2
\$\begingroup\$

I had an idea to transmitte a measured voltage of component X, from multiple slaves to one master device. Each slave's units would use the principle of time division multiplexing to set the data line to the corresponding analog level at a predetermined time and for a predetermined duration that the master unit would measure measure the line and separate the data. What kind of issue would I face doing this method rather than using a Data bus, or other serial interface.

The timings I'm considering are 100us cycles with a each device device holding the line for 5us and 5us gap.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ How would the master determine which voltage came from which slave? \$\endgroup\$
    – brhans
    Commented Feb 3, 2016 at 22:46
  • \$\begingroup\$ the master would send out a sync pulse, then slave 1 would set the live voltage at 5us, slave 2 at 15us, salve 3 at 25us and so on. this would all be pre programmed, to the master will be expecting the slaves at certain times. \$\endgroup\$
    – Lpaulson
    Commented Feb 3, 2016 at 22:50
  • \$\begingroup\$ How will the slaves know the difference between the sync pulse and a high value analog pulse from one of the other slaves? \$\endgroup\$
    – Transistor
    Commented Feb 3, 2016 at 22:59
  • \$\begingroup\$ The sync is a dedicated pin \$\endgroup\$
    – Lpaulson
    Commented Feb 3, 2016 at 23:27

2 Answers 2

Reset to default
1
\$\begingroup\$

Settling. See speed on page 2. It depends on your ADC, MUX and preamp. Signal degradation. How long are your runs from Component X to the MUX and to the ADC?

\$\endgroup\$
5
  • \$\begingroup\$ There isn't a MUX per say, all the slaves devices will be sharing a single line, so there will need to be some circuitry to isolate them when they are not setting the line voltage. as for distance the closest would would be ~1' followed by 6" increments. \$\endgroup\$
    – Lpaulson
    Commented Feb 3, 2016 at 22:48
  • \$\begingroup\$ I would be concerned that each slave could drive the bus with an analog signal without some sort of buffering. You'll likely have some losses associated with the large effective bus length due to parasitic inductance and capacitance. What does the slave output look like? \$\endgroup\$
    – theorifice
    Commented Feb 3, 2016 at 22:54
  • \$\begingroup\$ I haven't ironed out the exact details, but the idea is to have a sample/hold circuit to quantize an AC signal, so the voltage will be held steady(ish), while the master samples the line. \$\endgroup\$
    – Lpaulson
    Commented Feb 3, 2016 at 22:57
  • \$\begingroup\$ @Lpaulson how many slaves are there? What is the distance to the furthest slave? What voltage range is the signal? How accurately do you wish to measure it? What sort of cable connects the slaves back to the master? Are they daisy chained together? What processing do you have available on each slave? \$\endgroup\$
    – Steve G
    Commented Feb 4, 2016 at 10:21
  • \$\begingroup\$ @SteveG Like I said above I haven't ironed out these details, this is a hypothetical situation. I'm wondering about the effects of multiplexing multiple quantized signals over a single line would over a distance of 30' (at most) would be. \$\endgroup\$
    – Lpaulson
    Commented Feb 4, 2016 at 15:13
4
\$\begingroup\$

What you are asking for is possible, but at those speeds you won't get much signal to noise ratio. Of course it would help to know what your upper frequency of interest in the sampled signals is, and what resolution and accuracy you want to measure them with. Longer transmission distance will degrade the signal more.

All these reasons and more is why things are not done this way anymore. The phone system, for example, has long ago converted to digital.

Your question is a imagines solution instead of the real problem. Tell us what you really want to accomplish, not how you think it should be accomplished.

A deliberate gap between adjacent signals is just a waste of settling time. Each signal should start as soon as possible within its time slot. One problem with these systems is syncronizing the sender and receiver. One way is to send a sync pulse each frame. In old TV, this was a blip more negative than any valid signal. You could use that to sync a phase locked loop, which would then generate the clock to find the individual time slices.

Once you get all this working and discover all the various sources of error, you'll understand why these things are done digitally now.

\$\endgroup\$
1
  • \$\begingroup\$ The upper frequency will be much lower than the sampling frequency if you're hinting at the nyquist. The voltage being measured will also be run through a sample hold circuit as well to quantize the signal for the master unit to measure. What I'm trying to do is multiplex multiple quantized voltage signals on one line over a distance upwards of 30' (at most). You're right about phone system it would be similar to that, but each unit has a simple uC to control the timing of when it will put the data on the line. \$\endgroup\$
    – Lpaulson
    Commented Feb 4, 2016 at 15:07

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.