Could this be the new 555 timer ?

[mikeselectricstuff]

Just noticed these interesting devices:
Greenpak programmable  mixed signal arrays
Micro FPGA with analogue functions, cheap ($0.24 100x), 1.8-5V supply (GreenPak2 version), schematic based devtool, and unlike all other current FPGAs, internals are fully documented.
Only downside I can see is that it's OTP, presumably to save cost.
These guys really should do a design contest

[Balaur]

The "one-time programmable" aspect is not that ideal for a hobbist setting.

However, potentially useful for a production run.

[deephaven]

Nice and cheap, $24 would buy you 100 goes at getting it right.

I'm trying to see what it could do that I couldn't do with a small PIC. I was hoping there would be more analogue functionality to set it apart from a micro. It is very small so could have some uses when space is at a premium.

[mikeselectricstuff]

I've only skimmed through the info, but some advantages over PIC could be parallel operation/faster response, lower cost (cheaper than 10F) and probably lower power draw when sitting waiting for something to happen, as it wouldn't need a clock.

[westfw]

Pretty neat looking.  I wish it were in a package more conducive to experimenting with, and/or reprogrammable.

Are there similar products without the analog functionality?  I've often thought I'd like a low-pin-count CPLD-like device with more internal structure (counters and registers, like this has) than the usual GAL devices...

[EEVblog]

What started the microcontroller revolution was the first erasable chip (PIC16C84 IIRC), and they were available in DIP packages.
These new devices have neither, but they do offer low cost, and they have ZIF demo board, so they at least have that right.

Need to look in more detail...

Dave.

[EEVblog]

Wow, the voice-over on the video is BAD!

https://www.youtube.com/watch?feature=player_embedded&list=PLB8A1CD5F1654D661&v=KT6-WtSdvyU#!

Looks like a really nice software interface though.

Dave.

[EEVblog]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

Dave.

[deephaven]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

Dave.

Postage! I asked if they had a UK distributer and they haven't and their international shipping charge is $50

[sub]

The problem with this is that it has a bit of everything, but doesn't seem to do anything particularly well.  You have some LUTs, but only three (!) flip-flops, so filters are almost certainly out. The ADC is pretty poor, and there's only PWM instead of a DAC.

I suppose you could use it as a way to get a very small amount of configurable logic, but Microchip are doing that on-board, so that's something of a moot point.

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

With the amount of logic that they have in there, I can't imagine that it would cost them that much to get a chip tested and packaged.  You can get a PIC in the same package for 37c after all, and that has more than 100 bits of state.

[mikeselectricstuff]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

Dave.

Small die, high volume. The idea is instead of one big CPLD type device that inevitably contains a lot of unused silicon, you scatter them around the board where needed. This is obviously attractive for consumer products with 1 & 2 layer PCBs.
I originally saw these in an article in one of the paper trade press - they claim they've already shipped over 100 million units. They also had several design wins before they'd done the GUI software, so looks like there is a market niche there.

[FreeThinker]

Just noticed these interesting devices:
Greenpak programmable  mixed signal arrays
Micro FPGA with analogue functions, cheap ($0.24 100x), 1.8-5V supply (GreenPak2 version), schematic based devtool, and unlike all other current FPGAs, internals are fully documented.
Only downside I can see is that it's OTP, presumably to save cost.
These guys really should do a design contest

Yes you could make a 555 timer out of one but why would you want too? . Seriously, it looks a nice first step but the package is not the best for the hobbiest and IMO a little too limited in functionality. Having said that I well remember the early days of the 555 and people thinking up clever ways to use it, so some thinking out side the box may spark a new revolution. As an introduction to FPGA's it looks a nice tool and the free software is a big plus, certainly one to watch.

[Rufus]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

There is a couple of pages in this month's New Electronics on the part. They say they shipped 100 million parts in 18 months, pretty good volume for a part you never heard of.

I think it an interesting product, great if you can use most of its functionality and cheap enough to use instead of a 555 or odd logic gate here and there. I think the biggest obstacle to widespread use is device programming. The attractiveness of being able to hold one part to do dozens of different jobs fades when you have to program each part in production. Are there pick and place machines that could program these parts on the fly?

I would guess most of there product goes to customers buying pre-programmed parts by the reel.

[Psi]

It would be nice if they could implement some of this technology to add a FPGA section to regular micro controllers without raising the price much.
There's lots of stuff for which having an on board FPGA would be quite handy.

Or even, if you want to get crazy , it would be kinda interesting if someone came out with a micro which had smt pads built on top of the TQFP package for soldering on one of these FPGA chips.

[mikeselectricstuff]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

There is a couple of pages in this month's New Electronics on the part. They say they shipped 100 million parts in 18 months, pretty good volume for a part you never heard of.

I think it an interesting product, great if you can use most of its functionality and cheap enough to use instead of a 555 or odd logic gate here and there. I think the biggest obstacle to widespread use is device programming. The attractiveness of being able to hold one part to do dozens of different jobs fades when you have to program each part in production. Are there pick and place machines that could program these parts on the fly?

I would guess most of there product goes to customers buying pre-programmed parts by the reel.

Yes - their primary model is they program (or mask for high volume) and send you programmed and tested parts. Would be interesting to see what the cost and MOQ is.

[sub]

It would be nice if they could implement some of this technology to add a FPGA section to regular micro controllers without raising the price much.
There's lots of stuff for which having an on board FPGA would be quite handy.

http://www.microchip.com/pagehandler/en-us/press-release/microchip-launches-8-bit-mcus-with-configurable-lo.html

You only get a very small amount, so this isn't going to outdo an FPGA, but if you just want some combinatorial logic, it would probably do the job.  The cheapest is 50c for a single unit.

[EEVblog]

I would guess most of there product goes to customers buying pre-programmed parts by the reel.

Yes, every manufacturer offers this option if your volumes are high enough.

Dave.

[Hypernova]

I fail to see any reason why they couldn't release a DIP verson, even SOIC would make it much simpler. Unless the extra mataerial in the bigger package matters that much.

[slateraptor]

Ok, I'm curious, how are they making money at $0.23ea and $0.18ea for 100qty?

Dave.

At qty 100, it's $0.24 and $0.29 for greenpak1 and greenpak2, respectively. What's nice is if you order any of the programming boards, 50 of the respective sample units are included. Too bad they're sold out of the greenpak2 board...8 week lead time to rub salt in the wound.

I would guess most of there product goes to customers buying pre-programmed parts by the reel.

Yes, every manufacturer offers this option if your volumes are high enough.

Dave.

http://www.silego.com/uploads/Products/product_123/xSLG46200r108_09192011.pdf

Page 9 of the pak2 datasheet suggests their target market was customers seeking a pre-programmed, high-volume solution from the beginning...I'd even wager that "high enough" isn't as high as we might be led to think.

I fail to see any reason why they couldn't release a DIP verson, even SOIC would make it much simpler. Unless the extra mataerial in the bigger package matters that much.

Probably because DIP reeks of low-volume hobbyists...just sayin'. At the current price point, I can't see how they'd be able to sell any more than a single package.


P.S. Max 1uA quiescent current with I/Os held static...hmm.

[mikeselectricstuff]

I fail to see any reason why they couldn't release a DIP verson, even SOIC would make it much simpler. Unless the extra mataerial in the bigger package matters that much.

They could, but it isn't worth their time. The development costs of any design that would benefit from a $0.20 chip instead of a $0.40 MCU are typically such taht a few $100 on proto PCBs with  sockets are negligible.
BTW the sockets are only $18 for the 12 pin, and $12 for the 8-pin

BTW here is the New Electronics article

[Bored@Work]

I fail to see any reason why they couldn't release a DIP verson,

Because there isn't much money to be made. The total worldwide hobbyist market is less than what one or a few of their large customers buy. That is something hobbyists often forget. And for their large customers DIPs are apparently a thing of the past. So unless one of their large customers convinces them (by offering to buy DIPs in volume), they have no incentive to do a DIP.

And if you look at the price, they need to sell in bulk to make any money. What do you think, how much money do they make when they sell them for 15ct? Don't forget, this isn't just manufacturing cost, but they need to finance everything from development, service and support, marketing, and whatnot from selling 15ct pieces.

And one of their selling points is reduced board space. By adding  a DIP they would weaken this sales argument.

If you desperately want one in a hobbyist friendly package try to convince companies like sparkfun to do a breakout board with them. However, the cost of the total thing will be excessive compared to the cost of a single of these ICs.

[mikeselectricstuff]

P.S. Max 1uA quiescent current with I/Os held static...hmm.

And because it isn't an MCU, it's static when not doing anything, unlike an MCU, which as soon as you start the clock will be clocking a lot of internal nodes.
Even when clocking, I'd guess there will be a lot fewer parts being clocked.
In some applications I can see this thing giving power draw figures  orders of magnitude lower than an MCU, and probably  still giving faster response.

[slateraptor]

P.S. Max 1uA quiescent current with I/Os held static...hmm.

In some applications I can see this thing giving power draw figures  orders of magnitude lower than an MCU, and probably  still giving faster response.

Agreed with respect to power consumption, although I'm hesitant at this time about response since propagation delays haven't been published...albeit their datasheet is on v0.1 at the moment. Nevertheless, the simplicity of their fabric suggests sub-5ns delays. I wonder what technology node these guys are manufactured at.


P.S. In light of the DIP comments, I had to give myself a quick reality check (see attached image of 2.5mm TDFN beside a standard D package 8SOIC from TI). 12 pins at 0.4mm pitch...it's a good thing they sell sockets on their website too.

[mikeselectricstuff]

P.S. Max 1uA quiescent current with I/Os held static...hmm.

In some applications I can see this thing giving power draw figures  orders of magnitude lower than an MCU, and probably  still giving faster response.

Agreed with respect to power consumption, although I'm hesitant at this time about response since propagation delays haven't been published...albeit their datasheet is on v0.1 at the moment. Nevertheless, the simplicity of their fabric suggests sub-5ns delays. I wonder what technology node these guys are manufactured at.

Even if it's 25ns, this is still at least 10x faster than any MCU, more if it has to wake from sleep first.
I doubt these are made on a particularly small process, as for small dies, pad area tends to dominate size. The fact that they support 5V also suggests it isn't bleeding-edge tech.

[slateraptor]

The fact that they support 5V also suggests it isn't bleeding-edge tech.

Exactly my thought.