Would it work to attach an MCU to a +-9V supply like I have shown in the attached schematic?

If this were a subcircuit attached to a larger +-9V board with a ground pane, would it be best to create a Vcc plane for the MCU board, to shield it from RF noise, and keep the same shielding reference to the larger board?

I have never seen this, so I am uncertain whether it would work, be immune to noise, etc.

Thanks for any comments,
Dave

 

I searched for a 4905 regulator, but results were ambiguous. A ST Microelectronics L4905 regulator has two positive outputs, so it wouldn't work in your application.

If you're going to have analog mixed with digital, you will have a difficult time keeping things quiet.

You could operate a MCU with Vcc/Vdd at ground/0v and Vee/GND at -5v if the MCU is rated for 5v operation. As long as there is +5v across the Vcc/Vcc and Vee/GND terminals, it won't really care. Depending on what you're doing with the rest of the board and how you want to interface it with the MCU, it might complicate things - but you have not provided enough information to make a guess about that.

 

Sorry, for the typo! I meant 7905 voltage regulator - not 4905.

I re-uploaded the schematic on the previous post. It shows a Vcc plane connecting to GND of the outside circuit, and the MCU switching an audio signal with a JFET.

Dave

 

Well, I guess you could do it that way. However, why not use something like a CD4066B CMOS quad switch instead, and keep everything on the positive side?

Did you know that your NPN transistor on the output is wired as an emitter follower, and the voltage on the emitter will vary between roughly -0.63v when your MCU output is high, and -5.63v when it's low?

Also, you might think that the two 1N4148 diodes will have an 0.7v drop across them - that's not the case with the very low current flow they'll have. You'll see about 350mV-370mV across each diode, which will negate the Vbe of the NPN transistor; the FET will see 0v to -5v on it's gate.

Is that what you intended?

 

Sorry again!!! I copied an ltspice simulation of the jfet switch over to gschem and got the BJT symbol wrong - it is a pnp BJT.

I also tried a 3.3V zener in the simulation and it seemed to work well at holding the JFET on during the positive swing of the audio signal. The R2/C5 and R4/C6 networks smooth out the on/off transitions. By adding the shunt JFET, I double the "off isolation" from -60db to -120db. The shunt JFET is working without the diodes and with only -5V pull-down. I need to think about that more.

I know I could use the CD4066B CMOS quad switch, but I have read that JFET's are a very good transistor for audio. I like the low resistance of the J110 - 18 ohms. Also, this is a chance for me to learn how to use JFET's.

Where is the challenge in using the CMOS quad switch? They are too easy.

Thanks for the help,
Dave

 

Will the other I/O on the uC be used for anything? Even in the future when you forgot you made a positive ground system?

Running the uC from a +5V regulator with the output driving an opto-isolator would be cleaner by completely isolating the audio signal (±9V) from the logic (+5V).

Using an isolator also allows the uC to be interfaced with standard +5v logic, without having to invert all the voltages of modules you may add later, or change on this design if it doesn't work as a prototype as good as it does in simulation (which is a common).

Will it work without smoke as shown? Probably. Are there overall better design practices to accomplish the same thing? Yes.

 

Thatoneguy,
Thanks for the note. You make some good points. I really like the J110 JFET and would like to use it for some ideas I want to try. I have been scratching my head for a long time trying to figure out how to make it work. I agree that I don't like my solution very much. I do feel uncertain about having a Vcc plane and no ground plane with respect to the MCU.

I am concerned that, if I use an opto-isolator, it will really slowing things down. I have worked with them a little, and they are very slow compared to the speeds that an MCU or CMOS logic ic can handle.

The best thing might be to add a dc bias to the JFET and isolate the switch on both sides with caps. I do not want to add too much phase shift into the signal, though. I am reading about phase shift and feedback problems in amp circuits right now.

I would like to stick with as much direct coupling as I can, in order to avoid phase shift problems.

I also thought an opamp set as a comparator might work. Put the MCU at GND and +5; power the opamp with +/- 9V and set the inputs so that they saturate to GND, and -9V. I will have to think more about this.

Thanks,
Dave

 

Thatoneguy,

I am concerned that, if I use an opto-isolator, it will really slowing things down. I have worked with them a little, and they are very slow compared to the speeds that an MCU or CMOS logic ic can handle.

A low cost optoisolator (4N25) still has on and off times < 5μS, which translates to > 200kHz, or within 1/10th cycle of the highest typical audio signal.

Not as fast as logic, but for an audio application, fast enough. If looking for Mhz speeds, The H11L1M from Fairchild is rated for switching Mhz Signals.

I'm unsure what you'd be using the JFET circuit above for beyond those switching speeds.

 

Yes, I have used the H11L1M optos in a stepper motor controller box I made. I really learned a lot with that project. The optos isolate the computer's parallel port from the driver boards and the limit switches.

At 4 usec rise and fall times, that is very fast for an opto-isolator. I use AVR chips that can handle 20MHz clock speeds, so I would have to code some delays to make sure the signal passes through to the +/- 9V side.

This may be the solution I am looking for.

Thanks,
Dave