2
\$\begingroup\$

Circuit help!

I've been a lurking mechanical engineer for a while, metal fatigue specialty. I could really use some help with a simple circuit design please!

I have two pseudo-constant voltage sources that can arbitrarily vary between 0 and 10 VDC. I need to route either V1 or V2 to an input leg of a CD4051B multiplexer using a normally-closed switch. Vin pretty much needs to equal Vout, that is, very little voltage drop can be tolerated, say 0.010 VDC.

Also there needs to be a smooth transition between the two voltages, no spikes/overshoot upon activation/deactivation. Switching speed with stable voltages upon switching of < 0.5 millisecond would be grand.

Either +10 VDC or +15 VDC is available, but I would prefer to use the +15 VDC source.

Could someone please guide me down the path to success? I'm guessing some 3904/3906 transistors are involved, but maybe I need FET's? Obviously outside my knowledge base here... :( Thank you!

\$\endgroup\$
  • \$\begingroup\$ Ignoring the fact that you appear to be driving a CD4051B (which confuses me a bit, right now), it seems like the rest of what you say suggests an analog mux whose output goes to the "setpoint" input of an analog PI controller. What confuses me about the CD4051B is that it is already a mux and perhaps you mean to select one of your two inputs using it, but that you want its output to "move smoothly?" I'm just not sure, though. \$\endgroup\$ – jonk Jun 20 '18 at 19:19
  • \$\begingroup\$ What is the frequency of switching? How many total switching operations will be done over the life of the circuit? \$\endgroup\$ – Elliot Alderson Jun 20 '18 at 19:27
  • \$\begingroup\$ The multiplexer is already mounted in an existing circuit on a board. The voltage to the input leg in question is already being controlled via a potentiometer on a panel. I want to modify the circuit to add another panel-mounted potentiometer, and switch between the two voltages to the leg of the CD4051 using a mechanical switch. Frequency is slow, once every 150 ms would be fast. Call it <10,000 operations over the life. \$\endgroup\$ – scottrod Jun 20 '18 at 19:30
  • \$\begingroup\$ On thinking more about it, I guess switching could be as fast as 5 ms under the right situation. \$\endgroup\$ – scottrod Jun 20 '18 at 19:39
  • 1
    \$\begingroup\$ 5 ms times 10,000 operations means your operational product life is 50 seconds?! \$\endgroup\$ – Maple Jun 21 '18 at 7:16
0
\$\begingroup\$

What you want for solving the stated problem is a analog multiplexer. The CD4051 you are already using is such a multiplexer. The simplest solution would be to connect the two signal to two separate pins of that existing multiplexer.

If you really do need to select one of the signals onto a single pin of the CD4051, then you need a additional multiplexer. In that case, it's not clear why the CD4051 was even mentioned at all, since it is irrelevant to your requirements.

Since you are already using a CD4051, just use another one to select between the two analog voltages.

However

This smells strongly of a X-Y problem. The best system-level solution is likely at higher levels that what you are telling us about. For example, unless this is something on loan from a museum, something is ultimately going to digitize the analog signals, then do whatever further processing digitally. A possible solution might be to this new selecting in the digital domain.

Switching analog signals, especially ones this slow, went away with the 1980s for good reason.

\$\endgroup\$
  • \$\begingroup\$ Astute observation! It is a early '80's piece of equipment. The existing 4051 is already in service on all inputs and regardless, I can't mess with the firmware that's routing everything to the microprocessor, so I'm stuck with the one pin. I figured a couple of transistors would work somehow, so I added the 4051 information to let any responder know what the load on the circuit would be. If it's as simple as sticking another plexer in front of the pin, that sounds like a possible solution. With the NC switching limitation, would that work over trying to switch it with a transistor circuit? \$\endgroup\$ – scottrod Jun 20 '18 at 20:42
  • \$\begingroup\$ @scottrod I'm beginning to see the picture. More text in the question itself might have been advised. In any case, are you still focused also on "smooth transition between the two voltages, no spikes / overshoot upon activation / deactivation?" Is this still a requirement? Or are you perfectly fine with snap / sudden transitions if you switch between your two analog potentiometers? \$\endgroup\$ – jonk Jun 21 '18 at 5:50
  • \$\begingroup\$ A snap transition is great, as long as there isn't much overshoot. \$\endgroup\$ – scottrod Jun 22 '18 at 12:14
0
\$\begingroup\$

Add another multiplexer or switch (like ADG1419 for example) to select the voltage. Then send it to CD4051 through low-pass RC filter.

I was going to suggest a single capacitor filter first, but those caps on inputs and variable pot impedance will mess that up.

\$\endgroup\$
0
\$\begingroup\$

Multiplexer

I would connect the two 0.1uF capacitors in your diagram between the signals and 0V. As they are now, any ripple on the 10V supply will be coupled to the signal lines.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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