EMRFD Message Archive 5088
Message Date From Subject 5088 2010-09-12 10:36:56 ehydra mixers: strong signal immunity
strong signal immunity. What is the theory behind?
What are the other behaviour changes if the drive power level changes?
It is clear that a not saturated sinuoid waveform will reduce harmonics
down mixing.
Whats with noise figure?
- Henry
--
ehydra.dyndns.info5090 2010-09-12 11:00:17 Leon Re: mixers: strong signal immunity 5091 2010-09-12 11:24:52 victorkoren Re: mixers: strong signal immunity
High interfering signals at the RF port can affect this switching in this way:
When the LO signal is crossing the 0 voltage value, the LO current through the diodes is low and if the interfering signal current is then comparable in magnitude to the LO current, it modulates the exact point of zero crossing of the total current through the diodes. By doing so the interfering signal modulates all the signals that are converted by the mixer and causes spurious signals generation.
If the LO signal is larger, than its zero crossing time is much shorter and the interfering signal causes lower modulation of the zero crossing - higher mixer linearity.
Mathematically, such an operation of the mixer is like multiplying the input signal by a square wave that has just two values +1 and -1. Converting this signal to frequency domain gives the fundamental LO frequency and all its odd harmonics (with lower amplitude as harmonic order increases).
If the LO amplitude was much lower so that it only modulates the dynamic resistance of the diodes without switching them fully on, the mixer operation will be closer to taking two sine waves and mathematically multiplying them, giving mostly output signal products at Fin +/- Flo.
In such a mixer operation, the diodes will show significant series resistance to the signal, causing higher conversi5092 2010-09-12 12:14:50 Chris Trask Re: mixers: strong signal immunity
> We had a discussion about why mixers in saturated mode offer better
> strong signal immunity. What is the theory behind?
>
> What are the other behaviour changes if the drive power level changes?
> It is clear that a not saturated sinuoid waveform will reduce harmonics
> down mixing.
> Whats with noise figure?
>
discussion, I dragged out my box of notes and prototypes to reëxamine the
further improvement of my feedback mixers. I had more than once proposed
using a sinewave LO rather than the "traditional" squarewave so as to avoid
the 3dB NF penalty incurred by way of spectral folding, meaning the
conversion of higher frequency noise spectra to the IF by way of the
harmonic content of the squarewave LO.
The basis of my feedback mixers is to linearize the RF current passing
through the commutative switching quad. My approaches to this have been
useful, but there has been a limit as to how much IMD correction can be
done. In recent weeks, I've built no fewer than four prototypes of various
flavours and can now state the following:
1. There is a limit as to how much IMD improvement can be enjoyed by way
of linearizing the RF signal current. At some point, the switching
characteistics of the commutating quad become the limitation of IMD
performance. I had not reached this limitation until just this past week.
2. As many designers have found ever since the early 1950s, the symmetry
of the squarewave LO as well as the rise and fall times weigh heavily on the
IMD performance of the commutating devices, whether they be a diode/FET ring
or a commutative bipolr/FET quad. The manufacturer Marki Microwave is
seeing a 10dB improvement in OIP3 when they reduce the rise and fall times
down to the low picosecond range.
3. For now, a fully linearized analogue multiplying mixer is not in the
offing I was able to get OIP3 up to +32dBm, but it was tricky to adjust and
therefore not at all suitable.
4. I'm presently getting about +30dBm with a lossless feedback mixer
using a form of transformer-coupled negative feedback somewhat similar to
Norton. The design is not sensitive, and I think I can reduce the NF by way
of an interesting 1/f noise feedback method.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5100 2010-09-14 08:00:19 ehydra Re: mixers: strong signal immunity
Thank you for the nice explanation! Interesting aspect that reverse
mixing effect.
I want add two things:
1. The four active elements in the mixer will work more synchronuous if
saturated as switch. Their individual tolerances are then at the minimum
and this optimizes the output spectrum.
2. I can imagine that a driving LO waveform avoiding 3. and/or 5.
harmonic is maybe of benefit (sometimes referred as magic sinewave or
harmonic cancellation). But I never seen that for rf mixers. Here is a
link for stereo demods:
http://www.ham-radio.com/k6sti/ibocharm.htm
Of course, that would need very fast logic.
regards -
Henry
victorkoren schrieb:
> I we take for example a diode double balanced mixer, the LO signal
> causes the diodes in the mixer to connect the input signal at the RF
> port to the IF port but reversing the polarity of the connection
> according to the LO signal polarity. This is the mixing action of the
> mixer. High interfering signals at the RF port can affect this
> switching in this way: When the LO signal is crossing the 0 voltage
> value, the LO current through the diodes is low and if the
> interfering signal current is then comparable in magnitude to the LO
> current, it modulates the exact point of zero crossing of the total
> current through the diodes. By doing so the interfering signal
> modulates all the signals that are converted by the mixer and causes
> spurious signals generation. If the LO signal is larger, than its
> zero crossing time is much shorter and the interfering signal causes
> lower modulation of the zero crossing - higher mixer linearity.
> Mathematically, such an operation of the mixer is like multiplying
> the input signal by a square wave that has just two values +1 and -1.
> Converting this signal to frequency domain gives the fundamental LO
> frequency and all its odd harmonics (with lower amplitude as harmonic
> order increases). If the LO amplitude was much lower so that it only
> modulates the dynamic resistance of the diodes without switching them
> fully on, the mixer operation will be closer to taking two sine waves
> and mathematically multiplying them, giving mostly output signal
> products at Fin +/- Flo. In such a mixer operation, the diodes will
> show significant series resistance to the signal, causing higher
> conversion loss and its equivalent higher noise figure. Other kind of
> mixers, like Gilbert cell mixers (NE602) can be explained in similar
> ways with similar results. Victor - 4Z4ME5101 2010-09-14 08:08:05 ehydra Re: mixers: strong signal immunity
Hm. So if the parts counts should be low and the band around the carrier
is almost free of foreign energy (No need for big IMD), the answer is:
Use a sinus LO, do a mathematically multiplication! ??
>Seems that you on the road to a new better Trask mixer? If it is a low
> 4. I'm presently getting about +30dBm with a lossless feedback
> mixer using a form of transformer-coupled negative feedback somewhat
> similar to Norton. The design is not sensitive, and I think I can
> reduce the NF by way of an interesting 1/f noise feedback method.
part-count and simple structure I would try it out.
regards -
Henry
--
ehydra.dyndns.info5102 2010-09-14 08:51:40 Chris Trask Re: mixers: strong signal immunity
>Hm. So if the parts counts should be low and the band around the carrier
>is almost free of foreign energy (No need for big IMD), the answer is:
>Use a sinus LO, do a mathematically multiplication! ??
>
>
> >
> > 4. I'm presently getting about +30dBm with a lossless feedback
> > mixer using a form of transformer-coupled negative feedback somewhat
> > similar to Norton. The design is not sensitive, and I think I can
> > reduce the NF by way of an interesting 1/f noise feedback method.
>
> Seems that you on the road to a new better Trask mixer? If it is a low
> part-count and simple structure I would try it out.
>
Other interesting schemes that I've looked at over the years include using a pair of 4-way power dividers and having four class I (level 7) DRM mixers such as the SBL1-1 in parallel. That mixer has a +14dBm OIP3, but by having four in parallel you end up with +32dBm OIP3. However, you then require a +13dBm LO power and again there's no real advantage over a class III DRM mixer which uses the same LO power level, and it's far less expensive.
It seems to me that the goal should be to get +40dBm OIP3 or better, but with less than 1W of LO power. I'm not going to get there with any form of linearized active mixer as I now understand that the switching characteristics of the commutating quad are the limitation, and not the linearization of the RF current. It does help, but it appears that I probably will not reach the +40dBm OIP3 goal this way, at least not with readily available parts, and my oven doesn't get hot enough to use it as a silicon chip foundary.
I've set the linearized active mixer stuff aside for now. That box of prototypes and notes is pretty full after 10+ years, and I haven't achieved much beyond the original work. But, there's always that faint light at the far end of the tunnel. I'm now looking at an entirely different approach that uses a pair of SBL1-1 DRMs.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5104 2010-09-14 10:12:59 victor Re: mixers: strong signal immunity
What about this Idea: invert the mixer circuit upside down (using pnp transistors for the V to I converter) and use logic switches (FST3125 ,FST3253 etc) driven by logic level LO signal for the balanced mixing.
By using the mixer upside down the voltages around the switches will be near ground level and so they might perform well even though DC current is flowing through them.
This way you might get clean fast switching at the mixing stage and still use your lossless negative feedback for the RF signal.
Victor - 4Z4ME
5107 2010-09-14 12:49:24 ehydra Re: mixers: strong signal immunity
> It seems to me that the goal should be to get +40dBm OIP3 or better,As you wrote:
> but with less than 1W of LO power. I'm not going to get there with
> any form of linearized active mixer as I now understand that the
> switching characteristics of the commutating quad are the limitation,
> and not the linearization of the RF current. It does help, but it
> appears that I probably will not reach the +40dBm OIP3 goal this way,
> at least not with readily available parts, and my oven doesn't get
> hot enough to use it as a silicon chip foundary.
>
2. As many designers have found ever since the early 1950s, the
symmetry of the squarewave LO as well as the rise and fall times weigh
heavily on the IMD performance of the commutating devices, whether they
be a diode/FET ring or a commutative bipolr/FET quad. The manufacturer
Marki Microwave is seeing a 10dB improvement in OIP3 when they reduce
the rise and fall times down to the low picosecond range.
ECL. micrel has a whole line of very high-speed chips.
- Henry
--
ehydra.dyndns.info5108 2010-09-14 14:47:12 Chris Trask Re: mixers: strong signal immunity
> What about this Idea: invert the mixer circuit upside down (using pnp
> for the V to I converter) and use logic switches (FST3125 ,FST3253 etc)driven by
> logic level LO signal for the balanced mixing.near
> By using the mixer upside down the voltages around the switches will be
> ground level and so they might perform well even though DC current isflowing
> through them.use your
> This way you might get clean fast switching at the mixing stage and still
> lossless negative feedback for the RF signal.I've already done something similar to this, using NPN transistor arrays
> Victor - 4Z4ME
>
such as the CA3046 and CA3127. Four transistors are used for the
commutating quad and the fifth one is used as a diode for the biasing
reference, so that the emitters are all connected to DC ground. The circuit
is then reduced to three resistors, six transformers, a few capacitors, and
the IC transistor array. These work fairly well, but there is still the
+20dBm OIP3 threshold that I cannot cross with these. The NF is really good
since there is no driver transistor.
In this latest effort, I went back to the six transistor topology and
instead of using the Norton lossless feedback I instead used
collector-emitter transformer feedback, which provides a far greater
open-loop gain for linearizing the RF current. The OIP3 of the recovered RF
is approaching +40dBm this way, but the IF OIP3 is still in the high 20s and
can be coaxed to a little more than +30, but it's touchy. This was the
first time that I've seen the RF linearity appreciably higher than the IF
linearity, and it showed me where the next frontier is in these things.
It's not the switching speed of the LO signal itself that's the
limitation, but the storage time of the commutative quad devices. They turn
ON far faster than they turn OFF. When I use NEC monolithic dual devices I
get better IF OIP3, but that NEC product line is not stable. I'm better off
sticking to earlier monolithic transistor array devices such as the CA3046,
CA3054, CA3102, and CA3127.
The speed bump in linearizing the IF is that there is no distortion-free
signal to work with, and the mechanisms that create the distortion have to
be dealt with directly. But, there's an aspect to these and in fact all
forms of mixers that has me thinking of an entirely new approach.
Chris5109 2010-09-14 14:56:13 Chris Trask Re: mixers: strong signal immunity
> > It seems to me that the goal should be to get +40dBm OIP3 or better,
> > but with less than 1W of LO power. I'm not going to get there with
> > any form of linearized active mixer as I now understand that the
> > switching characteristics of the commutating quad are the limitation,
> > and not the linearization of the RF current. It does help, but it
> > appears that I probably will not reach the +40dBm OIP3 goal this way,
> > at least not with readily available parts, and my oven doesn't get
> > hot enough to use it as a silicon chip foundary.
> >
>
> As you wrote:
> 2. As many designers have found ever since the early 1950s, the
> symmetry of the squarewave LO as well as the rise and fall times weigh
> heavily on the IMD performance of the commutating devices, whether they
> be a diode/FET ring or a commutative bipolr/FET quad. The manufacturer
> Marki Microwave is seeing a 10dB improvement in OIP3 when they reduce
> the rise and fall times down to the low picosecond range.
>
> ECL. micrel has a whole line of very high-speed chips.
>
storage time of the commutative quad devices. That appears to be where the
problem is.
Chris5110 2010-09-14 15:02:10 victor Re: mixers: strong signal immunity
Anyway, don't you think that the fast logic transiti5111 2010-09-14 15:28:49 Lasse Re: mixers: strong signal immunity
Sorry if my comments are out of order.
Using the Steinbrecher mixer at HF we were pretty spoiled, it got it's
spectacular performance by using special drivers for the diode ring (if
I recall correct)
The new Peregrine MOSFET mixer (PE4140) looks very interesting with a
+36 dBm IP3
and very broadband too.
Recall Ulrich Rohde wrote about feedback amplifiers in mixer circuits in
HamRadio years ago. Think this wen't into production
in the R/S aircraft radio family
/Lasse SM5GLC
Chris Trask skrev 2010-09-14 23:56:
>>> It seems to me that the goal should be to get +40dBm OIP3 or better,
>>> but with less than 1W of LO power. I'm not going to get there with
>>> any form of linearized active mixer as I now understand that the
>>> switching characteristics of the commutating quad are the limitation,
>>> and not the linearization of the RF current. It does help, but it
>>> appears that I probably will not reach the +40dBm OIP3 goal this way,
>>> at least not with readily available parts, and my oven doesn't get
>>> hot enough to use it as a silicon chip foundary.
>>>
>> As you wrote:
>> 2. As many designers have found ever since the early 1950s, the
>> symmetry of the squarewave LO as well as the rise and fall times weigh
>> heavily on the IMD performance of the commutating devices, whether they
>> be a diode/FET ring or a commutative bipolr/FET quad. The manufacturer
>> Marki Microwave is seeing a 10dB improvement in OIP3 when they reduce
>> the rise and fall times down to the low picosecond range.
>>
>> ECL. micrel has a whole line of very high-speed chips.
>>
> Yes, but no degree of LO switching speed will overcome the charge
> storage time of the commutative quad devices. That appears to be where the
> problem is.
>
> Chris5112 2010-09-14 16:38:22 ehydra Re: mixers: strong signal immunity
>> ECL. micrel has a whole line of very high-speed chips.OK, you mentioned it in your other post. If faster than bipolar ist
>>
>
> Yes, but no degree of LO switching speed will overcome the charge
> storage time of the commutative quad devices. That appears to be where the
> problem is.
>
required, why not a small FETs like 2N7000 oder JFET U309 etc?
It was reported that they can switch easely in 1ns.
But I think you just alone as an mixer expert. Most use DRM, SA602 or
the newly CMOS switches.
Just for the records:
Potato Semi is interesting for very fast digital logic (500MHz).
- Henry
--
ehydra.dyndns.info5113 2010-09-14 16:51:14 ehydra Re: mixers: strong signal immunity
> Just for the records:Here are Bus Switches:
> Potato Semi is interesting for very fast digital logic (500MHz).
>
http://www.potatosemi.com/potatosemiweb/busswitch.html
You can order directly or using ebay. I know of friend using parts of
the manufacturer. So it is not vaporware.
- Henry
--
ehydra.dyndns.info5114 2010-09-14 17:08:25 Chris Trask Re: mixers: strong signal immunity
> >> ECL. micrel has a whole line of very high-speed chips.
> >>
> >
> > Yes, but no degree of LO switching speed will overcome the charge
> > storage time of the commutative quad devices. That appears to be where
> > problem is.If you went that route, you'd have to look at monolithic quad arrays
> >
>
> OK, you mentioned it in your other post. If faster than bipolar ist
> required, why not a small FETs like 2N7000 oder JFET U309 etc?
> It was reported that they can switch easely in 1ns.
>
like those made by Temex (nee Siliconix) so as to have four matched devices.
And then you need a much higher switching voltage, which defeats the
purpose.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5116 2010-09-14 19:28:35 Russell Shaw Re: mixers: strong signal immunity
>Using single-port devices for mixing (diodes), there's always the problem
>> Hm. So if the parts counts should be low and the band around the carrier is
>> almost free of foreign energy (No need for big IMD), the answer is: Use a
>> sinus LO, do a mathematically multiplication! ??
>>
>>
>>> 4. I'm presently getting about +30dBm with a lossless feedback mixer
>>> using a form of transformer-coupled negative feedback somewhat similar to
>>> Norton. The design is not sensitive, and I think I can reduce the NF by
>>> way of an interesting 1/f noise feedback method.
>> Seems that you on the road to a new better Trask mixer? If it is a low
>> part-count and simple structure I would try it out.
>>
>
> What I've been working on lately is not necessarily complex, but the parts
> count is such that I wonder if it's worthwhile as I can get +30dBm OIP3 with
> a class III (level 13) diode ring mixer (DRM) more easily.
>
> Other interesting schemes that I've looked at over the years include using a
> pair of 4-way power dividers and having four class I (level 7) DRM mixers
> such as the SBL1-1 in parallel. That mixer has a +14dBm OIP3, but by having
> four in parallel you end up with +32dBm OIP3. However, you then require a
> +13dBm LO power and again there's no real advantage over a class III DRM
> mixer which uses the same LO power level, and it's far less expensive.
>
> It seems to me that the goal should be to get +40dBm OIP3 or better, but with
> less than 1W of LO power. I'm not going to get there with any form of
> linearized active mixer as I now understand that the switching
> characteristics of the commutating quad are the limitation, and not the
> linearization of the RF current. It does help, but it appears that I
> probably will not reach the +40dBm OIP3 goal this way, at least not with
> readily available parts, and my oven doesn't get hot enough to use it as a
> silicon chip foundary.
>
> I've set the linearized active mixer stuff aside for now. That box of
> prototypes and notes is pretty full after 10+ years, and I haven't achieved
> much beyond the original work. But, there's always that faint light at the
> far end of the tunnel. I'm now looking at an entirely different approach
> that uses a pair of SBL1-1 DRMs.
that even if the diodes are ideal, the switching times are modulated
proportionally to the ratio of I_sig/I_lo, when I_lo is a sinewave.
A square wave LO reduces the problem as the rise/fall time is made
shorter.
Ideally, the diodes should instead be parametric devices such as an
LDR (light-dependant resistor), where the resistance is independent
of voltage/current imposed on it, and controlled only be light
incident upon it. A hall-effect or magnetic saturation device
is another option. Unfortunately, there are no current devices
fast enough i'm aware of that can be driven at low power levels.5117 2010-09-15 17:29:58 ehydra Re: mixers: strong signal immunity
eally, the diodes should instead be parametric devices such as an
> LDR (light-dependant resistor), where the resistance is independentHm. Whats about an hall plate? It is multiplicative and maybe one of the
> of voltage/current imposed on it, and controlled only be light
> incident upon it. A hall-effect or magnetic saturation device
> is another option. Unfortunately, there are no current devices
> fast enough i'm aware of that can be driven at low power levels.
>
multipliers (1. magnetic and 2. current) fast enough?
- Henry
--
ehydra.dyndns.info5120 2010-09-16 05:07:53 hanssummers2000 Re: mixers: strong signal immunity
What about the H-mode mixer? PA3AKE has a good website5121 2010-09-16 05:38:51 ehydra Re: mixers: strong signal immunity
never completed:
Gilbert cell converted to diode structure:
https://tspace.library.utoronto.ca/bitstream/1807/13265/1/MQ45606.pdf
DBM mixer modes:
http://arxiv.org/pdf/physics/0608211+2006-arxiv-0608211v1-mixer-tutorial.pdf
regards -
Henry
ehydra.dyndns.info
hanssummers2000 schrieb:
> Hi Chris
>
> What about the H-mode mixer? PA3AKE has a good website on this, see http://www.xs4all.nl/~martein/pa3ake/hmode/ , and reports OIP3's of over +45dBm with bus switches.
>5122 2010-09-16 06:22:41 jwharding3 Re: mixers: strong signal immunity
Another excellent source of info5123 2010-09-16 06:29:35 Chris Trask Re: mixers: strong signal immunity
> What about the H-mode mixer? PA3AKE has a good website on this, see
> http://www.xs4all.nl/~martein/pa3ake/hmode/ , and reports OIP3's of
> over +45dBm with bus switches.
>
One of the H-Mode flunkies showed up on a QRP list years ago boasting about how much better it was, which prompted me to do the comparison. He claimed that my evaluation was not valid because I had to use the exact same parts for both circuits. When I told him that was what I had done, he kept on claiming that comparing the two mixers with the exact same parts was not valid as I had to use the exact same parts in each circuit. This went on for a few days and he was never heard from again.
Chris5124 2010-09-16 06:41:55 k5nwa Re: mixers: strong signal immunity
>To be fair a H class mixer does not use diodes, it works better
>
> >
> > What about the H-mode mixer? PA3AKE has a good website on this, see
> >
> <http://www.xs4all.nl/~martein/pa3ake/hmode/>http://www.xs4all.nl/~martein/pa3ake/hmode/
> , and reports OIP3's of
> > over +45dBm with bus switches.
> >
>
>I was curious about those shortly after they came out. Saw a way to
>make them with diodes rather than the FETs. Made one, tested it,
>then disassembled it and made a conventional diode ring mixer (DRM)
>with the exact same parts. The DRM had better IMD performance by
>6dB. Now I know why they have never compared an H-Mode mixer and DRM
>with the exact same parts. Same thing for the infamous Pullen
>vacuum-tube mixer, which is a class-B unilateral switch.
>
>One of the H-Mode flunkies showed up on a QRP list years ago
>boasting about how much better it was, which prompted me to do the
>comparison. He claimed that my evaluation was not valid because I
>had to use the exact same parts for both circuits. When I told him
>that was what I had done, he kept on claiming that comparing the two
>mixers with the exact same parts was not valid as I had to use the
>exact same parts in each circuit. This went on for a few days and he
>was never heard from again.
>
>Chris
>
supposedly because of the switching FETS so you did not really have a
H mode mixer. It's like comparing a Volkswagen to a Corvette but
forcing the Corvette to use the VW engine.
Cecil
k5nwa
< www.softrockradio.org > < www.qrpradio.com >
< http://parts.softrockradio.org/ >
Never take life seriously. Nobody gets out alive anyway.5125 2010-09-16 06:48:04 Chris Trask Re: mixers: strong signal immunity
>Here are two other interesting docs. But the theme mixers is endless and
>never completed:
>
>Gilbert cell converted to diode structure:
>https://tspace.library.utoronto.ca/bitstream/1807/13265/1/MQ45606.pdf
>
I've also been wondering about using those fast-switching FETs for the commutating devices, but now that I've seen this thesis I'm not going to give that approach any further consideration as I can get far faster switching times requiring lower LO voltage swing with Schottky diodes.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5126 2010-09-16 06:52:39 Chris Trask Re: mixers: strong signal immunity
>Here are two other interesting docs. But the theme mixers is endless and
>never completed:
>
.
>I like this tutorial. He jumps right into the topic of "spectral folding", which he calls "multi-harmonic conversion" beginning on page 14. That aspect of commutating mixers made me consider using a linearized analogue multiplier, which proved to be more work than it was worth.
>DBM mixer modes:
>http://arxiv.org/pdf/physics/0608211+2006-arxiv-0608211v1-mixer-tutorial.pdf
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5127 2010-09-16 06:54:51 Chris Trask Re: mixers: strong signal immunity
>Speaking of high level mixers-I'd like some thoughts about the TLT
>Mixer (an H Mode mixer using Transmission Line Transformers (by
>Oleg Skydan, UR3IQO that appeared in the latest QEX- better yet
>someone might try it with the more readily (for US readers anyway)
>available cores.
>Another excellent source of info on practical H Modes mixers, with
>a very through set of results is at
>http://www.xs4all.nl/~martein/pa3ake/hmode/index.html by Martein PA3AKE. >Any thoughts out there???
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5128 2010-09-16 07:08:45 William Carver Re: mixers: strong signal immunity
UR3IQO, are FT43-37 (u=850). Other materials have u=300, etc, but they
are not readily available. Jerry Sevick says in Transmission Line
Transformers, there was little difference between different
manufacturers material of similar u. Of course he was not measuring
IMD.
Building Olegs topology with FT43-37, then FT61-37 (expect degradation
on 160-80m), both using FSA3157 SPDT switches would tell us a lot: can
his results be reproduced, how important is core u, how does it perform
with better switches, and find out if there's still a benefit to
per-band adjustable switch symmetry via bias adjustments of PA3AKE.
Bill - W7AAZ5129 2010-09-16 07:10:01 Chris Trask Re: mixers: strong signal immunity
>>I was curious about those shortly after they came out. Saw a way to
>>make them with diodes rather than the FETs. Made one, tested it,
>>then disassembled it and made a conventional diode ring mixer (DRM)
>>with the exact same parts. The DRM had better IMD performance by
>>6dB. Now I know why they have never compared an H-Mode mixer and DRM
>>with the exact same parts. Same thing for the infamous Pullen
>>vacuum-tube mixer, which is a class-B unilateral switch.
>>
>>One of the H-Mode flunkies showed up on a QRP list years ago
>>boasting about how much better it was, which prompted me to do the
>>comparison. He claimed that my evaluation was not valid because I
>>had to use the exact same parts for both circuits. When I told him
>>that was what I had done, he kept on claiming that comparing the two
>>mixers with the exact same parts was not valid as I had to use the
>>exact same parts in each circuit. This went on for a few days and he
>>was never heard from again.
>>
>>Chris
>>
>
>To be fair a H class mixer does not use diodes, it works better
>supposedly because of the switching FETS so you did not really have a
>H mode mixer. It's like comparing a Volkswagen to a Corvette but
>forcing the Corvette to use the VW engine.
>
I'd certainly like to try an H-Mode mixer with high-speed bipolar devices such as SiGe, but you cannot get those in monolithic quads.
The diode-based commutating mixer in that thesis has me scrambling for Schottky diode monolithic pairs. I have 1N4148 pairs, but they might not be good enough for this application.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5130 2010-09-16 07:12:09 Chris Trask Re: mixers: strong signal immunity
>The closest cores in my junkbox to the u=400 specified by Oleg Skydan,
>UR3IQO, are FT43-37 (u=850). Other materials have u=300, etc, but they
>are not readily available. Jerry Sevick says in Transmission Line
>Transformers, there was little difference between different
>manufacturers material of similar u. Of course he was not measuring
>IMD.
>
>Building Olegs topology with FT43-37, then FT61-37 (expect degradation
>on 160-80m), both using FSA3157 SPDT switches would tell us a lot: can
>his results be reproduced, how important is core u, how does it perform
>with better switches, and find out if there's still a benefit to
>per-band adjustable switch symmetry via bias adjustments of PA3AKE.
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5131 2010-09-16 07:48:19 David Stone Re: mixers: strong signal immunity
same at all as using switches. Whilst diodes can be made to work a bit like a
switch, they have a region where they are very non-linear on every switch cycle,
as the LO sine wave increases the current through the diode to the point where
it is mainly a fairly linear current carrier; whereas FET switches are still
linear even though higher resistance. You can see this directly by lowering the
LO power to a standard DBM with Schottky diodes and watch the intermodulation
get very bad. If diodes could be switched with a square wave LO of infinitely
fast rise times perhaps they would give improved performance, but the commercial
parts are pretty much state of the art at any LO drive level. None come anywhere
near +45dBm output third order intercept so far achieved by an H mode mixer,
which also uses low LO power. The very high LO drive required by diodes is a big
snag. Analog Devices have some pretty good active parts loosely following the
ideas you have expressed (Gilbert cell of optimum linearity driven with square
LO) but they are still not as good. I am rather hoping that AD may produce an H
mode part with fet switches on a Si-Ge process and possibly a higher power rail
which could take this performance well into the microwave region and even to +50
dBm or more.
Regards
David
GI8FNR
________________________________
5132 2010-09-16 08:05:45 William Carver Re: mixers: strong signal immunity
at that time, but for frequencies where MOSFET switches are fast enough
count ME as one, too. If you didn't get a +40 dBm intercept with
fractions of a mW LO you were doing something wrong Chris. "H-mode
flunkies", huh? Hmmmmmm.
Nice Masters thesis paper. His idea of providing some forward bias to
reduce the LO power requirement for diodes is very, very nice. But it's
NOT a Gilbert Cell mixer, and it's not more of a "commutating mixer"
than ANY diode mixer. Unless they are starved of LO power, the diodes in
a ring mixer are saturated by the LO over most of the LO cycle: look at
the voltage at the LO pin of a diode mixer with a wideband oscilloscope.
I use diodes, and diode mixers all the time. But for an HF receiver
front end they can't even come close to an H-mode mixer. I have driven
+23 diode mixers with 1/4 WATT of 3 nS rise/fall square wave LO and
STILL not obtained a +40 intercept.
If you don't find some matched diodes I can send you some matched quads
of Shottkey diodes that I picked up at the TRW swapmeet a few decades
ago.
Bill - W7AAZ
>5133 2010-09-16 08:09:54 William Carver Re: mixers: strong signal immunity
use a W7ZOI term) improves the balance it should make a better mixer.
What's more interesting is the reality: can one duplicate the magnetic
components with readily available cores, and is the result, in fact, an
improvement.
Bill - W7AAZ
> >Building Olegs topology with FT43-37, then FT61-37 (expect
> degradation
> >on 160-80m), both using FSA3157 SPDT switches would tell us a lot:
> can
> >his results be reproduced, how important is core u, how does it
> perform
> >with better switches, and find out if there's still a benefit to
> >per-band adjustable switch symmetry via bias adjustments of PA3AKE.
> >
>
> When I get to see that article, I'll be able to give you an educated
> guess.
>
> Chris Trask
> N7ZWY / WDX3HLB
> Senior Member IEEE
> http://www.home.earthlink.net/~christrask/5134 2010-09-16 08:23:29 ehydra Re: mixers: strong signal immunity
don't to many toroids there.
BTW: I have a big directory on the hard-disk but lost the overview.
What do you think about the Polyakov-Mixer with it's unique operation by
doubling the frequency of the LO ?
- Henry
--
ehydra.dyndns.info
Chris Trask schrieb:
>> Here are two other interesting docs. But the theme mixers is endless and
>> never completed:
>>
>> Gilbert cell converted to diode structure:
>> https://tspace.library.utoronto.ca/bitstream/1807/13265/1/MQ45606.pdf
>>
>
> Every once in a while you come across a thesis that grabs your attention, and this is definitely one of them. I had often wondered if you could make a commutating (aka Gilbert) mixer with diodes, and now I see that you can. You can bet that I will be burning lots of midnight oil to figure out how to implement this in my feedback mixer topologies.
>
> I've also been wondering about using those fast-switching FETs for the commutating devices, but now that I've seen this thesis I'm not going to give that approach any further consideration as I can get far faster switching times requiring lower LO voltage swing with Schottky diodes.5135 2010-09-16 08:26:25 Chris Trask Re: mixers: strong signal immunity
>Fine Chris. I'm interested of your final design later on. But please,
>don't to many toroids there.
>
>BTW: I have a big directory on the hard-disk but lost the overview.
>
>What do you think about the Polyakov-Mixer with it's unique operation by
> doubling the frequency of the LO ?
>
Regards,
Chris5136 2010-09-16 08:31:51 Chris Trask Re: mixers: strong signal immunity
>I can make educated guesses: if Olegs arrangement of "sortabaluns" (to
>use a W7ZOI term) improves the balance it should make a better mixer.
>What's more interesting is the reality: can one duplicate the magnetic
>components with readily available cores, and is the result, in fact, an
>improvement.
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5137 2010-09-16 08:33:02 Chris Trask Re: mixers: strong signal immunity
>Chris I'm surprised that you are suggesting using diodes, as they
>are not the same at all as using switches. Whilst diodes can be
>made to work a bit like a switch, they have a region where they
>are very non-linear on every switch cycle, as the LO sine wave
>increases the current through the diode to the point where it is
>mainly a fairly linear current carrier;
>
The problem that I've been facing recently with my feedback mixers is the excessive charge storage time of the bipolar devices used in the commutating quad, which sets the speed limit for turning the devices on and off. This idea of using much faster Schottky diodes in their place is intruging and well worth the effort of building prototypes. I've already bought Schottky diode pairs on eBay, and now it's off to the drawing board to fill in the details of an initial design.
I'm also looking at a method for cancelling IMD products that was intended for a power amplifier application, but with some modification can be made much simpler and less expensive for small-signal applications such as this.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5139 2010-09-16 16:10:41 Chris Trask Re: mixers: strong signal immunity
> >To be fair a H class mixer does not use diodes, it works better
> >supposedly because of the switching FETS so you did not really have a
> >H mode mixer. It's like comparing a Volkswagen to a Corvette but
> >forcing the Corvette to use the VW engine.
> >
>
> True, but I saw the topology as a possible means of reducing the
>effect that the RF signal has on the switching of the diodes. I do
>like the idea that the FETs in the H-Mode mixer do not have the RF
>signal interfering with the switching, but then you are faced with the
>speed limitation of the FETs and the requirement for the higher
>swtching voltage required.
>
> I'd certainly like to try an H-Mode mixer with high-speed bipolar
>devices such as SiGe, but you cannot get those in monolithic quads.
>
> The diode-based commutating mixer in that thesis has me scrambling
>for Schottky diode monolithic pairs. I have 1N4148 pairs, but they
>might not be good enough for this application.
>
I just hastily built a somewhat simplified version of the Cheung diode
mixer, using a pair of BAV70 dual silicon switching diodes (common cathode),
which are very similar to the 1N4148. Without any tweaking, I'm seeing an
OIP3 of greater than +20dBm, which equates to an IIP3 of greater than
+27dBm. The SBL1-1 from Mini-Circuits has a +14dBm OIP3.
Those dual Schottky diodes (BAT54A and BAT54C) can't arrive fast enough
now. Those should improve the IIP3 and OIP3 by at least 6dB.
I'll have a schematic for this on my web page later this afternoon.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5140 2010-09-16 17:01:36 Chris Trask Re: mixers: strong signal immunity
,----------------------. High Performance Mixers and
/ What's all this \ Amplifiers for RF Communications
/ extinct stuff, anyhow? /
\ _______,--------------' Chris Trask / N7ZWY
_5141 2010-09-16 18:00:29 ehydra Re: mixers: strong signal immunity
> I just hastily built a somewhat simplified version of the Cheung diodeSorry for asking that:
> mixer, using a pair of BAV70 dual silicon switching diodes (common cathode),
> which are very similar to the 1N4148. Without any tweaking, I'm seeing an
> OIP3 of greater than +20dBm, which equates to an IIP3 of greater than
> +27dBm. The SBL1-1 from Mini-Circuits has a +14dBm OIP3.
I guess that IIP3 - OIP3 = loss
And system impedance is 50 ohms.
So for optimum noise performance the LNA should have gain equal to
calculated loss?
Is this right?
Nice to see that all four transformers are equal.
How does it compare to your other designs?
- Henry
--
ehydra.dyndns.info5142 2010-09-16 18:37:36 Chris Trask Re: mixers: strong signal immunity
>
> Chris Trask schrieb:
> > I just hastily built a somewhat simplified version of the Cheung
> > mixer, using a pair of BAV70 dual silicon switching diodes (commoncathode),
> > which are very similar to the 1N4148. Without any tweaking, I'm seeingan
> > OIP3 of greater than +20dBm, which equates to an IIP3 of greater thanYes
> > +27dBm. The SBL1-1 from Mini-Circuits has a +14dBm OIP3.
>
> Sorry for asking that:
> I guess that IIP3 - OIP3 = loss
>
>Yes
> And system impedance is 50 ohms.
>
>Basically, but it depends to some degree on the NF of the LNA.
> So for optimum noise performance the LNA should have gain equal to
> calculated loss?
> Is this right?
>
BTW: While poking around with this I found that the RF and LO inputs
can be reversed with no noticable change in performance.
This is intense!
>Yes, and also that commercial parts such as from Mini-Circuits can be
> Nice to see that all four transformers are equal.
>
used. I typically wind my own as I can get better performance.
>Right now, the OIP3 is about equal with my earliest feedback mixer
> How does it compare to your other designs?
>
designs, and I may very likely be able to roughly equal my latest efforts,
provided that the improvement using Schottky diodes is what I expect to see.
Even so, I'm able to beat the garden-variety diode ring mixers such as the
SBL1-1 by about 6dB. I'll be using the BAT54C to start with, but I'm also
interested in locating some microwave grade diode pairs (common cathode or
anode).
You'll notice that I reduced Cheung's 4-transformer LO circuitry to just
a pair of transformers.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5143 2010-09-16 19:15:48 ehydra Re: mixers: strong signal immunity
>> So for optimum noise performance the LNA should have gain equal tothanks!
>> calculated loss?
>> Is this right?
>>
>
> Basically, but it depends to some degree on the NF of the LNA.
>For a bidirectional transceiver or what?
> BTW: While poking around with this I found that the RF and LO inputs
> can be reversed with no noticable change in performance.
>
> This is intense!
>I found it horrible how many tranformers/parts you typically used before :-)
>> Nice to see that all four transformers are equal.
>>
>
> Yes, and also that commercial parts such as from Mini-Circuits can be
> used. I typically wind my own as I can get better performance.
>"Comparable"? So what is the milestone reached? Less transformers in
>> How does it compare to your other designs?
>>
>
> Right now, the OIP3 is about equal with my earliest feedback mixer
> designs, and I may very likely be able to roughly equal my latest efforts,
> provided that the improvement using Schottky diodes is what I expect to see.
> Even so, I'm able to beat the garden-variety diode ring mixers such as the
> SBL1-1 by about 6dB. I'll be using the BAT54C to start with, but I'm also
> interested in locating some microwave grade diode pairs (common cathode or
> anode).
opposit to your earlier designs? Maybe I'm blind or just to be dead on
one's feet.
Will it work with a digital driver LO like the Si570 ?
>No. I will revisit the paper later.
> You'll notice that I reduced Cheung's 4-transformer LO circuitry to just
> a pair of transformers.
>
You know, I'm interested in the band between 10MHz and 500MHz and how it
compares to the trusted SA602.
And going now to bed.
Always close daily work with good news! Don't work a second more!
- Henry
--
ehydra.dyndns.info5144 2010-09-16 19:46:28 Chris Trask Re: mixers: strong signal immunity
> >
> > BTW: While poking around with this I found that the RF and LO inputs
> > can be reversed with no noticable change in performance.
> >
> > This is intense!
>
> For a bidirectional transceiver or what?
>
>Nah! Transformers are your friend. Once you understand how the hybrid (or bridge) transformer works, which is the same as a hybrid junction, you can do all sorts of wonderful signal processing things.
> >
> > > Nice to see that all four transformers are equal.
> > >
> >
> > Yes, and also that commercial parts such as from Mini-Circuits can be
> > used. I typically wind my own as I can get better performance.
>
> I found it horrible how many tranformers/parts you typically used before :-)
>
>It means that the performance I got with the early feedback mixers can now be had with a far simpler and less expensive circuit that can potentially be a drop-in replacement for existing commercial parts.
> >
> >> How does it compare to your other designs?
> >>
> >
> > Right now, the OIP3 is about equal with my earliest feedback mixer
> > designs, and I may very likely be able to roughly equal my latest efforts,
> > provided that the improvement using Schottky diodes is what I expect to see.
> > Even so, I'm able to beat the garden-variety diode ring mixers such as the
> > SBL1-1 by about 6dB. I'll be using the BAT54C to start with, but I'm also
> > interested in locating some microwave grade diode pairs (common cathode or
> > anode).
>
> "Comparable"? So what is the milestone reached? Less transformers in
> opposit to your earlier designs? Maybe I'm blind or just to be dead on
> one's feet.
>
>Dunno yet. Only got it working about six hours ago.
> Will it work with a digital driver LO like the Si570 ?
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
[Non-text portions of this message have been removed]5145 2010-09-16 19:52:21 Jim Miller Re: mixers: strong signal immunity
73
jim ab3cv
[Non-text portions of this message have been removed]5146 2010-09-16 20:01:46 Chris Trask Re: mixers: strong signal immunity
> any idea of the range of frequencies this will handle?
>
the Schottky diodes I'll see what the frequency limit is with the
transformers I used. I have at least gotten rid of the charge storage time
problem. The BAT54 diodes have a recovery time of 5nSec.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5147 2010-09-16 20:18:16 kb1gmx Re: mixers: strong signal immunity 5148 2010-09-16 20:24:30 Chris Trask Re: mixers: strong signal immunity
> Here's the schematic:
>
> http://www.home.earthlink.net/~christrask/Trask%20Cheung%20Mixer.pdf
>
> All of the transformers are 1:2CT, such as the Mini-Circuits T4-1 or
> T4-6T. Pay close attention to the phasing dots.
>
following:
http://www.home.earthlink.net/~christrask/X-Mode%20Diode%20Mixer.pdf
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5149 2010-09-16 20:32:27 Chris Trask Re: mixers: strong signal immunity
> T4 concerns me a bit in that IFs I've worked with range from high 144mhz,
> .455mhz and more than a few audio baseband IQ DCRX. For most of the RFrange
> it's core selection and turns, for the very low end The transformer startswith
> laminations (audio).I'll be testing this thoroughly once I have the better diodes installed.
>
> Drive required?
> Lo to RF isolation?
>
> Does the IF need greater/lesser attention that the typical DBM
> IF port? For example how does it's performance degrade with
> mismatched IF loads?
>
The transformers in it now are just my generic 4 turns of trifilar wire in a
2402 binocular core. I use those for just about everything when
prototyping.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5150 2010-09-16 20:57:03 k5nwa Re: mixers: strong signal immunity
pairs, but they are wired anode to cathode, I
bought them for use with anti-parallel diode
mixer, the thing is they have 1 nsc recovery and
are available in all sorts of configuration, but
the part number escapes me. This weekend I will
looking for some other parts and I will keep an
eye on those to get the part number.
It would be dandy to have a mixer like that one
on a tiny board, so that you could hook them up
in circuits like it was a chip and get decent performance.
At 10:01 PM 9/16/2010, you wrote:
>Cecil
>
> >
> > any idea of the range of frequencies this will handle?
> >
>
>The transformer design and the diodes will determine that. When I get
>the Schottky diodes I'll see what the frequency limit is with the
>transformers I used. I have at least gotten rid of the charge storage time
>problem. The BAT54 diodes have a recovery time of 5nSec.
>
>Chris Trask
>N7ZWY / WDX3HLB
>Senior Member IEEE
><http://www.home.earthlink.net/~christrask/>http://www.home.earthlink.net/~christrask/
>
>
k5nwa
< www.softrockradio.org > < www.qrpradio.com >
< http://parts.softrockradio.org/ >
Never take life seriously. Nobody gets out alive anyway.5153 2010-09-17 04:47:27 hanssummers2000 Re: mixers: strong signal immunity
> WOW!!!!!!!!!!Still it won't be like the H-mode mixer with OIP3 even as high as 45dB (see PA3AKE's work). Also with an H-mode mixer you're looking at much lower losses of typically 1-2dB at HF. Of course, the low drive level requirement too. I agree with Bill, at HF, I think the H-mode performance is hard to approach with anything else, let alone beat.
>
> I just hastily built a somewhat simplified version of the Cheung diode
> mixer, using a pair of BAV70 dual silicon switching diodes (common cathode),
> which are very similar to the 1N4148. Without any tweaking, I'm seeing an
> OIP3 of greater than +20dBm, which equates to an IIP3 of greater than
> +27dBm. The SBL1-1 from Mini-Circuits has a +14dBm OIP3.
>
> Those dual Schottky diodes (BAT54A and BAT54C) can't arrive fast enough
> now. Those should improve the IIP3 and OIP3 by at least 6dB.
73 Hans G0UPL
http://www.hanssummers.com5154 2010-09-17 04:55:32 hanssummers2000 Re: mixers: strong signal immunity 5155 2010-09-17 04:55:48 Fernando Krouwel Re: mixers: strong signal immunity
Congratulations and thank you very much for sharing your work with us!
What about the LO to RF ports isolation (very important mainly when building DC receivers for VHF and higher)?
73´s
Fernando - PY2ETT
----- Original Message -----
5156 2010-09-17 06:01:24 Chris Trask Re: mixers: strong signal immunity
>Congratulations and thank you very much for sharing your work with us!
>
>What about the LO to RF ports isolation (very important mainly when
>building DC receivers for VHF and higher)?
>
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5157 2010-09-17 06:16:01 Chris Trask Re: mixers: strong signal immunity
>> WOW!!!!!!!!!!
>>
>> I just hastily built a somewhat simplified version of the Cheung
>>diode mixer, using a pair of BAV70 dual silicon switching diodes
>>(common cathode), which are very similar to the 1N4148. Without any
>>tweaking, I'm seeing an OIP3 of greater than +20dBm, which equates to
>>an IIP3 of greater than +27dBm. The SBL1-1 from Mini-Circuits has a
>>+14dBm OIP3.
>>
>> Those dual Schottky diodes (BAT54A and BAT54C) can't arrive fast
>> enough now. Those should improve the IIP3 and OIP3 by at least 6dB.
>
>Still it won't be like the H-mode mixer with OIP3 even as high as 45dB
>(see PA3AKE's work). Also with an H-mode mixer you're looking at much
>lower losses of typically 1-2dB at HF. Of course, the low drive level
>requirement too. I agree with Bill, at HF, I think the H-mode
>performance is hard to approach with anything else, let alone beat.
>
While I'm waiting for the Schottky diodes to arrive, I have an IMD cancelling circuit that can be used with amplifiers or mixers and should be able to provide at least 20dB of IMD cancellation, and probably 30dB or more with some minor tweaking. It's called "The Matrix", a form of Neo-technology.
Too bad you can't make Lange couplers with lumped elements.
Chris
Regards,
Chris5158 2010-09-17 06:45:57 kb1gmx Re: mixers: strong signal immunity 5163 2010-09-17 08:53:33 ehydra Re: mixers: strong signal immunity
But I see a generic problem with the 'three port transformer interface'
if the IF is very low the transformers have to be very unequal realized.
What is the theoretical background for optimization of the three
interdependent frequencies of RF, LO, IF ? (Beside mirror frequency
suppression)?
Is there an optimum point as function of IMD, NF, phase noise, etc.?
hm.
- Henry
--
ehydra.dyndns.info
Chris Trask schrieb:
>> Congratulations and thank you very much for sharing your work with us!
>>
>> What about the LO to RF ports isolation (very important mainly when
>> building DC receivers for VHF and higher)?
>>
>
> Haven't gotten that far in the testing, but I'll make a preliminary measurement later today. I got up very early this morning and did a little more testing, finding that the three ports are universal. Any port can be used as LO, RF, or IF and the OIP3 performance remains the same.
>5165 2010-09-17 09:14:29 Chris Trask Re: mixers: strong signal immunity
>But I see a generic problem with the 'three port transformer interface'
>if the IF is very low the transformers have to be very unequal realized.
>
>What is the theoretical background for optimization of the three
>interdependent frequencies of RF, LO, IF ? (Beside mirror frequency
>suppression)?
>Is there an optimum point as function of IMD, NF, phase noise, etc.?
>
For instance, for the general coverage (0-30MHz) receiver that I have in mind, T3 would be designed to handle the input spectrum while the remaining three would be designed to accomodate the 40-70MHz LO and the 40MHz IF.
Designing all the transformers to accomodate 0.1-100MHz at all three ports is NOT unreasonable since they can all be made with trifilar twisted wires. In that approach, the amount of wire has a lot to do with the upper cutoff and the core size and material determines the low cutoff by way of the magnetizing inductance. But then there's the ferroresonance to take into consideration, which can impact high frequency losses, especially with ferrite materials.
Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/5166 2010-09-17 09:45:38 Helge Kyndbo Re: mixers: strong signal immunity
Could you help me to understand "ferroresonance"
a little better, when I google there only examples
in high power systems.
Helge5167 2010-09-17 11:13:46 Chris Trask Re: mixers: strong signal immunity
> Hi Chris,
> Could you help me to understand "ferroresonance"
> a little better, when I google there only examples
> in high power systems.
> Helge
>
imaginary part of the permeability of the magnetic material. At that point
you will see the highest degree of induced (not bulk) core losses. For
instance, Fair-Rite 73 material has a ferroresonance frequency just above
1MHz. If you were making a flux-coupled transformer with monofilar
windings, you would see an increase in the transmission loss there.
Fair-Rite material has it's ferroresonance above 10MHz, and 61 material is
above 60MHz.
If you make wideband transformers with twisted wires, where you minimize
the coupling to the core and the magnetic material serves to increase the
magnetization inductance and reduce the low cutoff frequency, you might not
see this effect at all.
Chris5223 2010-09-22 11:39:58 ehydra Re: mixers: strong signal immunity
>> I found it horrible how many tranformers/parts you typically usedI know this because I worked at german telekom. Yes, sometimes a
>> before :-)
>>
>
> Nah! Transformers are your friend. Once you understand how the
> hybrid (or bridge) transformer works, which is the same as a hybrid
> junction, you can do all sorts of wonderful signal processing things.
>
transformer is cheap and elegant.
- Henry
--
ehydra.dyndns.info