On Feb 21, 5:27�pm, YTach <y.tachw...@gmail.com> wrote:
[snip]
> Also regarding the equalizer, I am also sharing your opinion that I
> may probably need it and probably not, when I read the book of
> Rappaport "Wireless Communications" �He has mentioned in an example
> that an outdoor wireless channel has a coherent bandwidth of 146kHz
> and therefore, systems like GSM with 200 kHZ bandwidth do require
> equalizers. But I am not sure the coherent bandwidth for indoor
> environment and do not have any idea what could be the range of them.
> Do you have any clue on this.
Rather than argue qualitatively about this, how about actually
measuring
the effective channel in your system? If your DAC and ADC are fed off
the same clock, this is as simple as sending a proper pseudo-random
sequence and cross-correlating the receiver symbols.
Then you can try the same exercise with your hand in the way, and you
can then "see" the difference.
Julius
Reply by YTach●February 22, 20092009-02-22
On Feb 22, 3:36�pm, julius <juli...@gmail.com> wrote:
> On Feb 21, 11:38�am, cincy...@gmail.com wrote:
>
> > A dose of reality: if you are clocking the TX and RX off of the same
> > source, that implies some sort of physical connection between the two
> > devices. If that's available, why not use that connection to carry
> > your DBPSK signal instead?
>
> [snip]
>
> To add to Jason's comment, also check to make sure that your
> wired clock connection does not cause the overall system
> response to change (i.e., the wire does not conduct RF signals).
>
> This has happened to me before .... �and when I unplugged the
> clock cable to check the PHY, the antenna response changed
> dramatically and I had to start from scratch again!!
>
> Julius
Wooow Julius, very interesting hint. I am wondering if this is what is
happening in my case, never thought of this! I will probe the clocks
and see how they look with/without RF board.
Reply by julius●February 22, 20092009-02-22
On Feb 21, 11:38�am, cincy...@gmail.com wrote:
>
> A dose of reality: if you are clocking the TX and RX off of the same
> source, that implies some sort of physical connection between the two
> devices. If that's available, why not use that connection to carry
> your DBPSK signal instead?
[snip]
To add to Jason's comment, also check to make sure that your
wired clock connection does not cause the overall system
response to change (i.e., the wire does not conduct RF signals).
This has happened to me before .... and when I unplugged the
clock cable to check the PHY, the antenna response changed
dramatically and I had to start from scratch again!!
Julius
Reply by YTach●February 21, 20092009-02-21
On Feb 21, 12:34�pm, Eric Jacobsen <eric.jacob...@ieee.org> wrote:
> On Fri, 20 Feb 2009 18:31:38 -0600, "ytach" <ytachw...@ou.edu> wrote:
> >Hello Guys,
>
> >I have built a basic DBPSK transceiver and I was verifying the
> >implementation when I found out that when I put my hand on the antenna, the
> >signal error rate goes up big time. My question to you guys, is it possible
> >to implement a basic DBPSK transceiver with no error correction or
> >equalization or AGC and yet can sustain the multipath and attenuation
> >effect of placing hand around the antenna? When do I need really these
> >components (Please guys let me know if my assumptions below are wrong):
>
> Your physically touching the antenna? �Depending on how your receiver
> is built you're probably screwing up the input impedance or something.
> As Sebastian mentioned, your antennas may already be in near-field,
> and your hand certainly is if you're touching the antenna. � �At that
> point, all bets are off.
>
> >1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> >from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> >fixed in place I did not implement AGC.
>
> Even wired systems implement AGC. � AGC is pretty fundamental to
> getting deterministic receiver response. � Deleting it implies that
> the operating conditions are unusually stable, and for a wireless
> system that means nothing changes in the environment, no people moving
> around, nobody ever opens or closes a door, etc., etc.
>
> >2- Synchronizer : I did not implement carrier equalization because I am
> >using differential coding so I can live without carrier synch with loss 3
> >dB. and I did not implement the symbol synch because I am clocking both TX
> >and RX from the same source so I do not need timing extraction.
>
> As previously mentioned, use the timing connection to carry the data.
>
> >3- Equalizer
> >The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> >cosine pulse shaping) which I am assuming less than the channel bandwidth
> >coherency in indoor environment so an equalizer can be ignored.
>
> Maybe. � Maybe not.
>
> >Is it is really possible to implement DBPSK in an indoor environment and
> >receive good BER without the components mentioned above?
>
> IMHO not for any practical application. � If the connection exists for
> the synchronization, that's probably a good connection to use for the
> data, too. �AGC is fundamental. � Only in very unusual circumstances
> can you get away without an AGC if you want a practical system. Study
> "dynamic range" in receivers.
>
> Eric Jacobsen
> Minister of Algorithms
> Abineau Communicationshttp://www.ericjacobsen.org
>
> Blog:http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Hello Eric,
I have first to thank you for your post on pulse shaping, it is really
a great post and helped me to tune up my understanding of pulse
shaping. From the answers so far, it might be really scaling issues in
the receiving chain so problably I have to build the AGC.
Also, changing the impedence of the antenna is a good hint. does AGC
combat these problems (impedence change which I would guess cause a
drop in the wireless signal strength and hence AGC comes to play fix
this) ?
As a conclusion from your post, I should study the dynamic range of my
transceiver. However, as I said earlier, the errors are in burst
nature and sometimes lengthy ones. So do you think that AGC is the
only problem? Can I trust my modem design completely if I got it
working by looping the DAC of TX to the ADC of RX?
Also regarding the equalizer, I am also sharing your opinion that I
may probably need it and probably not, when I read the book of
Rappaport "Wireless Communications" He has mentioned in an example
that an outdoor wireless channel has a coherent bandwidth of 146kHz
and therefore, systems like GSM with 200 kHZ bandwidth do require
equalizers. But I am not sure the coherent bandwidth for indoor
environment and do not have any idea what could be the range of them.
Do you have any clue on this.
Sorry for too many questions. But again thank you for sharing your
experience and would be greatful for any hint that can help me going
in the right direction.
Thank you,
Reply by YTach●February 21, 20092009-02-21
On Feb 21, 10:38�am, cincy...@gmail.com wrote:
> On Feb 20, 7:31�pm, "ytach" <ytachw...@ou.edu> wrote:
>
>
>
>
>
> > Hello Guys,
>
> > I have built a basic DBPSK transceiver and I was verifying the
> > implementation when I found out that when I put my hand on the antenna, the
> > signal error rate goes up big time. My question to you guys, is it possible
> > to implement a basic DBPSK transceiver with no error correction or
> > equalization or AGC and yet can sustain the multipath and attenuation
> > effect of placing hand around the antenna? When do I need really these
> > components (Please guys let me know if my assumptions below are wrong):
>
> > 1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> > from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> > fixed in place I did not implement AGC.
>
> > 2- Synchronizer : I did not implement carrier equalization because I am
> > using differential coding so I can live without carrier synch with loss 3
> > dB. and I did not implement the symbol synch because I am clocking both TX
> > and RX from the same source so I do not need timing extraction.
>
> > 3- Equalizer
> > The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> > cosine pulse shaping) which I am assuming less than the channel bandwidth
> > coherency in indoor environment so an equalizer can be ignored.
>
> > Is it is really possible to implement DBPSK in an indoor environment and
> > receive good BER without the components mentioned above?
>
> > I am ready to provide any further technical details as needed and would
> > appreciate sharing your experience with me
>
> A dose of reality: if you are clocking the TX and RX off of the same
> source, that implies some sort of physical connection between the two
> devices. If that's available, why not use that connection to carry
> your DBPSK signal instead?
>
> What I'm getting at is that you will (almost) never have perfect time
> synchronization. Your other questions, as to what you can "get away
> without having," will have differing answers depending upon the exact
> topology of your system; there's no single answer. You seem to have a
> basic communications system in place, which will work in the presence
> of AWGN and no channel dispersion or attenuation. Those additional
> challenges require solutions to combat them, so the answers to your
> questions are "yes, you need AGC/equalization/time sychronization" if
> the structure of your system demands it.
>
> Jason- Hide quoted text -
>
> - Show quoted text -
I totally agree with you Jason, I am going to build all these blocks
after I verify that by transceiver is built correctly. My attempts to
test that were two:
1- looping back DAC of TX to ADC of RX (so this becomes a regular
modem and not a wireless one) and the transceiver is working fine.
2- go wireless! so I tried to control the environment as much as I
can. They transceivers are not moving, yet they are probably too
close. I eliminate the timing problem by clocking both sides from the
same source and here you go, a signal is received but a burst of
errors show up from time to time and some time lengthy busts. When I
see errors coming and going this way I feel it is clocking problem but
I could be wrong. I do not know for sure how wireless channel behave
in that short range. Do you have any debugging techniques to suggest.
Thank you for your reply Jason
Reply by YTach●February 21, 20092009-02-21
On Feb 21, 8:44�am, Sebastian Doht <seb_d...@lycos.com> wrote:
> ytach schrieb:
>
>
>
>
>
> > Hello Guys,
>
> > I have built a basic DBPSK transceiver and I was verifying the
> > implementation when I found out that when I put my hand on the antenna, the
> > signal error rate goes up big time. My question to you guys, is it possible
> > to implement a basic DBPSK transceiver with no error correction or
> > equalization or AGC and yet can sustain the multipath and attenuation
> > effect of placing hand around the antenna? When do I need really these
> > components (Please guys let me know if my assumptions below are wrong):
>
> > 1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> > from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> > fixed in place I did not implement AGC.
>
> > 2- Synchronizer : I did not implement carrier equalization because I am
> > using differential coding so I can live without carrier synch with loss 3
> > dB. and I did not implement the symbol synch because I am clocking both TX
> > and RX from the same source so I do not need timing extraction.
>
> > 3- Equalizer
> > The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> > cosine pulse shaping) which I am assuming less than the channel bandwidth
> > coherency in indoor environment so an equalizer can be ignored.
>
> > Is it is really possible to implement DBPSK in an indoor environment and
> > receive good BER without the components mentioned above?
>
> > I am ready to provide any further technical details as needed and would
> > appreciate sharing your experience with me
>
> Well I try to post something constructive:
> If the RX and TX are only 15cm apart I suspect the RX is in the
> near-field of the TX (depending on your used carrier frequency), so all
> your nice models for multi-path propagation that hold for plane waves
> won't fit any longer. By the way only comparing the BER does not give
> you much insight about what really happened to your signal. Try to
> compare the transmitted and received bitstream (in time and frequency
> domain) to gain some more insight about what happened to your signal on
> the way from the TX to the RX .
>
> Greetz,
> Sebastian- Hide quoted text -
>
> - Show quoted text -
That is a good clue regarding the model of wireless channel in that
range. I will be looking into it. I have simulated my design on a AWGN
channel and it is working fine. (yes AWGN is basic but for a
controlled setup like that where no mobility exists then I am assuming
it is a reasonable test).
and you are abosolutely right regarding looking at the bitstream. As a
matter of fact, I tried different bitstreams to transmit to have a
clue about what is going on. For example, I tried to send a stream of
bytes of the value 1 to check if I am having framing synchronization
problem, so if I received at the end a stream of 2 or 8 or 16 ... etc
then I am having a problem with my framing and the error I am getting
is due to frame synching. I am not doing yet any framing synch
techniques such as using preamble, I have just added a delay that
compensate the delay between TX and RX. I am sure that I need to do a
frame synch later but since I am building the transceiver step by
step, I am trying to verify that I am going to the right direction.
Having said that, it turns out that there is no frame synch problem
and the errors are quit random. Do you have any suggestions on what I
can conclude from the bit stream? Thank you for your reply.
Reply by YTach●February 21, 20092009-02-21
On Feb 21, 7:20�am, makol...@yahoo.com wrote:
> On Feb 20, 7:31�pm, "ytach" <ytachw...@ou.edu> wrote:
>
> > Hello Guys,
>
> > I have built a basic DBPSK transceiver and I was verifying the
> > implementation when I found out that when I put my hand on the antenna, the
> > signal error rate goes up big time. My question to you guys, is it possible
> > to implement a basic DBPSK transceiver with no error correction or
> > equalization or AGC and yet can sustain the multipath and attenuation
> > effect of placing hand around the antenna? When do I need really these
> > components (Please guys let me know if my assumptions below are wrong):
>
> > 1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> > from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> > fixed in place I did not implement AGC.
>
> you probably don't need an AGC if nothing ever moves, but if you're
> going to wrap your hand around on of the ants, then you need an AGC.
> On the other hand, AGC is easy and could help a lot, why not put it
> in?
>
Thank you for the advice. I will be looking into the implementation of
AGC. I am building the transceiver step by step and I want to make
sure that I will not be adding AGC on top of a transceiver that was
not built correctly. As a matter of fact, I have tried to connect the
output of the DAC of the TX to the ADC of the RX (Since TX and RX are
so close) and the receiver is working perfect with no problem (yes the
connection here is pretty stable) but when I go wireless by connecting
the digital converters to RF boards, then I have this problem at the
receive, but you probably right since sometimes I can fix the
receiption by tuning "manually" the gains of TX and RX but not always
that works which makes me skiptical alittle bit and wondering what
else could be wrong.
>
>
> > 2- Synchronizer : I did not implement carrier equalization because I am
> > using differential coding so I can live without carrier synch with loss 3
> > dB. and I did not implement the symbol synch because I am clocking both TX
> > and RX from the same source so I do not need timing extraction.
>
> probably correct
>
> > 3- Equalizer
> > The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> > cosine pulse shaping) which I am assuming less than the channel bandwidth
> > coherency in indoor environment so an equalizer can be ignored.
>
> probabyl correct
>
> > Is it is really possible to implement DBPSK in an indoor environment and
> > receive good BER without the components mentioned above?
>
> if the Tx and Rx are a fixed 15 cm apart, probably yes ...
> if by "indoor env" you mean moving around through various rooms �then
> probabyl no.
I am at the prototyping stage and both TX and RX are at the same room
facing each other. so I "feel" that I should be able to establish
pretty stable connection at such environment with noise immune
modulation such as BPSK. If any clues from your experience could be
shared I will be very greatful. Thank you very much for your reply.
>
>
>
>
>
> > I am ready to provide any further technical details as needed and would
> > appreciate sharing your experience with me- Hide quoted text -
>
> - Show quoted text -
Reply by Eric Jacobsen●February 21, 20092009-02-21
On Fri, 20 Feb 2009 18:31:38 -0600, "ytach" <ytachwali@ou.edu> wrote:
>Hello Guys,
>
>I have built a basic DBPSK transceiver and I was verifying the
>implementation when I found out that when I put my hand on the antenna, the
>signal error rate goes up big time. My question to you guys, is it possible
>to implement a basic DBPSK transceiver with no error correction or
>equalization or AGC and yet can sustain the multipath and attenuation
>effect of placing hand around the antenna? When do I need really these
>components (Please guys let me know if my assumptions below are wrong):
Your physically touching the antenna? Depending on how your receiver
is built you're probably screwing up the input impedance or something.
As Sebastian mentioned, your antennas may already be in near-field,
and your hand certainly is if you're touching the antenna. At that
point, all bets are off.
>1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
>from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
>fixed in place I did not implement AGC.
Even wired systems implement AGC. AGC is pretty fundamental to
getting deterministic receiver response. Deleting it implies that
the operating conditions are unusually stable, and for a wireless
system that means nothing changes in the environment, no people moving
around, nobody ever opens or closes a door, etc., etc.
>2- Synchronizer : I did not implement carrier equalization because I am
>using differential coding so I can live without carrier synch with loss 3
>dB. and I did not implement the symbol synch because I am clocking both TX
>and RX from the same source so I do not need timing extraction.
As previously mentioned, use the timing connection to carry the data.
>3- Equalizer
>The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
>cosine pulse shaping) which I am assuming less than the channel bandwidth
>coherency in indoor environment so an equalizer can be ignored.
Maybe. Maybe not.
>Is it is really possible to implement DBPSK in an indoor environment and
>receive good BER without the components mentioned above?
IMHO not for any practical application. If the connection exists for
the synchronization, that's probably a good connection to use for the
data, too. AGC is fundamental. Only in very unusual circumstances
can you get away without an AGC if you want a practical system. Study
"dynamic range" in receivers.
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org
Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Reply by ●February 21, 20092009-02-21
On Feb 20, 7:31�pm, "ytach" <ytachw...@ou.edu> wrote:
> Hello Guys,
>
> I have built a basic DBPSK transceiver and I was verifying the
> implementation when I found out that when I put my hand on the antenna, the
> signal error rate goes up big time. My question to you guys, is it possible
> to implement a basic DBPSK transceiver with no error correction or
> equalization or AGC and yet can sustain the multipath and attenuation
> effect of placing hand around the antenna? When do I need really these
> components (Please guys let me know if my assumptions below are wrong):
>
> 1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> fixed in place I did not implement AGC.
>
> 2- Synchronizer : I did not implement carrier equalization because I am
> using differential coding so I can live without carrier synch with loss 3
> dB. and I did not implement the symbol synch because I am clocking both TX
> and RX from the same source so I do not need timing extraction.
>
> 3- Equalizer
> The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> cosine pulse shaping) which I am assuming less than the channel bandwidth
> coherency in indoor environment so an equalizer can be ignored.
>
> Is it is really possible to implement DBPSK in an indoor environment and
> receive good BER without the components mentioned above?
>
> I am ready to provide any further technical details as needed and would
> appreciate sharing your experience with me
A dose of reality: if you are clocking the TX and RX off of the same
source, that implies some sort of physical connection between the two
devices. If that's available, why not use that connection to carry
your DBPSK signal instead?
What I'm getting at is that you will (almost) never have perfect time
synchronization. Your other questions, as to what you can "get away
without having," will have differing answers depending upon the exact
topology of your system; there's no single answer. You seem to have a
basic communications system in place, which will work in the presence
of AWGN and no channel dispersion or attenuation. Those additional
challenges require solutions to combat them, so the answers to your
questions are "yes, you need AGC/equalization/time sychronization" if
the structure of your system demands it.
Jason
Reply by Sebastian Doht●February 21, 20092009-02-21
ytach schrieb:
> Hello Guys,
>
> I have built a basic DBPSK transceiver and I was verifying the
> implementation when I found out that when I put my hand on the antenna, the
> signal error rate goes up big time. My question to you guys, is it possible
> to implement a basic DBPSK transceiver with no error correction or
> equalization or AGC and yet can sustain the multipath and attenuation
> effect of placing hand around the antenna? When do I need really these
> components (Please guys let me know if my assumptions below are wrong):
>
> 1- AGC : The TX and RX are so close (15 cm apart) and the TX power can go
> from -6dBm to 6 dBm and I am sending at -3 dBm. since the TX and RX are
> fixed in place I did not implement AGC.
>
> 2- Synchronizer : I did not implement carrier equalization because I am
> using differential coding so I can live without carrier synch with loss 3
> dB. and I did not implement the symbol synch because I am clocking both TX
> and RX from the same source so I do not need timing extraction.
>
> 3- Equalizer
> The bandwidth of my signal is around 400kHz (using BPSK and 0.5 sqrt
> cosine pulse shaping) which I am assuming less than the channel bandwidth
> coherency in indoor environment so an equalizer can be ignored.
>
> Is it is really possible to implement DBPSK in an indoor environment and
> receive good BER without the components mentioned above?
>
> I am ready to provide any further technical details as needed and would
> appreciate sharing your experience with me
Well I try to post something constructive:
If the RX and TX are only 15cm apart I suspect the RX is in the
near-field of the TX (depending on your used carrier frequency), so all
your nice models for multi-path propagation that hold for plane waves
won't fit any longer. By the way only comparing the BER does not give
you much insight about what really happened to your signal. Try to
compare the transmitted and received bitstream (in time and frequency
domain) to gain some more insight about what happened to your signal on
the way from the TX to the RX .
Greetz,
Sebastian