WTF at the bodging ...
You mean the small red jumper? I put it there to isolate power from RF module. To measure RF module current consumption.
WTF at the bodging ...
You mean the small red jumper? I put it there to isolate power from RF module. To measure RF module current consumption.
the bodging of the IC (the blue electro cap)
WTF at the bodging ...
You mean the small red jumper? I put it there to isolate power from RF module. To measure RF module current consumption.
the bodging of the IC (the blue electro cap)
I first wrote about that, and then edited my comment
I think they didn't have enough space on the board to put decoupling cap, so they put it on top of the micro.
I think they didn't have enough space on the board to put decoupling cap, so they put it on top of the micro.
There's much less parasitic series inductance between the cap and the microcontroller's power pins that way. For all its ugliness, it's actually not a bad way to decouple an IC.
Except: it's an electrolytic cap, which IS a bad way to decouple an IC - though even then, the board's designer might well know more about the characteristics of the power supply and its load which mean an electrolytic is exactly what's needed.
I wouldn't be too quick to judge. If space is tight, and labour is cheap, it may be a perfectly good solution to a problem.
Except: it's an electrolytic cap, which IS a bad way to decouple an IC
You said it yourself. And it looks like it's the cheapest cap they found. ESR should go through the roof.
I'm holding this board in my hands right now, and don't see any easier way to put a decoupling cap there, except for putting something on the bottom, which might not be a very desirable thing.
It's the second time I'm taking out this board from the trash can today.
Hi , tnks for the look inside .
I could not see a tilt sensor , that mean it aint got one?
I've seen a few on bay like these ans they all mention sound as a detector , no tilt.
Nope, it doesn't have one. Only shock(vibration) sensor.
CF745 is an "untested" version of the PIC16C54 intended for the Chinese domestic market only. There is also CF775 which is a similarly untested PIC16C57.
Untested means Microchip didn't test the parts before packaging and shipping, and that various functions might not work at all or be up to spec. The advantage is they sell these at really low cost, it's their strategy of selling "genuine" PIC parts in an attempt to displace the various Chinese clones around. (Don't know if they also add some rejects too, since these are intended to be imperfect parts for non-critical applications.)
Thanks a lot amyk! While I was googling for CF745, I've seen some references to PIC16C54 datasheet, but nothing in that datasheet that will point to CF745.
Yikes... no wonder they keep that quiet. I feel queasy.
I know there are Far Eastern manufacturers of microcontrollers which just don't have any presence in western markets at all, and I presume their parts are cheaper. Maybe this is how they do it for the price? Is 100% testing just not always expected?
Yikes... no wonder they keep that quiet. I feel queasy.
I know there are Far Eastern manufacturers of microcontrollers which just don't have any presence in western markets at all, and I presume their parts are cheaper. Maybe this is how they do it for the price? Is 100% testing just not always expected?
Probably depends on the exact manufacturer. But it does make sense from an economic standpoint if a part isn't 100% functional to make a profit selling the parts of it that may still be OK. Similar to how AMD produces multicore CPU dies but sells them cheaper with one or more cores disabled because they didn't pass testing. Likewise an MCU with broken subsystems would still be usable in an application that didn't make use of them.
Microchip also has
MCV08A/14A/28A and the similarly obscure HAxxxx series which are apparently for "home appliance" use.
It's known as chip harvesting
Intel does the same, the cheaper Celerons are often those that have a failed core or a bad half of cache. During test you find the partly faulty and have enough extra logic to bypass the bad parts and do so with either a fuse link or a non accessible program pad on the die. You see it on the datasheet where you have pins that are unlabelled or marked test but always have to be connected to a ground.
If you have a partly functional chip in a package ( passed the quick die test so had the most expensive step of packaging done) and you can batch them as to the failed portions then there is no good reason to not sell them to a user who will be able to use the limited functionality, and recover the cost of the testing and packaging. No brand of the manufacturer of course, but a house code and a lot number. you can then save on scrapping the defects.
It's better then doing a complete redesign which costs lots of money.
Right?
There was a time when Celerons were equal or actually faster than the Pentium parts
Can't really remember when but i think that was like PII or PIII?
If you shut off half the power dissipation you can run the remaining a lot faster to get the same dissipation. Intel now calls this Speedstep, basically if one core is idling along the other can be boosted to a higher speed for a while.
If you shut off half the power dissipation you can run the remaining a lot faster to get the same dissipation. Intel now calls this Speedstep, basically if one core is idling along the other can be boosted to a higher speed for a while.
Intel doesn't call it speedstep, speedstep existed since the original Pentiums and that was just to step the clock frequencies lower when not in use like what AMD calls Cool'N'Quiet but the one you are looking for is called Turbo Boost
Speaking of AMD TurboCORE, they did something marvellous with their trinity laptop processors
Firstly they increased the IPC by a large amount thanks to their piledriver cores
Secondly they increased the clock frequency to 2300MHz base but that goes up to 3.2GHz with 3 cores(!) and the GPU can be tuned up
http://www.anandtech.com/show/5831/amd-trinity-review-a10-4600m-a-new-hope/2And harvested lower core models ... The A8-4xxx series < Quad core but not that fast A6-4xxx < Dual core but as quick computing single core
I was wrong, but then again i run on the trailing edge of CPU's, just this year went dual core ( got it on fire sale as they were finishing the line) to replace the single core. The newer processors are a lot less reliable, I have seen a few that are dead along with the motherboard. I tried once to kill a first gen Celeron with an electric fence energiser. Still worked, even after a few minutes of being belted between pin sets. Only way to kill it was to connect to a 12V car battery and burn the pins off.
It is expected for robustness to decrease with each process shrink, as the structures like gate oxide become more delicate and sensitive to damage. The PICs are still on >0.1u process.
That's to be expected with a 22nm-32nm process!
Intel doesn't call it speedstep, speedstep existed since the original Pentiums and that was just to step the clock frequencies lower when not in use like what AMD calls Cool'N'Quiet but the one you are looking for is called Turbo Boost
And if you actually buy a good heatsink and fans, it's possible to tweak a 6 core i7 to run full turbo on all 6 cores for basically an unlimited time.
Intel doesn't call it speedstep, speedstep existed since the original Pentiums and that was just to step the clock frequencies lower when not in use like what AMD calls Cool'N'Quiet but the one you are looking for is called Turbo Boost
And if you actually buy a good heatsink and fans, it's possible to tweak a 6 core i7 to run full turbo on all 6 cores for basically an unlimited time.
Which is pretty much overclocking
Thanks for sharing,I will try
