Doubt about RC constant Low Pass Filter

Hi, Si-List friends,This is my first post. I do a very simple simulation in Hyperlynx about RClow pass filter to estimate scope bandwidth influence toward IO bufferoutput rise time.The time constant I choose is     22.5 ps = 1 K ohm * 23 fF = 50 * 450 fFand use the Kintex-7 IBIS file buffer SSTL15_F_HP. Actually, I expect thatthe rise time I got from C1 and C2 is same. But the result surprises me.The RC circuit of 50 ohm * 450 fF gives me a rise time of *169ps* while the1 Kohm * 23 fF gives me a rise time of *1.077ns*. Can anyone told me whydoes that happen?Regard* **Fan* Gaofenghttp://About.me/Fan.GaofengFan GaofengPHONE:+65 9130 5759about.me/Fan.Gaofeng/
cnmuggle 7 years 5 months 25 days

3 answers


The best answer


You can select the best answer for current question!
Answered byArpad_Muranyi 7 years 5 months 25 days
All,While it is true that circuit elements provided by the GUI toolbarin HyperLynx include mounting parasitics, you can also build circuitsin HyperLynx with "pure" SPICE circuit elements as in any other SPICEsimulator.  You just need to be familiar enough with HyperLynx to knowhow to do it.  While HyperLynx is an easy to use tool, it is also verysophisticated with zillions of features and capabilities.  If you needhelp with HyperLynx, please do not hesitate to ask for help from yourlocal Mentor technical support team.  Public discussion forums like thisone are not the place to get such help.Fan,To be honest, I really don't understand the details of your originalproblem description.  I suspect that there may be fundamental errors inthe calculations (for example omitting the PCB trace impedance, or thebuffer impedance, or the buffer's pad capacitance, etc...).  Pleasedescribe your circuit and calculations with a little more detail sowe can see better what you are trying to do.Thanks,Arpad=========================================================================-----
Answered byCharles.Grasso 7 years 5 months 25 days
One thing to keep an eye on is the parasitics in the models. Hyperlynx does not provide *pure* models for C and L - by that I mean that the models also include parastics for mounting inductanceand stray capacitance. So, for example, you may have a different parasitic mounting inductance (by accident) selected between models. Its unlikely that this is your problem - just one thing to check.I have found that Spice is the best tool for "simple" models! Try your calculations in Spice and confirm your calculations.-----
Answered byistvan.novak 7 years 5 months 25 days
Need to keep in mind that a series resistor of R and parallel capacitance of C gives anR*C time constant only as long as this RC filter is driven with a zero impedance sourceand terminated in infinite impedance.  With IBIS models at the driver and/or receiverthis condition is usually not met.Regards,Istvan NovakOracleOn 8/5/2013 5:09 AM, Fan Gaofeng wrote:> Hi, Si-List friends,>>> This is my first post. I do a very simple simulation in Hyperlynx about RC> low pass filter to estimate scope bandwidth influence toward IO buffer> output rise time.>>>> The time constant I choose is     22.5 ps = 1 K ohm * 23 fF = 50 * 450 fF> and use the Kintex-7 IBIS file buffer SSTL15_F_HP. Actually, I expect that> the rise time I got from C1 and C2 is same. But the result surprises me.>>>> The RC circuit of 50 ohm * 450 fF gives me a rise time of *169ps* while the> 1 Kohm * 23 fF gives me a rise time of *1.077ns*. Can anyone told me why> does that happen?>>>> Regard>> * *>> *Fan* Gaofeng>> http://About.me/Fan.Gaofeng> Fan Gaofeng> PHONE:+65 9130 5759> about.me/Fan.Gaofeng/>>>