Trouble simulating single ended to differential output using differential amplifier

[Hawaslsh]

Hello all,

I have an application where I need to create relatively low frequency differential signals from a single ended input. I have been exploring the option of using differential amplifiers and have been trying to simulate some circuits within LTspice. I have been specifically looking at the LT1993-10 (datasheet). The chips are available, they have a great BW, I can get free samples, and a LTspice model exists. I was able to get the circuit to function as expected using a single supply and pretty much copying from the datasheet.

To make the circuit above work I needed to apply a DC offset to the input equal to the common mode voltage supplied to the amplifier. However, in my real case I will have an input signal with zero DC offset. I would rather not add circuitry to apply an offset to the signal, and the datasheet for the LT1993-10 says split supplies are possible.

Split supplies are possible as long as the voltage between VCC and VEE is 5V.

I adjusted the circuit to accommodate split supplies, at least I think I did. However, as you can see from the output, the amplifier isn't working as expected, neither out of phase nor equal magnitude. I've tried numerous different topologies to no avail. Am I setting up the circuit work? Is it even possible to achieve my goal with this part? Any thoughts or advice would be much appreciated.
Thanks in advance!
Sami

 

[KlausST]

Hi,

I don't know why you use R2, R3, C2, C3.

Especially C2, C3 block any DC voltage.
DC reference to GND is undefined, you can not measure it.

And here I think is tye problem in understanding.

Single ended:
If you add here DC to the input ... (to comply with common mode input voltage range) ... you have to add it to your signal.
Input signal, as well as common mode input voltage is referenced to GND (your upper schematic)

Differential:
If you add DC here, (to comply with common mode input voltage range), you don't have to add it to your signal.
Because your signal is defined as V_diff = V_p - V_m

In either case: If you are interested in DC performance, then you need to omit the series capacitors.

*****
This is the benefit of a transmitting signals differentially:
At the output you may have have an unknown V_com but a known_V_DC.

With a practical circuit:
If you have two devices and you want to transmit DC signals (maybe combined with AC) via a long wire ... you may have a GND_bounce between both devices.
* With a single ended system the GND_bounce causes errors
* with a differential system the GND_bounce does not cause errors.

Klaus

 

[FvM]

Forgot to connect /EN pin correctly (to -2.5V)

 

[KlausST]

Added:
If you want to stay with C2 and C3:
Then replace R4 with two 200 Ohms resistors in series. (obviously giving 400 Ohms)
Now connect the center point of these two resistors with GND somehow.
It does not matter whether you connect it directly to GND or you use a 1000 Megohm resistor.
--> now you get a defined DC operation point.

 

[Hawaslsh]

Forgot to connect /EN pin correctly (to -2.5V)

Thank! sometimes the simplest things are the hardest.

I don't know why you use R2, R3, C2, C3.

I was taking a cue from the part's development board. Unless I am reading AD's schematic wrong, they have blocking caps on their outputs preventing any DC path to ground on their outputs?
Ultimately the diff amp will be driving a differential IF port of a passive double balanced mixer for upconversion. The datasheet for the mixer says the IF port is DC coupled with an input impedance of 25 ohms to each port. (mixer datasheet). I kept the series caps on the outputs because LTspice was predicting a DC offset between the outputs, and I wasn't sure if that would affect the mixer's performance, so I thought I'd play it safe and add the caps. See below for the two versions of the simulation.

Any further advice or comments would be much appreciated. Thanks to both of you!

 

[KlausST]

Hi,

Maybe I misunderstood your post#1.
You talk about
* "relatively low frequency" and
* "I will have an input signal with zero DC offset"

So my idea was:
* "low frequency" combined with "series capacitors" is contradicting somehow.
* since you mention "DC in the input" ... I thought you are interested in "this DC information" at the output, too. If this is the case, then you can't use series capacitors at all.

But if you are not interested in "DC input" to "DC output" performance at all --> then use the simplest circuit (single supply) and add the capacitors ... and you are done.

Klaus