Can a fpga replace external inverters in a crystal osc ?
Dan
My design has a common low cost crystal oscillator. It uses two inverters
HC7404 and a few caps & two res.
Can the inverter chip be replace by the FPGA. Can I simply put the crystal
across two IO pins ( I am using a Spartan IIE ) and configurae them as an
inverter ( while keeping the caps and the res(s) ) ?
Sincerely
Dan
Ray Andraka
be done with a simple inverter. An FPGA has too many gain stages between the
pins to be abused as an amplifier in an oscillator circuit. While it may work
on the bench, it will not reliably start and maintain the fundamental
frequency with variations in temp, process, and voltage. Why not use an
integrated xtal oscillator instead of a simple crystal? The costs are nearly
the same.
Dan wrote:
--
--Ray Andraka, P.E.
President, the Andraka Consulting Group, Inc.
401/884-7930 Fax 401/884-7950
email r...@andraka.com
http://www.andraka.com
"They that give up essential liberty to obtain a little
temporary safety deserve neither liberty nor safety."
-Benjamin Franklin, 1759
Dan
Dan
Peter Alfke
Ray Andraka wrote:
> Not reliably. The HC7404 works because it is used as an amplifier, which can
> be done with a simple inverter. An FPGA has too many gain stages between the
> pins to be abused as an amplifier in an oscillator circuit. While it may work
> on the bench, it will not reliably start and maintain the fundamental
> frequency with variations in temp, process, and voltage. Why not use an
> integrated xtal oscillator instead of a simple crystal? The costs are nearly
> the same.
Amen.
Could not be said better...
XC3000 has a dedicated single-stage amplifier between XTAL1 and XTAL2 pins. But
no other FPGA family has repeated that dubious feature.
Peter Alfke
Rick Filipkiewicz
Ray Andraka wrote:
> Not reliably. The HC7404 works because it is used as an amplifier, which can
> be done with a simple inverter. An FPGA has too many gain stages between the
> pins to be abused as an amplifier in an oscillator circuit. While it may work
> on the bench, it will not reliably start and maintain the fundamental
> frequency with variations in temp, process, and voltage. Why not use an
> integrated xtal oscillator instead of a simple crystal? The costs are nearly
> the same.
>
Do you happen to know if there's a 32.??Khz xtal osc. that uses sufficiently
little power that it can run off a 3V coin cell ? or with some power-down mode ?
The reason I ask is that I'd dearly like to replace the PIIX4 southbridge +
CombiIO functions in an FPGA/CPLD and about the only thing that's hard is the real
time clock. I've already got a CPLD on board so it could be replaced by a
CoolRunner (II ?) to take the RTC function.
Maybe this is just to perfectionist & I'll just bite the bullet and find a
stand-alone PC-compatible RTC.
Jim Granville
If you want to run of a coin cell, what's going to keep track of the
time ?
You could look at :
Dallas DS32KHz - Not cheap, but high precision, and with battery pin.
Intersil used to make a programmable OSC chip
Or, make your own using HEF4069, or HEF4007 and a 32Khz xtal.
Philips PCF8563 - docs suggest it defaults to 32KHz CLKOUT, so should
power up without i2c BUS activity ?
You could put a Read-Time state engine into the FPGA, and then the
PCF8563.RTC
can keep track of time, with << 1 uA Icc when in OFF mode.
Seems the OSC is less of a problem, than the question of who's in
charge of the time.
-jg
Ray Andraka
Rick Filipkiewicz wrote:
--
Jay
book the crystal by itself is about 50 cents and the full up
oscillator is $1.50, thats a dollar difference. So if is a low
quantity, $1000 Vertex kinda thing, buy the can, but its high volume
and real price sensitive, engineer and test your own.
Ray Andraka <r...@andraka.com> wrote in message news:<3D8C886A...@andraka.com>...
Hal Murray
>little power that it can run off a 3V coin cell ? or with some power-down mode ?
>The reason I ask is that I'd dearly like to replace the PIIX4 southbridge +
>CombiIO functions in an FPGA/CPLD and about the only thing that's hard is the real
Look at Dallas/Maxim. Probably others too. (Maxim bought Dallas)
In the old days, the RTC was a separate chip. Dallas was one of
the main suppliers. They still make them.
--
The suespammers.org mail server is located in California. So are all my
other mailboxes. Please do not send unsolicited bulk e-mail or unsolicited
commercial e-mail to my suespammers.org address or any of my other addresses.
These are my opinions, not necessarily my employer's. I hate spam.
Ray Andraka
board space, inventory costs, pick and place cost, increased failure rate costs, etc) the
price does get closer to parity. Engineering involves tradeoffs of course, but that is
not done properly unless all the costs and potential problems are considered. Using the
FPGA for the inverting gain in an oscillator is not good practice because of the
sensitivity to circuit parameters that you can't control.
Jay wrote:
> The costs are the same? Au contrair mon friar. According to my price
> book the crystal by itself is about 50 cents and the full up
> oscillator is $1.50, thats a dollar difference. So if is a low
> quantity, $1000 Vertex kinda thing, buy the can, but its high volume
> and real price sensitive, engineer and test your own.
>
> Ray Andraka <r...@andraka.com> wrote in message news:<3D8C886A...@andraka.com>...
> > Not reliably. The HC7404 works because it is used as an amplifier, which can
> > be done with a simple inverter. An FPGA has too many gain stages between the
> > pins to be abused as an amplifier in an oscillator circuit. While it may work
> > on the bench, it will not reliably start and maintain the fundamental
> > frequency with variations in temp, process, and voltage. Why not use an
> > integrated xtal oscillator instead of a simple crystal? The costs are nearly
> > the same.
> >
> >
Austin Lesea
I agree. Using an FPGA IOB as an amplifier is something you can make work one time, maybe
dozens of times, but with changes in process the parameters can change, and the oscillator
may sometimes not start up. The Gain and Phase (and delay) of an IOB ->LUT->IOB is pretty
hard to characterize!
Fairchild actually has a series of CMOS logic that consists of logic elements in tiny
packages (for those pesky need an inverter problems in a cell phones).
Austin