1. Home
  2. Electronics Design
  3. Replacing a transformer with opamps

Replacing a transformer with opamps

I=92ve been mucking about with double balanced mixers. I wanted something to work below 100kHz, so I ordered a few audio transformers. They worked fine, but it seemed to me that you should be able to replace the transformers with some opamps. I bashed my head against the wall* for a day or so trying to come up with a circuit of my own. Today I went searching online and found this idea from Tom Hay. Basically two cross-connected differential amps with the noninverting input =91ground=92 coming from the output of the other opamp. Slick! (LT spice file attached)

Any other ideas on how to do this? Maybe there is already an IC that does this?

George H.

*Sometimes I find something, but most of the time I just put a dent in the wall.

Version 4 SHEET 1 880 1008 WIRE -544 -176 -576 -176 WIRE -384 -176 -464 -176 WIRE -352 -176 -384 -176 WIRE -240 -176 -272 -176 WIRE -576 -144 -576 -176 WIRE -384 -80 -384 -176 WIRE -336 -80 -384 -80 WIRE -240 -64 -240 -176 WIRE -240 -64 -272 -64 WIRE -208 -64 -240 -64 WIRE -96 -64 -128 -64 WIRE 320 -64 -96 -64 WIRE 416 -64 320 -64 WIRE 528 -64 464 -64 WIRE -560 -48 -704 -48 WIRE -496 -48 -560 -48 WIRE -384 -48 -416 -48 WIRE -336 -48 -384 -48 WIRE 320 -32 320 -64 WIRE 416 -32 416 -64 WIRE -384 32 -384 -48 WIRE -352 32 -384 32 WIRE -128 32 -272 32 WIRE -704 48 -704 -48 WIRE 320 48 320 32 WIRE 320 48 256 48 WIRE 416 48 416 32 WIRE 464 48 464 -64 WIRE 464 48 416 48 WIRE 528 48 528 -64 WIRE -480 128 -512 128 WIRE -368 128 -400 128 WIRE -352 128 -368 128 WIRE -96 128 -96 -64 WIRE -96 128 -272 128 WIRE -704 144 -704 128 WIRE -704 144 -816 144 WIRE -512 160 -512 128 WIRE 528 160 528 128 WIRE 768 160 528 160 WIRE -816 176 -816 144 WIRE -128 176 -128 32 WIRE -16 176 -128 176 WIRE 64 176 64 96 WIRE 64 176 -16 176 WIRE 192 176 64 176 WIRE 192 192 192 176 WIRE 768 192 768 160 WIRE -704 208 -704 144 WIRE -368 208 -368 128 WIRE -336 208 -368 208 WIRE 528 208 528 160 WIRE -240 224 -272 224 WIRE -208 224 -240 224 WIRE -128 224 -128 176 WIRE -560 240 -560 -48 WIRE -480 240 -560 240 WIRE -368 240 -400 240 WIRE -336 240 -368 240 WIRE 768 288 768 272 WIRE 192 304 192 272 WIRE -368 320 -368 240 WIRE -336 320 -368 320 WIRE -240 320 -240 224 WIRE -240 320 -256 320 WIRE 320 320 320 48 WIRE 416 320 416 48 WIRE 320 400 320 384 WIRE 320 400 64 400 WIRE 416 400 416 384 WIRE 416 400 320 400 WIRE 256 432 256 48 WIRE 528 432 528 288 WIRE 528 432 256 432 WIRE -544 464 -576 464 WIRE -384 464 -464 464 WIRE -352 464 -384 464 WIRE -240 464 -272 464 WIRE -576 496 -576 464 WIRE -384 560 -384 464 WIRE -336 560 -384 560 WIRE -240 576 -240 464 WIRE -240 576 -272 576 WIRE -208 576 -240 576 WIRE -96 576 -128 576 WIRE 64 576 64 400 WIRE 64 576 -96 576 WIRE -496 592 -560 592 WIRE -384 592 -416 592 WIRE -336 592 -384 592 WIRE -384 672 -384 592 WIRE -352 672 -384 672 WIRE -128 672 -272 672 WIRE -704 704 -704 288 WIRE -560 704 -560 592 WIRE -560 704 -704 704 WIRE -480 768 -512 768 WIRE -368 768 -400 768 WIRE -352 768 -368 768 WIRE -96 768 -96 576 WIRE -96 768 -272 768 WIRE -512 800 -512 768 WIRE -128 832 -128 672 WIRE -16 832 -16 176 WIRE -16 832 -128 832 WIRE -368 848 -368 768 WIRE -336 848 -368 848 WIRE -240 864 -272 864 WIRE -208 864 -240 864 WIRE -128 864 -128 832 WIRE -560 880 -560 704 WIRE -480 880 -560 880 WIRE -368 880 -400 880 WIRE -336 880 -368 880 WIRE -368 960 -368 880 WIRE -336 960 -368 960 WIRE -240 960 -240 864 WIRE -240 960 -256 960 FLAG 192 304 0 FLAG 768 288 0 FLAG -816 176 0 FLAG -576 -144 0 FLAG -512 160 0 FLAG -576 496 0 FLAG -512 800 0 FLAG 64 96 Vout SYMBOL voltage -704 32 R0 WINDOW 123 0 0 Left 2 WINDOW 39 -157 61 Left 2 WINDOW 3 -195 24 Left 2 SYMATTR SpiceLine Rser=3D10 SYMATTR Value SINE(0 1 100k) SYMATTR InstName V1 SYMBOL diode 400 -32 R0 SYMATTR InstName D1 SYMATTR Value 1N4148 SYMBOL diode 400 320 R0 SYMATTR InstName D2 SYMATTR Value 1N4148 SYMBOL diode 336 384 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D3 SYMATTR Value 1N4148 SYMBOL diode 336 32 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D4 SYMATTR Value 1N4148 SYMBOL voltage 528 32 R0 WINDOW 123 0 0 Left 2 WINDOW 39 92 72 Left 2 WINDOW 0 79 13 Left 2 WINDOW 3 77 47 Left 2 SYMATTR SpiceLine Rser=3D10 SYMATTR InstName V2 SYMATTR Value SINE(0 .5 10k) SYMBOL res 208 288 R180 WINDOW 0 36 76 Left 2 WINDOW 3 36 40 Left 2 SYMATTR InstName R3 SYMATTR Value 1k SYMBOL res 784 288 R180 WINDOW 0 -39 78 Left 2 WINDOW 3 -47 33 Left 2 SYMATTR InstName R4 SYMATTR Value 1 SYMBOL voltage -704 192 R0 WINDOW 123 0 0 Left 2 WINDOW 39 -222 31 Left 2 WINDOW 0 -178 103 Left 2 WINDOW 3 -242 65 Left 2 SYMATTR SpiceLine Rser=3D10 SYMATTR InstName V3 SYMATTR Value SINE(0 1 100k) SYMBOL voltage 528 192 R0 WINDOW 123 0 0 Left 2 WINDOW 39 54 95 Left 2 WINDOW 0 75 4 Left 2 WINDOW 3 52 49 Left 2 SYMATTR SpiceLine Rser=3D10 SYMATTR InstName V4 SYMATTR Value SINE(0 .5 10k) SYMBOL Opamps\\opamp -304 -128 R0 SYMATTR InstName U1 SYMBOL Opamps\\opamp -304 288 M180 SYMATTR InstName U2 SYMBOL res -512 -32 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R5 SYMATTR Value 10k SYMBOL res -368 -160 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R6 SYMATTR Value 10k SYMBOL res -352 336 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R7 SYMATTR Value 10k SYMBOL res -496 256 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R8 SYMATTR Value 10k SYMBOL res -560 -160 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R9 SYMATTR Value 10k SYMBOL res -496 144 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R10 SYMATTR Value 10k SYMBOL res -368 144 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R11 SYMATTR Value 10k SYMBOL res -368 48 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R12 SYMATTR Value 10k SYMBOL res -224 -48 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R13 SYMATTR Value 10 SYMBOL res -224 240 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R14 SYMATTR Value 10 SYMBOL Opamps\\opamp -304 512 R0 SYMATTR InstName U3 SYMBOL Opamps\\opamp -304 928 M180 SYMATTR InstName U4 SYMBOL res -512 608 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R1 SYMATTR Value 10k SYMBOL res -368 480 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R2 SYMATTR Value 10k SYMBOL res -352 976 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R15 SYMATTR Value 10k SYMBOL res -496 896 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R16 SYMATTR Value 10k SYMBOL res -560 480 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R17 SYMATTR Value 10k SYMBOL res -496 784 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R18 SYMATTR Value 10k SYMBOL res -368 784 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R19 SYMATTR Value 10k SYMBOL res -368 688 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R20 SYMATTR Value 10k SYMBOL res -224 592 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R21 SYMATTR Value 10 SYMBOL res -224 880 R270 WINDOW 0 32 56 VTop 2 WINDOW 3 0 56 VBottom 2 SYMATTR InstName R22 SYMATTR Value 10 TEXT 456 568 Left 2 !.tran 400u TEXT 464 624 Left 2 !.include opamp.sub

Reply to
George Herold
Loading thread data ...

There's the simple +1/-1 opamp thing, discussed here before. Just a few parts. The only gotcha is that the LO is essentially a square wave.

John

Reply to
John Larkin

"George Herold" wrote

I?ve been mucking about with double balanced mixers. I wanted something to work below 100kHz, so I ordered a few audio transformers. They worked fine, but it seemed to me that you should be able to replace the transformers with some opamps. I bashed my head against the wall* for a day or so trying to come up with a circuit of my own. Today I went searching online and found this idea from Tom Hay. Basically two cross-connected differential amps with the noninverting input ?ground? coming from the output of the other opamp. Slick! (LT spice file attached)

Any other ideas on how to do this? Maybe there is already an IC that does this?

How about MC1496 or AD633?

Reply to
Andrew Holme

Or NE612.

--
www.wescottdesign.com
Reply to
Tim Wescott

It seems overly complicated. What are you trying to do? What are your real frequencies and signal levels? You do realize that with a diode- ring mixer you want the diodes to be either ON or OFF -- no namby-pamby in-between stuff. This means that the LO input can be driven with a square wave -- and a square wave means you can drive that input with logic.

And if you _can't_ accept driving the LO input with a square wave, then you need to rethink the whole "mixer" concept and ask yourself if you should be using a real multiplier, instead.

--
www.wescottdesign.com
Reply to
Tim Wescott

Ah, Thanks Andrew. I was not so much interested in the mixing aspects of the circuit. But how to make a transformer out of opamps. Or how to make a floating voltage source. (Of course with the opamp it can only float up to it=92s rails.)

Here=92s the cross connected differential circuit.

GND---RRR---+---RRR--+ | | | |\ | +-|-\ | | >---+-R1--+----(+) Out +-|+/ | | |/ | +-----RRR--+----RRR=97------)---+ | | | Vin-+ GND=97RRR---+---RRR-----+ | | | | | | |\ | | +-|+\ | | | >---+-R1---+-- (-) Out | +-|-/ | | | |/ | +----RRR------+---RRR--+

RRR=92s are like 10k and R1 10 ohms.

George H.

Reply to
George Herold

ng

r a

=92

ur

Hi Tim, Thanks. I 'think' the idea is to contrast diode mixers with other means of mixing signals. So there will also be an analog multiplier as part of the 'thing'. I'm afraid to say more about the 'thing' cause it's still a bit nebulous and I'm waiting for the various parts to congeal into some coherent whole.

The person pushing this project is coming into town this weekend so I=92ll have a clearer view of where we are going after that.

I just got intrigued on my own by the transformer to opamp idea. We could then have a diode mixer that operated from DC to ~1MHz. Of course the whole idea might end up in the trash can.

George H.

Reply to
George Herold

If it's just to take a differential input voltage and make a level-translated (floating) output voltage, there's the OTA (operational transconductance amplifier, like LM13700). For inputs in the few-to-50 mV range, it produces a proportional output current, that you can return through a resistor to any low-Z point. A true transformer produces two matched output currents, one positive and one negative, and you can center-tap it, so it's hard to see how an op amp trick can really do a good emulation. If you were to wire the two OTA sections in opposite input polarities, it would come close.

The same LM13700 can be wired to make a multiplier, though; it wouldn't generally do mixer duty by transformer emulation.

Reply to
whit3rd

I had this epiphany driving in to work today, when I realized that a double balanced mixer was just like a +1/-1 switching amp. Did I mention I am slow*? Thanks to all you others with similar hints, Jamie, Tim, Andrew, etc...

Any clues for finding the +1/-1 opamp thread? (The diodes like a square wave drive too, as Tim said.)

At what frequency do diodes start to beat out Fet switches these days?

George H.

*Until at least age 20, I was still connecting words between my two separate spoken and written vocabularies.
Reply to
patricia herold

Thanks Whit3rd, I've never used an OTA. A colleague reminded me today that Instrument amps have a floating output. I think we should do a switching amp thing. The easy solution is to plug in an AD630 that we already use elsewhere.

George H.

Reply to
patricia herold

Things like this, maybe:

ftp://jjlarkin.lmi.net/Plus_Minus_Mixer.JPG

Diodes can mix at 100 GHz.

There are integrated gaasfet mixers, too. They can probably go fast.

John

Reply to
John Larkin

Right. It's the same as driving a transformer and swapping the leads (quickly, at the modulation / chopping rate). The diodes are just switches, swapping the transformer leads: conducting =3D "ON", reverse- biased =3D "OFF."

At lower frequencies ordinary CMOS switches are very, very good in DBM.

"Switchable inverting/non-inverting opamp circuit?", 26-Jul-2011

formatting link

-- Cheers, James Arthur

Reply to
dagmargoodboat

Thanks John, (I assume a MUX is an analog switch.) I did the bottom one a week or two ago. There was this 'huge' (40nS) switching transient. I used a break before make switch, perhaps not the best choice.

Made me think I should try a single pole switch, maybe on the input (as in your middle sketch.) Or maybe on the output... Switch from (Vin) to (Vin - 2*Vin)

Cool

George H.

Reply to
George Herold

g

Ahh got it Thakns James, I found this long thread from 2005 about mixers....

George H.

Reply to
George Herold

What sorts of frequencies are you working at?

A lot of analog switches have nasty charge injection spikes, like 10s of pC, and the spikes can cause troubles in both upstream and downstream opamps. There are good low-charge-injection analog switches.

There are linear ICs that use differential current steering to do balanced mixing, Gilbert cell sorts of things. JT designed some of them about 150 years ago.

The single-switch version could be interesting with a gaasfet as the switch. The phemts I've tried work pretty much like jfets as far as polarities and such go, but with much lower capacitances, lower thresholds, and lower ON resistance.

John

Reply to
John Larkin

ing

Hi John, The question is only academic at the moment. I was mucking about with the stepped sine thingie and tried the +1/-1 switching idea. (simulating 1/2 of a sine wave and the inverting to get the other half.) So the output frequency was only 100kHz, but the steps were square waves at 2 MHz. I used an ADG419 as the switch. (I have an inordinate fondness for 8 pin dips.) The inputs were both at ground during switching, but I got this 40ns few volt =91spike=92 from the switch.. the amplitude seemed to be independent of how much I loaded the output. Perhaps a make before break switch would have been better. There were also some timing =91issues=92* with this approach. And I dropped it as the first circuit incarnation was just fine.

George H.

*the digital signal would beat the analog signal to the switch... it was fixed, but was another kludge that had to be tweaked.
Reply to
George Herold

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.