Audio switcher

[zathrasb5]

Hello,

This is my first post, and my first electronic project of any kind (aside from assembling kits in 20 years ago in high-school), so please be kind.

The goal is to add Bluetooth to my 1986 IROC, recently inherited from my mother.  The complication is that I want to keep the original head-unit.

After much research, I found out how the AC Delco radio boys add an aux output to the original radios, and which wires to splice in, (basically the FM signal from the radio to the amplifier), so that should not be a problem.  (They also recommend replacing some caps on the amplifier, so, with any project, the more you get into it, the bigger it becomes  ).

However, what they tend to do is add a five-wire aux port that switches between passing the existing FM signal though, or switching in the AUX signal.  Fine if all you want to do is to plug you iPhone in; not very elegant for a permanently installed Bluetooth receiver, as you would have to plug the headphone wire in and out all the time.

Fortunately, the Bluetooth receiver I found has a control wire that is on when the Bluetooth is on, so all I need is an audio switch.

Again after much research, and education, and following some of the debate on the internet about audio switchers (how IC's are great/good/in league with the devil), I decided that the better was the enemy of the good and settled on a plain old 4066.  But then I realized that there were about a thousand different plain old 4066's.  After more reseach I settled on the Max4066, due to the low Ron, and wide input voltage (which I though I would use, but in the end added a linear regulator).

More research about audio line level, coupling capacitors, decoupling capacitor, not gates, transistors, resistor networks, and diodes, (I really was starting at the basics), I came up with the following:

The voltage regulator is basically to provide power protection to the rest of the circuit.  I settled on 8 volts as it was the highest (I think) that the LM78xx series does, starting from 10.5V (low, but not knowing what power regulation, if any is in the car, I designed if for 10.5-14.5V). (please comment).

Then I decided it would be nice to add some leds to show if everything was working, so back to the books to learn about how not to blow up a led.  I stumbled upon a bi-colour led to  be blue when the bluetooth was on, or green for FM, but had to hit the books again becasue it was a common cathode led (figured out I needed a PNP transistor, not an NPN. 

In a case of the project growing in scope, It occurred to me that the automatic antenna should not be raised when the Bluetooth is on (which it otherwise would be as this is where I will splice into), so more research showed how to build an And gate (actually 1/2 of an And gate) with another transistor.  More research on how not to blow up a transistor when switching a relay.

Then, to avoid having a separate box for the fuses, I added fuses for the Bluetooth receiver as well (plus a USB port to charge the iPhone with) (I decided to call it quits here, and just buy a DC-USB power converter here, as opposed to more circuits).  (Again, the problem with project creep).  I know I will need wide traces at this end of the board.

I am going to breadboard it next, but I wanted to post what I have so far and see if anybody would like to comment.  Basically, I would like to know if I made any obvious mistakes as to where I will let all the magic smoke out, or if I am doing anything particularly hard or convoluted.

[Zero999]

I'm not quite clear on what this is supposed to do? Switch between two different audio sources?

F105 says 0.75mA, which is probably a typo. I've not seen a fuse that small!

The NOT gate has an output impedance of 750R when on, which will be loaded down by the LED. Check the logic level is still high enough for the MAX4066 to be high.

The part number for Q103 is incorrect because it's drawn as a PNP transistor, yet the 2N3904 is NPN. I assume you meant 2N3906?

https://www.sparkfun.com/datasheets/Components/2N3904.pdf
https://www.sparkfun.com/datasheets/Components/2N3906.pdf

The maximum base emitter voltage rating of the 2N3906 is 6V, which will be exceeded when AUX_ON is 14.5V.

The logic for the LEDs is flawed. When AUX_ON is high, LED-BLUE will be off because it's a PNP transistor and LED-FM will be off too because it's connected to the NOT gate. When AUX_ON is low, both the blue and green LEDs will light.

Wouldn't the 74HC4053 be more suitable? It contains thee SPDT switches: two of which can be used for the audio signal and the remaining one for the LEDs.
https://assets.nexperia.com/documents/data-sheet/74HC_HCT4053.pdf

D105 and D106 aren't needed. In fact, they make life harder for the IC because the inputs can now go 0.6V above and below the power supply rails.

The input voltages to the MAX4066 shouldn't exceed the power supply rails by more than 0.3V. AUX_ON can go up to 12V and the MAX4066 is powered from 8 - 0.6V = 7.4V! Connecting the inputs directly to 12V will damage the IC. The same is true for most logic ICs, including the 74HC4053 I mentioned previously.

You can connect the inputs to the MAX4066 to 12V, via a suitable current limiting resistor: 100k to 1M. The tiny current will be diverted into the supply rail, via the IC's internal ESD protection diodes. This will only work if there are other devices on the +V rail to absorb the current (there are in your case: LEDs) but you get rid of the diode in series with the positive supply, otherwise the supply voltage will rise considerably.

Q101 is an emitter follower, so you don't need the base resistor, as the base current is self limiting.

[Alex Nikitin]

That looks like a perfect job for a simple DPDT 12V telecom relay.

Cheers

Alex

[Zero999]

That looks like a perfect job for a simple DPDT 12V telecom relay.

Cheers

Alex

I agree, although id a relay is used, the resistance to vibration needs to be checked, since this is to be used in a car. A telecommunications relay might not be able to withstand that much vibration: the contacts could bounce around, if it's subjected to mechanical shock. An automotive relay might be a better match, but it needs to be suitable for switching small signals.

[Alex Nikitin]

That looks like a perfect job for a simple DPDT 12V telecom relay.

Cheers

Alex

I agree, although id a relay is used, the resistance to vibration needs to be checked, since this is to be used in a car. A telecommunications relay might not be able to withstand that much vibration: the contacts could bounce around, if it's subjected to mechanical shock. An automotive relay might be a better match, but it needs to be suitable for switching small signals.

Actually, a small signal telecom relay usually OK up to 10g vibration (operational, and some rated up to 20g), with automotive relays have about the same level of vibration resistance, so no problem there.

Cheers

Alex

[zathrasb5]

Hero999,

Thanks for the reply.

Yes, it  switches the Bluetooth audio into the amp line when the Bluetooth is on, and otherwise lets the existing FM signal though

1)  Yep, the fuse is a typo, I was planning on a 750mA, (0.75A) fuse. for F105

2)  I will check (at work now so I don't have my notes on hand).

3)  Q103.  Yes, should be 2N3906.  Looks like I also have the incorrect transistor for Q102

4)  I never thought/new that.  Thanks.  I think I can just use a voltage divider on the AUX_On, as my currents are small?

5)  I will check the logic on the leds.

6)  Thanks for the info on the 74HC4053, I never found that.  Yes, it would likely be easier.  I will consider this.

7)  The datasheet for the Max4066 includes the diodes if it is possible that there will be signal coming in without the IC having power.  As the radio and Bluetooth are outside my circuit, I included the diodes.  Upon further reflection,

  Thanks again for the 12V on input; something I had not considered.  I like your solution with only one resistor.  Another reason for no diodes.

9)  Thanks for the info on the emitter follower.  I will eliminate R107 and R108.

Thanks for your help.


Alex

God idea on the relay, something else I never knew about (I felt like I was recreating the wheel a bit).  I will consider this as well.  Would I still need the low pass filter and capacitors to eliminate pops?

[zathrasb5]

I was thinking some more, and I see now how the blue led won't turn on.    In order to use the PNP transistor, I think I would need another NPN transistor on the base, esentially as a not gate.

But...

Rather than that (and assuming I don't connect the led directly up to the AUX_On (which even though the manufacture says there is enough supply current, I would still like to power the led from the voltage regulator (it will also keep the led at a constant brightness)), I could put a not gate on the Not_AUX_On, with a single NPN transistor, and essentially have the blue led be, logically, powered off a Not_Not_AUX_On signal.

I also see how I don't need the diodes on U101.  It is voltage on the control signal that I can't have, not the input signals.  As Not_AUX_ON is dependent on my linear regulator, the only risk is on AUX_ON, which comes from the Bluetooth receiver.  This would only be on if the Bluetooth has power itself.  I was thinking that there would be issue if there was voltage on the signal path as well.  I could eliminate this as well by hooking U101 up to my new Not_Not_AUX_On signal if I wanted to, but this is probably not an issue (the AUX_on line, on a power-off from the Bluetooth would have to retain power in a capacitor longer than my IC does.  I think R104 should drain the AUX_on line in that case.

Thanks again; I will work on the schematic tonight.

[zathrasb5]

I was checking out the 4053 (sure wish I had found this earlier), and it really solves all of my problems.  It seams as though the 4053 was almost designed for my project.  Also kind of cool in that it will be used both as a multiplexer (for the audio) and a demultiplexer (for the common cathode leds).  I know I will need to watch how many amps I put though it for the leds, but except for that, I can see that the new version is really simple.  I don't even need a not gate, as the 4053 needs only one input (well three, but in my case, all tied together).

I kind of suspected I was doing something harder than need be; but really any project is more in the journey than the results.  If I wanted, I could simply hook up an FM transmitter, but that would be too easy.   I will post the updated schematic once I have it done; It will be much, much simpler.

As a question, is there a preference to using a voltage bias to bring the audio signal up to 1/2 Vin, or keep the audio at 0V, and add a negative rail (other than adding the negative rail itself). It seams straightforward to do in either case.  I know I already need 1/2 of the voltage bias to eliminate pops, so the other 1/2 (to +V) is easy to do.

Also, what happens if you give a LM780x less than the required input voltage?  Will it be happy, and just output a lower voltage, or will it stop working?  I could change the 7808 to a 7812  (with corresponding changes to the TVS), but was not sure what would happen if the car battery goes low. 

I know I don't really need the regulator, but it is a 30 year old car; I really have no idea about the input power.  Is there a better way of providing protection?  Right now I thing the LM78xx will handle persistent over-voltage, and the TVS will take the transient spikes and load dump; and the diode is if I hook the battery up backwards.  The polyfuse will protect the TVS, and the 10A fuse just in case everything goes wrong. 

Or, as the max4053 will take up to 17V, should I eliminate the regulator, and pick a tvs with a clamping voltage between 14.5V-17V, and a breakdown voltage above my 14.5V?  In that case, I don't think I would be protected against a persistent over voltage, but the TVS should then blow the polyfuse, I think.

Or am I making my life more complex than it needs be?

Thanks again for everybody's help.

[zathrasb5]

I have updated (redesigned) using the Max4053.  Much simpler.  Any comments are appreciated.

[Zero999]

A relay would simplify things because you'd no longer need the biasing diodes or coupling capacitors. The trouble is many power relays don't reliably switch small signals, because some current is needed to breakdown the oxide layer on the contacts. If you opt for a relay, you could use three SPDT (Single Pole Double Throw) relays or a singe 3PDT (3 Pole Double Throw) unit.

If you go for the MAX4053, I advise keeping the biasing resistors because it's easier than adding a negative power supply. If your circuit already had one, then it would be different but generating a negative rail is a pain and can be electrically noisy, as you need a switched mode power supply. There are some analogue switches which have a built-in negative power supply and allow you to switch negative voltages with only a positive supply, but I don't know of any part numbers off the top of my head, or if one as suited to your project as the '4053 exists.

A voltage regulator is a good idea. If the input voltage is too low to maintain regulation, then the output voltage will fall and be no longer regulated. Your circuit will be better with a regulated power supply because the voltage on the biasing resistors will be constant, preventing changes in power supply voltage from inducing noise on the audio lines.

I don't think you understood what I was saying about emitter followers. With 12V on ANTENNA-IN and on AUX_ON, R101 in parallel with R102, form a potential divider with R103,  making the base voltage on Q101, 8.67V (a bit less due to the loading due to the base current, which is unknown because I don't know what's connected to the emitter) and the voltage on ANTENNA-OUT 8V, which could be too low. If R101 to R103 are removed, the voltage on ANTENNA-OUT will be 12V - diode drop, or around 11.4V. The base current won't exceed the transistor's maximum rating, because as it increases, the emitter voltage rises, reducing the potential difference between the base and emitter, causing the base current to fall, until it reaches an equilibrium with the emitter sitting around a diode drop below the base.

[madires]

A relay would simplify things because you'd no longer need the biasing diodes or coupling capacitors. The trouble is many power relays don't reliably switch small signals, because some current is needed to breakdown the oxide layer on the contacts. If you opt for a relay, you could use three SPDT (Single Pole Double Throw) relays or a singe 3PDT (3 Pole Double Throw) unit.

The recommended relay for switching line level audio is a signal relay with gold plated contacts.

[Zero999]

A relay would simplify things because you'd no longer need the biasing diodes or coupling capacitors. The trouble is many power relays don't reliably switch small signals, because some current is needed to breakdown the oxide layer on the contacts. If you opt for a relay, you could use three SPDT (Single Pole Double Throw) relays or a singe 3PDT (3 Pole Double Throw) unit.

The recommended relay for switching line level audio is a signal relay with gold plated contacts.

I know, hence the words "telecommunications relay" mentioned previously. It may be challenge to find one with the necessary contact configuration, 3PDT, and 12V coil, though.

I suppose a DPDT unit could be used, with a semiconductor solution, (i.e. NOT gates) to control the LEDs.

[Zero999]

I had a quick look at RS Components (other distributors are available) and found this relay. It's 4PDT (4 Pole Double Throw) with a 12V coil and should be suitable. You could use the spare contacts to replace Q101 for ANTENN-OUT. The coil could be driven with an NPN transistor and reverse parallel diode.

http://uk.rs-online.com/web/p/non-latching-relays/8278318/
http://docs-europe.electrocomponents.com/webdocs/1398/0900766b81398a3d.pdf

[Alex Nikitin]

I had a quick look at RS Components (other distributors are available) and found this relay. It's 4PDT (4 Pole Double Throw) with a 12V coil and should be suitable. You could use the spare contacts to replace Q101 for ANTENN-OUT. The coil could be driven with an NPN transistor and reverse parallel diode.

http://uk.rs-online.com/web/p/non-latching-relays/8278318/
http://docs-europe.electrocomponents.com/webdocs/1398/0900766b81398a3d.pdf

This relay is not a signal relay though with gold-plated contacts may work OK. I would rather go for one or two proper DPDT signal relays, like this one:

http://uk.farnell.com/axicom-te-connectivity/8-1393792-8/relay-signal-dpdt-250vac-220vdc/dp/4219557

It will be cheaper even for two and will work better.

Cheers

Alex

[Zero999]

I had a quick look at RS Components (other distributors are available) and found this relay. It's 4PDT (4 Pole Double Throw) with a 12V coil and should be suitable. You could use the spare contacts to replace Q101 for ANTENN-OUT. The coil could be driven with an NPN transistor and reverse parallel diode.

http://uk.rs-online.com/web/p/non-latching-relays/8278318/
http://docs-europe.electrocomponents.com/webdocs/1398/0900766b81398a3d.pdf

This relay is not a signal relay though with gold-plated contacts may work OK. I would rather go for one or two proper DPDT signal relays, like this one:

http://uk.farnell.com/axicom-te-connectivity/8-1393792-8/relay-signal-dpdt-250vac-220vdc/dp/4219557

It will be cheaper even for two and will work better.

Cheers

Alex

Yes, that's a better option. It's designed for the purpose and even with the two coils in parallel, or you could the 5V or 6V variant with the coils in series, the coil current will be lower than the one I suggested.

[zathrasb5]

Hero,

Thanks, I am obviously still missing something on understanding the emitter follower on Q101.  Now I am determined to understand, but it will have to wait for tonight.  I think your explanation is clear; something is just not getting though my thick head.

Overall I like the design based on the 4053; I don't think I am going to do another design with a relay, I see the advantages of 2 DPDT or a 4PDT, but I should be able to get this to work. 

Thanks for the input on the voltage regulator; it is a cheap component that will make everything more robust.

[Zero999]

See the attached circuit of an emitter follower. The output voltage is one diode drop less than the base voltage and the base current is tiny, at only 23.6µA, even though there is no base resistor.

[zathrasb5]

Thanks, I see now one of my misunderstandings.  Vout is based on Vbase, not Vc.  In my nativity i was thinking too much like a switch, in that the Voltage would be the same as Vc.  The other name for this circuit now makes sence, a voltage follower.

THanks again.

[zathrasb5]

Light bulb moment.  I see where I was getting confused, looking at examples without understanding them.   

Some examples on the internet have voltage biasing on the base of the resistor to bring the voltage down so that it is not saturated (partially on), and will pass audio signal.  Other examples don't have that, as they operate saturated (fully on), and the full voltage of the base is used (less the 0.7 loss) at the emitter.  I was getting confused between the two different applications, as well as not doing the voltage biasing correct anyways.

Thanks hero for the explanation as to how the transistor would work.

I was making this more complex than I need, as in my application, I just need a switch. 

What I haven't found is how to calculate the value of the resistor.  I am thinking that the value doesn't matter that much (within certain bounds), but I think this is where your comment as to what is connected to the emitter matters.  I haven't taken the dash apart yet, but per the manual the antenna goes to an automotive relay (ohms currently unknown, but likely small).  As the resistor in your example is 1k (much bigger than the load), the load won't change the operations.  Or do I misunderstand again? 

I think the only thing, from a design perspective to watch, with the above assumption, is that the current though the transistor (which is the current in the resistor plus the current in the load), is within the capability of the transistor, and of course, the watts on the resistor.

[Zero999]

What I haven't found is how to calculate the value of the resistor.  I am thinking that the value doesn't matter that much (within certain bounds), but I think this is where your comment as to what is connected to the emitter matters.  I haven't taken the dash apart yet, but per the manual the antenna goes to an automotive relay (ohms currently unknown, but likely small).  As the resistor in your example is 1k (much bigger than the load), the load won't change the operations.  Or do I misunderstand again? 

I think the only thing, from a design perspective to watch, with the above assumption, is that the current though the transistor (which is the current in the resistor plus the current in the load), is within the capability of the transistor, and of course, the watts on the resistor.

The resistor in my previous schematic was only there to represent whatever is connected to the emitter of the transistor. You don't need any resistors, just connect the base directly to AUX_IN and the emitter to ANTENNA-OUT.

[zathrasb5]

That makes sense, I thought it didn't matter.   Thanks for all your help.   I think I am ready to order parts and breadboard it.  I may wait until the fall though to pull the stereo, there are too few nice weeks left to drive the car before it snows.

[zathrasb5]

One question just occurred to me.  Will bad things happen to the transistor if there is no load.  The antenna-out will be connected by a plug; is there a risk if the plug is not connected?

Also, it occurs to me that I should add a fuse (polyfuse) to the antenna out, in case there is a short in the wire or relay; the existing antenna out from the radio likely is protected, but I don't know of what value, and I should probably protect my transistor. 

[Zero999]

One question just occurred to me.  Will bad things happen to the transistor if there is no load.  The antenna-out will be connected by a plug; is there a risk if the plug is not connected?

Also, it occurs to me that I should add a fuse (polyfuse) to the antenna out, in case there is a short in the wire or relay; the existing antenna out from the radio likely is protected, but I don't know of what value, and I should probably protect my transistor.

No, nothing bad will happen to the transistor if there is no load.

I have just realised something: if there's no signal on ANTENNA-IN but there is on AUX_ON, then the transistor's base-emitter junction (this is electrically a diode) will pass current, from AUX_ON through to ANTENNA-OUT. If this can happen, then I advise having a base resistor after all, just to limit the current to a low enough level to not harm the transistor or activate the relay. Try 1k, which will only allow about 14mA worst case and shouldn't be enough to activate the relay.

[zathrasb5]

Hero,

Thanks, good catch, yes, antenna-it can be off.  Essentially, the transistor is being used as the last half of an and gate (antenna relay will only get power if both the existing head unit calls for it (on FM), and the bluetooth is off as well).  I think the examples of the and gates that I found (which include the resistor), confused me, trying to reconcile them with the voltage follower examples.  The resistor has nothing to do with the transistor, and only limits current to the relay.  I agree 14mA should keep the relay off.  One more reason to breadboard and test.

Just to confirm my understanding of how transistors work, if the DC gain of the transistor is 50, then 14ma will be enough to put the transistor into saturation, assuming the relay draws less than 700mA.

I should have been more clear initially, as I am splicing into the existing FM wires, the existing radio will think it is playing fm, but will actually be receiving the signal from the Bluetooth receiver.  This transistor is to just turn of the antenna signal, and bring the antenna down when the Bluetooth is on.  Purely unrelated to the rest of the circuit.  I know I could do it with another GM relay, or a relay on my board, but I can't see any reason not to use a transistor either (a transistor on the board is cheaper than a 30 year old stock relay, running for around $40, or a new signal relay, for a few dollars), but part of this project is just to learn how to actually design a circuit. 

The other way the AC Delco guys hook this up is though the cassette deck; the problem with that is that some head units require a blank cassette to be in the deck in order to activate the amplifier (I don't know if mine is one, but the FM line should always work).  Plus the wires for the FM are physically easier to splice into.

One last thing I know I need to do is check the watts on all my resistors, the 1K above should be a 1/2; I will check the rest.

[Zero999]

Just to confirm my understanding of how transistors work, if the DC gain of the transistor is 50, then 14ma will be enough to put the transistor into saturation, assuming the relay draws less than 700mA.

Remember it's an emitter follower, not a common emitter configuration. In an emitter follower, the transistor will never go into saturation. There will always be a fairly high voltage drop across the transistor. When ANTENNA-IN and AUX_ON are 12V, the base current will be equal to the load current divided by the transistor's Hfe and if that's 700mA it would be 14mA, in which case a 1k resistor base would give too higher voltage drop. If you need to drive a 700mA load, then you need a transistor with a higher current rating, as well as a much smaller base resistor. The 2N2222A won't do.