Determining SBL-1 Mixer LO Power level
ae...@mindspring.com
signal application and am having trouble getting +7dBm into the LO
port. The RF and IF ports are terminated with 50 Ohms.
I thought the LO impedence was 50 ohms but appears to be closer to 15
Ohms at 4MHz and also is very non linear. I found an app note with a
table showing a VSWR of around 3.5 at 1 MHz. The RF books I have
assume the LO port is 50 Ohms and just say set the voltage to 1.4Vp-p.
My 50 Ohm signal generator tops out at +18dBm and it still will not
drive the LO port to 1.4vp-p.
What am I missing?
Any thoughts on using the AD831 mixer and is it tempermental? Aside
from it's price and noise figure, it's specs look a lot better than
the diode mixers.
thanks,
Moe, ae4jy
John Miles
>
> Am trying to get a MiniCircuits SBL-1 mixer working for a high level
> signal application and am having trouble getting +7dBm into the LO
> port. The RF and IF ports are terminated with 50 Ohms.
> I thought the LO impedence was 50 ohms but appears to be closer to 15
> Ohms at 4MHz and also is very non linear. I found an app note with a
> table showing a VSWR of around 3.5 at 1 MHz. The RF books I have
> assume the LO port is 50 Ohms and just say set the voltage to 1.4Vp-p.
> My 50 Ohm signal generator tops out at +18dBm and it still will not
> drive the LO port to 1.4vp-p.
> What am I missing?
You're driving a pair of diodes, remember. You just want to get them
into a reasonably-saturated portion of their curves. The only thing
that matters is IMD -- if you can reach the manufacturer's IMD spec with
your particular LO, you don't need to lose sleep over exact impedances
and power levels.
If you ever see more than 1.4v p-p going into the mixer, it means you've
fried a diode or two.
-- jm
------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------
Richard Karlquist
<ae...@mindspring.com> wrote:
>Am trying to get a MiniCircuits SBL-1 mixer working for a high level
>signal application and am having trouble getting +7dBm into the LO
>port. The RF and IF ports are terminated with 50 Ohms.
>I thought the LO impedence was 50 ohms but appears to be closer to 15
>Ohms at 4MHz and also is very non linear. I found an app note with a
This topic is very confusing and annoyed me so much that I eventually
decided to investigate it and get to the bottom of it. Basically,
the "official" answer is that you need to drive the LO port from a
50 ohm source with available power of +7 dBm. You are correct that
the impedance is much lower than 50 ohms and non-linear. So don't
try to look at the LO voltage to figure out if you have enough drive.
It will simply saturate at about .6V P-P due to diode turn-on voltage.
You can get an idea of how much drive you have by measuring the
*current*. It turns out that at the "official" drive level, the
current will be 30 mA P-P (counting just the fundamental frequency,
and ignoring harmonics).
If you want to "civilize" the LO port, put a series resonant LC
tank in series with it, with the reactance of the L and C around
150 ohms (not critical as long as it resonates at the LO frequency).
Then the source will see a linear non-time varying impedance of
25 ohms or so. If you want, you can then transform this to 50 ohms.
The series tank will make both the source and the mixer "happier"
although things will work OK without it obviously.
Also, you don't lose much conversion efficiency even at +4 dBM LO
drive and the mixer will be usable with higher loss to around +1 dBm
so in all likelihood, you are OK right now. You can also overdrive
the mixer by 10 dB with no harm.
Rick Karlquist N6RK
ri...@area.com
>table showing a VSWR of around 3.5 at 1 MHz. The RF books I have
>assume the LO port is 50 Ohms and just say set the voltage to 1.4Vp-p.
>My 50 Ohm signal generator tops out at +18dBm and it still will not
>drive the LO port to 1.4vp-p.
>What am I missing?
>
Leon Heller
ae...@mindspring.com wrote:
> Am trying to get a MiniCircuits SBL-1 mixer working for a high level
> signal application and am having trouble getting +7dBm into the LO
> port. The RF and IF ports are terminated with 50 Ohms.
> I thought the LO impedence was 50 ohms but appears to be closer to 15
> Ohms at 4MHz and also is very non linear. I found an app note with a
> assume the LO port is 50 Ohms and just say set the voltage to 1.4Vp-p.
> My 50 Ohm signal generator tops out at +18dBm and it still will not
> drive the LO port to 1.4vp-p.
> What am I missing?
When I was doing this sort of thing some time ago, I simply checked the
voltage across a 51 ohm resistor with a scope, and assumed that this
would be correct for the mixer input.
Leon
--
Leon Heller, G1HSM
Tel: (Mobile) 079 9098 1221 (Work) +44 1327 357824
Email: leon_...@hotmail.com
Web: http://www.geocities.com/SiliconValley/Code/1835
Sent via Deja.com http://www.deja.com/
Before you buy.
Richard Karlquist
Dana Myers K6JQ <da...@source.net> wrote:
>
>
>While I can't point at a huge deal of experience to support my beliefs,
>I'm inclined to avoid reactances at any diode mixer port (especially the
>IF but the LO as well). My suggestion is to use a resistive attenuator at
>the LO port and increase the LO drive appropriately. In other words,
>if you want +7dBm of LO drive, and use a -6dB attenuator, the LO would
>be designed to produce +13dBm into a 50 ohm load.
>
>Dana
There is a distinction between the LO port and the RF/IF ports. The RF/IF
port terminations are an important issue and as you say the safest thing
to do is resistively terminate then. You wouldn't want to put a tank
circuit on one of them unless you were attempting some trick like
reflecting back the image to improve the conversion loss (which can be
made to work). OTOH, the source impedance used to drive the LO port
is an entirely different discussion. In this case, if you have good LO
isolation specs on the mixer, there will be no significant mixer products
at the LO port so you don't need to worry about how they are terminated.
In my experience, mixers work better if the LO source is a current source.
IMHO, there is no justification for padding down the LO input with a 6 db
attenuator unless there is a long cable between the source and the mixer
and the source match is poor.
Rick Karlquist N6RK
www.karlquist.com
Moe
questions as this project unfolds.
The Analog Devices AD831 still looks pretty good especially the 1 dB
compression level.
moe, ae4jy
Yvon
to keep the LO level to the lowest value because the 57 Mhz signal and its
harmonics were leaking into other parts of the design. I adjusted the source
to 7 DBM into a 50 ohm resistor and also measured the voltage across the 50
ohm resistor; then I connected the source to the mixer, but the voltage was
quite different. I wanted to experiment more but did not have time. The next
thing that I want to experiment when time is available is to put a series
resistor of let say 5 ohms in series with the input of the LO port and
connect a
balanced transformer across the 5 ohm resistor and read the voltage on a
spectrum analyzer.As the LO level increases, the voltage should start to
increase more rapidly when the diodes in the mixer are conducting. It should
also be possible to use a toroid core to make a current transformer with
only a one turn primary connected in series with the mixer. Overdriving such
a mixer at the LO port can raise problems if sensitive radio-frequency
circuits are close to the driving circuits which can produce very high level
of harmonics. Regards!
John Miles
That's fundamentally a problem with shielding, though, not LO drive.
Backing off on your LO drive is not a smart way to deal with leakage.
Richard Karlquist
Graeme Zimmer <gzi...@bigpond.com> wrote:
>I been wondering about this for ages.....
>
>If the diodes in a ring mixer are truly switching (and not just running in
>their square-law region)....... why don't we always drive a DBM with a
>square wave ?
>
>Or to put it another way, why is a sine wave important (eg why low harmonic
>content) if we must ensure sufficient LO level to make sure that the diodes
>are acting as switches ?
>
>Am following the thread with great interest
>
>direct replies would be appreciated.
>
>regards .............. Zim ............. VK3GJZ
>
>
This is a controversial topic. My opinion (I'm sure some will disagree)
is that you want to feed a sine wave of current into the diodes and
let them make a square wave of voltage out of it. The reason for this is
that you have to pass the LO wave through a transformer and transformers
are much better for handling sine waves than square waves. It is quite
possible that phase shifts will cause the square wave to turn into a
not so square wave by the time it goes through the transformer. This
cannot happen with a sine wave. Secondly, for optimum IM3, the LO
signal should contain no even order harmonics. For a sine wave all
this requires is that you low pass filter it. For a square wave, you
have to generate exactly 50% duty cycle, and the output impedance of
the LO source must be the same for a logic high and a logic low or you
must pad it down.
Also, the distortion in a mixer is caused by the nonlinear nonzero
on resistance of the diodes, not what happens during the instant they
switch. Thus there is nothing magic about shortening this time with a
square wave.
The above notwithstanding, I have sometimes used differential ECL outputs
to directly drive mixer LO's in noncritical applications.
Rick Karlquist N6RK
Graeme Zimmer
Jake Brodsky
wrote:
>In article <rJ5l4.2786$VJ1....@newsfeeds.bigpond.com>,
>Graeme Zimmer <gzi...@bigpond.com> wrote:
>>I been wondering about this for ages.....
>>
>>If the diodes in a ring mixer are truly switching (and not just running in
>>their square-law region)....... why don't we always drive a DBM with a
>>square wave ?
>>
>>Or to put it another way, why is a sine wave important (eg why low harmonic
>>content) if we must ensure sufficient LO level to make sure that the diodes
>>are acting as switches ?
>>
>
>This is a controversial topic. My opinion (I'm sure some will disagree)
>is that you want to feed a sine wave of current into the diodes and
>let them make a square wave of voltage out of it. The reason for this is
>that you have to pass the LO wave through a transformer and transformers
>are much better for handling sine waves than square waves. It is quite
>possible that phase shifts will cause the square wave to turn into a
>not so square wave by the time it goes through the transformer. This
>cannot happen with a sine wave. Secondly, for optimum IM3, the LO
>signal should contain no even order harmonics. For a sine wave all
>this requires is that you low pass filter it. For a square wave, you
>have to generate exactly 50% duty cycle, and the output impedance of
>the LO source must be the same for a logic high and a logic low or you
>must pad it down.
>
>Also, the distortion in a mixer is caused by the nonlinear nonzero
>on resistance of the diodes, not what happens during the instant they
>switch. Thus there is nothing magic about shortening this time with a
>square wave.
>
>The above notwithstanding, I have sometimes used differential ECL outputs
>to directly drive mixer LO's in noncritical applications.
One question, Rick:
Let's assume we are driving the mixer with the output from a sine wave
going through a limiter (perhaps this is from some sort of synchronous
detector scheme).
Even if the LO transformer saturates, aren't the diodes already biased
on? How would the LO transformer saturation make any difference?
Jake Brodsky, AB3A mailto:fru...@erols.com
"Beware of the massive impossible!"
Richard Karlquist
Jake Brodsky <fru...@erols.com> wrote:
>One question, Rick:
>
>Let's assume we are driving the mixer with the output from a sine wave
>going through a limiter (perhaps this is from some sort of synchronous
>detector scheme).
>
>Even if the LO transformer saturates, aren't the diodes already biased
>on? How would the LO transformer saturation make any difference?
When I was talking about the transformer not passing a square wave,
I was referring to phase shift problems (or you could describe it
as different group delay for different harmonics). This is a problem
even at low levels where the transformer is linear.
I can't imagine why you would want to saturate the transformer in
a mixer.
Rick Karlquist N6RK
Richard Karlquist
Lasse Moell <lasse...@swipnet.se> wrote:
>Hmm,
>I guess there more to this than one might thing at first sighting!
>The "Paramixer" made by Steinbrecher in the late 80s did use matched
>diodes for switching but used fet-drivers to get the square wave!
>
>This mixer acheived IP in the +50 dBm performing excellent!!
>
>Wish I had more information about this one.
>
>Anyone read the patent??
>
>/Lasse SM5GLC
Yes, I have read the Steinbrecher patent. He has a clever topology
that causes the transformer to see a constant linear resistive load
rather than the usual "funny" load of the diodes. In this case,
you actually have some hope of getting a square wave through the
transformer in decent shape, although you have to extend the
frequency response of the transformer considerably. This is a
great mixer but comparing it to an SBL-1 is apples vs oranges.
Rick Karlquist N6Rk
Lasse Moell
I guess there more to this than one might thing at first sighting!
The "Paramixer" made by Steinbrecher in the late 80s did use matched
diodes for switching but used fet-drivers to get the square wave!
This mixer acheived IP in the +50 dBm performing excellent!!
Wish I had more information about this one.
Anyone read the patent??
/Lasse SM5GLC
On 30 Jan 2000 17:55:30 -0800, ri...@vax.area.com (Richard Karlquist)
wrote:
>In article <rJ5l4.2786$VJ1....@newsfeeds.bigpond.com>,
>Graeme Zimmer <gzi...@bigpond.com> wrote:
>>I been wondering about this for ages.....
>>
>>If the diodes in a ring mixer are truly switching (and not just running in
>>their square-law region)....... why don't we always drive a DBM with a
>>square wave ?
>>
>>Or to put it another way, why is a sine wave important (eg why low harmonic
>>content) if we must ensure sufficient LO level to make sure that the diodes
>>are acting as switches ?
>>
>>Am following the thread with great interest
>>
>>direct replies would be appreciated.
>>
>>regards .............. Zim ............. VK3GJZ
>>
>>
>
>
>This is a controversial topic. My opinion (I'm sure some will disagree)
>is that you want to feed a sine wave of current into the diodes and
>let them make a square wave of voltage out of it. The reason for this is
>that you have to pass the LO wave through a transformer and transformers
>are much better for handling sine waves than square waves. It is quite
>possible that phase shifts will cause the square wave to turn into a
>not so square wave by the time it goes through the transformer. This
>cannot happen with a sine wave. Secondly, for optimum IM3, the LO
>signal should contain no even order harmonics. For a sine wave all
>this requires is that you low pass filter it. For a square wave, you
>have to generate exactly 50% duty cycle, and the output impedance of
>the LO source must be the same for a logic high and a logic low or you
>must pad it down.
>
>Also, the distortion in a mixer is caused by the nonlinear nonzero
>on resistance of the diodes, not what happens during the instant they
>switch. Thus there is nothing magic about shortening this time with a
>square wave.
>
>The above notwithstanding, I have sometimes used differential ECL outputs
>to directly drive mixer LO's in noncritical applications.
>
>Rick Karlquist N6RK
Jake Brodsky
wrote:
>When I was talking about the transformer not passing a square wave,
>I was referring to phase shift problems (or you could describe it
>as different group delay for different harmonics). This is a problem
>even at low levels where the transformer is linear.
>
>I can't imagine why you would want to saturate the transformer in
>a mixer.
I wouldn't *want* to saturate the transformer, Rick. I'm merely
pointing out that once it does saturate, wouldn't the diodes already
be switched in to conduction?
Let me restate the question: Would the group delay effects of a square
wave signal propagating through a transformer actually cause a diode
to drop out of conduction when it is actually expected to stay on?
Conversely, would group delay effects through the transformer actually
cause the diodes to stay in conduction when they're supposed to be
switched off?
I can envision a situation where the hysteresis effects of a
transformer might cause a phase shift of the LO once it gets to the
mixer diodes. And I can envision that using a square wave on the LO
port might result in a different phase shift. But I can't see how it
would necessarily be a "bad" thing if you consistently drove the input
with a square wave.
Tom Bruhns
with a square wave current source. It seems to me that there is no
danger of saturating any cores that way, any more than if you drive it
with the same current level some other way.
Seems like it would be an easy enough test to do. You'd want to know
what things you wanted to characterize: IP3, P1db, etc. Surely someone
out there has done tests like this before. Is there anything definitive
in Steven Maas's book on DBDM's? (I think that's the guy who has
written books and magazine articles on the subject.)
I'd love to be able to ask my uncle his thoughts on it; he got a patent
for at least one version of a diode DBM. Unfortunately, he's beyond
being able to discuss things like this.
Cheers,
Tom
John Miles
>
> I'm aware that there are those who would suggest you drive the LO port
> with a square wave current source. It seems to me that there is no
> danger of saturating any cores that way, any more than if you drive it
> with the same current level some other way.
>
> Seems like it would be an easy enough test to do. You'd want to know
> what things you wanted to characterize: IP3, P1db, etc. Surely someone
> out there has done tests like this before. Is there anything definitive
> in Steven Maas's book on DBDM's? (I think that's the guy who has
> written books and magazine articles on the subject.)
For what it's worth, the FAQ in the Mini-Circuits designer's guide
claims it's perfectly kosher to drive their DBMs with a square wave.
Don't have the exact quote handy, but it's one of their FAQs.
lar...@northwest.com
commutation mixing.
===================
Richard Karlquist
Jake Brodsky <fru...@erols.com> wrote:
>On 31 Jan 2000 19:04:14 -0800, ri...@vax.area.com (Richard Karlquist)
>wrote:
>
>wave signal propagating through a transformer actually cause a diode
>to drop out of conduction when it is actually expected to stay on?
>Conversely, would group delay effects through the transformer actually
>cause the diodes to stay in conduction when they're supposed to be
>switched off?
>
>transformer might cause a phase shift of the LO once it gets to the
The distortion of a square wave by a transformer can occur without
any nonlinear core effects such as hysteresis or distortion. If
you pass a square wave through any network, such as a transformer,
than does not have flat frequency response and linear phase, you
will get a distorted square wave out. It may have overshoot, it
may have rounded off corners, and/or it may have droop. Thus the
waveform that is presented to the diodes will not be a perfect
square wave and it will not turn the diodes on the same as it would
if it were a perfect square wave. Suppose the transformer winding
capacitance causes a funny suckout at some harmonic of the square
wave. If you delete that harmonic, the wave will acquire a bunch
of wiggles. If you had used a sine wave, you wouldn't have to
worry about a suckout at a harmonic.
I hope that answers the question, and again, I never said you can't
use a square wave, it just isn't generally the optimum thing to do.
Rick Karlquist N6RK
Richard Karlquist
Tom Bruhns <tom_b...@agilent.com> wrote:
>I'm aware that there are those who would suggest you drive the LO port
>with a square wave current source. It seems to me that there is no
>danger of saturating any cores that way, any more than if you drive it
>with the same current level some other way.
The reasons for not using a square wave have nothing to do with
core saturation.
>Seems like it would be an easy enough test to do. You'd want to know
>what things you wanted to characterize: IP3, P1db, etc. Surely someone
>out there has done tests like this before. Is there anything definitive
>in Steven Maas's book on DBDM's? (I think that's the guy who has
>written books and magazine articles on the subject.)
I don't remember seeing anything about that in his book. It is
excellent for what it covers, but it doesn't cover everything.
I did some characterization of a W-J M9E mixer in conjunction with my
1996 FCS paper, and I found that it was critical to heavily low pass
filter the LO drive to the mixer in order to get the best IMD.
So that is some kind of a data point.
If you propose to use a square wave, tell me how we generate this
square wave. How do we get exactly 50% duty cycle? I can get
arbitrarily close to 50% duty cycle if I use a sine wave by just
filtering the dickens out of it.
Rick Karlquist N6RK
Richard Karlquist
John Miles <jmi...@pop.removethistomailme.net> wrote:
>
>For what it's worth, the FAQ in the Mini-Circuits designer's guide
>claims it's perfectly kosher to drive their DBMs with a square wave.
>Don't have the exact quote handy, but it's one of their FAQs.
Well, W-J doesn't recommend this. I would tend to believe them.
Mini-Circuits' talent lies in low cost manufacturing, not mixer
theory.
Rick Karlquist N6Rk
William Janssen
as I remember it, a saturating core causes a pulsed output.
Transformers only have an output when the flux in the core is
changing. When the core saturates the output drops to zero.
Bill K7NOM
Jake Brodsky wrote:
>
> On 31 Jan 2000 19:04:14 -0800, ri...@vax.area.com (Richard Karlquist)
> wrote:
>
> >When I was talking about the transformer not passing a square wave,
> >I was referring to phase shift problems (or you could describe it
> >as different group delay for different harmonics). This is a problem
> >even at low levels where the transformer is linear.
> >
> >I can't imagine why you would want to saturate the transformer in
> >a mixer.
>
> I wouldn't *want* to saturate the transformer, Rick. I'm merely
> pointing out that once it does saturate, wouldn't the diodes already
> be switched in to conduction?
>
> Let me restate the question: Would the group delay effects of a square
> wave signal propagating through a transformer actually cause a diode
> to drop out of conduction when it is actually expected to stay on?
> Conversely, would group delay effects through the transformer actually
> cause the diodes to stay in conduction when they're supposed to be
> switched off?
>
> I can envision a situation where the hysteresis effects of a
> transformer might cause a phase shift of the LO once it gets to the
> mixer diodes. And I can envision that using a square wave on the LO
> port might result in a different phase shift. But I can't see how it
> would necessarily be a "bad" thing if you consistently drove the input
> with a square wave.
>