1
\$\begingroup\$

I've been trying to solve a problem, but a few days of searching has yielded nothing, and I'm inexperienced to create my own electronic circuit. First, let me explain the problem. We use data acquired (hijacked) with the monitor in our control loop to send setpoints to the source whose feedback must be connected to the sink (they are both from the same OEM, and this connection is "crucial").

![enter image description here

We can log feedback from the source by connecting the OEMs diagnostic tool and compare it to the feedback we measure by monitor. We noticed that when both Sink and Monitor are connected, an additional dynamic is introduced into the signal measured by the monitor with tau being up to 0.2 - 0.3 seconds. This dynamic disappears if we connect only monitor to the source (but the system doesn't work if the sink has no feedback).

My idea for this problem was to split the signal with either operational amplifiers or optocouplers. But I don't even know where to start, let alone choose the correct components, and I believe my initial ideas are very naive. For example, a wire going from source to sink to pass through an optocoupler to copy the signal for the monitor, or, copy signal twice (we have 24V and 12V in the system, and if something like this is plausible, I know we need to add resistors to divide the voltage for it to be 0-10V):

enter image description here

Am I on the right track? Are there ICs that can simplify this, or even solve it out of the box? What would be the best solution?

EDIT: additional info about the time scale. Here is a plot of measured values from the left case in the first picture. The red full line is a setpoint we've sent to the source, the blue dotted line is setpoint he is receiving, yellow dotted is the feedback he is sending, and the purple full line is the raw feedback we measure with the monitor (no LP filter, only scaled from raw voltage to deg). Logged data by OEMs tool has a small and variable sample rate, with the lowest sample time being 100ms; our sample rate is 62.5Hz. Timestamps are synced by hand, so dotted and full lines may be off by a few ms. enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ What is the timescale in your plots? Are you looking for 10 s risetime or 10 ns? \$\endgroup\$
    – The Photon
    Jun 19, 2020 at 14:27
  • \$\begingroup\$ Plots in the first picture are a symbolic representation. For a step function input where there is no overshoot (no TF saturation), rise time from logged data is ~0.5 sec, while monitored data had ~1 sec \$\endgroup\$
    – stribor14
    Jun 19, 2020 at 14:38

1 Answer 1

2
\$\begingroup\$

That solution of using opto-coupler wouldn't work for your problem if "linearity" matters in your system. The input resistance of SINK and MONITOR seems quite high enough, but their input capacitance might have caused the problem of delayed rising.

Anyway, your problem most likely will be solved if you use an Analog Buffer (which is linear). The simplest circuit to do that is to use an Emitter-follower as a buffer. You should connect the output of the Source unit to the input of this buffer and connect the output of buffer to your monitoring system. However, there's a problem that you have a 0.6v drop in your voltage through the emitter-follower buffer.

Emitter follower as a buffer:

A more accurate analog-buffer can be implemented by an OPAMP. The advantage of using an Opamp-buffer is that its input impedance is pretty high, its output impedance is pretty low and there's no voltage drop across it.

enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ Thank you! This was really helpful and what I was searching for (but I did not know correct term how to search for it) \$\endgroup\$
    – stribor14
    Jun 23, 2020 at 6:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.