1. Home
  2. Electronics Design
  3. Help with Mosfet Mixer

Help with Mosfet Mixer

I have constructed a Mixer for a VHF receiver (150MHz). The IF frequency is 45 MHz and LO is 105 MHz.

Having not made a mixer before, I copied a typology directly from one of my fundamental electronics books.

The layout is... RF signal into Gate 2 of BF998. Source has 100R Par with 1n to ground. G1 has 47K to ground and 100p to LO signal (I am using a sig gen for testing).

The drain has a Tuned circuit consisting of 100nH and 125pF. A 1k resistor is across the coil to provide approx. 800 Ohms needed to interface to a 4-pole crystal filter.

For testing I used a tap-C down to 50 Ohms from the 800R tank in the drain of the mixer BF998.

I observe the frequencies coming out of the mixer when I inject the 150 MHz RF signal and find I can tune the 45MHz for a peak no problem. There do not appear to be any instabilities. The gain of the preceding stages (Input single tuned filter, MOSFET RF amp and Double Tuned Circuit) is around 24dB.

When I add the mixer I find the total system gain is the same (24dB), The mixer seems to be mixing but adding no gain or loss. I also wonder where the mixer MOSFET is getting bias (from the LO ?) because the mixer stage doesn't appear to be drawing any current. I found the max. gain at the mixer output was when the LO level was anywhere from -5 to 0dBm.

Would someone please mind explaining how the bias of the mixer should be applied and how much gain/loss I should expect. ( i hope I am asking the right questions here as have no previous experience with a discrete mixer).

Thank in advance.

Regards

David

Reply to
David
Loading thread data ...

Normally, RF is applied to G1 and LO to G2. In "Introduction to Radio Frequency Design," Wes Hayward recommends biasing G2 with a 100k resistor from the source. The LO injection is then capacitvely coupled into G2 at about 5V p-p. He explains that "conversion gain is 12dB less than can be expected from the same device operated as an amplifier." It is not unreasonable to see a net 0dB gain; it depends on the device transconductance at the operating point, and the input / output matching. Much of the gain in a FET stage is achieved using a voltage step-up at the input, exploiting the high device input impedance. Your device should be drawin g a few mA of supply current.

Reply to
Andrew Holme

Andrew,

Thanks for your reply.

I was mistaken, The RF does go into G1 on both RF amp and Mixer.

The last thing I tried was grounding the source and now I have gain from RF in to Mixer out of 28dB (6dB at mixer stage). I also lowered the 1k across the drain coil in the mixer to 330R as the Q was too high.

I have EMRFD, I'll take a look at the biasing in there.

The circuit does not seem to far off what I want it do do.

The big issue now is matching to these 4-pole filters. I and getting about 10dB loss through them with huge amount of passband ripple and very sharp fall off but poor stop band attenuation. This tells me they are grossly mismatched.

Any ideas for getting this match right ?

Fo(IF) = 45 MHz The relevant circuitry is

100nH (Q=100)and 330R from mixer drain to Vcc and a 6.8 to 30pF trimmer par with 100 pF to ground for tuning. This should be giving Rp of around 220R. The 4-pole filters require 88R par with 3pF. These crystal filters comprise 2x 2-pole (with dots to centre for correct polarisation) with 4p7 to ground at centre of the filters). The output is to be matched to the front end of a SA605 that has 4500 Ohms par with 2.5pF at 45MHz.

I have tried simple "L" match from drain circuit of mixer to filter input and a 800R to 50R L/C-tap combination to my spectrum analyzer to view results but cannot get the correct response from the filters.

Thanks in advance

Kind Regards

David

Andrew Holme wrote:

Reply to
David

Hmmm. I don't know.

You could have a problem with leakage around the filter.

You could try inserting resistive pads to improve the broadband match.

You could validate your L/C-tap match using them at both ends, and driving one end from your signal generator.

Reply to
Andrew Holme

And they are. Crystal filters need a broadband match from DC to the speed of light on input and output. ;-)

Set up a resistive match with correct capacitance. Adjust C's for best response. Calculate or measure filter insertion loss. Note any difference to C opt and C match. Some passband ripple is to be expected depending on the filter design, bandwidth and published figures

Now you know what to expect to get close to.

There are still far to many unknowns for a good receiver design

You need a broadband buffer not any LC circuit. A diplexer may help.

Consider broad band transmission line transformers but generaly only useful up to a few K ohms since Zo must equal sqrt(zin*zout).

What is the highest level of interference with the lowest wanted signal the RF amplifier and mixer must handle without intermodulation and cross modulation?

You can follow the filter with a -3dB resistive pad as some aid in making a correct match if the additional loss is OK.

Your concern should be for noise figure, dynamic range and linearity, not just overall gain.

What is the published stopband attenuation. You may have to screen input and output if this is anywhere near -60dB.

Much has been written on good receiver design and much remains to be written.

Peter

Reply to
Jake

Nope mismatch is the probable cause.

Allways a good idea to get a feel for the response. Not a good idea for keeping front end gain with good dynamic range reasonable and noise figure defined by the front end amp.

One point, crystal filters ring like bells and a fast scan speed with give incorrect response 10*1/BW or slower through the passband if my memory is correct.

Peter

Reply to
Jake

Nope mismatch is the probable cause.

Allways a good idea to get a feel for the response. Not a good idea for keeping front end gain with good dynamic range reasonable and noise figure defined by the front end amp.

One point, crystal filters ring like bells and a fast scan speed with give incorrect response 10*1/BW or slower through the passband if my memory is correct.

Peter

Reply to
Jake

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.