Slightly O.T.: 90 degree phase shift of sinusoidal wave using analog components

"Jerry Avins" <jya@ieee.org> wrote in message
news:MImdndlYgs5EmPjYnZ2dnUVZ_vadnZ2d@rcn.net...
> Randy Yates wrote:
> > Jerry Avins <jya@ieee.org> writes:
> >
> >> Randy Yates wrote:
> >>
> >>   ...
> >>
> >>> If you're talking about a multiplier circuit in the phase detector, I
> >>> agree. Since the filtered output of the multiplier with cos(wc*t+p1)
> >>> and cos(wc*t+p2) as inputs is cos(p1-p2), and the loop drives this
> >>> error to zero, and cos(x) = 0 when p1-p2 = 90, the resulting output is
> >>> in phase quadrature.
> >>> Thanks for bringing this up - I never thought about it. As you said,
> >>> I would've thought a typical PLL locks direct, not in quadrature.
> >> Think of the XOR or balanced modulator as a multiplier. If the
> >> harmonics of a square wave don't hurt (in a PLL they actually increase
> >> the gain by 4/pi) just multiplying the sine bits is enough. XORing the
> >> analog signals does that.
> >
> > How in the HECK would you "XOR" analog signals? You don't seriously mean
> > inputting analog signals into a digital device?
>
> Why not? With CMOS, make amplifiers out of inverters with resistive
> feedback from output to input and a resistor and capacitor in the input,
> and run the clipped signal into a gate. Inside, all circuits (except
> maybe magnetic cores) are analog.
>

AMEN to that !

M.



-- 
Posted via a free Usenet account from http://www.teranews.com

Jerry Avins wrote:

> Why not? With CMOS, make amplifiers out of inverters with resistive 
> feedback from output to input and a resistor and capacitor in the input, 
> and run the clipped signal into a gate. Inside, all circuits (except 
> maybe magnetic cores) are analog.
> 
> 
>                +----\/\/\/\----+
>                |               |
>                |      |\       |   Gain ~= 23 dB
> __| |_/\/\/\___|______| \______|___
>   | |                 | /
>                       |/
> 


It works, it's a fact. You can demonstrate it however I would strongly 
not recommend using the tricks like that in the production.Vladimir 
Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com The operation strongly depends on the series and 
on the particular part. The questient current will be high and it will 
eventually damage the CMOS gate. Many gates have a hysteresis at the 
input, so this thing will oscillate.


> Without too much fuss, this works with TTL, too, but you need to watch 
> the dissipation.

It may work with the classic 74LS and CD40xx, however the results with 
the other series are unpredictable.


Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

Thanks a lot,

I think this may do, I just have to check the price here in South
Africa.

Kind Regards,
Jaco

Randy Yates wrote:
> Jaco,
>
> Why not just use something like the TRF 3701?
>
> http://focus.ti.com/paramsearch/docs/parametricsearch.tsp?family=analog&familyId=632&uiTemplateId=NODE_STRY_PGE_T&paramCriteria=no
> --
> %  Randy Yates                  % "She's sweet on Wagner-I think she'd die for Beethoven.
> %% Fuquay-Varina, NC            %  She love the way Puccini lays down a tune, and
> %%% 919-577-9882                %  Verdi's always creepin' from her room."
> %%%% <yates@ieee.org>           % "Rockaria", *A New World Record*, ELO
> http://home.earthlink.net/~yatescr

Hi,

I think you need the LO signal and the 90 degree shifted LO to multiply
with in QPSK.  But, I might have it wrong....

Kind Regards,
Jaco


Jerry Avins wrote:
> jaco.versfeld@gmail.com wrote:
> > Hi,
> >
> > I apologise if this is not strictly DSP related, but it is still signal
> > processing related.
> >
> > Basically my question is how one can get a 90 degree phase shifted
> > version of a sinusoidal wave using analog components.  The only
> > challenge is that the frequency will not be fixed.
>
> The legalistic answer is "no", but it can be done well enough. There are
> circuits that accept an audio input and produce a pair of outputs that
> are nearly 90 degrees apart from one another. Such circuits become very
> difficult to trim if the bandwidth much exceeds a decade or if
> quadrature must be maintained to better than a few tenth of a degree.
> The phase response of either output relative to the input is well
> defined, but not very interesting. The one I built used six op-amps.
>
> > For a fixed frequency, one could use an all-pass filter.  (Are there
> > any other solutions for a fixed frequency.)
> >
> > Basically, I am trying to build a QPSK modulator, and would like to
> > implement frequency hopping as well.
> >
> > Is there a way that this could be done digitally at say 433 MHz?
> >
> > Any thoughts, suggestions and/or pointers to literature would be
> > greatly appreciated.
>
> What signal do you need to shift? The RF oscillator can have a pair of
> quadrature outputs at all frequencies.
>
> Jerry
> --
> Engineering is the art of making what you want from things you can get.
> =AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF

Hi,

I think the 433MHz band is an unlicensed band, or at least here in
South Africa...

Kind Regards,
Jaco


Vladimir Vassilevsky wrote:
> jaco.versfeld@gmail.com wrote:
>
> > Basically my question is how one can get a 90 degree phase shifted
> > version of a sinusoidal wave using analog components.  The only
> > challenge is that the frequency will not be fixed.
>
> This is indeed possible. However it all depends on how accurate should
> be the phase shift over the bandwidth of interest.
>
> >
> > For a fixed frequency, one could use an all-pass filter.  (Are there
> > any other solutions for a fixed frequency.)
>
> There are many possible solutions depending on the application. Perhaps,
> the simplest is to make a Chebyshev approximation of the desired phase
> shift using a bunch of allpass filters.
>
>
> >
> > Basically, I am trying to build a QPSK modulator, and would like to
> > implement frequency hopping as well.
>
> I see. Basically, you want to suppress the unwanted sideband.
>
> >
> > Is there a way that this could be done digitally at say 433 MHz?
>
> It is possible, however why would you need to do it at 433 MHz?
>
>
> Vladimir Vassilevsky
> 
> DSP and Mixed Signal Design Consultant
> 
> http://www.abvolt.com

Thanks a lot for the answers.  I certainly learned a lot.

I hope that one day I can also be as knowledgable...

Kind Regards,
Jaco

jaco.versfeld@gmail.com writes:

> Thanks a lot for the answers.  I certainly learned a lot.
>
> I hope that one day I can also be as knowledgable...

Hi Jaco,

I have observed your postings to this group for a few years and from
my perspective you appear to be one of the most knowledgable and
polite people here. You are definitely an asset to this group, and I
would love to meet you in person some day. Good luck in your work
and studies!
-- 
%  Randy Yates                  % "And all that I can do
%% Fuquay-Varina, NC            %  is say I'm sorry, 
%%% 919-577-9882                %  that's the way it goes..."
%%%% <yates@ieee.org>           % Getting To The Point', *Balance of Power*, ELO
http://home.earthlink.net/~yatescr

Vladimir Vassilevsky wrote:
> 
> 
> Jerry Avins wrote:
> 
>> Why not? With CMOS, make amplifiers out of inverters with resistive 
>> feedback from output to input and a resistor and capacitor in the 
>> input, and run the clipped signal into a gate. Inside, all circuits 
>> (except maybe magnetic cores) are analog.
>>
>>
>>                +----\/\/\/\----+
>>                |               |
>>                |      |\       |   Gain ~= 23 dB
>> __| |_/\/\/\___|______| \______|___
>>   | |                 | /
>>                       |/
>>
> 
> 
> It works, it's a fact. You can demonstrate it however I would strongly 
> not recommend using the tricks like that in the production.Vladimir 
> Vassilevsky
> 
> DSP and Mixed Signal Design Consultant
> 
> http://www.abvolt.com The operation strongly depends on the series and 
> on the particular part. The questient current will be high and it will 
> eventually damage the CMOS gate. Many gates have a hysteresis at the 
> input, so this thing will oscillate.

You would not want to use B (buffered output) series CMOS. The A series 
withstands continuous shorts to ground or +5. They get warm, but 
survive, at +12.

>> Without too much fuss, this works with TTL, too, but you need to watch 
>> the dissipation.
> 
> It may work with the classic 74LS and CD40xx, however the results with 
> the other series are unpredictable.

So don't use other series. As with all other crafts, understanding how 
the materials work helps engineers immensely.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������