

11 hours ago, Czauto said:

That's a stupid assumption ...

Yes it is but what is not is that you don't want the hydrogen sucked into the machine more than what is happening already IF it is mounted above the batts.

I never thought of that.

Had my T105RE's in my office (dommer as grond). I smelt the hydrogen at times (even though I have good ventilation) so after seeing P,lonksters problem his batts caused in his garage, I checked and saw my tools starting to get a coating of fine rust in my office. 

So I would hazard and say ... makes sense to blow out the bottom to disperse more of the hydrogen for electronics can take a wee bit more heat than hydrogen - IF your inverter is directly above the batts.

[Czauto]

It does make sense, BUT........seeing that you can decrease the inverter's temperature by avg 20deg (which I think can only be good for the inverter), why not rather put a big sticker on the inverter saying "请勿直接挂在电池上方" or for the rest of the world using their product "Do not mount this unit directly above batteries" 

I fact, why don't they just ad a bookmark of this forum to their engineer's desktops? I'm sure they will learn A LOT from this forum and it's members. Who knows, with a little improvement they can maybe start painting the Axpert's blue

[PaulF007]

47 minutes ago, Czauto said:

It does make sense, BUT........seeing that you can decrease the inverter's temperature by avg 20deg (which I think can only be good for the inverter), why not rather put a big sticker on the inverter saying "请勿直接挂在电池上方" 

 - nou het ek lekker gelag!!

 

[SilverNodashi]

1 hour ago, Czauto said:

It does make sense, BUT........seeing that you can decrease the inverter's temperature by avg 20deg (which I think can only be good for the inverter), why not rather put a big sticker on the inverter saying "请勿直接挂在电池上方" or for the rest of the world using their product "Do not mount this unit directly above batteries" 

I fact, why don't they just ad a bookmark of this forum to their engineer's desktops? I'm sure they will learn A LOT from this forum and it's members. Who knows, with a little improvement they can maybe start painting the Axpert's blue

Hahaha!

[Guest]

8 hours ago, Czauto said:

I'm sure they will learn A LOT from this forum and it's members.

I sense a bit of something, irritation at a product or some such vague thoguht ... that is after I laughed at the  "请勿直接挂在电池上方" instruction. 

[Czauto]

I sense a bit of something, irritation at a product or some such vague thoguht ... that is after I laughed at the  "请勿直接挂在电池上方" instruction. 

Believe me, none of us would have Axperts hanging against their walls if we could get the blue stuff for the same price. Yes there is irritations, etc but hey, if you're stingy.....shut up![emoji23]

The Chinese translation was courtesy of Google translate. [emoji12]

Sent from my S60 using Tapatalk

[___]

17 hours ago, Czauto said:

Believe me, none of us would have Axperts hanging against their walls if we could get the blue stuff for the same price. Yes there is irritations, etc but hey, if you're stingy.....shut up!

I still think we need to consider the Infinisolar in this comparison, because that is the blue inverter's natural competitor. The very fact that people choose the Axpert over the Infini actually answers this question perfectly without having to resort to brand snobishness :-)