For validation, probing at die is better. For field measurement correlation, probing at pin is better.
My DDR is having data transfer speed of 1600 MT/S.
In Hyperlynx Digital Oscilloscope window in oscillator section
i have to take frequency 800 MHz. Am i correct?
When i have taken 800MHz & Interactively simulated then i have got
a different result as compared to the batch mode.
At 800 MHz signal looks somewhat distorted but at 133MHz in digital oscilloscope
window it looks fine.
I am mailing you the simulation results as attachment is not allowed to be sent.
1)I want to know why there is difference between interactive & Batch simulation at same
2)I am taking 800MHz in digital oscilloscope window as my DDR is having data transfer rate of 1600MT/s
is this frequency what i am feeding in oscilloscope is correct ?
Yes, the fequency of stimulus for data is 800MHz for data lines. The selection of simualtor engine and timing step may have some impacts on waveforms's apperances. Without the testcase data, I have no more comments on the difference between result you get from DDRX batch run and
result from individual run. You may donwoad the DDR2 ket from Mentor's support net and check it to see whether there is difference between result you get from DDRX batch run andresult from individual run by varifying all settings and corners.
From: yu.yanfeng email@example.com
Sent: 29 January 2014 11:18
To: Sandeep Kumar(Sandeep Kumar)/Engineer I (사원)/MSI.Engineering/(주)만도
Mentor Graphics Communities<http://communities.mentor.com/index.jspa>
Location of probes at the die or at the pin
reply from yu.yanfeng<http://communities.mentor.com/people/yu.yanfeng> in Signal and Power Integrity, Thermal, EMI, EMC, 3D EM - View the full discussion<http://communities.mentor.com/message/47540#47540
Simulation Results.docx 38.9 KB
I will also like to know for the address & control signal the frequency simulation will be same i.e. 800 MHz or it will be different if 1T timing is used.
I am pasting the different results obtained in different scenarios,
for 1t timing, the sitimulus for Address/ctrl lines should be 400MHz(1/2 clock frequency), not 800MHz
Somewhere i have read that if we are using 2T timing then address frequency is half of the clock frequency.
I have sent u the simulations results in MS word file that is attached above in this thread.
Kindly see it & reply.
For DDR 2T timing , the maxium stimulus should be set to 1/4 clock frequency.
Thanks a lot for your valuable support......
What i understood from the discusssion is that :-
I am having DDR3 having a data rate of 1600 MT/s so Clock Frequency will be 800 MHz.
I will choose 1T timing then for ADDRESS/CONTROL signals frequency will be 1/2 * 800 = 400 MHz &
if i will choose 2T timing then for ADDRESS/CONTROL signals frequency will be 1/4 * 800 = 200 MHz.
Am I Correct ?
also i want to know for DATA/DQS/DM Signals what will be the frequency of simulation if DDR is having transfer rate of 1600MT/s i.e. clock = 800 MHz
Kindly Clarify .
DDR 3 1600MT, Fclock=800MHz
Address/Command=400MHZ@ 1T timing and =200MHz@ 2T timing
Control = 400MHz whatever 1T timg or 2T Timing
Thanks for the support & clarification
While interactively simulating the different signals i.e. address/data/command.we can manually change the required frequency
needed for that signal .
But in BATCH MODE simulation after choosing the Timing either 1T or 2T does the tool itself adjusts
frequecy for different group of signals like address/data/command ? if not then how can we feed the different frequencies to different
group of signals.
here already Data Clock is getting half to 800 . For Address/Control it is again getting halved why ?