EMRFD Message Archive 8604

Message Date From Subject
8604 2013-05-01 02:43:17 101 How Much L.O. Power Is Needed?
Hi,

How do we know how much L.O. power is needed for any mixer to convert
OPTIMALLY i.e. without prior referring to any generic type of mixer?

Regards.
8605 2013-05-01 16:40:02 kb1gmx Re: How Much L.O. Power Is Needed?
8606 2013-05-01 16:52:26 101 Re: How Much L.O. Power Is Needed?
Dear Alison,

1. What data do we need?
2. Can't we get it by in-situ measurement?

Regards.

> >
>
> I'd love to answer the question but the extreme lack of data makes it
impossible.
>
> It's critical to know the mixer used. Every mixer topography has its
own needs and there is no one pat answer.
>
> Allison
>
8608 2013-05-01 21:43:18 Todd F. Carney / ... Re: How Much L.O. Power Is Needed?
8609 2013-05-01 23:08:35 101 Re: How Much L.O. Power Is Needed?
Hi Todd,

Thanks for your advice.

But, I cannot get any datasheet because it is homebrew projects. I've
built 2 set of mixers, one is a H-Mode Mixer as they use in CDG2000
while the other is a H-Mode Mixer but using diodes (Schottky) as
proposed by Christ Trask. Both are followed by diplexers.

For the LO, I plan to use Si570.

For the RF, it is gone through a homebrew very narrow bandwidth
preselector.

Now, it should be the final step of front-end hook up which I still
wonder how much the LO power is appropriate to optimize the whole
system.

Regards,
Amin


8610 2013-05-02 04:04:58 Eduardo Alonso Re: How Much L.O. Power Is Needed?
Hello Amin,

Check the figure 16 (Conversion loss measured at various LO power
levels) in this document:

http://arxiv.org/pdf/physics/0608211v1.pdf

and some results in this webpage

http://users.tpg.com.au/nfieraru/Electronics/DoubleBalancedMixers.htm
(discussed here http://groups.yahoo.com/group/emrfd/message/7989 )

73 eduardo
8611 2013-05-02 04:48:34 Ashhar Farhan Re: How Much L.O. Power Is Needed?
Power requirement to drive the h mode mixer using mosfets is,
techincally speaking, zero as the gates draw no current. However, you
must get enough level to drive the gates from on to off,
instantaneously. Which means you need a sqaure wave, something that
the cmos version of the Si570 will provide, though only upto 3V level.
You may need a very fast level shifter (or just make one up with a
very fast switching transistor.
- farhan

8612 2013-05-02 05:11:46 Ron McCurdy Re: How Much L.O. Power Is Needed?
I believe I saw that the LO should be 7dB. That would be a good starting
place.



Ron

KE5QDA



_____

8613 2013-05-02 05:39:38 William Carver Re: How Much L.O. Power Is Needed?
Now with some specific information on your mixer it's possible to
comment on your original question.

H-mode is a switching (or "commutation") mixer using three terminal
(MOSFET) switches. Ed Oxner, in Siliconix application note AN85-2
(October/January 1986) described the conversion loss, intercept and
disortion characteristics of such a mixer as a function of switch
resistance and speed. I would suggest trying to find a copy of this
application note.

It is referenced as prior art by everyone who patents THEIR mixer
design, but I can't find the application note itself. My copy is on FAX
paper and is so old and faded you can barely see the text. Does anyone
have an original, or a .PDF of AN85-2 that they can post here?

The Si570 has a square wave digital output with rise/fall times of about
1nS. The CMOS bus switches have comparable speed. For Si570 driving bus
switches the issue of LO amplitude in Oxner's application note is
eliminated: you are handed very fast LO switching on a silver platter.
TO say it a different way, LO "power" is not an issue, you just connect
the square wave LO VOLTAGE to the bus switch inputs.

Have you looked at PA3AKE's website? It has EXTENSIVE measured data on
H-mode mixers with various CMOS bus switches. Note the majority are like
the FST3125 used in the CDG2000 have "TTL" gate threshold (about 1.5
volts), FSA3157 threshold is one-half of Vcc. Also note that Martein's
tests used a sinusoidal LO with a variable DC bias so he can change the
square wave from perfect 50/50 to optimize third order intercept. Si570
is fixed at 50/50 so if its output directly drives the switches you
cannot "tweek" the mixer for optimum intercept.

Using diodes instead of switches is a different matter. The diodes are
switched by LO current flowing in them. Chris Trask can best answer your
question for his diode mixer.

W7AAZ
8614 2013-05-02 05:50:56 William Carver Re: How Much L.O. Power Is Needed?
On Thu, 2013-05-02 at 17:18 +0530, Ashhar Farhan wrote:
>
> Power requirement to drive the h mode mixer using mosfets is,
> techincally speaking, zero as the gates draw no current. However, you
> must get enough level to drive the gates from on to off,
> instantaneously. Which means you need a sqaure wave, something that
> the cmos version of the Si570 will provide, though only upto 3V level.
> You may need a very fast level shifter (or just make one up with a
> very fast switching transistor.
> - farhan

The CMOS switches gate input, like most logic circuits, switches from on
to off with a very small change in gate voltage. The FSA3157 switches at
0.5 Vcc. The data sheets says this could be as low as 0.3 Vcc or as high
as 0.7 Vcc. I found this is "boiler plate" specification, I tested a
handful and found them to be withing 1% of Vcc/2.

The 3V "swing" of the Si570 is more than enough to do the switching. You
can "center" that 3V switch at Vcc/2 simply by capacitor coupling the
square wave from Si570 to the gate input, and bias the gate input at its
threshold (about 1.5 volts for FST3125, or 0.5 Vcc for FSA3157). If you
like potentiometers to play with, make that bias adjustable. The gate
draws no current so go from resistive divider/pot to the gate through a
high value resistor...say 100K, 10K, etc. Couple the Si570 square wave
through a capacitor...probably .001 uF is enough, but 0.01 or even 0.1
uF is definitely enough. IT DOESN'T MATTER.

Since the Si570 has a rise/fall time of about 1nS, and the bus switches
are similarly fast, a potentiometer doesn't hardly seem necessary, but
neither would it hurt anything.

An active level shifter needs to preserve, or even improve, the
rise/fall time of the Si570. This isn't impossible, but it isn't very
likely. Much more likely it would degrade the rise/fall times. So
personally I would avoid an active level shifter.

W7AAZ



>
8616 2013-05-02 07:11:35 Todd F. Carney / ... Re: How Much L.O. Power Is Needed?
8621 2013-05-02 20:53:47 101 Re: How Much L.O. Power Is Needed?
Hi Everybody!
The answers & arguments to this very questi
8624 2013-05-04 16:14:20 kb1gmx Re: How Much L.O. Power Is Needed?
8625 2013-05-04 16:23:19 kb1gmx Re: How Much L.O. Power Is Needed?
8626 2013-05-04 18:05:00 William Carver Re: How Much L.O. Power Is Needed?
The Si570 output voltage swing is more than enough to drive bus switches
in an H-mode mixer.

The only precaution is to center the peak-to-peak Si570 square wave
around the switching threshold of the bus switch. That threshold could
be "TTL" or Vcc/2 depending on what switch is being used.

W7AAZ

On Sat, 2013-05-04 at 23:23 +0000, kb1gmx wrote:
>
>
>
>
8627 2013-05-05 05:27:38 kb1gmx Re: How Much L.O. Power Is Needed?
8628 2013-05-05 08:36:54 William Carver Re: How Much L.O. Power Is Needed?
I agree Allison, in logic design you don't operate parts at different
supply voltages. But this is NOT logic or computer design, this is a
mixer. No "maybe": the critical issue in the switching mixer is that the
switch duty cycle is 50/50.

Since the rise and fall times of the logic LO may not be identical you
can improve the intercept somewhat by being adjustable around the
threshold voltage of the bus switch. Vcc/2 in the case of FSA3157, or
"TTL" (about 1.4 volts) in the case of FST3125, FSAV450, FSAV332, and
every other bus switch whose data sheet I've read.

I tested several FSA3157 switches from two different purchases and found
that the threshold was within 1% of 0.5*Vcc, despite the data sheet
specification of 0.3*Vcc to 0.7*Vcc.

W7AAZ




8629 2013-05-05 08:37:24 101 Re: How Much L.O. Power Is Needed?
Dear Allison,

I think this is the directi
8642 2013-05-07 16:17:50 kb1gmx Re: How Much L.O. Power Is Needed?