MediaTek MT5389

[jimmyboy]

Anyone have specs or details on the pinout? I sent Mediatek a few emails and no response

a datasheet would be even better but I am not going to get my hopes up
the best I could find was a few vague details in this Service Manual

Code: [Select]

http://www.lcd-television-repair.com/newsletter/1-sharp-led.pdf
Thanks

[Rasz]

forget about this datasheet, you will never get one

[diyaudio]

Anyone have specs or details on the pinout? I sent Mediatek a few emails and no response

a datasheet would be even better but I am not going to get my hopes up
the best I could find was a few vague details in this Service Manual

Code: [Select]

http://www.lcd-television-repair.com/newsletter/1-sharp-led.pdf
Thanks

You need to be a partner to obtain datasheet and reference material for MediaTek, same story with Rockchip, Allwinner. Get a Chinese hookup try this guy https://www.youtube.com/user/Charbax
 

[Stigaard]

Bunnie had a talk at CCC where he talks about what he had to go through to get something usable for the MT6260, you may want to try some of his methods. But as Rasz said, it is not going to be easy. But to summarize, first step is http://www.baidu.com/s?wd=MT5389 together with google translate.

[amyk]

Your best bet on pinouts would be to find other service manuals for TVs using the same part; it seems to be a relatively rare one though, as even the Chinese forums are unusually silent on it.

[chickenHeadKnob]

Bunnie had a talk at CCC where he talks about what he had to go through to get something usable for the MT6260, you may want to try some of his methods. But as Rasz said, it is not going to be easy. But to summarize, first step is http://www.baidu.com/s?wd=MT5389 together with google translate.

I went through the long thread of comments on Hackaday about the MT6260 and only two commenters pointed out the real death-blow for reverse engineering one of these  Chinese wonders: their limited production lifetime. By the time you figure out one chip the fly by night fabless producer has gone bankrupt or is making something else. The whole thing Bunnie is doing is a fools' errand, and Bunnie is no fool, so I don't get it. I also laugh at the market strategy of these Chinese companies that chase the same consumer markets and price under-cut each other so in the end no one makes book. Definition of madness.

[amyk]

Bunnie had a talk at CCC where he talks about what he had to go through to get something usable for the MT6260, you may want to try some of his methods. But as Rasz said, it is not going to be easy. But to summarize, first step is http://www.baidu.com/s?wd=MT5389 together with google translate.

I went through the long thread of comments on Hackaday about the MT6260 and only two commenters pointed out the real death-blow for reverse engineering one of these  Chinese wonders: their limited production lifetime. By the time you figure out one chip the fly by night fabless producer has gone bankrupt or is making something else. The whole thing Bunnie is doing is a fools' errand, and Bunnie is no fool, so I don't get it. I also laugh at the market strategy of these Chinese companies that chase the same consumer markets and price under-cut each other so in the end no one makes book. Definition of madness.

MTK is a big manufacturer, and there are still tons of MT6260 available.

[Stigaard]

Bunnie had a talk at CCC where he talks about what he had to go through to get something usable for the MT6260, you may want to try some of his methods. But as Rasz said, it is not going to be easy. But to summarize, first step is http://www.baidu.com/s?wd=MT5389 together with google translate.

I went through the long thread of comments on Hackaday about the MT6260 and only two commenters pointed out the real death-blow for reverse engineering one of these  Chinese wonders: their limited production lifetime. By the time you figure out one chip the fly by night fabless producer has gone bankrupt or is making something else. The whole thing Bunnie is doing is a fools' errand, and Bunnie is no fool, so I don't get it. I also laugh at the market strategy of these Chinese companies that chase the same consumer markets and price under-cut each other so in the end no one makes book. Definition of madness.

I'd say both yes and no, the chips change often, but as bunnie points out in his talk MediaTek's specialty is to cheaply create complex chips using complex bonding of a lot of standard cores.
I'm guessing here, but when you have support for a certain amount of cores achieved by researching a number of chips very little will change from chip to chip because of reuse of cores by the manufacturer.

That is not to say I would necessarily use it in a design with custom pcb as stuff will still be locked to the pinout, but it it really much different than any other ARM chip for the cell phone industry?
Would it be much different in terms of available life if it was a Broadcom chip for instance?

[chickenHeadKnob]

@Stigaard

For me the comparison wouldn't be a broadcom or Rockchip part, they might be as bad or worse, I would compare to Freescale i.Mx or TI. soc. that are intended for the embedded industrial control market and have some promise of continuity of supply and good documentation. In TI case they claim 10 year availability or similar. I suppose I am risk averse and I view time spent designing in and committing to a new chip as a big risk. If you don't mind throwing that  effort away you may feel differently.

 Olimex is selling Allwinner boards into this type of market and their customers often ask, "how long will this be available". Tsevtan recently replied "don't worry Allwinner is stopping production of A10 but as long as we order 50,000 we can get them forever into the future".  Well I would worry, I was with a small manufacturer and we got put on allocation a few times. That was pain I will long remember. That promise from Olimex is unrealistic for two reasons. First the mask is tied to a particular scale  "node"  and when the fab changes, and they change on their own schedule not Olimex's whim, then that's it. In addition there is the negative feedback effect, Olimex tells customers wait until we get to 50,000 ordered and the customer will say we can't wait and then designs out the part. That will mean it never gets to 50000.

[Stigaard]

@Stigaard

For me the comparison wouldn't be a broadcom or Rockchip part, they might be as bad or worse, I would compare to Freescale i.Mx or TI. soc. that are intended for the embedded industrial control market and have some promise of continuity of supply and good documentation. In TI case they claim 10 year availability or similar. I suppose I am risk averse and I view time spent designing in and committing to a new chip as a big risk. If you don't mind throwing that  effort away you may feel differently.

 Olimex is selling Allwinner boards into this type of market and their customers often ask, "how long will this be available". Tsevtan recently replied "don't worry Allwinner is stopping production of A10 but as long as we order 50,000 we can get them forever into the future".  Well I would worry, I was with a small manufacturer and we got put on allocation a few times. That was pain I will long remember. That promise from Olimex is unrealistic for two reasons. First the mask is tied to a particular scale  "node"  and when the fab changes, and they change on their own schedule not Olimex's whim, then that's it. In addition there is the negative feedback effect, Olimex tells customers wait until we get to 50,000 ordered and the customer will say we can't wait and then designs out the part. That will mean it never gets to 50000.

The reason i choose Broadcom for comparison is that they are an American vendor producing chips for the same kind of products (cell phones). If the comparison is against classical embedded chips/products I fully agree with you. But if you look into the market for which Bunnie is proposing this chip for it is not for the long term designs, it is as an alternative to Arduino, Raspberry PI and the likes and compared to Raspberry PI I have a hard time seeing the difference (except from that the sole amount of people using Raspberry PI today will probably ensure availability for a long time). My understanding is that this is a tool suggested for the hobbyist / hacker not for industry and therefore long term availability is less important.

[rizzy]

The interesting point from Bunnies talk on the CCC is that he showed how things work in China and that people there do not really care for NDAs and stuff like that.
Nowadays chinese vendors are pretty much overcautious when it comes to releasing datasheets and source code because they are afraid that their competitors will copy their work.
You can already see slight changes in that manner e.g. Allwinner and Mediatek became a member of Linaro in early 2014.
Well, joining an organization and releasing source code are definitely two different things but I think they are heading towards a more open direction as you can see by Mediatek releasing the source for Android One http://www.xda-developers.com/android/mediatek-android-one-source/ .