rf design

[Active8]

On Sat, 20 Nov 2004 02:42:27 GMT, John Fields wrote:

On Sat, 20 Nov 2004 00:42:09 GMT, Rich Grise <rich@example.net> wrote:

It sounds like John is just talking about looking at the waveform
on a scope. The FM is straight across the top and bottom, the
AM has humps.

---
What I'm talking about is what happens to a carrier when it's
frequency modulated. Its amplitude doesn't change, its frequency
does. Take a look at an FM final sometime and you'll see that it's
running class C, and balls-to-the-wall, and if its amplitude changes
at all, it's incidental and certainly not intentional. Think about
why it's called carrier "deviation". It's because the modulation
causes the carrier frequency to deviate about its unmodulated center
frequency, and the depth of modulation is set by causing, for a
particular amplitude variation of the modulating signal, a particular
variation in the carrier frequency.
---

Naturally, when you transform to the frequency domain, things
look a little different. ;-)

And I wasn't speaking loudly enough to convey that I was thinking in

the freq domain.

carrier is modulated, it decreases in level and the sidebands contain the
energy of the information. So you'd be trying to detect

See? It was buried in a previous post.
--
Best Regards,
Mike

 

[rex]

On 9 Apr 2005 16:11:49 -0700, snbaer@msn.com wrote:

will probably need to convert a 10-100 mhz sinewave to some L
band frequency

Sorry, I could go look it up, but what is L band in MHz or GHz?

 

[Joerg]

Hello Rex,

Sorry, I could go look it up, but what is L band in MHz or GHz?

Roughly 800-1500MHz, for most apps.

Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

Hello,

We are open to suggestions concerning off-the-shelf type
solutions, as well as custom design.

That is unlikely as the market for such a device would be small.

Also, we have been bitten in the past by 'out-of-house' designs
that just don't seem to work as advertised. Any suggestions? Is there
such a thing as a contract based upon a working prototype? Am I
dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

That can be done. But first it requires a detailed and finalized
specification. Fixed price contracts are also an option but then the
spec must be pretty much "cast in stone" and there needs to be a clause
that handles diversions from the spec by the client.

Good consultants don't release designs that do not work properly. Else
their reputation would be toast pretty soon ;-)

There is another option and sometimes that is better in the long run. If
there is an engineer in-house who could do the job to some extent he or
she could be paired up with a consultant. The consultant then suggests
all the critical stuff or sometimes designs the nastiest sections, but
always explaining what was done and why it was done that way. This will
build up know-how where it should be (in house). Done it many times,
even across a few thousand miles or an ocean.

Regards, Joerg

http://www.analogconsultants.com

 

[Joerg]

Hello,

I like the last option you presented. I agree, an engineers
reputation should be determined by the quality of their designs, but
have found the pool of engineers with known reputations to be rather
small. I have recently worked on designs that 'do-not-work' even after
the initial design, and a design review by a 'reputable consultant'.
Who's responsible? I guess it depends on the structure of the group you
work with, or the details of some legal agreement.

Known reputation is a problem. How can you tell? That is what I ran into
when looking for a consultant who could design for injection molding
because that was clearly out of my realm. Got a few bruises there, too.

Calling former clients is one method. However, most contracts clearly
prohibit the consultant from revealing the identity of clients.
Sometimes that is because the group that needed the consultant was kind
of embarrassed that they couldn't do it on their own. So they'd rather
not have it broadcast. Other times the matter is just so confidential
that the client doesn't want the slightest risk of divulgement. Both are
very understandable but will put you in a dilemma figuring whether that
consultant is really going to pan out for you.

I believe the best strategy is to enter into a mutual NDA and let the
consultant make some suggestions. Many will offer an hour or so of
discussion free of charge, just like some attorneys do. If you have
knowledgeable folks sitting in (but who don't hedge an NIH syndrome...)
they could then voice their opinions afterwards. Other times you might
just want to send the docs over and have the consultant propose what he
or she would do. When it comes down to the mat, invite the consultant.
That's going to cost, of course, but you could still turn them down if
uneasy about the person.

I have compiled some thoughts about this on our web site:

http://www.analogconsultants.com/howtofind.htm

Regards, Joerg

http://www.analogconsultants.com

 

[Ban]

snbaer@msn.com wrote:
snip

Also, we have been bitten in the past by 'out-of-house' designs
that just don't seem to work as advertised. Any suggestions? Is there
such a thing as a contract based upon a working prototype? Am I
dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

It will be hard if there isn't a person who has deep insight into the matter
himself. Just like finding a good dentist or plumber is easier if you are
familiar with the respective field, otherwise you will need some good luck.
On the other hand a well educated and experienced person in that field will
be always the better choice. If you are in doubt, it might be of advantage
to hire two persons and compare their advices.
--
ciao Ban
Bordighera, Italy

 

[RST Engineering (jw)]

Well, sir, I'd suggest that your past "bittens" are possibly of your own
making

10 to 100 millihertz (mhz) are rather low frequencies. Perhaps you meant
MEGAhertz, but that is not what the "specification" said.

LIkewise the gate signal between 50 nanoseconds to 1 milliSiemens. That
doesn't make a lot of sense. And "some" L-band frequency is pretty
vague...like from 1 to 3 Gigs.

Perhaps the bitten might be avoided by writing the specification so that a
competent consultant can bid on it properly.

Jim



<snbaer@msn.com> wrote in message
news:1113088309.259466.157900@o13g2000cwo.googlegroups.com...

I am tasked with evaluating the feasablility of having an
upconverter designed. The requirements are not clear yet, but the
design will probably need to convert a 10-100 mhz sinewave to some L
band frequency. The output will need to be gated on with an externally
supplied signal that has programmable output amplitude, and will
support most 'positive-logic' families. The gate signal will be 50ns to
1mS. The IF input will be CW and linear-sweep type waveforms.
We are open to suggestions concerning off-the-shelf type
solutions, as well as custom design.
Also, we have been bitten in the past by 'out-of-house' designs
that just don't seem to work as advertised. Any suggestions? Is there
such a thing as a contract based upon a working prototype? Am I
dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

 

[Ken Taylor]

<snbaer@msn.com> wrote in message
news:1113088309.259466.157900@o13g2000cwo.googlegroups.com...

I am tasked with evaluating the feasablility of having an
upconverter designed. The requirements are not clear yet, but the
design will probably need to convert a 10-100 mhz sinewave to some L
band frequency. The output will need to be gated on with an externally
supplied signal that has programmable output amplitude, and will
support most 'positive-logic' families. The gate signal will be 50ns to
1mS. The IF input will be CW and linear-sweep type waveforms.
We are open to suggestions concerning off-the-shelf type
solutions, as well as custom design.
Also, we have been bitten in the past by 'out-of-house' designs
that just don't seem to work as advertised. Any suggestions? Is there
such a thing as a contract based upon a working prototype? Am I
dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

You could look at Miteq for upconverters. Not sure what you're on about with
the gate signal with programmable output amplitude (what's that doing to the
upconverter output?). Anyway Miteq have L-band stuff, have a hunt.

Ken

 

[John Woodgate]

I read in sci.electronics.design that snbaer@msn.com wrote (in
<1113088309.259466.157900@o13g2000cwo.googlegroups.com&gt about 'rf
design', on Sat, 9 Apr 2005:

Is there such a thing as a contract based upon a working prototype? Am
I dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

Any design work that I do contracts for (normally) EIGHT units:

One Engineering model, that may not be mechanically to drawings;

One prototype, that is to drawings apart from a list of agreed
deviations (e.g. moulded parts won't be available);

Six pre-production samples, ***made and tested by the client from parts
approved by me (so no purchasing the cheapest trash)***, so I can see
that they are competent to make and test, and to show that the design is
manufacturable.

Things CAN still go wrong after that, but then they can go wrong after
10 000 units have been built. For example, a manufacturer may 'improve'
an IC out of all recognition. (8-O(
--
Regards, John Woodgate, OOO - Own Opinions Only.
There are two sides to every question, except
'What is a Moebius strip?'
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

 

[Mac]

On Sat, 09 Apr 2005 16:11:49 -0700, snbaer wrote:

I am tasked with evaluating the feasablility of having an
upconverter designed. The requirements are not clear yet, but the
design will probably need to convert a 10-100 mhz sinewave to some L
band frequency. The output will need to be gated on with an externally
supplied signal that has programmable output amplitude, and will
support most 'positive-logic' families. The gate signal will be 50ns to
1mS. The IF input will be CW and linear-sweep type waveforms.
We are open to suggestions concerning off-the-shelf type
solutions, as well as custom design.
Also, we have been bitten in the past by 'out-of-house' designs
that just don't seem to work as advertised. Any suggestions? Is there
such a thing as a contract based upon a working prototype? Am I
dreaming, or is there a way to guarantee a contractor or consultant
delivers as promised?

This is a strange post.

Are you trying to figure out whether the technical spec's can be met or
whether there is a methodology for finding good consultants?

From what you've said, the technical spec's don't sound problematic. I am
aware of systems which very closely resemble what you describe. One point
I'm not too sure about is whether you want the consultant to design the
external gating signal, or just accept it. Either way, this part is easy,
and could be done with an off-the-shelf RF switch which just takes a TTL
input to control the switching.

--Mac

 

[Joerg]

Hello Steve,

(1) The electronic design would have to convert a programmable output
of 10 MHz to 100 MHz to L band frequencies ( 1 GHz to 2 GHz ). The
programmable output design is done - we do not have engineers that can
handle the up-conversion part of the design. Also, there is a need to
gate the L band output with about 1 ns resolution. The gate signal
design that will be supplied to the L band output section is done - it
is a digital signal with output levels programmable from 0 V to 5 V.

What a consultant would have to know are details such as:

- Precision, stability of output frequency?
- Desired sideband rejection?
- Output signal level (dBm)?
- Output range freely programmable? If yes, in what steps?
- Required gating spectral purity?
- EMC concerns (for example if 2.4GHz or 5GHz gear is nearby)?
- Production cost budget?

Up-conversion in itself isn't such a big deal but probably something has
to control the exact frequency range where you want to see your
10-100MHz signal converted to. The frequency stability of the oscillator
will to a large extent determine the effort. Also, unless you can
completely forego sideband rejection this scenario may require several
conversions or at least I-Q techniques.

There is a lot of other stuff that needs to be known but which would
have to be dealt with outside a public forum. A good consultant will
always ask if a spec isn't complete. At an early stage of a concept that
would be quite normal.

(2) The business and legal concerns are about ways to ensure that the
result of our business arrangement is working hardware.

Usually that's no problem but it might help if you talk this over with
someone who has done business with consultants before and whom you
trust. It can also be someone with such experience from another
technical discipline, like mechanical engineering.

Regards, Joerg

http://www.analogconsultants.com

 

[Mac]

On Mon, 11 Apr 2005 07:20:24 -0700, snbaer wrote:

Mac,
My post was not very clear. Let me try to clarify. The post was
intended to as questions about two areas: (1) electronic design, and
(2) business and legal arrangements. As I am new to this area I am
trying to get some advice on both.
(1) The electronic design would have to convert a programmable output
of 10 MHz to 100 MHz to L band frequencies ( 1 GHz to 2 GHz ). The
programmable output design is done - we do not have engineers that can
handle the up-conversion part of the design. Also, there is a need to
gate the L band output with about 1 ns resolution. The gate signal
design that will be supplied to the L band output section is done - it
is a digital signal with output levels programmable from 0 V to 5 V.
(2) The business and legal concerns are about ways to ensure that the
result of our business arrangement is working hardware.

Well, there are still many details unspecified, but if you don't need high
power, don't need ultra-high spectral purity or ultra-low phase noise, the
design shouldn't be hard for an experienced microwave guy (which I am NOT).

The gating switch part should be easy, unless you need it to be tiny or
really cheap, and you may opt to NOT have the contractor do the gating.
Instead, just put an RF switch after the contractor-designed mixing /
multiplying section. You can buy switches with TTL inputs. The switch
will have some additional turn-on time (e.g., 10 ns), but it should be
more or less constant, and can therefore be compensated for. (Or,
if I correctly infer your application, you can just ignore the
additional gating delay.)

I am sort of assuming that there is some linear relationship between the
input and output frequency.

I.e., Fout = m * Fin + Fbase
where Fout is the output frequency, Fin is the input frequency, and m and
Fbase are constants.

As for how to find a good contractor, I don't really know, but it seems
like Joerg's advice is pretty good. (Maybe Joerg could tackle the job?)

regards,
Mac