Oscillator driving two ICs, possible?

[FreshmanNewbie]

I'm having this oscillator IC which I plan to drive KSZ8795 and LAN9500A wihtout clock buffer.

TG2016 output load drive is 10pF for the capacitor

KSZ8795 can have a crystal load capacitance of 27pF max and LAN9500A can have a crystal load capacitance of 8pF to 12pF.

My question is whether the TG2016 oscillator has enough drive for driving both the device without a clock buffer.

Whether this arrangement will work? Has anyone tried such an architecture before?

 

[KlausST]

Hi,

basically all you need to know is written in the datasheets ... and given/required by the application.

As long as the "output specifications" of the oscillator meets the "input specifications" of the two connected devices ... including the application conditions ... there is no problem in driving multiple devices from one source.

And yes, I´ve rather often used one clock source for several ICs.

****
You´re a quite experienced forum user and already did some electronics projects.
Thus I´m a bit surprised about the question and your given/missing informations.

* you miss to give rather important informations like: clock frequency and supply voltage..
* you use an "oscillator" .. so you clearly don´t connect a crystal, but still you refer to "crystal load capacitance"
* also you should know that the schematic is a good source of information what you are trying to do. Without it we need to guess a lot. (Pin names, wiring, termination ...)
* also you should already know that PCB layout has big influence in signal quality. Thus - even if the oscillator is able to drive both devices - it still depends on the PCB layout whether it will work on your board or not.

And "work" not only in the meaning that both IC´s operate correctly in your laboratory, but reliably work in the field with variable power supply, temperature... And also in the meaning: your application complies with the regulations (EMI, EMC, ....and so on)

One can say "we" need to "expect" that you correctly decided to use a 25MHz source and correctly decided to use the XI pins as clock input (There are other clock inputs, too). In the end we can not be sure about this. So if we have to guess this, do we also have to expect that you visited the manufacturer internet site for related douments? Like: did you read the "Hardware Design Checklist"?
I mean: you did not give much informations what you did so far.

Klaus

 

[FreshmanNewbie]

Hi,

basically all you need to know is written in the datasheets ... and given/required by the application.

As long as the "output specifications" of the oscillator meets the "input specifications" of the two connected devices ... including the application conditions ... there is no problem in driving multiple devices from one source.

And yes, I´ve rather often used one clock source for several ICs.

****
You´re a quite experienced forum user and already did some electronics projects.
Thus I´m a bit surprised about the question and your given/missing informations.

* you miss to give rather important informations like: clock frequency and supply voltage..
* you use an "oscillator" .. so you clearly don´t connect a crystal, but still you refer to "crystal load capacitance"
* also you should know that the schematic is a good source of information what you are trying to do. Without it we need to guess a lot. (Pin names, wiring, termination ...)
* also you should already know that PCB layout has big influence in signal quality. Thus - even if the oscillator is able to drive both devices - it still depends on the PCB layout whether it will work on your board or not.

And "work" not only in the meaning that both IC´s operate correctly in your laboratory, but reliably work in the field with variable power supply, temperature... And also in the meaning: your application complies with the regulations (EMI, EMC, ....and so on)

One can say "we" need to "expect" that you correctly decided to use a 25MHz source and correctly decided to use the XI pins as clock input (There are other clock inputs, too). In the end we can not be sure about this. So if we have to guess this, do we also have to expect that you visited the manufacturer internet site for related douments? Like: did you read the "Hardware Design Checklist"?
I mean: you did not give much informations what you did so far.

Klaus

Thank you for your response.

Yes, I missed informations to share.

Regarding the load capacitance, the crystal oscillator also has a load capacitor parameter. So, I assumed the load capacitance topic is applicable to crystals as well as oscillators, right?

Can you clarify whether the load capacitance parameter is applicable to the oscillator parts like in crystals?

 

[KlausST]

Hi,

If you use an crystal oscillator then the manufacturer already took care that the internal crystal, driver, Rs and Cs do their job reliably. You don´t need to take care anymore.

--> Just take care about the output signal, regarding frequency, levels, (load = your application) impedance, waveform...

Since your two ICs don´t incolve any crystal, just omit all crystal referred informations, they don´t apply. Just refer to the "external clock input specifications". Same as above: frequency, levels...

Just to be clear: you simply don´t use a crystal.
***

Compare it with this: Maybe your car has a trailer hitch. As long as you don´t use a trailer .. just ignore all "trailer related informations" .. in the car´s operator manual.

Klaus

 

[FreshmanNewbie]

Hi,

If you use an crystal oscillator then the manufacturer already took care that the internal crystal, driver, Rs and Cs do their job reliably. You don´t need to take care anymore.

--> Just take care about the output signal, regarding frequency, levels, (load = your application) impedance, waveform...

Since your two ICs don´t incolve any crystal, just omit all crystal referred informations, they don´t apply. Just refer to the "external clock input specifications". Same as above: frequency, levels...

Just to be clear: you simply don´t use a crystal.
***

Compare it with this: Maybe your car has a trailer hitch. As long as you don´t use a trailer .. just ignore all "trailer related informations" .. in the car´s operator manual.

Klaus

Thank you very much for the clarification.

So you mean to say that I need not bother about the load capacitance parameters since I am using an oscillator and not a crystal? In that case, why is the load capacitance value of 10pF given in the oscillator datasheet?

 

[KlausST]

Hi,

don´t mix
* "crystal load capacitance" . which you need for resonance --> see the according schematic or any crystal datasheet "recommended circuit".
* with "load capacitance" for an output (may be analog or digital)

--> a crystal without a "crystal load capacitance" may have no reliable resonance and thus may not oscillate at all.
--> a load capacitance at an output will draw current when the signal voltage changes. A load capacitance is hard to drive. The higher the frequency and the higher the capacitance the more current the output needs to provide.

Lets say you have a 25MHz sinewave with 1.5V amplitude. Now imagine there is a 1uF capacitor connected. It has 6.3mOhms impedance .. and needs a current of 230A amplitude. It should be quite obvious that a tiny oscillator can´t provide 230A.
On the other hand every PCB trace, every solder pad, every connector, every IC pin causes some stray capacitance. This is not zero, but much lower than 1uF.
(@10pF it is 630 Ohms and 2.3mA)
So obviously the oscillator needs to be able to drive some amount of capacitance properly and reliably.
So the datasheet gives a recommendation.

***
The oscillator datasheet is for a total of 64 different oscillator devices. We still don´t know which one you decided to use.
But from reading some pieces of the datasheet, it doesn´t look like a "logic level" oscillator at all. But indeed you want to drive some digital logic.. or at least the reference for a PLL.

The datasheet says the output waveform is "clipped sine" with an amplitide of 0.8 x VCC. Which does not define LOGIC_LOW and LOGIC_HIGH levels, nor rise- and fall -rate, nor duty_cycle.

LAN950x datasheet says XI input is fine with LOW:<0.5V and HIGH > 0.8V. Additionally it specifies: "
XI should be driven with
a nominal 0-3.3V clock signal. The input clock duty cycle is 40% minimum, 50% typical and 60% maximum."
I guess your oscillator can´t guarantee this.

Klaus

 

[danadakk]

The OSC Load C spec insures it meets its Tr and Tf spec, and its duty cycle spec.
If you violate that those specs are no longer guaranteed.

The 27 pF is related, I think, to use of a xtal for clock, and allowed max loading
of external caps for the xtal osc circuit. To insure there is enough G in osc internal
G element at frequency to run the osc, have it start up properly. Consult the
architecture manual for this part when using external vs xtal as clock source, the
configuration and recommendations. Then I think you will find using external
clock it only sees a "normal" or slightly higher Cload. Not the 27 pF show in specs.

MCP talks about external clock here, but still no spec on X1 Cinput.....

http://ww1.microchip.com/downloads/en/DeviceDoc/KSZ8795CLX-Hardware-Design-Checklist-00003579B.pdf

Regards, Dana.