The ubiquitous IC company you probably never heard about.....

[schmitt trigger]

Or perhaps you have.
I am talking about QX Micro Devices. **broken link removed**

If you have ever used or seen a solar-cell powered LED garden light, it will most likely utilize one of its ICs, QX5251/5252
The datasheet is appalling and has the barest details, but the minimalist IC concept is actually clever.

 

[dknguyen]

How many garden lights do you have to produce and sell in order to be able warrant a custom ASIC like that?

 

[schmitt trigger]

millions and millions

 

[JonSea]

If you can SELL a solar yard light for 97 cents, they have to be chuning them out in the millions.

 

[ChrisP58]

And if the same chip maker is supplying that part to hundreds, if not thousands, of yard light makers....

Or, more likely, the PCB assembly. Ready to drop into the plastic case.

 

[JonSea]

I saw the 97 cent yard lights at Walmart - they probably come 100 in a tray and there's a stack of at least 5 trays. Let's say 500 lights per store. In the US, there are 4177 stores, so if each one has the same stock, that's over 2 million just to stock WalMart! Things do get cheaper in volume

 

[ronsimpson]

millions and millions

much less now days. That part has very little silicon in it.

 

[schmitt trigger]

You are right
The IC is quite simple. Should be dirt (Silicon ) cheap.

Anyway the point is that a semiconductor company does not require an ultra sophisticated device to have a runaway hit.

 

[audioguru]

If you can SELL a solar yard light for 97 cents, they have to be chuning them out in the millions.

They must be losing money. Walmart sells a solar garden light for $0.99CAN which is $0.77US. It comes with a clear plastic lens and a plastic cover on the solar panel, both get sunburned and get cloudy then don't work in a few months. The solar panel is poorly sealed causing the electronics underneath to corrode away in a couple of months. The battery is little so either the light is dim or lasts only for 2 hours at night. The shipping probably costs another $0.77US.

 

[ronsimpson]

They must be losing money.

There is a art form of copying some one's design, using questionable parts, ending up with a almost working product.
Price is hard to understand:
Five years ago, there was a import duty on batteries. (not on batteries in a product) We sold products at Walmart, with a battery inside for less than you can get a battery for.
We sold products for less than the cost of the parts. (and made money)
I do not understand the price.

 

[Nigel Goodwin]

We sold products for less than the cost of the parts. (and made money)

Presumably you weren't paying as much as that for the parts?, that's how you made a profit.

As for the chip in question, it's presumably a simple chip - so fairly cheap to manufacture - obviously numbers need to be pretty high to get it down nice and cheap, but probably not as high as you might imagine. Once you reach a certain figure prices don't really drop much above that - and plants in China are set up for relatively small production runs, and quick changes to the next low cost job.

My daughter could perhaps make you one?, here's a picture she posted earlier today:

 

[schmitt trigger]

To your point about simplicity, another low-cost driver is minimal testing with wide open specs such that the yields are relatively high.

As you can see from the datasheet, I'm sure that the minimal testing strategy is in place.

 

[ronsimpson]

Presumably you weren't paying as much as that for the parts?,

In the US: If I get one million 10k resistors and three 1k resistors there is a big price difference.
In China: our supplier pays 100 million price for both type of resistors. Volume pricing is counted differently.
Also: There is "China Friend" pricing. Our China supplier clearly pays a different price for the parts. I think a China assembly house gets their parts at no profit so the US assembly houses can not survive.