Single transistor mixer

At 40kHz a jelly-bean op-amp for a few pennies, and John's classic* +1/-1 circuit would almost certainly work, with whatever is cheapest for an analog switch.

DigiKey lists a 74VHC1GT66 at 7 cents each in qty 1000, which is pretty cheap and should do.

• I invented one of those when I was in grad school. I tend to invent a lot of stuff that's been around for decades. It's trying.

--
www.wescottdesign.com

What about the LM13700 OTA?

I've used them to make a mixer at 500kHz or so - works great! They should have enough bandwidth to do 1MHz no problem.

And you also get a Darlington transistor with each section, which is nominally intended for use as a buffer, but which I used to build a single transistor Sallen-Key filter to low pass filter and get the difference signal.

If those are your concerns, you'll need a double balanced image reject mixer. If you can sample and process fast enough, you can do it in software. Elsewise, you'll need a quadrature local oscillator, two balanced mixers and possibly a Hilbert transformer to sum the I and Q signals so you can gain a sqrt(2) improvement in s/n.

--
Grizzly H.

A single transistor single ended mixer will have a poor performance. In addition to the sum and difference frequencies (select either with a filter on output side) it will also pass through the original tones. It will probably also cause mixing with the harmonics of both tones. If the input signals are not clean sine waves but also contains some other spurs, these spurs will also be mixed with the other frequency.

With _very_ careful selection of frequencies, you might be able to (low pass) filter the desired frequency and reduce the spurious frequencies.

A better solution is to use a double balanced mixed such a 1496 or a newer mixer. It quite effectively blocks the original frequencies and only produce the sum and difference tones. You still need to be careful with any odd harmonic products if hard driven.

With a double balanced mixer the post mixer filtering is much simpler.

Mixers are always driven with a square wave local oscillator to guarantee linear response to the signal and eliminate LO amplitude variations as a factor.

--
Grizzly H.

On a sunny day (Sun, 21 Feb 2016 16:22:20 -0600) it happened ChesterW wrote in :

No. but I have used dual gate MOSFETs any time I needed just a simple mixer. Oscillator on gate 2 biased with a trimpot to get the best conversion point.

But there are great 4 quadrant multiplier chips that work into the GHz range with a good large signal performance (google).

Anyways, in the dual gate scenario gate 2 controls the gain to put it simply.

Even cheaper use a normal BJT, signal on base, oscillator input on emitter.

The classic diode ring modulator is what I have used many times in large signal cases, but likely needs transformers in your case.

As somebody in some other group once mentioned: 'almost anything that is non -linear will 'mix'' A single diode works too.

So, not sure about your setup, get a 4 quadrant multiplier chip?

Or maybe in a 64 MHz PIC in software with sine lookup tables, but I have not tried, well actually I did to about 200 kHz for FT display

Here you see me using XOR gates as multiplier in a quadrature modulator:

Also note the GHz RMS42-N diode ring modulator, will not work at low frequencies, but you can use cheap audio transformers (like from a transistor radio) for that.

And note the AD8347 QAM modulator on board... That should? work for LF too. There are so many of these type of chips. This would become a very long posting, as again, anything non-linear will mix, even a bad contact.

Yep, should work fine; connect signal to (+) input on one section, and (-) input on the other, grounding the unused inputs, and feed the current source inputs differentially with the local oscillator. Wire both outputs together, of course. You can also do it with a single section, but that doesn't balance out the LO.

If your microcontroller has a sigma-delta ADC covering the audio range, you may find that it responds at multiples of the oversampling clock and nearb y frequencies. All you need to do is to remove the RC anti-aliasing filter from the ADC input. If the ADC does respond in this way, it will behave l ike a very good quality digitising mixer.

I have made this work with Cirrus sigma-delta audio ADCs in the past.

John

If the DUT doesn't like a square wave then what about just reducing the amplitude of the square wave until the DUT treats it linearly? Or how about a triangle wave - that would be easier than a sine for your micro to generate?

Won't you need a low pass filter or anti-alias filter between DUT and demod anyhow, if so then that would take care of 3rd order harmonics and allow use of a simple commutating switch mixer, in its simplest incarnation just one transistor as you wanted?

piglet

The mixers today are either analog multipliers or switching mixers (which means multiplication with the oscillator signal sign only).

If you do not need a perfect four-quadrant multiplication with a sine, probaly the best of today's constructions is called a Tayloe mixer (Google for it), which uses CMOS switches or multiplexers. The construction is for RF signals, but it is not difficult to adjust for lower frequencies.

--

-Tauno Voipio

There's nothing new under the sun!

That's very similar to something I built in 1979 using 4066s. It had a Q of, IIRC, 4000 at 8kHz and enabled me to detect a

1pW optical signal with a 1mm2 photodiode - which was 20dB better than competing equipment, I'm pleased to say.

I "researched" the design from a 1960 paper in the Bell System Technical Journal, "An Alternative Approach to the Realization of Network Transfer Functions: The N-Path Filter"

On a sunny day (Mon, 22 Feb 2016 13:33:44 +0200) it happened Tauno Voipio wrote in :

Yep, I like this AM modulator:

+5 | 100k [ ] ........ | . . audio -- |[]--------0 . C2 0-5Vpp C1 | . 0-----------||--- RF AM - filter --> | . 0 / . [ ] ..|..... 1/3 4053 / 74HC[T]4053 |100k | |control /// /// | RF square wave

I also did a PAL quadrature modulator like that (4.43 MHz ) in the seventies, generated 90 degrees with a counter from an 8.86 MHz crystal.

So the way this works: If no audio, then the output switches at RF frequency between 0 and 2.5 V, a 2.5 Vpp square wave. If audio is minium, at 0 V, than the output is 0 Vpp RF. And if audio is at maximum, +5V, then the output is 5 Vpp. Cannot be more perfect that that.

Don't forget that all commutating mixers (and many others too) have conversion loss associated with them so if low signal levels and noise are an issue some other approach may be better - do you even need to frequency change at all?

piglet

[snip]

I used that scheme in a variety of hybrids I made at Dickson Electronics ~1972, and later, ~1980 in an FSK discriminator at OmniComp/GenRad, see Fig 5 of...

...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Just 'cause >:-} ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

I wrote a whole chapter in this book (~1964)...

about filtering using active filters and the N-path method (as well as mixers... ;-)

I actually implemented an N-path 455kHz IF stage using that method. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

(I'm a bit late to the discussion...)

You mean a ring modulator type of thing?

It would be cool (and perhaps exists) if someone made synchronous FET (analog) switches that just flipped the (output) polarity back and forth.

George H.

Single transistor mixers are the pits. You can make a single-balanced one using a diff pair, e.g. a BC817DS, which at least keeps the LO out of the output to some degree.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Lock-ins often have pretty stiff performance requirements, so that e.g. mildly asymmetrical slewing or Ron variations between the +1 and -1 arms can be a serious issue.

To the OP: I gather this isn't a super-high performance application, but how good does it need to be?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net