Question for Win Hill

Hi Win,

I can't seem to find where I stashed your gate-charge subcircuit. (*)

Pointer or re-post?

Thanks!

(*) I have a PhD that needs an education ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson
Loading thread data ...

Give me more of a hint, doesn't ring a bell. Are you talking about subthreshold MOSFET modeling?

Reply to
Winfield Hill

Maybe I remembered wrong. I thought you had developed a gate-charge subcircuit add-on. I have a situation where I need to model fast turn-on and turn-off of an FQN1N60C.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

What you may be thinking of is how I broke a power MOSFET into three or four dissimilar regions, in a reasonably- successful attempt to make spice modeling match my bench measurements. The factor I considered was gate spreading resistance. This is generally a single resistor in simple MOSFET models, and is often improperly rather large, such as 50 ohms. I found that simply reducing the resistance could work in modeling a MOSFET's observed fast turn-on, but worsened modeling of its slow turn-off.

So I broke the MOSFET into say five pieces, assigning a very low gate resistance to one-fifth, slightly higher to the 2nd, etc., and the highest to the last section. The fast portions would turn the MOSFET on quickly, and the slow ones, being further from the gate connections, would keep a portion on longer, thereby with experimentation, nicely modeling the observed slowly-rising Ron at turnoff.

Different types of MOSFETs behave very differently with regard to turnoff, so it's necessary to make extensive bench measurements to get good data to refine the model.

I have no experience with the FQN1N60C, although I see it's a small-die low-capacitance 600-volt TO-92 part. Fairchild also has the same die in a convenient D-pak, the fqd1n60c. These appear similar to the small MOSFETs in my high-voltage amplifier designs, but gate spreading resistance has not been an issue in these designs.

I've found it necessary to use large-die parts for fast high-voltage switching, and this is where gate modeling becomes an issue. For example, a well-driven fcd4n60 D-pak could more likely rapidly switch a high voltage.

What are you working on?

Reply to
Winfield

A specialized ASIC for high efficiency flyback applications, but with a 1 cent die cost target :-(

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

It sounds like it may be worth it to ask the foundry if they can give you information about the base spreading resistance, or at least the ohms per square of the gate polysilicon (if that's what they're giving you).

I wonder if you could speed this up by either slapping a layer of aluminum over the (presumably) polysilicon gate, or by splitting up the last two stages into multiple, smaller driver/final sets running in parallel. (I wouldn't know. I have no IC design experience whatsoever, but I know how to make impertinent suggestions).

--
Tim Wescott
Control systems and communications consulting
 Click to see the full signature
Reply to
Tim Wescott

I suppose you're joking, but seriously, how would one work toward a penny-scale IC die, by making it really, really small? I see the FQN1N60C you were talking about costs 15 cents, qty 5k, from Mouser. That's considerably cheaper than the same die in an SMD package, so it appears that very high production quantities are also required to get the price down.

Reply to
Winfield
[snip]

A penny for your thoughts?

Bob

Reply to
Bob

What's the cheapest IC-based consumer product on the market? Blinking LED pin? Musical greeting card? Kitchen timer?

I saw a scan-tuned AM/FM radio at Walgreens, with headphones, for $1.50.

John

Reply to
John Larkin

The FET is pre-defined. I'm designing the driver chip; and am looking for good characterization.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

No joke. Quantities exceeding 1 million.

We already have located a bipolar foundry which will do a 1mm X 1mm chip for 1¢.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

That's only $10K for a run of a million chips! What's that, 100 wafers? I bet your engineeering will cost a lot more, per chip, than making them!

John

Reply to
John Larkin

Not including testing, die-scoring, and packaging?

Reply to
Winfield

I don't know the details... the client did the arrangements. I'm only the doobie ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

I've seen them in the dollar store (no display though). Do CF lamps currently have ICs in them? They're down around a dollar and change, with the electronics only a bit of that. Cheap calculators are under a dollar ex-factory.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
 Click to see the full signature
Reply to
Spehro Pefhany

Trouble is, that's the price for really large quantities. ;-)

--
Paul Hovnanian	paul@hovnanian.com
-----------------------------------------------------------------------
 Click to see the full signature
Reply to
Paul Hovnanian P.E.

Actually, this can make sense. A small complete low- power off-line converter with total production costs under say 10 to 20 cents -- that's got to to be good for up to 100M produced units, or more, for the right applications. That would turn an intial $10k fab bill into $1M + and a clear smash success for the fab bidder. What's more, if the fab obtains the right to market the IC chip, it could have an additional lush sales life.

Reply to
Winfield

Solved....

formatting link

Rather elegantly, maybe up to paper submission quality ;-)

I need to add some more parameterization to make it easer to set the breakpoints and then I'll publish how I did it.

(No voltage-controlled switching like in some papers I've found... note the smooth curves ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

John Larkin snipped-for-privacy@highNOTlandTHIStechnologyPART.com posted to sci.electronics.design:

Lets see, 1 mm by 1 mm die on 200 mm circular wafer. 3/4 of 40,000 square mm; say about 30,000 dice per wafer. That would be 30 wafers, about half a boatload. NRE still dominates.

Reply to
JosephKK

[snip]

Only because bipolar parts are getting rarer and rarer.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
 Click to see the full signature
Reply to
Jim Thompson

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.