Re: [BP] thermal noise question for signaling conference call today
Thanks for clearing that up for me (and thanks to Charles Moore as well). I
was thinking about the resistor as being the source of the noise rather than
the load of the noise.
jts
***************************************
***************************************
John T. Stonick, Ph.D.
Scientist
Synopsys, Inc.
Synopsys Technology Park
2025 NW Cornelius Pass Road
Hillsboro, OR 97124
(P) (503)547-6535
(F) (503)547-6017
(C) (503)330-4858
***************************************
***************************************
> -----Original Message-----
> From: owner-stds-802-3-blade@listserv.ieee.org [mailto:owner-stds-802-3-
> blade@listserv.ieee.org] On Behalf Of Vivek Telang
> Sent: Friday, October 29, 2004 2:08 PM
> To: STDS-802-3-BLADE@listserv.ieee.org
> Subject: Re: [BP] thermal noise question for signaling conference call
> today
>
> John,
>
> Since I brought it up, I'll take a shot at answering your question.
> The 100x multiplier comes in when you convert W to V^2, assuming that
> the load is a 100ohm resistor. This is valid for a 100ohm differential
> impedance system, which I was assuming the backplane channel to be.
>
> But like I said at the end of the call, I was proposing the -140dBm/Hz
> number as the background noise level, not the thermal noise level. I
> didn't realize that Mike had broken out those two parameters.
>
> Vivek
>
> Vivek Telang
> Broadcom Corp.
>
>
> -----Original Message-----
> From: owner-stds-802-3-blade@listserv.ieee.org
> [mailto:owner-stds-802-3-blade@listserv.ieee.org] On Behalf Of John
> Stonick
> Sent: Friday, October 29, 2004 3:29 PM
> To: STDS-802-3-BLADE@listserv.ieee.org
> Subject: [BP] thermal noise question for signaling conference call today
>
>
> I am confused what the -140dBm number refers to in the discussion today
> (I do not recall who put it forth).
>
> -140dbm/Hz is equivalent to 3.2nV/sqrt(Hz).
>
> [-140dBm/Hz = -170dBw/Hz = 1e-17W/Hz = 3.2nV/sqrt(Hz)].
>
> You only get to 32nv/sqrt(Hz) (the number that was said to result) if
> you multiply the original number by 100 (1e-17*100 W/Hz = 32nV/sqrt(Hz))
> The factor of 100 was Mike's resistor value. If this is why this was
> done then I am assuming that the person who mentioned this is using the
> standard spot noise equation for noise from a resistor:
>
> Pspot = 4kTR = 1e-15 (derived from 32nV/sqrt(Hz))
>
> 4(1.38e-23)T(100) = 1e-15W/Hz
>
> T = 181e3
>
> This seems bizarre. Can someone explain why we multiplied -140dBm =
> 1e-17W by 100? If we do not do this and use 3.2nV/sqrt(Hz) then we end
> up with 224uV of noise. Mike's value would have led to about 100uV of
> noise.
>
> I am not sure what the proper number to include for thermal noise, but I
> think that 2.2mV is a bit high.
>
> jts
>
> ***************************************
> ***************************************
>
> John T. Stonick, Ph.D.
> Scientist
> Synopsys, Inc.
> Synopsys Technology Park
> 2025 NW Cornelius Pass Road
> Hillsboro, OR 97124
> (P) (503)547-6535
> (F) (503)547-6017
> (C) (503)330-4858
>
> ***************************************
> ***************************************