IK1ODO QSD frontend : mixer dynamics

Some work is required to understand the dynamic behaviour of the QSD mixer employing bus-switch integrated circuits.

From the various Fairchild application notes, the simplified circuit of the FST3125 should be like this:

That is, a simple NMOS transistor, with a 0.65V turn-on voltage. Typically, the Ron of our transistor is as low as 4 Ohm, but varies considerably with the Vgs: from the FST3125 data sheet, it varies between 4 and 20 Ohm, depending on the input voltage and Vcc.

Considering that this non linearity of Ron may affect the operation of the switch, I tried to characterize the performance of my mixer. I applied a fixed RF input voltage, so to obtain a 100mV audio output from the OPA1632, and started to measure the audio output while varying Vref.



From the graph is clear that the Rdson increases quite linearly from 0.5 to 4V of Vref, then drops quickly. The more linear zone seems to be around 3V, so I measured the 2nd harmonic level at this voltage. It exibits a minimum where the curve has maximum flatness.

Now, to find a better switch to be used as a sampler we have to consider at least the following factors:

Several authors suggested to try analog switches, like the Fairchild FSAV332. I presume that the internal structure of this device is as follows:



I am not sure that it will perform better than the FST3125, since it has two pass transistors, one P and one N channel, with added capacitance (30pF when the channel is ON). There may be timing problems if the two transistors turn on/off in different times; the switching transient will be different, since the P transistor has approx. three times the area (and gate C) of the N one. I expect to try it in next few weeks, and then to publish the results.


In the mean time, this is my current protype:





Marco IK1ODO ik1odo at spin-it.com 


Circuit description
Measurements