Terminating HC4066 Mixer
J M Noeding
level, and it may simplify some of my applications. Several different
applications have been found, but some points remain unclear.
Some application uses 50 ohm on the RF/IF ports, while some say
200-300 ohm or higher is better compromise based on series resistance
and possible supply voltage. Is it really necessary to have any
defined load at all on the output? Could it just be an LC-circuit of
which impedance can vary?
Some applications have DC-termination on IF port, but I don't
understand the need as long as it is properly DC-biassed on the RF
port.
An interesting application is found on
http://home.snafu.de/graff/sbt40.pdf it seems that the IF termination
is 400 ohm for the AC component, but it is no DC-termination. The
solution to avoid input transformer or link also seems to be a good
idea.
73,
Jan-Martin
LA8AK
--
remove ,xnd to reply
Steve Ratzlaff
I have done rather extensive tests with both the HC4066 and the HC4053
used as a mixer. These tests include checking mixer output (IF) IMD at
various impedances. Best IMD is achieved with both IC's when all ports
are terminated in 50 ohms, exactly like the passive DBM requires. My
particular application was for an LF to HF upconverter. I initially
tried the "standard" diplexer at the mixer output, but further tests
showed I could not achieve a flat IF response--the diplexer had a
narrower bandpass than the 500kHz I wanted. I ended up terminating the
mixer with a simple wideband one-transistor amp, with 50 ohm in/out
impedances, and adjusting its gain to give an overall mixer conversion
loss of 0 dB.
The mixer input was fed with a wideband ferrite trifilar transformer;
the oscillator port from an HC04 oscillator output (not necessarily 50
ohms).
Performance of this upconverter is good.
Steve AA7U
Steve Ratzlaff
by a +7dBm LO input, not a high-level DBM.
Steve
Richard Hosking
20-30 ohms, depending on supply volts etc. They could be regarded as a
resistor in the circuit consisting of source impedance/mixer/load impedance
at the LO frequency. Thus their response will depend very much on load
impedance. For example, using the Tayloe setup with a large capacitor as the
mixer load (ie an audio low pass filter) the mixer acts like a very sharp
bandpass filter. The mixer presents a relatively high impedance to the RF
input at LO frequency (and harmonics thereof) and low impedance at other
frequencies. It would presumably look like a switched capacitor filter
(which it is) at audio.
Rod VK6KRG and I has done some experiments with the Tayloe circuit (ie 4 of
these mixers switched by a quadrature commutated LO). If the 4 phases are
mixed back up again to RF with a second set of quadrature mixers, the result
is a filtered version of the input with a *very* sharp bandpass - -20 dB at
+\- 50 KHz, regardless of LO frequency. These results could be achieved at
20 MHz. (and presumably higher with the right mixer) There were opposite
sideband effects but this could presumably be improved with good circuit
balance. We made no attempt to get ideal balance.
I would be interested to know if anyone has done a rigorous analysis of this
circuit using a Spice package or in real life
Richard
"Steve Ratzlaff" <srat...@flash.net> wrote in message
news:3CDB32E5...@flash.net...
J M Noeding
wrote:
>Hi,
>I have done rather extensive tests with both the HC4066 and the HC4053
>used as a mixer. These tests include checking mixer output (IF) IMD at
>various impedances. Best IMD is achieved with both IC's when all ports
>are terminated in 50 ohms, exactly like the passive DBM requires. My
>particular application was for an LF to HF upconverter. I initially
>tried the "standard" diplexer at the mixer output, but further tests
>showed I could not achieve a flat IF response--the diplexer had a
>narrower bandpass than the 500kHz I wanted. I ended up terminating the
>mixer with a simple wideband one-transistor amp, with 50 ohm in/out
>impedances, and adjusting its gain to give an overall mixer conversion
>loss of 0 dB.
>The mixer input was fed with a wideband ferrite trifilar transformer;
>the oscillator port from an HC04 oscillator output (not necessarily 50
>ohms).
>Performance of this upconverter is good.
>Perhaps I should add, the IC mixer was comparable to a passive DBM fed
>by a +7dBm LO input, not a high-level DBM.
>Steve AA7U
>
thanks for the reply.
Want to use Philips or Motorola HC4066 on 12V supply. The maximum
signal level is 4.28V RMS. Using 200 ohm impedance the power level is
max 92mW or +19.6dBm, 6dB higher with 50 ohm RF/IF termination, and
still 16.6dBm with 400 ohm terminations.
My idea is to use a grounded gate amplifier with MPF105 or BF245B.
In the application as mixer from 136kHz to 60kHz I don't really see
that such amplifier could cause termination problems provided the
mixer sees 200 ohm for all mixer products, HF and VHF would be
different.
Understood that you ended up without diplexer and the 50 ohm amplifier
is capacitively coupled to the mixer output(?). How important is
DC-return from mixer output?
wonder if somebody has tried 2/3 4053 instead of 4/4 4066?
Steve Ratzlaff
My experiments were with 5 and 6volts, VCC on the IC; I didn't try
12volts. Performance was somewhat improved (better IMD) with 6 volts, so
I would assume significantly better at 12VCC. That is a huge max input
signal you're expecting. A high-level DBM (+23dBM LO) would be straining
to stay linear with such a large input. I would guess this IC mixer will
not be linear with such a large input, but perhaps your application is
not as a receiving mixer but as a transmitting mixer, where strict
linearity may not need to be as desirable.
The following IF amp has to be capacitively coupled, in order not to
short out the transistor's bias.
The IC mixer must also see a DC return on the mixer output. It's biassed
at 1/2 VCC. I used 10kohm to the mixer output, for isolation, as the
following IF amp provides the 50 ohm termination for the mixer.
(The complete circuit can be seen in the article in April 2002 QST
magazine.)
My experiments used both the HC4066 and 4053, with identical results.
Steve
Rodney Green
years or so, and I have found them to be very insensitive to termination
impedance. I developed a mixer using 74hc4053 called The Ned Kelly mixer
(Ned kelly was an armour wearing bullet proof bush ranger in the 1800s) This
was published in Electronics World Jan 1999 P59.
I aklso have a new receiver architecture using these mixers in a number of
variations. It is called The Bedford Receiver and was published by the ARRL
in QEX magazine Sept/ OCT 1999. These articles could prove useful
Regards Rod green of Bedford.
J M Noeding
tnx.
this is what I expected, but it seems often to be lot of experience
presented by hams which is not based in any facts, so you may wonder
what is the secret in 50 ohm terminations. In my opinion 200 ohm is
better for RF, and for my purpose - on 136kHz - I don't see the need
to use a diplexer for proper termination on 100MHz or above
So I am pleased to learn about your experience regarding termination.
A friend who is experimenting with this says he cannot note any
intermodulation products provided the peak-to peak signal is lower
than the supply voltage, so it seems difficult to overload the mixer.,
it would mean a signal level of +15dBm on the antenna, but no signals
below few MHz has be found to be stronger than -50dBm on the active
antenna used for LF
73
Jan-Martin, LA8AK
on the Rubber Coast
Jim Pennell
tradeoffs.
I would make sure to match the output of the Antenna to the same
impedance. The reasoning is, the ON resistance of a 4066 is quite a few
ohms, and so the insertion loss in a 200 ohm system will be lower.
You do lose a little of the maximum input level, due to the impedance
transform raising the RF input voltage.
As for terminating the LO frequencies at the mixer output with a diplexer,
I wouldn't worry too much about it. The internal isolation of the 4066 is
good enough that the amount of LO energy on the output can reflect back into
the switch and not degrade the performance very much at your frequencies.
----------------------
So, the tradeoff choice is, lower conversion loss/lower intermod point
with 200 ohms or higher insertion loss/higher intermod point with 50 ohms.
You can more or less predict the distortion point, since the RF will clip
as you approach VCC. Obviously the switch has to be DC biased to half of
VCC for maximum dynamic range.
I'd supply DC on the output side of the mixer, and AC couple the input.
The input will go to 1/2 VCC as the switches turn on. Leaving the input
floating to DC will minimize the tenancy for the DC level to change as the
switches turn on and off.
Given the noise level at the VLF frequencies you want to use, I'd be
inclined to go with 50 ohms since you can afford the conversion loss....
Jim Pennell
N6BIU
Markus Wolfgart
as far as I have not enough expirance with a 4066 mixer, but
what about the adg711-712-713 ic from AD.
The ON resistance is between 4 to 8 ohm depending on the
supply voltage.
The ic is much more expansiv, but is suitable up to 200MHz.
Would be great to find someone to make some test with it.
Markus
DL8MBY
Harold E. Johnson
> as far as I have not enough expirance with a 4066 mixer, but
> what about the adg711-712-713 ic from AD.
> The ON resistance is between 4 to 8 ohm depending on the
> supply voltage.
> The ic is much more expansiv, but is suitable up to 200MHz.
> Would be great to find someone to make some test with it.
The FST 3125 from Fairchild, Texas Instruments, Perricom and others exhibits
under 4 ohms of Rdss, works to somewhere well above 30 MHz, has the switch
drivers on board, and an Ip3 of above +42 dBm, (With square wave LO) . It
costs under a buck in single quantities from DigiKey. (TI is $1.02). I've
measured under 5 dB conversion loss in the "H" mode configuration with both
MCL TT4:1A transformers at 200 ohms and MCL T1:1T's at 50 ohms, both at 20
meters. Slightly more using T4:1's. It all gets better (But not a lot) when
you raise Vdd from 5 volts to 7 volts, max ratings. Available in both SSOP
and TSSOP.
W4ZCB
Markus Wolfgart
is such a switch ic suitable for qrp (means power consumption)
and also suitable for direct conversion to audio by a pair of
this ic (I and Q chanal) to avoid the sideband filter?
Have anybody made experience with this technique?
Would be great to get some hints what to pay attention for when
designing such a rx.
Markus
DL8MBY
"Harold E. Johnson" schrieb:
Harold E. Johnson
"Markus Wolfgart"
> Hallo Harold and the world,
>
> is such a switch ic suitable for qrp (means power consumption)
> and also suitable for direct conversion to audio by a pair of
> this ic (I and Q chanal) to avoid the sideband filter?
> Have anybody made experience with this technique?
I don't know the power consumption. Not being green or into QRP, I've never
measured it. My board layout includes 3 digital 74AC IC's in a variable
divider chain and all 4 chips draw 60-100 mA at 5 volts, depending on LO
frequency. (110 mA at 150 MHz, about the upper limit of AC logic). I use the
3125 for RF mixing to an RF IF frequency. I have used the FST 3253 (Only
takes one for both I and Q) in my construct of Rod Greens Bedford receiver.
As far as either of us know, it was the first one built. (After his, and his
great article in QEX) Works fine. I use the Burr Brown INA 103 op-amp
amplifiers post detection ahead of the phasing networks.
>
> Would be great to get some hints what to pay attention for when
> designing such a rx.
Layout Symmetry!!! And square wave LO. And a steady hand if using the TSSOP
package with it's 0.025 leads!
W4ZCB
>
J M Noeding
now I've built the new converter using 74HC4066 with 10V supply. Use a
TIS88 as buffer to 60kHz two stage BPF (4.5mH coils) and S042P as
mixer up to 3627kHz to be fed into Yaesu FT-902. An old 4069UBE is
used as buffer for the 195-198kHz VFO
Lowest CW signal level easily tuned is -120dBm (75 ohm) and lowest
signal level heard when you know it is there is -126dBm, so the mixer
seem reasonable well without RF stage.
The idea for input circuit (although it seems to be somewhat wrong as
it was used on 7MHz) is found at http://home.snafu.de/graff/sbt40.pdf,
so I use 1.5mH high-Q coils with capacitor values for the 2nd circuit
are 1137pF, 10nF + 10nF (no trimmer capacitor).
A xtal filter described by SM5BSZ was planned, but I didn't succeed
building it for 60kHz, so in the meantime LC-circuit is used, I hope
to visit some friends in the summer holiday to get some help in
trimming it, it isn't so easy when everybody who could help is at
least 200 miles away or in another country, hi.
73
Jan-Martin
LA8AK
--
remove ,xnd to reply
Markus Wolfgart
Would like to make my own experience, but by the way
did you ever try a DDS IC for the LO for example
the ad9854 (10bit DAC). This chip provide tow outputs,
one for I and one for Q channel.
I was told, that it's not so easy to make the low pass
filter following the DDS IC to reduce the frequencies
above clock/2 when taking the phase into account especially
when trying to construct a I&Q-DC-Rx. But this is another story.
Markus
DL8MBY
--
=============================================
Markus Wolfgart
DLR
=============================================
PS.: remove the xx_ from email adr. to replay
=============================================
"Harold E. Johnson" schrieb:
Harold E. Johnson
> Would like to make my own experience, but by the way
> did you ever try a DDS IC for the LO for example
> the ad9854 (10bit DAC). This chip provide tow outputs,
> one for I and one for Q channel.
> I was told, that it's not so easy to make the low pass
> filter following the DDS IC to reduce the frequencies
> above clock/2 when taking the phase into account especially
> when trying to construct a I&Q-DC-Rx. But this is another story.
>
> Markus
I've used a pair of 9851's (Without the 6X PLL, too much jitter in that for
the application) in a VNA. Fine for that purpose, but I wouldn't want to use
a straight DDS for a LO without PLLing it to a VCO to kill the spurs. I sure
wouldn't want my neighbor at the foot of the hill to be using one at any
power level above a watt or two, his phase noise at a couple hundred KHz is
bad enough now.
You can make a perfect quadrature output with a pair of 7474's, but it cuts
the frequency by four. If you can live with that, it doesn't do a lot for
the spurs, but cuts any LO phase noise by 12 dB. You get automatic high
level (~+17) dBm square wave LO out of them as well.
W4ZCB
Markus Wolfgart
I have read some articles in the German ham magazine Funkamateur
about this concept with the pair of 2xD-FF (74x74) for generating
clock for a 74HCT4066 Mixer for a I and Q channel DC-RX.
I have made some scanned copies, if you are keen on it, I can
mail them as a attachment to you.
In the case someone use a DDS like the AD9854/9857 for the switching
of the 4066 he avoid it to make different LO's with four times higher
frequencies for the different sw-bands.
When using the ad9854/57 you get a better phase noise (about 80 dB SFDR)
because of the 12/14bit DAC.
When someone avoid to use the internal multiplier of the DDS chip like
you
mentioned before and take a stable oscillator for clocking the DDS he
could
achieve a good performance over a wide range of frequencies.
I have not tried it before but this will be may next fovored project in
the future.
73 & gd dx
Markus
DL8MBY
J M Noeding
wrote:
>Hi,
>My experiments were with 5 and 6volts, VCC on the IC; I didn't try
>12volts. Performance was somewhat improved (better IMD) with 6 volts, so
>I would assume significantly better at 12VCC. That is a huge max input
>signal you're expecting. A high-level DBM (+23dBM LO) would be straining
>to stay linear with such a large input. I would guess this IC mixer will
>not be linear with such a large input, but perhaps your application is
>not as a receiving mixer but as a transmitting mixer, where strict
>linearity may not need to be as desirable.
in my circuit CD4069UBE draws only 10mA. I wasn't sure how much
voltage was safe to use with such inverter in linear mode, so I
reduced the voltage with a 220 ohm resistor from 12V. Suppose it is
the first limiter which draw current and the next two inverter stages
of CD4069UBEand HC4066 draw very little current, so I really don't
understand the 23dBm LO level you mention.. LO frequency is 196kHz
73,
Dan Tayloe
quadrature detectors. A pair of 9851's draw much less power than a
single 9854, and it allowed me to run at 1x on both 9851s, but still
generate the two bit clocking sequence that you are discussing with a
7474.
However, my experience was that the 9851's generated a fair number of
very narrow birdie spurs. I was stepping the 9851's at a 0.014 Hz
resolution for continuos tuning. If a larger step size like 25 or 50 Hz
were used, most of these birdies would be skipped right over.
- Dan Tayloe, N7VE; Phoenix, Az; Az ScQRPions
John Miles
>
> I used a pair of 9851s phased 90 degrees apart to run one of my DC
> quadrature detectors. A pair of 9851's draw much less power than a
> single 9854, and it allowed me to run at 1x on both 9851s, but still
> generate the two bit clocking sequence that you are discussing with a
> 7474.
>
> However, my experience was that the 9851's generated a fair number of
> very narrow birdie spurs. I was stepping the 9851's at a 0.014 Hz
> resolution for continuos tuning. If a larger step size like 25 or 50 Hz
> were used, most of these birdies would be skipped right over.
>
Those 10-bit DDS parts are junk (IMNSHO). Fine for a BFO, but I
wouldn't use them for anything else.... certainly not to drive any mixer
that doesn't have a sharp filter in front of it.
The 9852/9854 generation is much better, and they're not really that
much harder to use.
-- jm
------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------
Dan Tayloe
seem to gain it very much, at least from the advertised specifications.
Narrow band spurs are 80 to 90 db down for both parts (wideband ~ 50 db
down), with the 9854 being only a couple db better than the 9851.
My experience showed me that the 9851 was not a good part for direct use
as a VFO for a high performance radio, and the 9854 is at best a few db
better (from the specs). I understand that they can be cleaned up
significantly by using it as a reference to a VCO and have seen much
exciting work on that topic on this list.
On the other hand, the even less complex 9835 does well in the DSW
series QRP rigs because it provides a simple, stable frequency source
and not everyone needs the ultimate experience in high end receiver
performance.
I myself am trying to pursue very high end receiver performance with
very low end current drain and DDSs don't provide what I am looking for.
One good thing I noticed when using the dual 9851s, the quadrature drive
I got from that arrangement was superior to using a pair of very fast
74LCX74 flip-flops clocked at 4x, even when both flip-flops were driven
simultaneously from the same 4x clock signal. When using the flip-flop
approach, the DC quadrature balanced shifted from band to band, making
it necessary to provide a balance adjust for the different bands to get
the best DC image rejection. This was not necessary when using the dual
9851s as the balance was rock solid from 1.8 to 30 MHz.
- Dan, N7VE
J M Noeding
<W4...@mchsi.com> wrote:
>
>The FST 3125 from Fairchild, Texas Instruments, Perricom and others exhibits
>under 4 ohms of Rdss, works to somewhere well above 30 MHz, has the switch
>drivers on board, and an Ip3 of above +42 dBm, (With square wave LO) . It
>costs under a buck in single quantities from DigiKey. (TI is $1.02). I've
>measured under 5 dB conversion loss in the "H" mode configuration with both
>MCL TT4:1A transformers at 200 ohms and MCL T1:1T's at 50 ohms, both at 20
>meters. Slightly more using T4:1's. It all gets better (But not a lot) when
>you raise Vdd from 5 volts to 7 volts, max ratings. Available in both SSOP
>and TSSOP.
>
>W4ZCB
>
>
A TI 74HC4066 has ON-resistance of 15 ohm typical, see
http://www-s.ti.com/sc/ds/cd74hct4066.pdf while Philips 74HC4066 has
20 ohm at 9V supply, see
http://www.philipslogic.com/products/hc/pdf/74hc4066.pdf
You normally connect two and two switches in parallel, then you have
typical resistance of 7.5-10 ohm, and it has only academical interest
to choose at better device, particularly when the better device is an
SMD type which isn't very much suitable for amateurs using dead-bug
mounting method.
The CD4069UBE buffer draws 10mA@10V, not certain if the 74HC86 will
draw more current on 9 or 10V. This may work with -10dBm oscillator
level (-30dBm on 200kHz), so I still believe the HC4066 is the right
device to choose for QRP up to 7MHz. Believe the HC86 stage is
described as 74ac86.jpg (never understand why somebody scan and save
as jpg when gif makes much smaller file size and better readability)
at http://www.geocities.com/CapeCanaveral/Lab/9595/fst3125.html .
DL7JGR's notes will give some valuable information to simplify the
construction, but I suspect he looses some gain in the front-end, it
is fairly easy to make a construction with -120dBm sensitivity, at
least it worked for me on 136kHz to hear a signal down below -126dBm
with no RF-stage ahead of 74HC4066, and otherwise quite similar
construction to what I would have used on 7MHz
It is so easy to wind your own transformers, so I don't see the point
in using MCL, they probably saturate long before the mixer.
One problem remain. Since everybody is only concerned about using 50
ohm, I have a problem finding suitable 200 ohm termination amplifier
designs. Suppose it isn't just to use a lower gain JFET or J310 with
low current, hi
73