Is the circuit shown below (view with fixed font such as Notepad) an acceptable way of connecting an opamp to a SA602 or SA612? Would the performance (whatever that might be) be improved with a high impedance load on the mixer?
That depends on what performance degradation you'll find acceptable. It doesn't present a balanced load to the mixer -- assuming that the cap values are high, it forces the AC values of pin 4 voltage to be equal to that of pin 5. But it may well work OK -- if you're building a DC receiver with an SA612, you're already compromising performance over a diode ring mixer, so the question that remains is whether you've gone too far.
Unfortunately I don't recall NXP (well, it was Philips last time I looked) as saying how much degradation you can expect for an unbalanced load. It may not degrade the performance much, particularly if the voltage swing at the output isn't high -- any problems will be from second-order or higher effects.
So if it's _not_ in the data sheet, just give it a whirl and see how it does. Or you may see if you can get the attention of an NXP applications engineer and ask -- the worst that'll happen is that they won't return your mail.
It has current-steering outputs, your best error budget results from a LOW impedance load (like the pseudoground (-) input of an op amp with resistor in negative feedback). That mainly just minimizes the Miller effect on the output transistors, though, which is rarely an issue. As I read the data sheet, the outputs are most pulled up to the (+) supply rail, but the circuit you give is ground-referenced (that means (+) power ripple is not rejected).
Thanks very much for your reply. The SA612 spec doesn't say anything about unbalanced loads.
I'm attempting to build a VLF receiver. Actually, I've already built one and it doesn't work very well. An overstatement, in fact I can't see a hint of the signal I'm looking for. This is to be a SID (sudden ionosphere disturbance) monitor; the frequency of the signal I'm interested in is 24 KHz and is coming from a station on the east coast (I'm about 100 mi. NE of Seattle). My problems are compounded by a similar station about 50 miles away transmitting on 24.8 KHz. The mixer clock frequency is 24096 Hz (a frequency I could conveniently derive). The rest of the receiver consists of some gain, a 5-pole 150 Hz LP filter and a detector (a couple of diodes and an integrator).
[front end with transformer and commutating switches]
The RF input would benefit from a bit of resonance at the frequency of interest; consider a double-tuned transformer instead of a straight coupling transformer. If you have any uncertainty of the phase or frequency of your source, you NEED to get both the sine and cosine phases of the target frequency detected. The single-phase detector you sketched out is not adequate (because of the Nyquist limit). Nyquist didn't quote a maximum-allowable condition, he was talking of a disaster-guaranteed condition.
Where, if anywhere, is the automatic-gain-control amplification?
Thus, the desired output is at 96 Hz, the strong unwanted signal at
704 Hz (and a strong image at 48.896 kHz if inadequate front end selectivity).
If the OpAmp has a considerable gain, it will easily be overloaded by the 704 Hz signal. If the 49 kHz image signal is strong, the OpAmp open loop gain has dropped significantly, thus, the spectral purity is not very good even with feedback.
At least some passive low pass RC or LC filtering should be inserted between the mixer and the OpAmp to prevent overloading the OpAmp.
Of course, some frequency selective filtering in front of the mixer would be even better. Assuming a loaded Q=100, the -3 dB bandwidth would be 240 Hz at 24 kHz, thus, attenuating the 24.8 kHz slightly. A tuned magnetic loop could deliver at least part of this selectivity.