[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Crystal Controlled Oscillator

I would add a strong caution against using that transistor + Schmitt
arrangement if you can avoid it. The problem isn't so much the Axis part
it is the specifications of the Broadcom Ethernet chip.

If you read the Broadcom specs they want a +/- 50 ppm clock with a 50%
cycle (+/- 10%)and edge rates from 1ns to 4ns(10% to 90%). 
If you use the transistor + Schmitt Trigger clock circuitry I think 
you will find that you will have a difficult time meeting 
those specs especially as regards variation in parts, temperature, etc.
We switched to a master clock oscillator circuit from CTS (CB3LV @
20MHz) and 
have had no problems with clocks since and no issues with FCC

I am glad the Axis folks don't have any problems with that clock
arrangment but
I don't like it and wouldn't recommend it.


David Kilp

> -----Original Message-----
> From: owner-dev-etrax@xxxxxxx.com 
> [mailto:owner-dev-etrax@xxxxxxx.com] On Behalf Of Per Zander
> Sent: Wednesday, September 22, 2004 8:10 AM
> To: Ferit Cosan
> Cc: dev-etrax@xxxxxxx.com
> Subject: Re: Crystal Controlled Oscillator
> Hi,
> In ETRAX 100LX Designer's Reference chapter 20, Electrical 
> Information, there is a DC electrical specification on page 29-30.
> There you can see:
> clkin Vih min: 0.7 x Vdd, max: 5.5 V
> clkin Vil min: 0 V, max: 0.3 x Vdd
> Iin (all inputs): min: - 10 uA, max: 10 uA
> This is for the ETRAX 100LX itself.
> Additionally, there is one other component connected to the 
> clkin internally in the MCM. This one has normal LVTTL levels 
> and also around the same input current.
> So, effectively clkin Vil max is 0.8 V and Iin is +/- 20 uA 
> for the MCM.
> Generally rise and fall times are not extremely critical as 
> long as you meet the clkin high and clkin low time (min 15 
> ns) specified in page 55.
> Axis has recommended the transistor + Schmitt Trigger 
> solution because we have long experience with it and get good 
> results both regarding reliability and EMI. It is also fairly 
> low cost. But it isn't mandatory. You can choose another solution.
> What you should avoid is a clock circuit with jitter, a 
> circuit with a lot of over/undershoot and ringing or a 
> circuit that causes significant 20 MHz ripple on the power 
> supply. But I think you can easily avoid all of this. Another 
> concern is EMI. Many integrated oscillator component are 
> rather noisy and may cause problems with FCC and EC compliance.
> best regards
> Per Zander
> On Wed, 22 Sep 2004, [iso-8859-9] Ferit Co?an wrote:
> > Hello,
> > Does anybody have a recommendation for a 20.000MHz Crystal 
> Controlled 
> > oscillator for the ETRAX 100LX MCM 4+16. I've found some 
> samples from 
> > Connor Winfield firm (HSM643), however i can't match the 
> requirements 
> > of ETRAX clock signal in a healthy way because of the lack of 
> > information about Etrax clocks electrical specifications (only 
> > information i've found about clock signal is in Etrax "Electrical 
> > Information" tutorial on the 55th page, yet there isn't
> > any electrical information about the clock signal-clkin's current
> > requirements, voltage level of high and low levels of the 
> clock signal,
> > rise/fall time of the clock, etc.)
> >
> > ?s it an obligation to use a clock signal circuit as shown in the 
> > Serial Server reference design, can't we minimize the board size by 
> > using one chip instead of using the Schmiddt Trigger and the 
> > transistor circuit.
> > If we can, how much current does Etrax require through the "clkin"
> > pin for a stable clock operation, what must be the high and 
> low level
> > voltage values of the clock signal, and what is the 
> rise/fall time of
> > the clock.
> >
> > I'll be appreciated if someone guide me about this problem...
> >
> > Best regards
> >
> > Ahmet Ferit Cosan