Pulse Attenuation Problem in AWR

Hi everyone,
I made a TDR simulation in AWR with short time pulses and also an TDR experiment in laboratory. The measured time delays were similar,however the magnitudes of reflected pulses were highly different.
As a incident signal i used a pulse with 8ns pulse width, 1V voltage and at 1Mhz frequency. In simulation the magnitude of reflected pulse was 0.96V,however it was 0.65V in experiment.
I think the difference caused by AWR. In AWR, pulse attenuation is calculated with project frequency 1Mhz. Its calculated like the pulse width is 500ns.
But actually pulse width is 8ns and the attenuation should be more because narrow pulse have less energy. If the project frequency is 1MHz , pulses, with 1ns-500ns pulse width attenuate at same rate,however it can't be in real world.
How can i solve this problem?

Do you have any idea of this problem?
AWR calculate the attenuation of signals according to project frequency. If the project frequency 1MHz it doesn' matter that pulse width 10ns, 100ns or 500ns. They all are thought as a pulse with a 500ns pulse width because the project frequency is 1MHz.

I think, the basic question is, if AWR has a suitable cable model and is using it in time domain simulation? Did you calculate the expectable pulse response by hand?

Hi again FvM,

Yes, i calculated the expectable pulse response and also made an experiment with pulse generator and digital oscilloscope. The results are similar.
I used a 8ns pulse in experiment at 1V voltage and 1MHz. I became 0.62V reflection from the open circuited far end of 25 m RG58 coaxial cable. At the theorical part; there is a formula for frequency of cable insertion loss in MHz according to %50 pulse width in ns (f = 250 / tp) So for insertion loss calculation, the frequency of 8ns pulse width is 31.25MHz. The coax calculator on the internet gave for RG58 at that frequency 2.292 db/25 m. Of course for round trip the loss would be 4.584 dB for 50 m. If we calculate we expect 0.59V . The results are nearly similar.
Also for pulse with 500ns pulse width, the attenuation is calculated in same way and the result is 0.532dB for 50 m and it is 0.94V expectation. The practical measurement is 0.945V
However in AWR the situation is different. There are kinds of coaxial transmission line element. I used COAXI2 and COAXP2 and configured them according to manufacturers data (RG58-CU);
The values that i gave to COAXP2 are: Di: 0.9mm, Do:2.95mm, L:25m, K:2.295(dielectric), A:0.177 dB/m , F:100MHz
The values that i gave to COAXI2 are: Z:50, L:25m, K:2.295(dielectric), A:0.177 dB/m , F:100MHz

I gave them the same attenuation value 0.177 according to manufacturers data. But the result in 8ns pulse width at 1MHz is for COAXI2 0.9697V, for COAXI2 0.1012V.
I don't understand why COAXP2 behaves totally stupid, but then i got the idea that COAXI2 behaves true but the 0.9697V response is not for 8ns pulses.Its calculated for 500ns pulses. Then i make a simulation with 500ns pulse and the result is the same 0.97V.
The point i got from this things, AWR calculates the attenuation according to project frequency 1MHz (normally 500ns pulse width for 1Mhz pulse), doesn't look to actual pulse width...