Simple analog bandpass filter of the incoming signal, syncronize a ADC from=
a PWM Local Osc generated by the micro, use a low frequency FIR filter to =
regenerate the shifted down signal from the undersampled output of the ADC.=
http://www.hqew.net/product-data/74HC4052
Reply by Phil Hobbs●May 5, 20122012-05-05
Joerg wrote:
>
> Phil Hobbs wrote:
> > Tim Wescott wrote:
> >> On Fri, 04 May 2012 10:47:31 -0500, Tim Wescott wrote:
> >>
> >>> On Fri, 04 May 2012 08:13:29 -0700, Joerg wrote:
> >>>
> >>>> miso wrote:
> >>>>> On 5/3/2012 7:38 PM, Tim Williams wrote:
> >>>>>> Hell... these days, it's more like... ADC, Hilbert transform, lowpass
> >>>>>> filter. And it'd probably come out way less than a watt. That's kind
> >>>>>> of sad.
> >>>>>>
> >>>>>> Tim
> >>>>>>
> >>>>>>
> >>>>> If you have access to the clock, actually 4x the clock, the demod can
> >>>>> be quite simple. Split the sample stream in half, then with those two
> >>>>> streams, alternately multiply by -1. Then you have I and Q channels.
> >>>>>
> >>>>>
> >>>> I know, I've done it many times that way using discrete parts. Since
> >>>> the late 80's, pretty much. But that consumes lots of real estate and
> >>>> power. This time I don't have either.
> >>> You could reduce power (maybe) by using NE612 parts -- but not space, I
> >>> think.
> >> Come to think of it, I've done this using CMOS switches as the "mixer" --
> >> but that was in a spot that could stand being low dynamic range, and it
> >> was at audio-ish frequencies.
> >>
> >
> > CMOS switches make pretty good mixers. Good MOSFET mixers have a higher
> > IP3 than just about anything--I still miss the late lamented Si8602.
> >
>
> If you ever come across the absolute cat's meouw for any sort of analog
> chip get a tube of them. That's what I did with the mother of all
> mixers, the Plessey SL6440. Just in time before it all went under. Same
> with the LH0063 and LH0033, because they never made anything even
> remotely comparable and I kind of knew they wouldn't.
>
Well, there was the LH4008/4009, but they went away first.
I have a reel of BF862s, and will probably get one of NE38143s. I also
have several hundred BFG25A/W's and a bunch of MC1496s in various
packages.
(I also just got another SD-24 for $100 even.)
Cheers
Phil Hobbs
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Reply by Joerg●May 5, 20122012-05-05
Phil Hobbs wrote:
> Tim Wescott wrote:
>> On Fri, 04 May 2012 10:47:31 -0500, Tim Wescott wrote:
>>
>>> On Fri, 04 May 2012 08:13:29 -0700, Joerg wrote:
>>>
>>>> miso wrote:
>>>>> On 5/3/2012 7:38 PM, Tim Williams wrote:
>>>>>> Hell... these days, it's more like... ADC, Hilbert transform, lowpass
>>>>>> filter. And it'd probably come out way less than a watt. That's kind
>>>>>> of sad.
>>>>>>
>>>>>> Tim
>>>>>>
>>>>>>
>>>>> If you have access to the clock, actually 4x the clock, the demod can
>>>>> be quite simple. Split the sample stream in half, then with those two
>>>>> streams, alternately multiply by -1. Then you have I and Q channels.
>>>>>
>>>>>
>>>> I know, I've done it many times that way using discrete parts. Since
>>>> the late 80's, pretty much. But that consumes lots of real estate and
>>>> power. This time I don't have either.
>>> You could reduce power (maybe) by using NE612 parts -- but not space, I
>>> think.
>> Come to think of it, I've done this using CMOS switches as the "mixer" --
>> but that was in a spot that could stand being low dynamic range, and it
>> was at audio-ish frequencies.
>>
>
> CMOS switches make pretty good mixers. Good MOSFET mixers have a higher
> IP3 than just about anything--I still miss the late lamented Si8602.
>
If you ever come across the absolute cat's meouw for any sort of analog
chip get a tube of them. That's what I did with the mother of all
mixers, the Plessey SL6440. Just in time before it all went under. Same
with the LH0063 and LH0033, because they never made anything even
remotely comparable and I kind of knew they wouldn't.
--
Regards, Joerg
http://www.analogconsultants.com/
Reply by Phil Hobbs●May 5, 20122012-05-05
Tim Wescott wrote:
>
> On Fri, 04 May 2012 10:47:31 -0500, Tim Wescott wrote:
>
> > On Fri, 04 May 2012 08:13:29 -0700, Joerg wrote:
> >
> >> miso wrote:
> >>> On 5/3/2012 7:38 PM, Tim Williams wrote:
> >>>> Hell... these days, it's more like... ADC, Hilbert transform, lowpass
> >>>> filter. And it'd probably come out way less than a watt. That's kind
> >>>> of sad.
> >>>>
> >>>> Tim
> >>>>
> >>>>
> >>> If you have access to the clock, actually 4x the clock, the demod can
> >>> be quite simple. Split the sample stream in half, then with those two
> >>> streams, alternately multiply by -1. Then you have I and Q channels.
> >>>
> >>>
> >> I know, I've done it many times that way using discrete parts. Since
> >> the late 80's, pretty much. But that consumes lots of real estate and
> >> power. This time I don't have either.
> >
> > You could reduce power (maybe) by using NE612 parts -- but not space, I
> > think.
>
> Come to think of it, I've done this using CMOS switches as the "mixer" --
> but that was in a spot that could stand being low dynamic range, and it
> was at audio-ish frequencies.
>
CMOS switches make pretty good mixers. Good MOSFET mixers have a higher
IP3 than just about anything--I still miss the late lamented Si8602.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
Reply by Joerg●May 5, 20122012-05-05
krw@att.bizzzzzzzzzzzz wrote:
> On Fri, 04 May 2012 09:39:03 -0700, Joerg <invalid@invalid.invalid> wrote:
>
>> miso wrote:
>>> On 5/4/2012 8:13 AM, Joerg wrote:
>>>> miso wrote:
>>>>> On 5/3/2012 7:38 PM, Tim Williams wrote:
>>>>>> Hell... these days, it's more like... ADC, Hilbert transform, lowpass
>>>>>> filter. And it'd probably come out way less than a watt. That's kind of
>>>>>> sad.
>>>>>>
>>>>>> Tim
>>>>>>
>>>>> If you have access to the clock, actually 4x the clock, the demod can be
>>>>> quite simple. Split the sample stream in half, then with those two
>>>>> streams, alternately multiply by -1. Then you have I and Q channels.
>>>>>
>>>> I know, I've done it many times that way using discrete parts. Since the
>>>> late 80's, pretty much. But that consumes lots of real estate and power.
>>>> This time I don't have either.
>>>>
>>> Lots of patents on that scheme. It is really a pity since the scheme is
>>> totally obvious to those skilled in the art.
>>
>> If such silly patents haven't run out all it takes is proof that you
>> dunnit 25 years ago and ... poof ... blows it out the water.
>
> That and a megabuck or two.
Depends on how compelling your evidence is. I've so far never had to go
to the mat on any legal stuff. The (very) few times it came to a head
there the guys looked at the evidence and decided to bury the hatchets.
--
Regards, Joerg
http://www.analogconsultants.com/
Reply by Joerg●May 5, 20122012-05-05
Klaus Kragelund wrote:
> On 5 Maj, 03:06, n...@puntnl.niks (Nico Coesel) wrote:
>> Kvik <klaus.kragel...@gmail.com> wrote:
>>> On 4 Maj, 21:10, Joerg <inva...@invalid.invalid> wrote:
>>>> Glenn wrote:
>>>>> On 04/05/12 02.53, Joerg wrote:
>>>>>> While looking at a single sideband receiver project that may be coming
>>>>>> up I checked the usual suspects for quadrature demodulators. Many won'=
>>> t
>>>>>> (technically ...) go much below 100MHz. Some do and I need to be more =
>>> in
>>>>>> the 5-10MHz region. But they are massive guzzlers when it comes to
>>>>>> draining the battery. Like this one which slurps a whole watt (!):
>>> Why not use a cheap microcontroller, low current like the Cortex M0
>>> NXP series.
>>> Simple analog bandpass filter of the incoming signal, syncronize a ADC
>> >from a PWM Local Osc generated by the micro, use a low frequency FIR
>>> filter to regenerate the shifted down signal from the undersampled
>>> output of the ADC.
>>> The ADC needs to be triggered with low jitter, but that should be
>>> simple.
>>> The Cortex M0 can run at full speed below 1mA, add another 1mA for a
>>> 1MSa/s ADC. A number of other microcontrollers can do the same, the
>>> Cortex is just a very nice part.
>> Sounds like a nice idea. IIRC the ADC on NXP ARM devices can be
>> triggered by a pin
>
> Yes, you can use a PWM output routed back to a pin to trigger the ADC
>
> It's 12bit, 2MSa, with a couple of LSB max error
>
With a direct conversion receiver that won't work too well. One issue is
that in many applications there is a substantial DC bias on the signal
which is essentially useless and must be stripped. Done with a capacitor
in analog but when using the ADC as the downconverter the DC part eats
up a lot of its dynamic range.
The other challenge is in-band noise from other signals that can be much
stronger than the one you want to listen to. The pre-filter can only do
so much. 16 bits would be nice. The more the better, of course :-)
--
Regards, Joerg
http://www.analogconsultants.com/
Reply by Klaus Kragelund●May 5, 20122012-05-05
On 5 Maj, 03:06, n...@puntnl.niks (Nico Coesel) wrote:
> Kvik <klaus.kragel...@gmail.com> wrote:
> >On 4 Maj, 21:10, Joerg <inva...@invalid.invalid> wrote:
> >> Glenn wrote:
> >> > On 04/05/12 02.53, Joerg wrote:
> >> >> While looking at a single sideband receiver project that may be coming
> >> >> up I checked the usual suspects for quadrature demodulators. Many won'=
> >t
> >> >> (technically ...) go much below 100MHz. Some do and I need to be more =
> >in
> >> >> the 5-10MHz region. But they are massive guzzlers when it comes to
> >> >> draining the battery. Like this one which slurps a whole watt (!):
>
> >Why not use a cheap microcontroller, low current like the Cortex M0
> >NXP series.
>
> >Simple analog bandpass filter of the incoming signal, syncronize a ADC
> >from a PWM Local Osc generated by the micro, use a low frequency FIR
> >filter to regenerate the shifted down signal from the undersampled
> >output of the ADC.
>
> >The ADC needs to be triggered with low jitter, but that should be
> >simple.
>
> >The Cortex M0 can run at full speed below 1mA, add another 1mA for a
> >1MSa/s ADC. A number of other microcontrollers can do the same, the
> >Cortex is just a very nice part.
>
> Sounds like a nice idea. IIRC the ADC on NXP ARM devices can be
> triggered by a pin
Yes, you can use a PWM output routed back to a pin to trigger the ADC
It's 12bit, 2MSa, with a couple of LSB max error
/klaus
Reply by Allan Herriman●May 5, 20122012-05-05
On Fri, 04 May 2012 11:11:22 -0500, Tim Wescott wrote:
> On Fri, 04 May 2012 10:47:31 -0500, Tim Wescott wrote:
>
>> On Fri, 04 May 2012 08:13:29 -0700, Joerg wrote:
>>
>>> miso wrote:
>>>> On 5/3/2012 7:38 PM, Tim Williams wrote:
>>>>> Hell... these days, it's more like... ADC, Hilbert transform,
>>>>> lowpass filter. And it'd probably come out way less than a watt.
>>>>> That's kind of sad.
>>>>>
>>>>> Tim
>>>>>
>>>>>
>>>> If you have access to the clock, actually 4x the clock, the demod can
>>>> be quite simple. Split the sample stream in half, then with those two
>>>> streams, alternately multiply by -1. Then you have I and Q channels.
>>>>
>>>>
>>> I know, I've done it many times that way using discrete parts. Since
>>> the late 80's, pretty much. But that consumes lots of real estate and
>>> power. This time I don't have either.
>>
>> You could reduce power (maybe) by using NE612 parts -- but not space, I
>> think.
>
> Come to think of it, I've done this using CMOS switches as the "mixer"
> --
> but that was in a spot that could stand being low dynamic range, and it
> was at audio-ish frequencies.
I've done that using CMOS switches as well, about 19 or 20 years ago.
455kHz with a 74HC4053 running from +/- 5V. I had a ring counter made
from 74HC74 flip flops to make the quadrature switching signals for the
'4053.
This was the last IF in a 30GHz tracking receiver.
Regards,
Allan
Reply by krw...@att.bizzzzzzzzzzzz●May 4, 20122012-05-04
On Fri, 04 May 2012 09:39:03 -0700, Joerg <invalid@invalid.invalid> wrote:
>miso wrote:
>> On 5/4/2012 8:13 AM, Joerg wrote:
>>> miso wrote:
>>>> On 5/3/2012 7:38 PM, Tim Williams wrote:
>>>>> Hell... these days, it's more like... ADC, Hilbert transform, lowpass
>>>>> filter. And it'd probably come out way less than a watt. That's kind of
>>>>> sad.
>>>>>
>>>>> Tim
>>>>>
>>>>
>>>> If you have access to the clock, actually 4x the clock, the demod can be
>>>> quite simple. Split the sample stream in half, then with those two
>>>> streams, alternately multiply by -1. Then you have I and Q channels.
>>>>
>>>
>>> I know, I've done it many times that way using discrete parts. Since the
>>> late 80's, pretty much. But that consumes lots of real estate and power.
>>> This time I don't have either.
>>>
>>
>> Lots of patents on that scheme. It is really a pity since the scheme is
>> totally obvious to those skilled in the art.
>
>
>If such silly patents haven't run out all it takes is proof that you
>dunnit 25 years ago and ... poof ... blows it out the water.
That and a megabuck or two.
Reply by Nico Coesel●May 4, 20122012-05-04
Kvik <klaus.kragelund@gmail.com> wrote:
>On 4 Maj, 21:10, Joerg <inva...@invalid.invalid> wrote:
>> Glenn wrote:
>> > On 04/05/12 02.53, Joerg wrote:
>> >> While looking at a single sideband receiver project that may be coming
>> >> up I checked the usual suspects for quadrature demodulators. Many won'=
>t
>> >> (technically ...) go much below 100MHz. Some do and I need to be more =
>in
>> >> the 5-10MHz region. But they are massive guzzlers when it comes to
>> >> draining the battery. Like this one which slurps a whole watt (!):
>>
>
>Why not use a cheap microcontroller, low current like the Cortex M0
>NXP series.
>
>Simple analog bandpass filter of the incoming signal, syncronize a ADC
>from a PWM Local Osc generated by the micro, use a low frequency FIR
>filter to regenerate the shifted down signal from the undersampled
>output of the ADC.
>
>The ADC needs to be triggered with low jitter, but that should be
>simple.
>
>The Cortex M0 can run at full speed below 1mA, add another 1mA for a
>1MSa/s ADC. A number of other microcontrollers can do the same, the
>Cortex is just a very nice part.
Sounds like a nice idea. IIRC the ADC on NXP ARM devices can be
triggered by a pin as well.
--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------