Dual-frequency quartz oscillator with a FPGA ?

Hello,

John Adair wrote:
> On cost the new MEMS oscillators very often compete well with crystals.=

MEMS oscillators are not as widely available as quartz :-/

> If you are worried about the power why not control the
> oscillator power with a mosfet and simply turn it off when not in use.
In fact i'll keep it constantly on,
and the PLL will derive other / higher frequencies.
for the rest of the system.

> There a lot of the newer MEMS oscillators that don't use mush power
> and are physically very small. Spot the 25MHz MEMs oscillator on here
> http://www.enterpoint.co.uk/component_replacements/craignell2.html and
> that's not the smallest one we are using.

Nice board. My issue though is : is your oscillator available
in 11.2896MHz frequency, and at what cost ? The solution I have
found costs around 0.6=80 in small (10-100) quantities.
The can is a bit larger (H49 size) than today's miniature
standards but it's not an issue here.

> John Adair
> Enterpoint Ltd.
yg

--=20
http://ygdes.com / http://yasep.org

> Nice board. My issue though is : is your oscillator available
> in 11.2896MHz frequency, and at what cost ? The solution I have
> found costs around 0.6� in small (10-100) quantities.
> The can is a bit larger (H49 size) than today's miniature
> standards but it's not an issue here.

It took me less than 30 seconds on the web to find an oscillator at that 
frequency: 
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=ASML-11.2896MHZ-ET-ND

hi !

David Spencer wrote:
> It took me less than 30 seconds on the web to find an oscillator at that frequency: 
> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=ASML-11.2896MHZ-ET-ND

thank you.
However, I see 2 things :
  - no stock available
  - quantity by 1000 which is not "prototype-friendly"
Ok, I know, this can change and I can look elsewhere etc.

I think that i'll quit this thread here since
my initial question has been answered previously,
and it's now an uninteresting matter of component sourcing...

regards,
yg

-- 
http://ygdes.com / http://yasep.org

On Apr 18, 11:13=A0am, "David Spencer" <davidmspen...@verizon.net>
wrote:
> > Nice board. My issue though is : is your oscillator available
> > in 11.2896MHz frequency, and at what cost ? The solution I have
> > found costs around 0.6=80 in small (10-100) quantities.
> > The can is a bit larger (H49 size) than today's miniature
> > standards but it's not an issue here.
>
> It took me less than 30 seconds on the web to find an oscillator at that
> frequency:http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&nam=
e=3DASML...

If you want MEMS, go to the Ecliptek page
Here's one with factory stock:

http://www.ecliptek.com/stocksearch/Stock.aspx?Partnumber=3DEMK33G2H-11.289=
6M

I've used the 2 x 2.5mm (smallest) version of these and they
work very well and have better vibration tolerance than most
crystal oscillators.

Regards,
Gabor

On Sat, 18 Apr 2009 15:16:04 +0200, whygee <whygee@yg.yg> wrote:

>Hello,
>
>John Adair wrote:
>> On cost the new MEMS oscillators very often compete well with crystals.
>MEMS oscillators are not as widely available as quartz :-/

>Nice board. My issue though is : is your oscillator available
>in 11.2896MHz frequency, and at what cost ? 

Heh, thought as much.
The other crystal is 12.288MHz then.

- Brian

On Apr 18, 9:42=A0pm, whygee <why...@yg.yg> wrote:
> I've also found a working example's schematic using a couple of 74HC14 ga=
tes,
> two 1M resistors and the usual couple of capacitors...
> and I wonder again why a couple of FPGA pins wouldn't work :-/

You can certainly make a FPGA oscillate.
The problem is getting reliable, desired frequency output!.
A common issue with such circuits is 'edge fur' which is high speed
transistion oscillations, near the logic thresholds. Sometimes not
directly visible on a scope, but will show up as double-clocking.


> > so you may prefer to run them all the
> > time, and switch inside the FPGA.
>
> This means continuous power draw...
> even a few mA can matter, not for the proto but in the end... :-/
> I'll probably switch to the Igloo family later, too.
>
> It looks like i'll have to choose one fixed frequency
> and give up the others, or i'll have to end up with 3 or 4
> Xtal operating =A0simultaneously...
> Fortunately, FPGAs have integrated PLLs today.

The cheapest is obviously to use the PLL if you can, but if power
matters, then you also want to avoid a FPGA type oscillator, as they
do
not specify the transistion currents. (Icc vs Vi)

xxU04 gates are a mainstream part, and if you use one inverter as the
oscillator, and then cascade a couple more to square-up the edges, so
you are not driving
the FPGA with a sine-wave.

You then choose a logic family, and Vcc, to give the Icc you need.
10-12MHz is lowish, so you should be able to get lower Icc in a custom
design than from a Module.
You can supply the Vcc from a FPGA port pin, to enable
the oscillator (add a RC to adjust Icc )

-jg

Brian Drummond wrote:
> On Sat, 18 Apr 2009 15:16:04 +0200, whygee <whygee@yg.yg> wrote:
>> Nice board. My issue though is : is your oscillator available
>> in 11.2896MHz frequency, and at what cost ? 
> Heh, thought as much.
> The other crystal is 12.288MHz then.
good guess :-)
Now if you have a smart idea about how to switch between them
without a glitch... Oh well, when i'll add a video output,
i'll need yet another oscillator with another frequency :-/

I'm waiting for ecliptek's answer about a sample request.
However, I wonder how i'll be able to solder it in
the first prototypes (i'm not at ease with the hot air gun yet).
I'll have to measure how much it consumes, too. So many questions...

> - Brian
yg

-- 
http://ygdes.com / http://yasep.org

-jg wrote:
> On Apr 18, 9:42 pm, whygee <why...@yg.yg> wrote:
>> I've also found a working example's schematic using a couple of 74HC14 gates,
>> two 1M resistors and the usual couple of capacitors...
>> and I wonder again why a couple of FPGA pins wouldn't work :-/
> You can certainly make a FPGA oscillate.
I'm sure, too ;-)

Oh well, what if I tried now and checked the different frequencies
that it can generate ? I'm feeling hacky now... The only limit is that
I have no frequency meter above 20MHz, which is well below what
is theoretically possible...

> The problem is getting reliable, desired frequency output!.
> A common issue with such circuits is 'edge fur' which is high speed
> transistion oscillations, near the logic thresholds. Sometimes not
> directly visible on a scope, but will show up as double-clocking.
I imagine that the pin capacitances will smooth it a bit ?

Oh, if I understand correctly, if it's too smooth,
it can create problems... But in FPGA-land, that should not
be impossible ?...


>> It looks like i'll have to choose one fixed frequency
>> and give up the others, or i'll have to end up with 3 or 4
>> Xtal operating  simultaneously...
>> Fortunately, FPGAs have integrated PLLs today.
> The cheapest is obviously to use the PLL if you can, but if power
> matters, then you also want to avoid a FPGA type oscillator, as they
> do not specify the transistion currents. (Icc vs Vi)

An oscillator circuit that I have found uses a single
HC14 (schmitt trigger inverter) gate as oscillator.
The output goes to another digital input, as well as a 1M resistor,
whose other side is tied to a grounded capacitor (100pF), and
the crystal with a 1M resistor in parallel (both go to the HC14 input).

  (sorry, i can't find the original location of the PDF)

With this system, the current that goes through 1M must be very small...
Also, concerning your previous point about spurious signals
("edge fur"), I see that the 1M + 100pF network is a low-pass
that should reduce the impact of the "fur"...

Unfortunately, the Actel chips I currently use don't have Schmitt triggers.
However, the latest / newest Actel chips have, including some Igloo,
so the idea may resurface later in the design/development, so I can fall back to
"traditional" methods now. I'll be careful to select ST-capable parts later.

> xxU04 gates are a mainstream part, and if you use one inverter as the
> oscillator, and then cascade a couple more to square-up the edges, so
> you are not driving the FPGA with a sine-wave.
Shouldn't one inverter be enough to square the edges ?
As I re-looked at the 74LVC1GX04 (I had a few samples already)
I remarked that there are 3 inverter gates, this looks like
overkill to me...

> You then choose a logic family, and Vcc, to give the Icc you need.
> 10-12MHz is lowish,
it's on purpose.

> so you should be able to get lower Icc in a custom
> design than from a Module.
This shouldn't be too difficult to verify.

> You can supply the Vcc from a FPGA port pin, to enable
> the oscillator (add a RC to adjust Icc )
I would even use a LC network...

But what about the "enable" input of the canned oscillators ?
Normally, the osc's output is tristated so (taking into account
various capacitances) two oscs should be able to share a single
FPGA input pin (then 2 output pins control the enable pins)...


regards,
> -jg
yg

-- 
http://ygdes.com / http://yasep.org

On Apr 19, 1:48=A0pm, whygee <why...@yg.yg> wrote:
>
> Oh well, what if I tried now and checked the different frequencies
> that it can generate ? I'm feeling hacky now... The only limit is that
> I have no frequency meter above 20MHz, which is well below what
> is theoretically possible...

Set up some simple dividers and scope the outputs, and Freq count,
and look for phase jitter, and erratic jumps...
Don't try and look directly for the edge oscillations.

> I imagine that the pin capacitances will smooth it a bit ?
>
> Oh, if I understand correctly, if it's too smooth,
> it can create problems... But in FPGA-land, that should not
> be impossible ?...

In extreme cases, these transistion oscillations are VERY high, and
I've
seen them disturb UNRELATED logic, due to common mode inductive
effects.

Such things are outside the testing realms of the IC vendors :)


> An oscillator circuit that I have found uses a single
> HC14 (schmitt trigger inverter) gate as oscillator.
> The output goes to another digital input, as well as a 1M resistor,
> whose other side is tied to a grounded capacitor (100pF), and
> the crystal with a 1M resistor in parallel (both go to the HC14 input).
>
> =A0 (sorry, i can't find the original location of the PDF)
>
> With this system, the current that goes through 1M must be very small...
> Also, concerning your previous point about spurious signals
> ("edge fur"), I see that the 1M + 100pF network is a low-pass
> that should reduce the impact of the "fur"...

Ah, yes Schmitt-assisted oscillators shift the problem :)

 They remove the  'fur' issue, but now you are never sure if the
Schmitt RC-fb, or the Xtal is locking. (unplug the Xtal, and they
usually still oscillate!)

 I have had promising looking bench results, that set up a RC schmitt
lower than
the Xtal, and then hope that the Xtal rings enough, to build amplitude
and lock the schmitt freq. We never brave enough to actually ship a
product using this :)

 Where a Schmitt Osc _does_ work very well, is in LC mode, so if you
can get Schmitt pin option, (and can tolerate LC precision!), then a
SMD
inductor is by far the lowest cost clocking option.

> Shouldn't one inverter be enough to square the edges ?
> As I re-looked at the 74LVC1GX04 (I had a few samples already)
> I remarked that there are 3 inverter gates, this looks like
> overkill to me...

If you are doing this properly, a schmitt buffer is better, and you
chose
a schmitt with lowest transistion currents (ie lowest Vcc when sine-
driven)

You may trade-off to a single-component, but the unbuffered gates have
low
gain, so a single stage is only 'squarish'.
I like the 3GU04, as they come in many logic families.

It is surprising the tiny logic vendors DON'T make 2G=3D(1GU04+1G17)
pairing
as that would cover many clocking choices: Xtal/LC/RC 2 Pin & 3 Pin.
I guess they are stuck in legacy-mode...

> > You can supply the Vcc from a FPGA port pin, to enable
> > the oscillator (add a RC to adjust Icc )
>
> I would even use a LC network...

The R drops voltage, and almost runs the part in constant current mode
oscillation.
You can drive Logic short of the rails, and save osc power.
You  do need to watch the IP Cell transition current, but usually
around 1V on
the PFET and 0.6V on the Nfet is OK.

>
> But what about the "enable" input of the canned oscillators ?
> Normally, the osc's output is tristated so (taking into account
> various capacitances) two oscs should be able to share a single
> FPGA input pin (then 2 output pins control the enable pins)...

Yes, but what is clocking the FPGA, when both are off  ?  ;)

-jg

On Sun, 19 Apr 2009 03:08:39 +0200, whygee <whygee@yg.yg> wrote:

>Brian Drummond wrote:
>> On Sat, 18 Apr 2009 15:16:04 +0200, whygee <whygee@yg.yg> wrote:
>>> Nice board. My issue though is : is your oscillator available
>>> in 11.2896MHz frequency, and at what cost ? 
>> Heh, thought as much.
>> The other crystal is 12.288MHz then.
>good guess :-)
>Now if you have a smart idea about how to switch between them
>without a glitch... Oh well, when i'll add a video output,
>i'll need yet another oscillator with another frequency :-/

I would (a) use external oscillators, as others have suggested, for the same
reasons. I'd be happy with either modules for simplicity, or something amenable
to oscillator use, like 74HCU04 in some flavour, if economics demand it.

If you need a glitch free switch you need both oscillators running at once, and
something like a BUFGMUX or one of Peter Alfke's clever little circuits to
switch between them.

If power or radiated noise was important I'd consider killing the unused osc (if
it's a 74HCU04, connect the input to a tristated FPGA output, or use a diode
switch to reduce stray C), and a state machine to start it; confirm it had
restarted; and then switch over. 

Last time I did it I could let the system go silent for a second when switching;
I used diode switches between two crystals on an oscillator in a rather nice
Philips chip (SAA7274, probably no longer made - that other brand didn't seem to
like crystals attached)

Time before that was pure 74LS series...

Incidentally I found I could pull the crystal over about +/-150ppm, and there
was measurably less phase noise (jitter by another name) from a reputable brand
of crystal.

If you're interested to talk more, email...

>I'm waiting for ecliptek's answer about a sample request.
>However, I wonder how i'll be able to solder it in
>the first prototypes (i'm not at ease with the hot air gun yet).
>I'll have to measure how much it consumes, too. So many questions...

Heh. Kapton tape a thermocouple to the top of a board and check out the
temperature profiles you can get with a 2kw Black&Decker hot air paint stripper
applied underneath...

I've mounted these connectors this way (note the power pins down the middle) 
http://www.samtec.com/technical_specifications/overview.aspx?series=QTH
but haven't tried the next logical step of BGAs yet.

- Brian