Go Down

Topic: how to do low voltage analog switching? (Read 10707 times) previous topic - next topic

dc42


Do you think your little -3v generator be able to provide -3v to all my analog switches?  I need 10 quad analog switches (most likely the MAX4677-79) for my full application.. 


Yes. Those ICs need a maximum of 1uA from the negative rail.


I've seen that IC's seem to exist to provide negative voltage as well.


Yes, ICs such as the MAX828 and MAX829 work on a similar principle to my schematic. However they use synchronous rectification, so they produce a greater negative voltage (i.e. very nearly -5V out for 5V in). Of those two, the MAX829 would be best for your application because its switching frequency is above the audible range.


Before forget, an analog signal expert told me that I should put a zero gain buffer on the output of each pickup for some reason. I believe he was concerned by noise in all the IC's.  What do you think?


I can see what he is getting at, however I think you will be OK without since you won't (I presume) be turning the switches on and off at audible frequencies. If you want to use the switches to connect pickups in series, then you can't buffer them all anyway.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

gratefulfrog

#16
May 04, 2013, 10:05 pm Last Edit: May 04, 2013, 10:07 pm by gratefulfrog Reason: 1
Hi Thanks so much! 

I'm trying to avoid PWM if possible.  But on your circuit, by the way, I couldn' read the values on the Resistor and the 2 capacitors?  Was that  100ohm  resistor and 100 µF for both caps ?

This is really getting exciting! I can feel the circuit laying itself out before my eyes, or I should say, I can see you laying the circuit out before me ;-)

I saw that this circuit is required for the MAX829.  Do you know what capacitor values it requires?



I am starting to feel a bit uncomfortable asking so many questions of a single person! 

You are really saving my project!

Thanks,
Bob


gratefulfrog

#17
May 04, 2013, 10:36 pm Last Edit: May 04, 2013, 10:41 pm by gratefulfrog Reason: 1
Ooops. I may have spoken too soon. Those MAX chips only seem to come in surface mount which is nearly impossible for a hacker like me....

Do you have any suggestions for similar with through hole mount?   I saw the TC-1044 on uk-rs http://docs-europe.electrocomponents.com/webdocs/11b5/0900766b811b5b9f.pdf that can be boosted to 45kHz by connecting the frequency boost pin to Vin...

Or maybe this baby the TC7660H http://ww1.microchip.com/downloads/en/devicedoc/21466a.pdf running at 120kHz ..

I wish I knew what any of this means ;-)

Cheers,
Bob

dc42


I'm trying to avoid PWM if possible.  But on your circuit, by the way, I couldn' read the values on the Resistor and the 2 capacitors?  Was that  100ohm  resistor and 100 µF for both caps ?


Yes. Although with only 5uA or so total load, you could reduce the caps to 10uF.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

Docedison

A Cd4053 might be a better choice it's a triple pole 2 position switch that was designed to switch bi-polar signals  below ground by connecting the Vee lead to a voltage a volt or so lower that the most negative signal being switched... I usually use an L1044/ICL7660. The Maxim part is a little easier to use and about 10 years newer.. Will give you a -5V supply from a + 5V supply with a couple of .1uF caps and there is another non obvious +/- 10- 12 V source that can be made from a MAX232 RS232 adapter and is good for the current that the CD4053 draws... data sheets for the CD4053, MAX232 and LT1044 attached..

Doc
--> WA7EMS <--
"The solution of every problem is another problem." -Johann Wolfgang von Goethe
I do answer technical questions PM'd to me with whatever is in my clipboard

gratefulfrog

    Hi DC42 & DocEdision!

    I'd like to summarize what I've learned from all this and put forward my "solution" set of components for you opinions before I start ordering parts all over Internet...

    My understanding of the solution architecture is:

    • 1x Arduino Micro,

    • 1x 9v batter for power,

    • 1x JY-MCU bluetooth module for comms; connected to tx/rx, +5v & gnd pins,

    • 3x AD5206 Digital Potentiometer sharing the 3 PSI pins + 1 digitalOutput pin per pot for SlaveSelect;  these pots are used to control the LED current on the Vactrols, (3 channels are unused)

    • 15x Vactols:

      • 9x VTL5C6 for volume & tone

      • 6x VTL5C4 for coil dosing rheostats


    • 15x various resistors to protect, ie. limit current to the Vactrol LEDs

    • 3x 0.022µF tone capacitors, one per pair of coils,

    • 10x 4-channel SPST analog switches: MAX4677-79  (4x SPST @ 1.6ohm)

      • require -5v input and +5v input to be able to switch my +/- 1.5v audio signals from the coils

      • require 1 data pin per channel, ie 40 data pins (is SPI an option???)


    • 1x Charge Pump to create negative voltage needed by the analog switches; it is connected to Arduino +5v.

      • 1x TC1044S or TC7600H

      • 2x 10 µF electrolytic caps


    • 5x 8-channel 74HC595 shift register to provide data signals to the analog switches

      • work with 3 shared SPI pins, plus 1 SlaveSelect pin per 74HC595? = 5 Arduino pins

      • or requires 3 Arduino pins for the chain?

      • requires 5v power from Arduino




    So what do y'all think?  Does that make sense? Will it at least function all together? Or will I simply end up with a pile of smoking silicon ;-)

    This is by far the most sophisticated electronics I've ever even looked at, so my profound thanks goes to you for your help!

    Please don't hesitate to tell me all that's wrong with this thinking now, before I start shopping....

    Thanks so much,
    Bob
    ps. Thanks for your ideas as well, DocEdison! Sorry if I wasn't able to understand them to incorporate them into my so called, "solution". ;-)


gratefulfrog

Hey guys?
Any last thoughts before I start buying parts???  Please?

Thanks,
Bob

dc42

Digital potentiometers are not ideal for controlling LEDs. Then ones you have specified have rather high resistances for controlling LEDs. DACs with constant-current outputs would be better, however I didn't find any.

One possibility is to use a DAC feeding a an op amp to control each LEDs. You can configure the op amps to act as constant current sources for the LEDs. If you use an octal DAC (e.g. http://www.ti.com/lit/ds/symlink/tlc5628.pdf) and quad op amps such as LM324, you can keep the component count quite low.

Also consider whether 8-bit resolution in the LED current is sufficient, especially at low volume levels. If not, you can use 10- or 12-bit DACs.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

gratefulfrog

Hello DC42 !

Thanks for coming back to me!

I had run some tests with the digital pots and my Vactrols and they seemed to work well together.  Could you tell me some more about why they might not work right? With 8 bits control I had no problem getting good resistance variability on the Vactrol's photoresistor, but I did not test them in a full audio circuit because for the moment I don't have a full test platform available - it's being built in parallel with the progress in circuit design.  I hesistate to invest in a guitar for tests until I have a good idea that my plan can work ...

Also, I will have 15 Vactrols to control independently so that would mean 2 DACs + 4 Quad op amps, right? If I were to use the digital pots, I would only need 3 ICs.

Please excuse me if I wonder, I'm not doubting your expertise, far from it. I'm trying to understand and learn...

Thanks again, and again, for all your help!
Ciao,
Bob

dc42

The reasons I am doubtful about using digital pots for controlling LEDs are:

- You need to be careful to limit the LED current to the device rating (11mA for the 10K version) using an external series resistor
- The wiper resistance of those pots is 50 ohms typical, 100 ohms max, so this will affect the LED brightness too
- The whole arrangement is highly nonlinear, so you may find that the 8-bit resolution available from the digital pots is sufficient at one end of the scale but not at the other. This may also apply if you use 8-bit DACs.

But if you are limiting the current to below the device maximum and you are satisfied you are getting enough range and resolution, go ahead and use them.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

dc42

PS - why use the Vactrols at all? Can't you use the digipots directly?
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

gratefulfrog

I use Vactrols because they behave as ordinary resistors so positive and negative voltages go through equally well. The pickup coils produce +1.5to -1.5v.



- You need to be careful to limit the LED current to the device rating (11mA for the 10K version) using an external series resistor


Thanks for that tip!  I hadn't realized that the AD5206 had such limits since I just followed the Arduino tutorial http://arduino.cc/en/Tutorial/SPIDigitalPot  where they do not limit the current to 11mA since they only put a 220 ohm series resistor into the circuit.  If there's a 1.5v drop on the led the I = 3.5/220 = 16mA, which would be  too much current?

In my 1st prototype circuit, I had up to 20mA running through the vactrol's LEDs so as to get minimal resistance on the photoresistor.

I'm happy to do something else, but have no idea what circuit you mean with  a DAC +op-amp... sorry, again to be so unknowledgeable...

Thanks,
Bob

dc42

You can get digipots that also take a negative supply (just like analog switches) so that they can handle negative voltages too. For example, http://www.farnell.com/datasheets/1697953.pdf.
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

gratefulfrog


You can get digipots that also take a negative supply (just like analog switches) so that they can handle negative voltages too. For example, http://www.farnell.com/datasheets/1697953.pdf.


Yet more testimony to my ignorance...  I looked at that reference, but it is for surface mount and limited to 100K ohm.  I need a range of 500 K ohm and don't think I could ever solder surface mount components on the kitchen table...

Do you think I can get away with using the Digital pots to directly control the LEDs, even at 20mA which is nearly twice the 11mA max?

Or how would I use the DAC+op amp?

Or maybe I should just admit defeat and give up and stop bothering you?

Thanks,
Bob

dc42

Running those pots at 20mA when their absolute max rating is 11mA is a bad idea. Are you sure that you can't get a low enough resistance using 10mA?

Attached is a schematic for using a DAC + op amp to make a DAC-controlled constant current driver. The LED current is equal to the DAC output divided by R. Many DACs have a voltage reference input that you can use to set the output range. Others have a built-in voltage reference.

The +ve supply voltage to the op amp needs to be at least the maximum DAC output, + the maximum forward voltage of the LED, + the voltage drop in the op amp (probably around 2V for the LM324 @ 10 to 20mA, maybe a little more).
Formal verification of safety-critical software, software development, and electronic design and prototyping. See http://www.eschertech.com. Please do not ask for unpaid help via PM, use the forum.

Go Up