EMRFD Message Archive 5938

Message Date From Subject
5938 2011-03-26 13:14:09 kb1gmx high level mixers
I'm working on a HF low gain SSB design and one area I'm playing on paper with is the translation from IF to operating frequency. In the pas with earlier designs I've used sa602/612, DBM and high level DBMs
and always had to limit the transmit IF gain to keep from overloading
the mixer. That usually ends up with a desired signal in the -6 to -20DBm range and lots of gain in the power amplifier chain even for QRP power typically 47DB.

I've been looking at using a passive high level mixer that is tolerant
of square wave (digital RF from SI570) and can handle IF to target RF
level in the 0 to maybe +7 DBm range. One circuit is the H-mode (I7SWX) but that uses a lot of ferrite and supporting circuits but offers a doubly balanced design. The alternate single ended or single balanced version of the Maas mixer it's simpler and tolerant of a square wave source.

with background as stated here are some details. If likely in the 9-11mhz range, output in the HF below 20mhz range and specific targets of 40M and 17M. What I need to know is some info on what the Maas mixer can handle for levels at the upper limit using common enhancement mode IGFETS (2n7000 and kin). I have to test the design to verify but it would help to know if what I want to do is reasonable rather than complain it can't do xyz.

Another design I've looked at is Ulrich Rhode single balanced mixer
that used 2n5179 but using 2n7000 instead with direct digital drive to the gates. That would be a hybrid of the Maas and Rhode deisgn.
this has a better starting point as the bipolar version has data associated with it for signal handling.

Thoughts?


Allison
5939 2011-03-26 15:47:20 KK7B Re: high level mixers
Hi Allison,

I described some interesting high level series FET balanced mixer designs and experiments a few years ago that sound like they might help you get started with your experiments. The work was published by IEEE, so there are copyright issues with posting it
5940 2011-03-26 15:57:09 ehydra Re: high level mixers
>> Thoughts?

A relative unknown structure is the diode switching bridge. More common
with IC test circuits.

- Henry


--
ehydra.dyndns.info


KK7B schrieb:
> Hi Allison,
>
> I described some interesting high level series FET balanced mixer designs and experiments a few years ago that sound like they might help you get started with your experiments. The work was published by IEEE, so there are copyright issues with posting it
5941 2011-03-26 18:46:22 kb1gmx Re: high level mixers
5942 2011-03-26 18:48:10 kb1gmx Re: high level mixers
5943 2011-03-27 06:08:10 Tim Re: high level mixers
Isn't it a little unfair to assume that a Gilbert cell type mixer must have exactly the same limitations as the SA602?

You already know the above, because we had this discussion just last week :-). SA602->MC1496->discrete parts at higher bias.

I swear, some of the tube SSB circuits in the 50's and 60's did in fact do high-level mixing but I'm unsure as to topology right now. I do remember some transmitters using as many as 4 7360's and some of those must've been hetrodyne mixers after the IF. Others used dual-triodes in either singly balanced or doubly balanced bridges. These I think come closest to what you want to do.

I looked at the K2 schematics and the transmit heterodyne mixer is just a SA602 followed by a LT1252 high-gain high-bandwidth buffer. It's only a couple bucks a pop but in my head a LT1252 is not quite a jellybean component, maybe I should learn about it.

How I feel about using a SA602 as a transmitter mixer, is that I have to cut down my very low phase noise +10dBm (or more) output VFO or LO to a much smaller level just to drive the SA602 cleanly. That doesn't seem right but I have no measurements (e.g. phase noise) to say that it's wrong, or how wrong it is. Very much in violation of EMRFD principles for me to say that it feels wrong :-).

They canceled the cherry blossom kite festival for today, maybe what I'll do this afternoon is build a high level (K7HFD, fig 4.18 in EMRFD) 2MHz VFO mixed with a 5MHz or 9MHz crystal via a +17dBm mixer to use
5944 2011-03-27 07:07:46 jeffthom99 Re: high level mixers
Hi Rick,

Would you be able to share your IEEE paper on high-level mixers with
me as well as Allison? I have also been looking for a simple, stable
mixer topography that will allow relatively high power output.

I have greatly enjoyed your work since retiring and 'getting
technical' again after many years in the lands of software design and
systems engineering at NASA.

72's and many thanks for your contributions,

Jeff, W3HVU

5945 2011-03-27 07:18:35 James Duffey Re: high level mixers
Allison - You wrote:

"That usually ends up with a desired signal in the -6 to -20DBm range and lots of gain in the power amplifier chain even for QRP power typically 47DB."

Is there anything really wrong with using lots of gain, used of course with proper design consideration, either in the LO or transmitter chain? One of the mantras of the microwave homebrewer is that gain is cheap.

Rick - You wrote:

"The work was published by IEEE, so there are copyright issues with posting it on line."

No problem with that, but there are no restrictions on posting the reference to the work on the reflector. Many of us have access to the IEEE archives, and nearly everyone has access through a library, so if you could post the reference, we could access the material in the manner IEEE the authors prefer. Usually if one person is interested in a topic, more than one are interested, and just posting the reference is one way to disseminate it widely.

Thanks - Duffey KK6MC/5

--
KK6MC
James Duffey
Cedar Crest NM
5946 2011-03-27 08:10:41 AD7ZU Re: high level mixers
There are several examples of using switching mixers .. well maybe they are not
actually mixers for frequency conversions.  The most common is from the RF to
baseband but the same circuit can be used as a converter.  Here is an example of
a 500khz to 3.5mhz conversion done with a switching converter.
http://www.nikkemedia.fi/juma-tx500/juma-tx500-main-sch-1.pdf
The use of a switch will operate well with the Si570 square wave output, is low
power, and does not require high drive levels.  The example uses an ft3253,
however the sn74auc2g53? spdt switches from TI have faster switching times and
better on/off symmetry but may require a more complex clocking scheme being a
spdt switch.
Just a couple thoughts
Randy




________________________________
5947 2011-03-27 08:31:28 Chris Trask Re: high level mixers
>
> The example uses an ft3253, however the sn74auc2g53 spdt
> switches from TI have faster switching times and better
> on/off symmetry but may require a more complex clocking
> scheme being a spdt switch.
>

Actally, it wouldn't be complicated at all. In a diode ring mixer (DRM), the diodes are switched on and off as SPDT pairs, so you could use a pair of these to make a demodulator or modulator. The on resistances are a bit high. Is there a dual version (DPDT) of this device?



Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5948 2011-03-27 08:44:56 kb1gmx Re: high level mixers
5949 2011-03-27 08:50:19 kb1gmx Re: high level mixers
5950 2011-03-27 09:05:31 kb1gmx Re: high level mixers
5951 2011-03-27 09:26:52 Chris Trask Re: high level mixers
>
> There was a discussion going back a few months about
> switching mixers where you put up an interesting design
> but I can't find the circuit.
>

I removed it (the X-Mode mixer) from the files section as there was a serious error in it. When I got it taken care of, I ended up with a diode version of the H-mode mixer, and it did not perform as well as other topologies that I had experimented with. I had done this H-mode diode mixer many years ago and the results were the same.

I experimented further, evaluating numerous single and dual Schottky diodes in three different configurations (DRM, H-mode, and "split-ring"). To no surprise I found that IMD was directly related to switching speed, and that is a major culprit in commutating (aka Gilbert Cell) mixers.

The "split ring" mixer performed best, having a conversion loss of only 3.5dB instead of the 6.0-6.5dB of the DRM (IMD remained the same). I'm not certain yet as to what this improved conversion loss is a result of, nor if it is universal for all manner of switching diodes, though I do know that it was consistent for all of the Avago dual Schottky diodes that I evaluated. The only significant difference between the "split-ring" and DRM is that the diodes see an RF source impedance of 200 ohms rather than 100 ohms, so it may be an impedance matching issue.




Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5952 2011-03-27 09:42:08 Tim Re: high level mixers
My impression is that CW keying of the K7HFD oscillator is a perfect way to get thermal whoops. But keeping it on all the time it ought to be fine. My plan is to do RIT shifts in the crystal oscillator, and keep varicaps out of the high-tank-energy K7HFD oscillator itself. Will find out tonight :-).

My feeling is that the Maas mixer has best isolation when output frequency is substantially (like an order of magnitude or more) lower than the input frequencies. Like microwave downconversion applications. Use
5953 2011-03-27 10:34:59 KK7B Re: high level mixers
Hi Jeff and all,

Here's the reference on the IEEE Paper. I'm on the road at the moment without access to my digital archive, so I can't do anything more than pass along the reference. Those of you with access to IEEE Xplore should be able to find it, and it will also be in technical libraries.

Richard L. Campbell, High Frequency Differential Passive FET Direct Conversi
5954 2011-03-27 11:53:43 Chris Trask Re: high level mixers
>
> Here's the reference on the IEEE Paper. I'm on the road at the moment
> without access to my digital archive, so I can't do anything more than
> pass along the reference. Those of you with access to IEEE Xplore should
> be able to find it, and it will also be in technical libraries.
>
> Richard L. Campbell, High Frequency Differential Passive FET Direct
> Conversion Mixer/Modulator,
> IEEE Microwave Theory and Techniques Society International Microwave
> Symposium Proceedings, pp. 922-925, June 2006
>

Darn, I don't have that one here. I would certainly appreciate a PDF
copy.

Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5955 2011-03-27 13:00:08 ha5rxz Re: high level mixers
Allison, we appear to be working on the same thing but for slightly different reasons.

I am currently working on a 160m-6m transceiver design and trying to find the best method of converting a 45 MHz signal to the wanted frequency. Current efforts are centered around the TI TS5A23159 DPDT chip which has a 1 ohm
5957 2011-03-27 16:27:41 kb1gmx Re: high level mixers
5958 2011-03-27 19:50:19 AD7ZU Re: high level mixers
Chris,
 
I have not seen any comparable dpdt switches from TI ..
 
Now for some off topic commentary...
 
My efforts have been focused on image rejecting sampling detectors. The diode
ring configuration is deceptive in that it requires 2 diodes in series to be
conducting at any given time...therefore the switching time is doubled and an
asymmetric condition occurs as the first diode is turned on while the second is
off (and visa-versa).
 
The design I favor (at least for the moment) consists of 1 spdt switch in each
leg of a transformer secondary.  The switches are clocked from a 1f quadrature
clock.  The top leg providing the 0 and 180 degree components, the lower leg
provides the 90 and 270 phases.  An inhibit signal alternating between top and
bottom each 90 degrees of the LO is derived from the same 1f quadrature clock
and disables the either the top or bottom lane at any given time.  This avoids
using a 4f clock. I have looked at creating the clocking and inhibit signals
using a high speed 3:8 decoder and some counter configurations…the problem is
again in the switching asymmetry of these devices. .. a very few ns at 30mhz is
significant!
 
Integrating detectors with asymmetric switching times result in a phase leakage
that I don’t believe can be fully corrected downstream?  I haven’t been able to
prove this.. but in several attempts phase correction by adding or subtracting a
fraction of the opposite lane do not seem to provide the needed function since
the phase leakage is cumulative into the baseband signal.  A better approach may
be to limit the duty cycle of the LO?
 
 
Randy
 

________________________________

5959 2011-03-27 21:16:21 Chris Trask Re: high level mixers
>
> > Actally, it wouldn't be complicated at all. In a diode ring mixer (DRM),
the
> > diodes are switched on and off as SPDT pairs, so you could use a pair of
these
> > to make a demodulator or modulator. The on resistances are a bit high.
Is there
> > a dual version (DPDT) of this device?
>
> I have not seen any comparable dpdt switches from TI ..
>

I found that they only make an SPDT and a dual SPST in that series.

>
> Now for some off topic commentary...
>
> My efforts have been focused on image rejecting sampling detectors. The
diode
> ring configuration is deceptive in that it requires 2 diodes in series to
be
> conducting at any given time...therefore the switching time is doubled and
an
> asymmetric condition occurs as the first diode is turned on while the
second is
> off (and visa-versa).
>

You might want to look at that short paper I uploaded into the files
section. The diodes are turned on and off as series pairs, and that might
account for the improved conversion loss. The common connection between
them is a virtual ground, so switching is more symmetrical and without the
doubling of the switching time.

>
> The design I favor (at least for the moment) consists of 1 spdt switch in
each
> leg of a transformer secondary. The switches are clocked from a 1f
quadrature
> clock. The top leg providing the 0 and 180 degree components, the lower
leg
> provides the 90 and 270 phases. An inhibit signal alternating between top
and
> bottom each 90 degrees of the LO is derived from the same 1f quadrature
clock
> and disables the either the top or bottom lane at any given time. This
avoids
> using a 4f clock. I have looked at creating the clocking and inhibit
signals
> using a high speed 3:8 decoder and some counter configurations…the problem
is
> again in the switching asymmetry of these devices. .. a very few ns at
30mhz is
> significant!
>

Hmmm... An interesting configuration. Do you have a schematic for
that?

The switching time and asymmetry problem sent me back to using diodes,
which are much faster. But making a feedback loop without getting
complicated is presently ellusive. You also can't have DC bias current
running through the diodes.

>
> Integrating detectors with asymmetric switching times result in a phase
leakage
> that I don’t believe can be fully corrected downstream? I haven’t been
able to
> prove this.. but in several attempts phase correction by adding or
subtracting a
> fraction of the opposite lane do not seem to provide the needed function
since
> the phase leakage is cumulative into the baseband signal. A better
approach may
> be to limit the duty cycle of the LO?
>

You might want to look at the phase detector that National uses in their
present PLL devices. They recovered a blanking pulse and applied it
downstream to greatly reduce the phase detector reference noise. I came up
with a simpler one with less propagation time here:

http://www.home.earthlink.net/~christrask/Paper023.html

Not that it's immediately usable in your problem, but it may be
inspirational.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5960 2011-03-28 05:25:43 Tim Re: high level mixers
Skew and asymmetries in decoder outputs is a big problem, you are right, when using a 3-bit ripple counter followed by a 3:8 decoder.

The solution is the ring counter. In many cases the Johnson counter variant is a perfect match. There's some introductory information on Wikipedia about ring counters but it's written by a newbie. You will find better informati
5963 2011-03-28 07:58:46 Corey Minyard Re: high level mixers
> Hmmm... An interesting configuration. Do you have a schematic for
> that?
>
> The switching time and asymmetry problem sent me back to using diodes,
> which are much faster. But making a feedback loop without getting
> complicated is presently ellusive. You also can't have DC bias current
> running through the diodes.
>
I'm curious, why can't you have DC bias current running through mixer
diodes?

I'm still learning a lot of basics about RF design.

I ask because I'm playing with a little though experiment using DC bias
through some diodes so that the diodes are normally on and the LO turns
them off. I've scratched a little schematic on the back of a Dilbert
calendar:
Diode mixer
The idea is to bias D1 and D2 with R1, R2, and R4 so that one of them
will turn off at the peak of the LO drive when D3 and D4 turn on. R3
and R5 control when D3 and D4 turn on. C1 is there to prevent the DC
bias from flowing through the transformers. C2 is there just to match
it, though the capacitance from D3 and D4 will be four time the
capacitance of D1 and D2.

I think this is sub-harmonic, driven at half the desired frequency, as
switching occurs four times every clock cycle:
When the diodes switch
If you set R4 and R5 correctly, one diode should turn on right when
another turns off, so you shouldn't need that much LO drive to make this
work. You could even computer-control the bias with digital
potentiometers and have a processor automatically tune the circuit for
best performance, especially since the setting is probably frequency
dependent. A good sine wave or triangle wave would be required for
this, obviously.

I'm sure this circuit has some problems. I know the LO will feed into
the RF and IF ports. It could be balanced, though since it's
sub-harmonic it may not matter so much. Though what feeds into the RF
and IF may be first harmonic in one direction. However, if DC biasing a
diode make something bad happen for a mixer, or if there's something
else that won't make this work, I can just abandon this idea without
building anything :).

-corey


[Non-text portions of this message have been removed]
5964 2011-03-28 08:06:38 Chris Trask Re: high level mixers
>
> > Hmmm... An interesting configuration. Do you have
> > a schematic for that?
> >
> > The switching time and asymmetry problem sent me back
> > to using diodes, which are much faster. But making a
> > feedback loop without getting complicated is presently
> > ellusive. You also can't have DC bias current running
> > through the diodes.
> >
>
> I'm curious, why can't you have DC bias current running
> through mixer diodes?
>

It places an extra burden on the switching times and symmetry. This is a serious problem with active switching mixers.



Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
5965 2011-03-28 12:52:44 Gian Re: high level mixers
5966 2011-03-28 21:14:07 longjohn119 Re: high level mixers
5967 2011-03-29 02:44:05 Gian Re: high level mixers
Hi JR,

thanks for your questions.

1) RF In at the 3 Windings Transformer is OK at HF and IF Output OK up to 70MHz. Conv loss -5/-6dB.

2) I have tested the 2T HMM up to VHF. For 50MHz input it is best to fedd RF in from the 5 winding transformer and IF output at the 3 windings one. This present a conv loss not more than -6dB.
I use this configuration when replacing the 2nd mixer in commercial equipments (Yaesu, Icom, Kenwood) where the IF is from 70MHz to 45MHz.

3) I have tried a test, but not assembled a converter or RX,
5968 2011-03-29 15:08:53 kb1gmx Re: high level mixers
5969 2011-03-29 17:39:07 ehydra Re: high level mixers
AD7ZU schrieb:
> Chris,
>
> I have not seen any comparable dpdt switches from TI ..

The datasheets says 15ns break-time for switch-over. This implies the
circuit is limited to the low MHz-range.


>
> Now for some off topic commentary...
>
> My efforts have been focused on image rejecting sampling detectors. The diode
> ring configuration is deceptive in that it requires 2 diodes in series to be
> conducting at any given time...therefore the switching time is doubled and an
> asymmetric condition occurs as the first diode is turned on while the second is
> off (and visa-versa).
>
> The design I favor (at least for the moment) consists of 1 spdt switch in each
> leg of a transformer secondary. The switches are clocked from a 1f quadrature
> clock. The top leg providing the 0 and 180 degree components, the lower leg
> provides the 90 and 270 phases. An inhibit signal alternating between top and
> bottom each 90 degrees of the LO is derived from the same 1f quadrature clock
> and disables the either the top or bottom lane at any given time. This avoids
> using a 4f clock. I have looked at creating the clocking and inhibit signals
> using a high speed 3:8 decoder and some counter configurations…the problem is
> again in the switching asymmetry of these devices. .. a very few ns at 30mhz is
> significant!
>
> Integrating detectors with asymmetric switching times result in a phase leakage
> that I don’t believe can be fully corrected downstream? I haven’t been able to
> prove this.. but in several attempts phase correction by adding or subtracting a
> fraction of the opposite lane do not seem to provide the needed function since
> the phase leakage is cumulative into the baseband signal. A better approach may
> be to limit the duty cycle of the LO?
>

Can you post the schematic?

What about a DC correction loop measuring the output DC voltage of the
switch bridge and using this as a duty-cycle corrective.

For a radio the DC has no information in the antenna signal.


- Henry


--
ehydra.dyndns.info
5970 2011-03-29 20:14:51 AD7ZU Re: high level mixers
I hope not to hijack Allison's topic..

switch time looks to me like max of 2.2ns using a 2.5v supply.
http://focus.ti.com/lit/ds/symlink/sn74auc2g53.pdf

currently my schematic is only in my notebook .. and hopefully soon on a proto
board

its relatively simple. so maybe in a few days i will make a pdf version.

A relatively simple but accurate closed loop control over the phase and
amplititude error correction usable for both modulator / demodulator might
be the holy grail for DC phasing rig designs.


Randy




________________________________
5971 2011-03-30 06:33:49 ehydra Re: high level mixers
AD7ZU schrieb:
> I hope not to hijack Allison's topic..
>
> switch time looks to me like max of 2.2ns using a 2.5v supply.
> http://focus.ti.com/lit/ds/symlink/sn74auc2g53.pdf
>

My comment was referenced to the mentioned TS5A23159. I looked again in
the datasheet: It is 12ns specified. A higher ohmic version should have
less time.
Searching the fastest switch is the way to go - but the answer is simply
a Schottky diode.
This 12ns can be seen as time where the receiver is muted but it adds
(in part) to noise. Overall the S/N goes down with increasing break time.

And for the IMD I think the transient switch timing (waveform) is important.


> currently my schematic is only in my notebook .. and hopefully soon on a proto
> board
>
> its relatively simple. so maybe in a few days i will make a pdf version.

I don't think we have probs with a scanned version.


>
> A relatively simple but accurate closed loop control over the phase and
> amplititude error correction usable for both modulator / demodulator might
> be the holy grail for DC phasing rig designs.
>

Never seen such.


- Henry


--
ehydra.dyndns.info
5972 2011-03-31 01:12:37 Gian Re: high level mixers
Hi,

there are many switches available, particularly for audio and video but for low switching speed. These have very low Ron.

Fast switching switches are not too many. Today the best one is the FSA3157, being a break-before-make, but it is small to handle.

If you visit PA3AKE web site you will be able to see all the studies Martein did to select the best IC switch for the very high performance H-Mode Mixer. Also you will see the comparis
5973 2011-03-31 06:28:06 ehydra Re: high level mixers
I already know this interesting site.
BUT the problem of not having any suitable component for h-mode over
50MHz remains. Or do I miss something?
For the 100+ MHz range one need a diode bridge mixer or sampler. Or the
active variants as SA602 etc. Some FETs exists.
And then higher there are single or double schottky diode variants
(rats-race etc) mostly with delay-lines on the pcb.


- Henry


--
ehydra.dyndns.info


Gian schrieb:
> Hi,
>
> there are many switches available, particularly for audio and video but for low switching speed. These have very low Ron.
>
> Fast switching switches are not too many. Today the best one is the FSA3157, being a break-before-make, but it is small to handle.
>
> If you visit PA3AKE web site you will be able to see all the studies Martein did to select the best IC switch for the very high performance H-Mode Mixer. Also you will see the comparis
5976 2011-03-31 23:01:36 longjohn119 Re: high level mixers
5977 2011-04-01 01:12:28 Gian Re: high level mixers
Guten tag Henry,

5978 2011-04-01 01:35:10 Gian Re: high level mixers
Hi JR,

I am glad you are Ok with the answers and with your switches.

As you may have seen with PA3AKE measurements, you will have to compromise with the performances, particularly for coverage from 160m to 6m. The H-Mode mixer will be your best solution.

The QSD is critical to cover such a wide bandwidth, particularly using the FST3253 as you cannot adjust phase balance/unbalance. In caseyou like to experiment a QSD you should use the FST3125/6 with a quadrature generator with switching duration, for each phase, of 180°. I believe I posted, some time ago, a file in the softrock group.

Please keep the group posted
5991 2011-04-01 19:50:14 ehydra Re: high level mixers
Gian schrieb:
> Guten tag Henry,
>

Ciao Gian!


>
5994 2011-04-02 07:56:08 owe_frisk high level mixers
Hi and thanks for an interesting subject.
But how does these H-mode mixers behave in the VLF band?
I´m planning to build a VLF-converter de-luxe and include a H-mode mixer, input frequency around 10-150 kHz.
The transformers maybe have to be tailor-made to that low frequency but is it possible to use some of the switches Martein has testedin his very good work?
73
Owe, SA0ASO
5996 2011-04-02 09:34:51 w4zcb Re: high level mixers
Very very well Owe. Many years ago I built a complete rceiver for
0-500 KHz, and since noise down there precludes any need for
sensitivity, I used the old original SD5000 chip in H mode. 10 dB
conversion loss in that chip, and you can get as low as 4 dB using the
FST 3125/6. But it's not needed.

I basically up-converted to 8 MHz where I had some filters for it, and
used MCL T4:1's for the transformers, and still find an occasional use
for it. Interesting stuff down there.

W4ZCB

Hi and thanks for an interesting subject.
But how does these H-mode mixers behave in the VLF band?
I´m planning to build a VLF-converter de-luxe and include a H-mode
mixer, input frequency around 10-150 kHz.
The transformers maybe have to be tailor-made to that low frequency
but is it possible to use some of the switches Martein has testedin
his very good work?
73
Owe, SA0ASO
5998 2011-04-02 13:43:15 w4zcb Re: high level mixers
BTW Owe, I can't recommend more highly, the article on VLF in the
Winter 1991 issue of Communications Quarterly. (The most wonderful rag
since old Skip Tenney and Jim Fisk did Ham Radio. RIP guys.) Titled a
VLF-LF receiver, and authored by VK5BR, it's the cadillac for those
frequencies. Might be a shortcoming somewhere in the receiver, I never
built that part, preferring the "H" mode and up conversion. But his
"Antenna resonating device" can't be improved upon.

Figure 11 on page 67 of his front end tuner will resonate anything
from a 10 foot piece of wire to a 1000 foot beverage. I recently got
rid of a bunch of cup core inductors on this forum. I use 24 of those
in my tuner and can readily tune in stations below 10 KHz with my 160
meter inverted L for an antenna. At 10-50 KHz or so, the Q is so high
the tuner will actually ring on noise if you don't use some resistive
damping. Much better selectivity than the 8 pole crystal filter for
the IF up to about 200 KHz or so.

Regards

W4ZCB


Hi and thanks for an interesting subject.
But how does these H-mode mixers behave in the VLF band?
I´m planning to build a VLF-converter de-luxe and include a H-mode
mixer, input frequency around 10-150 kHz.
The transformers maybe have to be tailor-made to that low frequency
but is it possible to use some of the switches Martein has testedin
his very good work?
73
Owe, SA0ASO
5999 2011-04-02 18:30:54 kb1gmx Re: high level mixers
Hi all,

I've kind of leaned back and copied the mail.

To restate what I am looking at.

Transceive mixer from IF to RF:

* able to take 3dbm to as high as 7dbm inputs as a transmit mixer
* not require more than 10db more LO than input (50mW of LO is
painful!!)
* reasonable as RX mixer (7db loss works if quiet).
* Square wave drive friendly (LO may be SI570 which is 3.3V
square wave)
* bi-directional for RF and IF in the HF range
* frequencies.. 7, 18 and 21mhz RF, IF in the 9.6 to 12mhz range,
LO at RF+IF (minimally 16.6 to 33mhz).
* no exotic parts, is easily replicated.

Diode rings.. I've used the level-7(SBL1, TUF1, MD108) stuff for years, same for higher level (10, 17dbm) ones as well. They are known and predictable devices. Along these lines I've built level-13 mixers using 1n4148 matched and this does make a decent higher level mixer (about 3dbm input max). Tentec has in several designs used matched 1n4148s and binocular cores for mixer with good result (Century-21, 1208 20m to 6M transverter, many others).

Some of Chris Trask's work with alternate style diode mixers are interesting. But from my perspective while they have a few interesting features they are along the same lines as diode rings of the DBM type. The lower loss one is appealing for a no RF amp RX.

H-mode, These are getting a look but VHF requires specialized devices and none of those are applicable. For HF the driving circuits are not what I find attractive. The inverters and XOR logic used is generally worrysome to me as propagation delays is asymetric. I've used these in pure digital designs and even the fastest parts are clock rate limiting when combinational. However, if driven with a a 2F LO into a D-flop in /2 configuration that can be acceptable. I need to test as at below 30mhz this may not be too hard. Parts choice for the digital side is critical and the use of varied CMOS types(74AC, ACT, AH, C) does make a big difference as not all of those clock well at over 25mhz!

Switching mixers using FST3125 and 74hct4053(KD1JV) I find interesting but often look oversold in some configurations.
Providing the switching drive has been hit or miss in some
cases.

The Maas mixer as a balanced pair on paper looks good but the ratios of frequencies and port isolation is hard to model and evaluate. Seems to look better when the conversion RF to IF or baseband is greater than an octave. In my case with IF at 11-12mhz and signal at 7 or 18mhz it complicates the input and output match. I have to try this For a DC radio though.

Rohde did a single balanced bipolar design but I found the RF port had way too much LO. It even looks like a push pull amp for the LO.

FYI: Of the linear flavor devices I've used 1496, NE612, and formerly common as house flies CA3028 diff pair with good result if proper levels are adhered to. My 10M QRP rig uses both 1496 balanced modulator and CA3028 for the TX mixer. I've also used the CA3028 for TX mixers at 6M and 20M as well, it works. For this exercise they
do not qualify.

I need to proto and test a bit. The winner is one I like and has no relationship to anyone. Favorites, none as all the ones I'd go with
naturally like the 1n4148 ring does not like a 3.3V square wave for best results. However those designs thta do not like square waves
will work well with a level matched signal out of a low pass filter.

There has been some interesting work pointed out for review and possible testing. As they say keep it coming.


Allison/KB1GMX
6001 2011-04-02 18:52:42 ehydra Re: high level mixers
If you want test h-mode I repeat request for the diode-bridge switch.
Works 100+ MHz.

- Henry

--
ehydra.dyndns.info


kb1gmx schrieb:
> Hi all,
>
> I've kind of leaned back and copied the mail.
>
> To restate what I am looking at.
>
> Transceive mixer from IF to RF:
>
> * able to take 3dbm to as high as 7dbm inputs as a transmit mixer
> * not require more than 10db more LO than input (50mW of LO is
> painful!!)
> * reasonable as RX mixer (7db loss works if quiet).
> * Square wave drive friendly (LO may be SI570 which is 3.3V
> square wave)
> * bi-directional for RF and IF in the HF range
> * frequencies.. 7, 18 and 21mhz RF, IF in the 9.6 to 12mhz range,
> LO at RF+IF (minimally 16.6 to 33mhz).
> * no exotic parts, is easily replicated.
>
> Diode rings.. I've used the level-7(SBL1, TUF1, MD108) stuff for years, same for higher level (10, 17dbm) ones as well. They are known and predictable devices. Along these lines I've built level-13 mixers using 1n4148 matched and this does make a decent higher level mixer (about 3dbm input max). Tentec has in several designs used matched 1n4148s and binocular cores for mixer with good result (Century-21, 1208 20m to 6M transverter, many others).
>
> Some of Chris Trask's work with alternate style diode mixers are interesting. But from my perspective while they have a few interesting features they are along the same lines as diode rings of the DBM type. The lower loss one is appealing for a no RF amp RX.
>
> H-mode, These are getting a look but VHF requires specialized devices and none of those are applicable. For HF the driving circuits are not what I find attractive. The inverters and XOR logic used is generally worrysome to me as propagation delays is asymetric. I've used these in pure digital designs and even the fastest parts are clock rate limiting when combinational. However, if driven with a a 2F LO into a D-flop in /2 configuration that can be acceptable. I need to test as at below 30mhz this may not be too hard. Parts choice for the digital side is critical and the use of varied CMOS types(74AC, ACT, AH, C) does make a big difference as not all of those clock well at over 25mhz!
>
> Switching mixers using FST3125 and 74hct4053(KD1JV) I find interesting but often look oversold in some configurations.
> Providing the switching drive has been hit or miss in some
> cases.
>
> The Maas mixer as a balanced pair on paper looks good but the ratios of frequencies and port isolation is hard to model and evaluate. Seems to look better when the conversion RF to IF or baseband is greater than an octave. In my case with IF at 11-12mhz and signal at 7 or 18mhz it complicates the input and output match. I have to try this For a DC radio though.
>
> Rohde did a single balanced bipolar design but I found the RF port had way too much LO. It even looks like a push pull amp for the LO.
>
> FYI: Of the linear flavor devices I've used 1496, NE612, and formerly common as house flies CA3028 diff pair with good result if proper levels are adhered to. My 10M QRP rig uses both 1496 balanced modulator and CA3028 for the TX mixer. I've also used the CA3028 for TX mixers at 6M and 20M as well, it works. For this exercise they
> do not qualify.
>
> I need to proto and test a bit. The winner is one I like and has no relationship to anyone. Favorites, none as all the ones I'd go with
> naturally like the 1n4148 ring does not like a 3.3V square wave for best results. However those designs thta do not like square waves
> will work well with a level matched signal out of a low pass filter.
>
> There has been some interesting work pointed out for review and possible testing. As they say keep it coming.
>
>
> Allison/KB1GMX
6002 2011-04-02 19:22:45 w4zcb Re: high level mixers
Allison, I've liked what you've had to say many times in the past, but
you totally confuse me at this point. You write your specs as both I
and O being in the HF range, and then worry about VHF requiring
specialized devices.

Then remark that some D flops don't trigger well above 25 MHz. I'd
agree that there may be one that doesn't, but I've never found a
74AC74 that wouldn't trigger with -10 dBm of drive up to a bit over
150 MHz. With +10, most will still trigger around 175 MHz before they
give up.

The buss switch as previously mentioned can't be beat or even
approached at your frequencies of interest with ANYTHING else, and if
you buy just one from Mouser, it'll cost you under $0.50 not counting
postage. A tad less expensive than an MCL diode mixer, and I dare say,
costs appreciably less than building your own diode mixer out of
discrete components. It's conversion loss including the diplexer
losses will be well under 5 dB.

IF you haven't already tried one, I have a proposition for you. I have
several of the logic circuits shown in figure 6.92 of EMRFD that
aren't doing anything but sitting in a box in my shop. (I have
several more that are in service ). These are sandwiched with a second
board, comprising the circuit of figure 6.90 of EMRFD. The diplexers I
have in these are likely not your choice of IF, you specify your
choice and if I don't already have one that covers your needs, I'll
modify the diplexer to cover your specified IF. (If you're not choosy,
I have a couple with 5.2 MHz diplexers installed)

Pipe in your SI 570 to the LO input, (2X frequency, but most of the
570 boards have provisions for division of the display by 2 or 4, a
Signal generator to the RF in and something to detect the IF out and
you're ready to go.

Dimensions are 2-7/8 x 1-3/4 x 1 inch high, but there's a lot of air
in the boards due to using DIP chips. If you need tiny, rework the
design for SOIC or some such.

Give me a snail mail address to send to, and I'll give you one, all
you have to do is publish your results on this forum after your
independent evaluation.

Regards
W4ZCB



Hi all,

I've kind of leaned back and copied the mail.

To restate what I am looking at.

Transceive mixer from IF to RF:

* able to take 3dbm to as high as 7dbm inputs as a transmit mixer
* not require more than 10db more LO than input (50mW of LO is
painful!!)
* reasonable as RX mixer (7db loss works if quiet).
* Square wave drive friendly (LO may be SI570 which is 3.3V
square wave)
* bi-directional for RF and IF in the HF range
* frequencies.. 7, 18 and 21mhz RF, IF in the 9.6 to 12mhz range,
LO at RF+IF (minimally 16.6 to 33mhz).
* no exotic parts, is easily replicated.

Diode rings.. I've used the level-7(SBL1, TUF1, MD108) stuff for
years, same for higher level (10, 17dbm) ones as well. They are known
and predictable devices. Along these lines I've built level-13 mixers
using 1n4148 matched and this does make a decent higher level mixer
(about 3dbm input max). Tentec has in several designs used matched
1n4148s and binocular cores for mixer with good result (Century-21,
1208 20m to 6M transverter, many others).

Some of Chris Trask's work with alternate style diode mixers are
interesting. But from my perspective while they have a few interesting
features they are along the same lines as diode rings of the DBM type.
The lower loss one is appealing for a no RF amp RX.

H-mode, These are getting a look but VHF requires specialized devices
and none of those are applicable. For HF the driving circuits are not
what I find attractive. The inverters and XOR logic used is generally
worrysome to me as propagation delays is asymetric. I've used these in
pure digital designs and even the fastest parts are clock rate
limiting when combinational. However, if driven with a a 2F LO into a
D-flop in /2 configuration that can be acceptable. I need to test as
at below 30mhz this may not be too hard. Parts choice for the digital
side is critical and the use of varied CMOS types(74AC, ACT, AH, C)
does make a big difference as not all of those clock well at over
25mhz!

Switching mixers using FST3125 and 74hct4053(KD1JV) I find interesting
but often look oversold in some configurations.
Providing the switching drive has been hit or miss in some
cases.

The Maas mixer as a balanced pair on paper looks good but the ratios
of frequencies and port isolation is hard to model and evaluate.
Seems to look better when the conversion RF to IF or baseband is
greater than an octave. In my case with IF at 11-12mhz and signal at 7
or 18mhz it complicates the input and output match. I have to try
this For a DC radio though.

Rohde did a single balanced bipolar design but I found the RF port had
way too much LO. It even looks like a push pull amp for the LO.

FYI: Of the linear flavor devices I've used 1496, NE612, and formerly
common as house flies CA3028 diff pair with good result if proper
levels are adhered to. My 10M QRP rig uses both 1496 balanced
modulator and CA3028 for the TX mixer. I've also used the CA3028 for
TX mixers at 6M and 20M as well, it works. For this exercise they
do not qualify.

I need to proto and test a bit. The winner is one I like and has no
relationship to anyone. Favorites, none as all the ones I'd go with
naturally like the 1n4148 ring does not like a 3.3V square wave for
best results. However those designs thta do not like square waves
will work well with a level matched signal out of a low pass filter.

There has been some interesting work pointed out for review and
possible testing. As they say keep it coming.


Allison/KB1GMX
6003 2011-04-02 20:02:19 dave Re: high level mixers
Allison:
In my UHFSDR, I use ADE-2ASK mixers(a level 7 DBM). I derive I/Q signals
from a LVPECL Si570 driving a pair of SY10EP52 LVPECL dividers that are rated to
4 GHz. I drive both sides of the LO input transformer with the 2 LVPECL outputs.

Dave - WB6DHW
<http://wb6dhw.com>

6009 2011-04-03 12:57:25 kb1gmx Re: high level mixers
6010 2011-04-03 13:06:04 kb1gmx Re: high level mixers
6011 2011-04-03 13:38:59 dave Re: high level mixers
Allison:
I have not looked at 3-7dBm at the RF input. When in transmit mode, it gets
fed audio at 1 to 2 volts P-P. I am deriving the I/Q digitally from 1750 KHz up
to 700 MHz (limited by the Si570) using the 2 LVPECL flip flops. With a CMOS
Si570, you would have to get a different flip flop as the 10E52 requires a
differential input.

Dave - WB6DHW
<http://wb6dhw.com>

6012 2011-04-03 13:52:55 victor Re: high level mixers
Chris,
In your split ring article you say that the split ring mixer has 3.5dB insertion loss.
I guess you know that the minimal attenuation of any passive double balanced mixer of any kind is 3.92dB , when using ideal switches instead of diodes.
This insertion loss can be minimized only by using frequency selective circuits that reflect unwanted mixing products back into the mixer to be re-mixed and increase the signal of the wanted conversion, but this solution is narrow band.
Victor - 4Z4ME
6013 2011-04-03 13:53:08 kb1gmx Re: high level mixers
Sounds all wrong. ADE2 is only level 7 and if you are hitting it with 2V (even 1V) across 50 ohms thats a lot of input and I can't imaging it's going to be clean unless there is attenuation on the audio. I'd expect the audio drive to be under 100millivolts at the IF pin.

I'm not doing SDR, the IF is in the mid hf (11mhz or so).

Allison


6015 2011-04-03 15:07:37 Chris Trask Re: high level mixers
>
> In your split ring article you say that the split ring mixer has 3.5dB
> insertion loss. I guess you know that the minimal attenuation of any
> passive double balanced mixer of any kind is 3.92dB , when using ideal
> switches instead of diodes.
>

Yes, I'm aware of the 3.92dB minimum loss. It goes all the way back to
Tucker in 1953, which he refined in a 1971 paper on the use of square-wave
LO signals:

LOSSmin = -20*Log(pi/2)

so I wonder if it is related to the use of square-wave LO signals, which has
additional conversion products at harmonics of the LO, all of which have to
add up to unity in an ideal mixer. This is the exact same number for the NF
penalty in a communtating (aka Gilbert Cell) mixer caused by the harmonics
of the LO folding the Nyquist noise of the driver device down to the IF
frequency.

I was surprised when I saw the conversion loss in the earlier
measurements, and when I go over everything again I'm going to get out the
micrometer. An HP141T at 10dB/division vertical resolution leaves a bit to
be desired. Using the 2dB/division resolution takes a lot of patience, and
at the time I was making a large number of measurements on almost a dozen
different mixer circuits. Now, the choice of circuits is narrowed down to
just one, but there's a good deal of experimentation to do with regard to
the LO source resistance issue.

>
> This insertion loss can be minimized only by using frequency selective
> circuits that reflect unwanted mixing products back into the mixer to
> be re-mixed and increase the signal of the wanted conversion, but this
> solution is narrow band.
>

Which happens to be similar to the mechanism (reflecting unwanted [and
wanted] mixing products back into the mixer) that causes rapid degredation
of IMD due to mismatched loads (and sources) in passive DBMs. There must be
some phasing involved in that approach.

I would be hesitent to say that it's the "only" method available. I
would, however, accept that it is a practical solution for minimizing the
conversion loss of DRMs.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6017 2011-04-03 17:19:07 ehydra Re: high level mixers
Chris Trask schrieb:
>> In your split ring article you say that the split ring mixer has 3.5dB
>> insertion loss. I guess you know that the minimal attenuation of any
>> passive double balanced mixer of any kind is 3.92dB , when using ideal
>> switches instead of diodes.
>>
>
> Yes, I'm aware of the 3.92dB minimum loss. It goes all the way back to
> Tucker in 1953, which he refined in a 1971 paper on the use of square-wave
> LO signals:
>
> LOSSmin = -20*Log(pi/2)
>
> so I wonder if it is related to the use of square-wave LO signals, which has
> additional conversion products at harmonics of the LO, all of which have to
> add up to unity in an ideal mixer. This is the exact same number for the NF
> penalty in a communtating (aka Gilbert Cell) mixer caused by the harmonics
> of the LO folding the Nyquist noise of the driver device down to the IF
> frequency.

It looks connected to the S/N of a 1-bit digitizer. There it is related
to a sine or say almost sine signal.


- Henry


--
ehydra.dyndns.info
6023 2011-04-04 05:54:01 Tim Re: high level mixers
6031 2011-04-04 12:24:47 victor Re: high level mixers
Check the results at a number of frequencies because it might be that is the measurement circuit has reflections, it can decrease the insertion loss at one frequency and increase it
6034 2011-04-04 13:15:11 Lasse Question about W1FB AGC design
Can anyone help me to understand how the venerable "An AGC system for
CA3028A" works??
I see it in my Handbook from 1977 page 267 and it is still in my 1989
Handbook at page 12-30.

IF is fed into the gate of a MPF102, and a PNP (2N4126) transistor has
its base to the drain and these two is fed to Vcc through a RFC. The
emitter of the PNP is coupled via a 100 ohm resistor to the Vcc end of
the RFC.

So, as far as I can see it, there will be no current flowing through the
PNP as the base will always have higher potential, and the PNP will
always be shut off.
I must oversee something... please enlighten me!

Why did I bring this up? I found I had made some notes in my old
Handbook about this circuit, must have been 30 years ago, and I was
looking for information about Cohn bandpass filters :)

73 Lasse SM5GLC
6035 2011-04-04 13:15:58 Chris Trask Re: high level mixers
>
> Check the results at a number of frequencies because it might be that is
> the measurement circuit has reflections, it can decrease the insertion
> loss at one frequency and increase it on another by "re-use" of harmonic
> mixing products and re-mixing them.
>

I looked into the 3.92dB minimum loss issue in depth, and found that it
is the minimum loss for a ring mixer with ideal diodes and a perfect square
wave LO. If it was a perfect sine wave LO, the minimum loss would be
3.02dB. The difference is that the square wave LO has odd integral harmonic
components that convert the RF into a number of IF frequencies, and the
power converted to the unwanted products is subtracted from the desired Frf
+/- Flo products.

So, with ideal diodes the conversion loss can vary from 3.02dB (perfect
sine wave LO) to 3.92dB (perfect square wave LO), and everywhere in between
as you progressively saturate the diodes and clip a sine wave LO. But, in
all the testing I did earlier the only noticable difference I saw was that
the OIP3 was much better for a square wave LO.

I've added a few paragraphs and equations about this to the theory
portion of the paper since yesterday, and this morning I carefully tested an
SRM (split ring mixer) to see if the reverse voltage across the diodes was
higher than the forward voltage, since each diode pair has a separate LO
balun. To no surprise, the two voltages were equal, but it did need to be
looked into as a probable cause. So, I'm back to looking at the LO source
impedance as being the reason for the lower conversion loss.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6036 2011-04-04 13:41:09 victor Re: high level mixers
Chris,
I guess that mean that for a Sine wave LO case you get at the mixer output only the IF and its "output image" (Frf +/- Flo), so each has half the power (or -3dB).
However I can't imagine ideal diodes that can do that. in the case of diodes using a sine signal for LO, the LO signal will modulate the resistance of the diodes in order to get "clean" multiplicati
6037 2011-04-04 14:28:24 Chris Trask Re: high level mixers
>
> Chris,
> I guess that means that for a Sine wave LO case you get at the mixer
output
> only the IF and its "output image" (Frf +/- Flo), so each has half the
power
> (or -3dB).
>

Yes. I'll be uploading the revised paper with that section added
shortly.

>
> However I can't imagine ideal diodes that can do that. in the case of
diodes
> using a sine signal for LO, the LO signal will modulate the resistance of
the
> diodes in order to get "clean" multiplication between the LO sine signal
and
> the RF sine signal. In such a case the diodes will contribute significant
> additional resistive loss that you can not disregard. Maybe you invent a
> lossless varactors mixer?
>

Ideal diodes are available at your local Unobtanium department store,
right next to the noiseless bipolar transistors. :{b

I'm just waiting for someone to suggest using tunnel diodes in series
with the Schottky diodes.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6041 2011-04-04 15:14:15 ehydra Re: high level mixers
Why not test it with cheap DIAC?
It is not mentioned as joke. Sometimes one don't oversee all
consequences and can have a interesting observation.
Most invention are made by accident zhen and careful exmination what was
going on.

- Henry


--
ehydra.dyndns.info




Chris Trask schrieb:
> I'm just waiting for someone to suggest using tunnel diodes in series
> with the Schottky diodes.
>
6043 2011-04-04 15:59:13 kb1gmx Re: high level mixers
6045 2011-04-04 17:49:06 kb1gmx Re: Question about W1FB AGC design
6046 2011-04-05 09:58:35 Lasse Re: Question about W1FB AGC design
kb1gmx skrev 2011-04-05 02:48:
>
>
6047 2011-04-05 13:38:38 Tim Re: high level mixers
Also on the subject, in the "Files" section of this group, there's a writeup by G4COL describing a dual-CA3046 Gilber cell mixer at 17mA tail current with 20-25 dBm IIP3, and a MC1496 pushed past its specced current to run at 11mA tail current and 20ish dBm IIP3.
6048 2011-04-05 17:05:07 Chris Trask Re: high level mixers
>
> Also on the subject, in the "Files" section of this group, there's a
> writeup by G4COL describing a dual-CA3046 Gilber cell mixer at 17mA
> tail current with 20-25 dBm IIP3, and a MC1496 pushed past its specced
> current to run at 11mA tail current and 20ish dBm IIP3.
>

In one of my many efforts to make a better mixer, I tried using the
CA3046 as a translinear analogue multiplier. It didn't work very well,
despite almost three months of tinkering. It does, however, make a fairly
nice lossless feedback mixer that minimizes the component count.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6060 2011-04-06 15:25:39 kb1gmx Re: high level mixers
6062 2011-04-06 22:30:52 victor Re: high level mixers
You mean 4007. I have used it too as a low IF amplifier.
Victor - 4Z4ME

6064 2011-04-07 00:41:53 ehydra Re: high level mixers
Hi Victor -

Is this somewhere documentated? Looks interesting.

- Henry


victor schrieb:
> You mean 4007. I have used it too as a low IF amplifier.
> Victor - 4Z4ME
>
>
6066 2011-04-07 03:05:09 victor Re: high level mixers
I'm talking about a circuit that I have used almost 30 years ago.
I remember it was replaced by a cascade of the 6 inverters in a 74hcu04 (The "u" means unbuffered single stage inverters), each having a feedback resistor to linearize the inverter and a series resistor from amplifier input to inverter input. This way it's an inverting amplifier with G= Rfeedback/Rinput, if gain is low. It was designed to have broadband band pass response by scaling the resistors and the coupling capacitors. The bias of all the 6 inverters was done with a single resistor (decoupled by a large capacitor) from the IC VDD pin to the +5V power supply.
I hope it is clear enough.
Victor - 4Z4ME

6071 2011-04-07 12:23:23 ehydra Re: high level mixers
Hi Victor -

Thank you for the details!

cheers -
Henry


victor schrieb:
> I'm talking about a circuit that I have used almost 30 years ago.
> I remember it was replaced by a cascade of the 6 inverters in a 74hcu04 (The "u" means unbuffered single stage inverters), each having a feedback resistor to linearize the inverter and a series resistor from amplifier input to inverter input. This way it's an inverting amplifier with G= Rfeedback/Rinput, if gain is low. It was designed to have broadband band pass response by scaling the resistors and the coupling capacitors. The bias of all the 6 inverters was done with a single resistor (decoupled by a large capacitor) from the IC VDD pin to the +5V power supply.
> I hope it is clear enough.
> Victor - 4Z4ME
>
>
6073 2011-04-07 15:17:27 Gian high level mixers
Hi Chris and all,

I have read you comments about my last message. Sorry for showing up later but I had other priorities.

I feel I did upset you but it was not my intention on "advertising the G3SBI H-Mode Mixer". I felt to repeat some story as some members seemed not well informed.

I normally try to learn and this group is one of the best for amateur radio technologies and homebrewing. As I feel interesting what you pablished about diode mixers, I spent some time fiddling with the circuit you reported in your paper posted in the EMRFD files.
Please forgive me if I did the trial of the Diode H-Mode Mixer and the Split-Diode DB Mixer modifying your configurations from a 3 transformer to a 2 transformer, something like I did for the G3SBI H-Mode Mixer.

I do not recall any EMRFD group member that has commented on assembled ciorcuit and its performances.

I have posted my report, to be taken as informative of my trials, in Files under I7SWX-Experim-Diode-HMode&SplitMixers. I hope other people will test your Split-Diode Mixer and see if the results are repeatable or can be better than my measurements.
I Have done CL test from 1.75 to 220 MHz and a Dual Tone IMD measurement on 10MHz.

My report shows your diagrams and my 2T simplifications. I could not do LO driving with square wave signal, only sinewave.
I have copied the Spectrum analyzer screen copy of the 2T IMD measurement and Tables with Conversion losses at different bands.

Best 73

Gian
I7SWX