On Fri, 27 Jun 2014 07:28:08 -0700 (PDT), Rakesh Joshi
<rakesh@saankhyalabs.com> wrote:

>On Tuesday, June 17, 2014 1:29:16 AM UTC+5:30, harishm wrote:
>> why center tone is null in  input of IFFT in OFDM systems ?
>> 
>> 
>> 
>> is it for reducing PAPR ...?
>> 
>> 
>> 
>> or 
>> 
>> 
>> 
>> to reduce complexity in OFDM receiver ?  	 
>> 
>> 
>> 
>> _____________________________		
>> 
>> Posted through www.DSPRelated.com
>
>Hello Eric,
>Im unable to catch words like 'mixer bleed through', although I feel like I understand something vaguely. Can you point me to some literature that briefly explains this situation so that its more clear to me.
>Thanks for your valuable guidance!

A better search term would be "direct conversion carrier leakage". 

In a heterodyne system the local oscillator mixing products other than
the desired image wind up out-of band.   Since the local oscillator
frequency in a direct conversion system equals the rf carrier
frequency, any leakage of the local oscillator into the signal at rf
winds up at DC, and exactly at DC if the receiver phase-locks to the
carrier.

So it's really just a technology problem in that we don't yet have
low-leakage mixer implementations that are cheap enough to use in
consumer products like WiFi, etc.  

Eric Jacobsen
Anchor Hill Communications
http://www.anchorhill.com

On Tuesday, June 17, 2014 1:29:16 AM UTC+5:30, harishm wrote:
> why center tone is null in  input of IFFT in OFDM systems ?
> 
> 
> 
> is it for reducing PAPR ...?
> 
> 
> 
> or 
> 
> 
> 
> to reduce complexity in OFDM receiver ?  	 
> 
> 
> 
> _____________________________		
> 
> Posted through www.DSPRelated.com

Hello Eric,
Im unable to catch words like 'mixer bleed through', although I feel like I understand something vaguely. Can you point me to some literature that briefly explains this situation so that its more clear to me.
Thanks for your valuable guidance!

On Mon, 23 Jun 2014 21:08:55 -0700 (PDT), Rakesh Joshi
<rakesh@saankhyalabs.com> wrote:

>On Tuesday, June 24, 2014 8:49:05 AM UTC+5:30, Eric Jacobsen wrote:
>> On Mon, 23 Jun 2014 11:59:16 -0700 (PDT), Rakesh Joshi
>> wrote:
>>=20
>>=20
>>=20
>> >Thanks Eric. Does having a dc carrier also affect the dagc response if i=
>ts employed?=20
>>=20
>>=20
>>=20
>> Typically DC offset and AGC are detected and adjusted separately from
>>=20
>> each other.   They can be made independent regardless of whether the
>>=20
>> DC subcarrier is populated or not.=20
>>=20
>>=20
>>=20
>>=20
>>=20
>> Eric Jacobsen
>>=20
>> Anchor Hill Communications
>>=20
>> http://www.anchorhill.com
>
>As per my understanding, DC offset that is corrected is due to the ADC comp=
>onents. A long window averaging is taken over the signal, and the estimated=
> DC offset is compensated. But, if there is a dc as part of the ofdm symbol=
>, its effect would last for an entire symbol duration. This gets seen by th=
>e AGC. If the AGC is quick to react to this dc, the AGC gain can falsely va=
>ry. Hence for the AGC to lock to the proper gain values, one may have to de=
>sign the AGC to react slowly. Does this increase the AGC settling time? If =
>so, by knocking off the DC carrier component in the ofdm symbol, does it he=
>lp reduce the AGC settling time? I hope Im clear. Please correct me if you =
>find me wrong.

The DC offset due to the ADC(s) is not the same as the
DC-bleed-through from the mixer into the center carrier.  The carrier
bleed-through is DC relative to the other subcarriers, so even after
the overall DC offset at the ADCs is removed, there can still be a
residual offset contaminating the center subcarrier.   This is why the
center subcarrier cannot be used reliably with direct conversion
(until somebody comes up with a downconverter with zero mixer
bleed-through).

So ADC DC offset and AGC in a direct-conversion OFDM demodulator
generally don't need to be any different than for any other system.
There are always tradeoffs and issues around developing both DC-offset
removal and AGC loop bandwidths, and they are often specific to the
system or even the implementation.   

Eric Jacobsen
Anchor Hill Communications
http://www.anchorhill.com

On Tuesday, June 24, 2014 8:49:05 AM UTC+5:30, Eric Jacobsen wrote:
> On Mon, 23 Jun 2014 11:59:16 -0700 (PDT), Rakesh Joshi
> wrote:
> 
> 
> 
> >Thanks Eric. Does having a dc carrier also affect the dagc response if its employed? 
> 
> 
> 
> Typically DC offset and AGC are detected and adjusted separately from
> 
> each other.   They can be made independent regardless of whether the
> 
> DC subcarrier is populated or not. 
> 
> 
> 
> 
> 
> Eric Jacobsen
> 
> Anchor Hill Communications
> 
> http://www.anchorhill.com

As per my understanding, DC offset that is corrected is due to the ADC components. A long window averaging is taken over the signal, and the estimated DC offset is compensated. But, if there is a dc as part of the ofdm symbol, its effect would last for an entire symbol duration. This gets seen by the AGC. If the AGC is quick to react to this dc, the AGC gain can falsely vary. Hence for the AGC to lock to the proper gain values, one may have to design the AGC to react slowly. Does this increase the AGC settling time? If so, by knocking off the DC carrier component in the ofdm symbol, does it help reduce the AGC settling time? I hope Im clear. Please correct me if you find me wrong.

On Mon, 23 Jun 2014 11:59:16 -0700 (PDT), Rakesh Joshi
<rakesh@saankhyalabs.com> wrote:

>Thanks Eric. Does having a dc carrier also affect the dagc response if its employed? 

Typically DC offset and AGC are detected and adjusted separately from
each other.   They can be made independent regardless of whether the
DC subcarrier is populated or not. 


Eric Jacobsen
Anchor Hill Communications
http://www.anchorhill.com

Eric Jacobsen <eric.jacobsen@ieee.org> wrote:

(snip)
> Systems that are expected to use a direct-conversion rf architecture
> in order to reduce cost have to leave the center carrier unpopulated
> since it will be subject to a DC offset due to bleed-through of the
> mixing oscillator.   
(snip)

> Since televisions have historically used super-heterodyne
> architectures, for those cases the center carrier can be used for
> information.   

Though for vestigial side-band, the carrier isn't in the center.

-- glen

Thanks Eric. Does having a dc carrier also affect the dagc response if its employed?

Thanks Eric. Does having a dc carrier also affect the dagc response if its employed?

On Mon, 23 Jun 2014 06:10:51 -0700 (PDT), rakesh@saankhyalabs.com
wrote:

>On Tuesday, June 17, 2014 2:05:25 AM UTC+5:30, Peter Mairhofer wrote:
>> On 2014-06-16 12:59, harishm wrote:
>> 
>> > why center tone is null in  input of IFFT in OFDM systems ?
>> 
>> > 
>> 
>> > is it for reducing PAPR ...?
>> 
>> 
>> 
>> Not sure what you deem "center tone" but if it's frequency 0 (mirrored
>> 
>> around y-axis), it is to remove DC components.
>> 
>> 
>> 
>> Peter
>
>I dont see such a measure taken in DVB-T or ISDB-T DTV stds. Is it there as part of WiMax/WiFi stds? 
>
>-Rakesh

Systems that are expected to use a direct-conversion rf architecture
in order to reduce cost have to leave the center carrier unpopulated
since it will be subject to a DC offset due to bleed-through of the
mixing oscillator.   WiFi falls into that category, as do many other
cost-sensitive consumer products.

Since televisions have historically used super-heterodyne
architectures, for those cases the center carrier can be used for
information.   So DVB-T and other standards that are not expected to
use direct-conversion architectures will populate the center
subcarrier.

Eric Jacobsen
Anchor Hill Communications
http://www.anchorhill.com