I am trying to use the ADG419 in a FSK configuration at the out of 2 oscillators, but I am experiencing severe crosstalk to the output making it unusable to drive a transmitter. Data rates are typically 9.6 KB/s or higher. The RF frequencies are in the 2 MHz area. I've tried varying supply voltages and resistive loading on the input and output without any success. Standard TTL logic levels are used. All I am doing is switching between 2 separate oscillators to generate FSK in a very small package. Am I using the wrong device and need a different one. Any suggestions?
MIT Lincoln Lab
What level of crosstalk do you see?
According to the datasheet, the ADG419 B version has crosstalk of -90 dB and the ADG419 T version has crosstalk of -70 dB at a frequency of 1 MHz. Thus, when using a frequency of 2 MHz, the crosstalk in your system will be worse than the one shown in the datasheet.
Can you tell me more about your application so I can recommend a different device if this is the case? (Things such as supply voltages and a schematic snippet would help).
Thank you for responding. The crosstalk I was witnessing was only 10 to 15 dB down on my spectrum analyzer depending on frequency shift and data rate. On a scope it looked like something resembling analog modulation of the 2 MHz signal. On the spectrum analyzer you could see sidebands that looked like Frequency modulation or some artifact of such. Wasn’t expecting anything like this at all. I have attached a PDF of the test setup. Osc. levels are around 2.5 v P/P and are amplified after the ADG419. The UART is setup for standard 3.3v levels and the the data rate can be 9.6 KB/s or higher. If I could get the crosstalk down to at least -40 or -50 dB I would be happy. Any suggestions?
I will be waiting for the extra information you have and I will get back to you as soon as possible.
We are going to figure out what might be wrong in your setup, and if it is not the setup fault, I am going to point you for a more suitable part.
Sorry for not responding earlier, had to set up to generate data. Hope this info can give you a better idea on what I am trying to do. The switch does work, its just all of this spurious output above and below the carrier which is getting coupled in that bothers me that feeds the transmitter. Years ago we used to use actual relays but that was a long time ago and they couldn't pass high data rates. Isolation was superb though.
The above shows output carrier without any digital information, only single carrier without input to Vin
This shot shows what happens on output when PRBS7 data is supplied to Vin or even a square wave input.
This shows the 419 shifting to the other input. The switch is doing what it is supposed to be doing. RF level is. RF input to switch is 5.5 and 3.5 v P/P for clarity so freq. shift can be seen.
Digital signal into Vin at 9.6 KB/s from function generator or from computer, it still produces the results as shown in second picture with spectrum analyzer.
MIT Lincoln Laboratory
What I suspect that could be the cause, for now, is when the ADG419 switches between the two oscillators, the switch is not settled yet, thus, you see this kind of interference in your output. You can see this in your plot of the output signal in the time domain.
I will set this up in the lab and try to replicate your results. I hope this will be enough to determine what is the root of the problem you are seeing. It will probably be a couple of weeks before I get back to you.
In the meantime, do you want to try a different switch? (The ADG772 seems to be more than enough for your application, but you can also have a look at our portfolio in this link if you want.).
For the time being, I cannot discard the ADG419 since it seems to meet all your requirements as stated in the datasheet, but maybe, a different switch might suit you better.
I am willing to try anything that has a chance of working, I'll look at your portfolio and see what else is readily available. I chose the 419 because we had some in the building leftover from something else, but were not used for this type of application. Will be waiting to see your results. Seems like other people must of seen this too but maybe not.
I'm not in a position to comment on the actual component choice here but I can perhaps comment on your method in general. I'm assuming that the two RF oscillators are independent and not phase correlated (coherent). And in addition they are not going to be phase correlated to the modulating digital signal. So the switching point from one RF source to the other can, could, occur at any point in the sine wave. So it is very likely that most of the transitions from frequency A to frequency B will not be smooth or continuous. If Osc A is at a positive peak when the digital signal switches and Osc B happens to be at a negative peak the output seen at the switch will take a very fast transition between the two. These large, nearly instantaneous steps with result in a very wide spectrum as seen in a spectrum analyzer. Most FSK analysis / tutorials I've seen assume smooth phase (i.e. continuous and even switched at zero crossings) as the frequency is changed/shifted.
Have you perhaps run simulations using Matlab/Simulink, using uncorrelated sources for RF and data to see what the ideal spectrum will look like? I guess ideal sources and switches in Spice could even be used to simulate this.
Just trying to help separate the expected ideal response from your test set-up results.