Were you terminating that feed coax in 50 Ohms? If not, the higher frequency dips and bumps probably are not 100% real. I would have terminated the coax with 50 Ohms right at the input to the multiplexer. Also some resistance at the output of the mux is probably a good idea also, maybe 10K??
Maxim also used to do a combined analogue multiplexer and buffer amplifier for video applications as the MAX453, 454 and 455 which are better for high frequency work. They would be worth a video if you can get them - they work up to 25MHz or so and have good isolation between channels - sadly they are not still in production. The switching speed of these - 120ns or better - was also very good and in my work allowed timesharing of a single ADC for low MHz waveforms.
Very recently I was working with another Maxim 8:1 analog mux and discovered an apparent silicon bug. Basically, whenever the mux switches, the chip sinks a large (sometimes nearly 10mA) current for a brief time (around 0.5us) from the path being turned off. This happens before the switch breaks, so it's effectively from the COM pin to GND. Datasheet makes no mention of this, and describes itself as break-before-make (technically true if you don't consider this glitch current to be a "make"). Might want to check if these chips suffer a similar issue.
The 14 MOhm on the Keithley look fishy. The 2015 supplies about 7V (Uref) via a ~10Meg resistor and measures the voltage drop in the high resistance ranges. With 20 pA leakage current, you should be in the GOhms, which the 2015 can't measure.
The Keithley was reading about the same @
Were you terminating that feed coax in 50 Ohms? If not, the higher frequency dips and bumps probably are not 100% real. I would have terminated the coax with 50 Ohms right at the input to the multiplexer. Also some resistance at the output of the mux is probably a good idea also, maybe 10K??
Always great how casual the device gets critically inspected using the test-setup. Thank you, it's a great way of learning and I really enjoy it.
Intercontact capacitance in the breadboard is something I've ran into before.
Maxim also used to do a combined analogue multiplexer and buffer amplifier for video applications as the MAX453, 454 and 455 which are better for high frequency work. They would be worth a video if you can get them - they work up to 25MHz or so and have good isolation between channels - sadly they are not still in production. The switching speed of these - 120ns or better - was also very good and in my work allowed timesharing of a single ADC for low MHz waveforms.
Would be fun to sweep it on the nanoVNA
Put a resistor from out to gnd, to improve off leakage.
Very recently I was working with another Maxim 8:1 analog mux and discovered an apparent silicon bug. Basically, whenever the mux switches, the chip sinks a large (sometimes nearly 10mA) current for a brief time (around 0.5us) from the path being turned off. This happens before the switch breaks, so it's effectively from the COM pin to GND. Datasheet makes no mention of this, and describes itself as break-before-make (technically true if you don't consider this glitch current to be a "make"). Might want to check if these chips suffer a similar issue.
The 14 MOhm on the Keithley look fishy. The 2015 supplies about 7V (Uref) via a ~10Meg resistor and measures the voltage drop in the high resistance ranges. With 20 pA leakage current, you should be in the GOhms, which the 2015 can't measure.
You lucky one, grab bag.
You didn't show it as a mux/demux. Use two, put four signals into one and see if you get them out of the other.
That's an $8 part. Not particularly cheap. Compare specs with ISL8405 or ADG series if you don't need 30V supplies.
What store are you going to? I used to go to Fry’s and some surplus stores but they all closed. I’m in San Jose.
Would the characteristics be the same for both directions?