EMRFD Message Archive 8982
Message Date From Subject 8982 2013-08-19 03:37:10 Ashhar Farhan KISS mixer with 2N7000
called Mixer Musings) with FST3157 and it works quite well.
Then I made a second mixer with 2N7000s. A trifilar transformer serves as
the basis for this mixer. The drains of the two 2N7000s are connected to
opposite ends of the transformers' primary winding. the center tap has a 1K
resistor connected to the +5v supply. The source leads of both the 2N7000s
are grounded. Their gates are driven by Q and Q' of a 74HC74 configured as
a divide-by-2 (to keep the square waves symmetric.
On sweeping the output of the mixer, it does appear to work well, quite
well actually. However, the noise is quite high. I am hardly able to
discern any signals. The noise vanishes as soon as the drive to the 2N7000s
is taken off.
The circuit is so simple that I am at a loss to know where to start
looking. I am, probably, making a mistake with the 2N7000s' drain
biasing. Any ideas?
- farhan
[Non-text portions of this message have been removed]8983 2013-08-19 03:44:17 bobtbobbo Re: KISS mixer with 2N7000
K1AO
8984 2013-08-19 06:28:15 Chris Trask Re: KISS mixer with 2N7000
>I have made a KISS mixer (refer to Trask's paper in the files of the group
>called Mixer Musings) with FST3157 and it works quite well.
>
>Then I made a second mixer with 2N7000s. A trifilar transformer serves as
>the basis for this mixer. The drains of the two 2N7000s are connected to
>opposite ends of the transformers' primary winding. the center tap has a 1K
>resistor connected to the +5v supply. The source leads of both the 2N7000s
>are grounded. Their gates are driven by Q and Q' of a 74HC74 configured as
>a divide-by-2 (to keep the square waves symmetric.
>
>On sweeping the output of the mixer, it does appear to work well, quite
>well actually. However, the noise is quite high. I am hardly able to
>discern any signals. The noise vanishes as soon as the drive to the 2N7000s
>is taken off.
>
>The circuit is so simple that I am at a loss to know where to start
>looking. I am, probably, making a mistake with the 2N7000s' drain
>biasing. Any ideas?
>
Chris8987 2013-08-19 10:09:00 Ashhar Farhan Re: KISS mixer with 2N7000
http://groups.yahoo.com/group/emrfd/files/VU2ESE/
I have uploaded the circuit and the sweep (on my sweeperino). I suspect
that the filter is doing much better than the -70dbm floor of the
sweeperino reveals. I have kept the filter open to 4 KHz because i like
broader bandwidths. There is a filter termination
problem in the sweep of the mixer. The 12 MHz ssb filter is quite flat on
it's own.
I probed the 2N7000 mixer a little. There is a massive ringing on the
transformer when the FET switches on and off. While this is to be expected
if there is DC current flowing through the transformer. I must switch back
to the FST3257 mixer and compare the results.
my crystal oscillator pair for measuring IMD are misplaced. I will report
the IMD measurements once this noise problem is sorted out.
- farhan
- farha
8988 2013-08-19 12:00:50 Dana Myers Re: KISS mixer with 2N7000 8989 2013-08-19 12:13:30 victor Re: KISS mixer with 2N7000 8990 2013-08-19 12:19:04 William Carver Re: KISS mixer with 2N7000
As their gate was driven through the active region, I would think.
Rather than throw in the towel so soon, I'd put a ferrite bead over each
2N7000 drain lead.
Bill
On Mon, 2013-08-19 at 22:38 +0530, Ashhar Farhan wrote:
> pse check the scans and the schematic in my folder. They are all here :
> http://groups.yahoo.com/group/emrfd/files/VU2ESE/
> I have uploaded the circuit and the sweep (on my sweeperino). I suspect
> that the filter is doing much better than the -70dbm floor of the
> sweeperino reveals. I have kept the filter open to 4 KHz because i like
> broader bandwidths. There is a filter termination
>
> problem in the sweep of the mixer. The 12 MHz ssb filter is quite flat on
> it's own.
>
> I probed the 2N7000 mixer a little. There is a massive ringing on the
> transformer when the FET switches on and off. While this is to be expected
> if there is DC current flowing through the transformer. I must switch back
> to the FST3257 mixer and compare the results.
>
> my crystal oscillator pair for measuring IMD are misplaced. I will report
> the IMD measurements once this noise problem is sorted out.
>
> - farhan
>
> - farha
>
>
>8991 2013-08-19 15:36:30 kb1gmx Re: KISS mixer with 2N7000
Looks like the Rohde Bipolar mixer in that there were a pair of
2n5179s instead of the 2n7000s. Note the Rohde dessign used
a trifiler transformer to get the drive polarity right.
The problem is there is a lot of feed forward from the FF to the transformer via Cdg (drain gate capacitance). With the low rise/fall time of the FF output I'd expect to see spikes8992 2013-08-19 15:55:11 kb1gmx Re: KISS mixer with 2N7000 8993 2013-08-19 16:30:24 Dana Myers Re: KISS mixer with 2N7000 8994 2013-08-19 16:57:45 William Carver Re: KISS mixer with 2N7000
spectrum analyzer you can SEE it.....
If it's a VHF resonance in the drain circuit a bead on the MOSFET drain
lead will de-Q it. It's quick and easy to try. A -61 bead won't
introduce significant inductance compared to leakage inductance of the
transformer: at VHF it's primarily resistive.
If the noise disappears you've found the culprit and can consider
whether the bead is the best solution.
I have some Calogic SD5000....four MOSFET switches in a DIP package.
They have ON resistance in 10's of ohms, but only 0.3 pF Cgs. They've
been used in switching mixer designs. You're welcome to several to try
Farhan, I owe you for that 7360!
W7AAZ8995 2013-08-19 18:56:43 Ashhar Farhan Re: KISS mixer with 2N7000
I hadn't thought of oscillations as this was supposed to be a passive
mixer in my mind.! I don't have a spectrum analyzer that goes into
VHF, just the ol 465. I will scope it today and find out. I think
putting the bead is a great idea to rule out the vhf oscillations.
Allison, I must remove the bias entirely as per your suggestion and
try out the circuit. I am not sure of how the MOSFET works as a
switch. Being a bidirectional device, I guess the gm is independent of
Vds? Incredibly, I have been unable to find literature on the net to
model MOSFET as a switch. The usual models portray FET as a controlled
norton source - Not much help in understanding them as a switch.
I will make the changes suggested by dana, william and allison and report back!
- Farhan
On 8/20/13, William Carver <bcarver@safelink.net> wrote:
> broadband noise sounds like VHF oscillation to me. If you have a
> spectrum analyzer you can SEE it.....
>
> If it's a VHF resonance in the drain circuit a bead on the MOSFET drain
> lead will de-Q it. It's quick and easy to try. A -61 bead won't
> introduce significant inductance compared to leakage inductance of the
> transformer: at VHF it's primarily resistive.
>
> If the noise disappears you've found the culprit and can consider
> whether the bead is the best solution.
>
> I have some Calogic SD5000....four MOSFET switches in a DIP package.
> They have ON resistance in 10's of ohms, but only 0.3 pF Cgs. They've
> been used in switching mixer designs. You're welcome to several to try
> Farhan, I owe you for that 7360!
>
> W7AAZ
>
>
>
--
Sent from my mobile device8996 2013-08-19 19:16:17 Dana Myers Re: KISS mixer with 2N7000 8997 2013-08-19 19:17:04 Dana Myers Re: KISS mixer with 2N7000 8998 2013-08-20 06:22:35 Chris Trask Re: KISS mixer with 2N7000
>Looks like the Rohde Bipolar mixer in that there were a pair of
>2n5179s instead of the 2n7000s. Note the Rohde dessign used
>a trifiler transformer to get the drive polarity right.
>
>The problem is there is a lot of feed forward from the FF to the transformer via Cdg (drain gate
>capacitance). With the low rise/fall time of the FF output I'd expect to see spikes on the Drains.
>
>It might actually work better without bias(the 1k and V+)
>as a switching mixer.
>
Chris8999 2013-08-20 07:14:43 Ashhar Farhan Re: KISS mixer with 2N7000
are allow DC current to flow in just one direction. That current has to be
substantially higher than the signal AC current to avoid IMD products.
However, such limitations shouldn't exist in MOSFETs that have symmetrical
channel.
The noise still continues to plague the mixer without any DC bias on the
drains. The noise is clearly generated through the LO drive.
I will try replacing the drive from the Si570 with that of a free running
signal generator.. though, I am pretty certain that Si570 is quite a clean
source.
- farhan
9000 2013-08-21 05:29:23 victor Re: KISS mixer with 2N7000
Did you check that you don't have ground noise problem? Please short the tip and ground of the scope probe and touch them to the circuit ground. Do this with and without the connection to the Si570 RF generator. You might find that the ground connecti9074 2013-09-05 11:04:10 ashhar_farhan Re: KISS mixer with 2N7000 Some update on the KISS mixer.
1. I gave up on using 2N7000s. Instead, I tried with J310s. The sources are connected to +5v line and bypassed. the drains are without any bias at all.
2. I am using a 74HC74 to divide by two and provide a 50% duty cycle to both the switches alternatively through Q and Q'.
3. I measured the DC resistance of J310 as 50 ohms when gate and source are at the same voltage.
4. The mixer is now working. I am able to hear the entire band using the FT817 as the IF sub-system.
5. The mixer has a gain of about -10db. This is probably due to the high on-resistance.
- farhan
9076 2013-09-05 16:01:21 kb1gmx Re: KISS mixer with 2N7000 Farhan,
2n7000 (mosfet) does not have a symmetrical channel. The J310 (Jfet) does. The difference lies in
substrate diode in the 2n7000, A reverse biased (D>>S) mosfet will show conduction as the diode is then forward biased.
The FST3125 switch has a far lower
9077 2013-09-05 18:01:36 Ashhar Farhan Re: KISS mixer with 2N7000
I had asked chris about the termination sensitivity of the KISS mixer.
He thought that the IMD will not be as bad as in a diode ring mixer if
the termination is improper, simply because the reflected signals will
not go through any non-linear device.
In a few days I should be able to test IIP3 and report it as well.
Thanks for looking into this small problem!
- farhan
On 9/6/13, kb1gmx@arrl.net <kb1gmx@arrl.net> wrote:
>
>
> Farhan,
>
>
> 2n7000 (mosfet) does not have a symmetrical channel. The J310 (Jfet) does.
> The difference lies in
>
> substrate diode in the 2n7000, A reverse biased (D>>S) mosfet will show
> conduction as the diode is then forward biased.
>
>
> The FST3125 switch has a far lower9101 2013-09-08 13:36:30 kb1gmx Re: KISS mixer with 2N7000 There is the yabut....
The MOSFET 2n7000 has the parasitic diode that is a very non-linear device that is part of that
fabrication technology, The older MOSFETS like the 3N128 and D5000 types are a different
structure. Same for the MOSFETS used in the 4016/4066 FST3125 switches however there
are parasitic diodes in all of them along with input and output protection diodes.
With the J310 the gate if driven more than .65V (one diode drop) positive with respect then
the source, it will also be non linear (at the input) as it goes from reverse biased diode to
forward biased diode. That is problematic as ideally you want to drive the gate right up to
that point but not cross it for best on state (minimum drain-source resistance).
Allison
9102 2013-09-08 13:43:35 Ashhar Farhan Re: KISS mixer with 2N7000