There are at least three ways to do the basic +-1 thing...
All have interesting charge injection issues. Some opamps go nuts if you switch at their inputs, or outputs.
A really good analog lockin is hard to do.
Got a favorite among those?
I've got some other (crazier) circuits scribbled in a note book.
An instrument amp with a SPDT analog switch that grounds each side in turn. (Input to both sides through a resistor.) (I think someone else suggested that one...)
George H.
Just a single diode. Tube type diodes were used initially. Some Ge and SiC diodes were used through the 1930s but they weren't highly evolved. The MIT Rad Lab guys developed some point contact Si diodes that were useful to several GHz (google 1N21) - you couldn't find a radar that didn't have one or a very close cousin.
One end of the diode went to ground and the other had a local oscillator (a volt or two) and RF connected to it. For low frequencies it was all lumped element and for microwave it was put into a waveguide.
The diode ring stuff came along later - some during WWII in military gear but it really wasn't until the early 60s that balanced mixers became common.
Done all the time. Even a lowly XOR gate in CMOS actually works rather well - below a volt or so the signal path is fairly linear. Can be used as a frequency doubler as well.
INA 0101 (LO) INB 0011 (SIG)
-------- OUT 0110
If a 40kHz-ish carrier is good enough for you, consider that there are sub-$1 Arm Cortex M0 parts out there that should be able to keep up with
200kHz-ish sampling rates if you don't mind that one task consuming a good portion of the processor's time. If you're lucky there'll even be one with a good-enough ADC.You'd have to do some benchmarking to see if the multiplication is fast enough -- I suspect that it is, though.
Sounds interesting. I sent off my $33 today. They said I could download the paper .... on Thursday! Haven't the Brits discovered the internet yet :)
ChesterW
You can do better with something that species how well the two transistors are matched.
It's probably two order of magnitude more expensive - $A13.63 from my local broad-line supplier, but they've got 14 in stock.
The MAT01 is even better, at twice the price, but they've got 77 in stock.
There are less tightly specified duals, but I can't be bothered finding them - somebody else will know off the top of their head.
One of the Australian electronics rags published a design in the mid-70's for a "synchrodyne" direct conversion AM receiver, which used a
4066 I/Q demodulator; Q was used to lock the local oscillator at 4x the receive frequency, divided down to produce symmetric I and Q signals.
That's not tying, that's encouraging!
ChesterW
200 uV rms error is plenty good enough. It's harsh conditions though, unknown and varying EMI, possibly poor power, big temperature swings. I thought I might switch a couple of calibration components into the loop to manage shifts in zero and proportion.
ChesterW
John's bottom one is generally best unless you can edit out the slew artifacts. An HC4017 or the PLD/FPGA/uP equivalent can turn off the integration during the teansients, which is generally a win.
Cheers
Phil Hobbs
Looks like you are doing some sort of side band modulation demodulatio etct.... I guess the carrier freq could be what ever, as long as you use the same one in decoding.
I remember using the LM1496, double balance mixer..gibert cell.
you can also use 4 diodes, ring diode mixer.. couple of op-amps from a dual supply to form the bridge.
Jamie
Thanks, (I know we've talked about his before.. I've got these DG611's in my part draw...(DIPs)... waiting.) I guess my mind went to the top circuit first. I drew something like the bottom one, but with an opamp in the top line... So many circuits, so little time... George H. (I'm trying to finish a digital thing... not my favorite.)
If you can't wait to Thursday you can search for patents in their names, they patented several variants of the p.d. in the late 1960s early 1970s
- all GB as I recall.
piglet
Bargain!
Random tangential question: Is there any technical reason that one couldn't design an IC OTA in the vein of the CA3080/LM13700 that has a much higher bandwidth, into the 50-100MHz range?
Perhaps in a modern BiCMOS process. ...Jim Thompson
Nope!
And if you really need to spend money:
Tim
They aren't. If the amplitude of the local sine wave oscillator is stable - which can be arranged - there aren't any local oscillator amplitude variat ions, and people who are worried about detecting odd harmonic content in th e signal they are looking at do go to the trouble of building clean and sta ble sine wave local oscillators.
Nice. Thanks. Looks like 2011 was a good year for lock-in amps.
ChesterW
Thanks everyone for your generosity in time and solutions. The customer is willing, so I plan to do everything in the micro - generate the carrier, digitize the signals and do the demod and filtering. They can pay me to reduce cost by designing some analog circuitry that needs only a cheaper micro later if they want.
ChesterW
I know that, but try to find a mixer that conversion gain, IIP3 and NF depend significantly on LO power. They put limiters in 'em and tell you feed 'em with enough voltage to make things nice and square. If you want a linear multiplier, you make or buy a multiplier.
I think it was an ADI app note (or dialog article) where that distinction was made by Barry Gilbert and Bob Clarke.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required