AD75019 input loading

[wwfeldman]

circuit attached

when the wires A, B or C are connected between their respective op amps and the AD75019, the voltage on the output
of the op amp reads about 0.5 V

when the wires A, B or C are disconnected from their respective op amps and the AD75019, the voltage on the output
of the op amp reads correctly, per the resistor divider.

the AD75019 appears to be loading the op amps or shorting them to ground(?).
a second AD75019 behaves the same way.
all power and ground are correct.

if the output of the AD75019 is disconnected, the op amp output behaves the same way.

the data, the serial clock, and the parallel clock are ok.

what did I do wrong?
what needs to be done to fix it?

 

[KlausST]

Hi,

why the detour via the PDF? The forum provides features to insert images.

We don't know R1, R2, R3, R4, thus we can´t know the expectable voltages A, B, C.
Also we don't know the internal switch setup.

Thus we are not able to find out what´s wrong.

Klaus

 

[barry]

This is a crosspoint switch-you can connect any port to any other port. Like, connecting an input "A" to one of the grounded inputs....

I suspect this device is doing exactly what you're telling it to do: connecting inputs to ground.

 

[danadakk]

circuit attached

when the wires A, B or C are connected between their respective op amps and the AD75019, the voltage on the output
of the op amp reads about 0.5 V

when the wires A, B or C are disconnected from their respective op amps and the AD75019, the voltage on the output
of the op amp reads correctly, per the resistor divider.

the AD75019 appears to be loading the op amps or shorting them to ground(?).
a second AD75019 behaves the same way.
all power and ground are correct.

if the output of the AD75019 is disconnected, the op amp output behaves the same way.

the data, the serial clock, and the parallel clock are ok.

what did I do wrong?
what needs to be done to fix it?

Are all the Yx lines disconnected when you do your measurements ?

Keep in mind if you do not program this right its possible to connect all
the X together on a Yx line, which because some of the Xx inputs are grounded
would ground the OpAmp outputs.

Dont forget the LM324 suffers from this problem :

Regards, Dana.

 

[wwfeldman]

@barry - any x can go to any y but an x cannot be sent to an x

@danadakk - interesting idea - multiple x to the same y, so that what barry said happens indirectly

basically, most of the switches should be open.
i will review the programming
thanks

 

[KlausST]

any x can go to any y but an x cannot be sent to an x

not directly, but if you activate switches: [X0Y0] and [X7Y0] then basically X0 is connected to X7 (and Y0)
Thus you are able to send signals from X0 to X7

(I see you already mentioned this)

So is you ensure that only one (or no) switch per X-line is activated you will prevent chort circuiting.

And possibly the problem is not only the GNDed pins.
Maybe just signal A, B, and/or C are "short circuited" this way...and OPAMPS are fighting against each other,

Klaus

 

[wwfeldman]

it seems to be working
and it is connected to the next gadget
the guy doing that will let me know if its ok

i think the real issue was the data i was shifting in
it was mostly the correct data, but it was shifted in LSB first instead of MSB first.
and there were a few wrong bits

thank you all for you input

 

[danadakk]

One last thought, when switching loads on the output of OpAmps you have transient
issues to deal with, especially if downstream you are trying to digitize or compare to some
reference value. So allow for settling time -

And the switch R, depending on config, can be in the 300 ohm area further compounding
settling and delay.


Regards, Dana.