Hi All,

I'm currently trying my hand at creating a device I can use in a loud vehicle to communicate internally with the driver/co-driver, or externally via a UHF radio. My inspiration is pretty much the kind of intercoms that helicopters use. Pilot/Co-pilot have headsets that they can talk to eachother with, as well as to others via a radio. But after a few goes at it, it's clear there are many large gaps in my knowledge, and I don't know what I don't know... So I thought I would ask here to at least confirm I'm somewhere near the correct track.

The main parts I'm struggling with is handling the audio signals. I've had experience before with microcontrollers, digital circuits, and programming. But the audio/analog stuff is doing my head in. It's my first time dealing with op-amps and my lack of an actual education on the subject is showing.

I'll focus on the two parts I think I'm having the most trouble with, because I think if I solve those then I can figure the rest out from that solution.

The design is as I think you'd expect; The two drivers have a headset each. When a driver presses the intercom button, their voice is heard in both headsets. The headsets are essentially some earmuffs with speakers in them, and an electret boom mic.
The signal from the electret is tiny, so presumably I need some kind of pre-amplifier to get it to line level (Correct term? About 1v p-p?). My current design revolves around using LM380 power amps for both the electret pre-amp and the headphone output amp. I think this was a mistake but I'm not really sure why. The audio I get out is pretty badly clipped by the time it is loud enough to hear over the vehicle. And the background noise is very noticeable, so ideally I'd like less gain but more volume. I don't know if 'gain' in an op-amp sense is the same as 'gain' from like an audio mixing desk.

So that's my first issue. Audio out of this circuit is seemingly clipping and not-great and I'm not sure why, other than 'This probably isn't what lm380 is for' pretty much. Half the reason I picked it was to avoid another BoM item, and it was readily available from down the road in DIP...
I spied a TI reference design TIPD181 reference design | TI.com (Single-Supply, Electret Microphone Pre-Amplifier) in my travels that I think I will remorselessly plagiarise in the next design.
Other questions on this circuit include:
- Why 10k on the output? Is this an impedance thing? What is impedance in the case of audio signals? The same?
- Do the types of capacitor matter? As in tant vs al-ellectro vs poly, but also as in polarised vs non-polarised.
- Do the values of the capacitors on the input and output really matter? Or are they just there to stop signals going backwards or something? iunno.
- Should volume control be done on the input signal, or as a function of gain on the op-amp?
- The reference design uses the op-amp as a 'transimpedance amplifier' vs whatever the lm380 is doing. What is the significance of that?

The second issue I think I'm having is merging all the different audio signals into one signal that then gets amplified out to the headsets.
In addition to the intercom circuit, there is also the ability to plug in a UHF radio, as well as a music source via 3.5mm jack. So we end up with essentially four individual line-level (hopefully) signals that need to be merged into one and then shot off to each headset's output amplifier. I came across the circuit for a 'summing amplifier' which looked like it would do that task, based on a tl071 op-amp.

But the audio coming out of it was hot garbage. It functioned better without the amp in place, so the signals just essentially went through a few resistors.
The purpose of the amp was to limit an issue in a previous version of the design where the volume in the headset would change depending on if anyone was using the intercom, if the music was playing, or if the UHF was receiving. Like if you had music playing and then used the intercom, the volume of both would be about half. And then half again if the UHF chimed in. This circuit was supposed to buffer the signals I think so that didn't happen. And sure enough it didn't happen, but the clipping was much worse with the IC in the socket. The overall output volume was about the same, though. Which I think proves the gain as indeed 1x? I ended up removing it as I mentioned previously, and the volume sag as more channels chimed in was notably less but not entirely eliminated.
So:
- I can't remember why I chose this op-amp as the adder. It was probably available down the road as well in DIP... Does it look correct? Is the concept at least in the ballpark or is there a better solution for this? I think it at least had a good enough slew rate to do audio signals at a 1x gain.

In addition to all that rubbish, there was also this other issue I couldn't explain/diagnose.
There are volume pots on the input to the pre-amps, as well as on the input of the output amplifier (The pre-amp and output amp circuits are identical, except the output amp doesn't have the PTT circuit). In certain combinations of knob orientation (Usually to the high or low extremes), there is incredibly bad interference. High pitch buzzing and whining, and the power consumption of the lm380s would plateau or even start to go down as volume was increased. These pots are linear, so I can only assume that if I change them to the log pots that they should be, that I'll end up with only a very narrow band of 'usable' volumes. I originally thought it was some picked up interference from the breadboard pins, but even after getting a PCB made, the issue was almost identical.
- Any hints as to what may be causing this, or what I can do to narrow the cause down? Is it just interference between the op-amps? Or some critical design mistake I made? Is there a way to make it not do this?

In the end, the block diagram looks something like this:

Would you recommend any way I could do this better? Is there critical concept I'm missing that could help me work through these issues or build the circuit better?

Thank you so much for making it all the way to the end. This post ended up a lot longer than I was anticipating, and it got a bit rambly. I realise it's a lot to ask of internet strangers to freely spend so much time potentially solving for my inadequacies, so any help or advice is very greatly appreciated.

Many Thanks, Tangence.

 

- Why 10k on the output? Is this an impedance thing? What is impedance in the case of audio signals? The same?

I can't image a good reason for the 10k. By the way, a 1 watt audio amplifier does not sound like a great choice for an audio preamp. The TL071 (perhaps replaced by a two channel TL072) might be a better choice.

- Do the types of capacitor matter? As in tant vs al-ellectro vs poly, but also as in polarised vs non-polarised.

Only use polarized capacitors if the average DC level (average based the time constant of the capacitor and the load) is relatively constant.

Ceramic capacitors are often microphonic and generally shunned in good quality audio circuits but in a small aircraft I doubt you will l hear the difference.

- Do the values of the capacitors on the input and output really matter? Or are they just there to stop signals going backwards or something? iunno.

They determine the low frequency rolloff. Just keep the rolloff below about 20 Hz and you should be fine for voice.

- Should volume control be done on the input signal, or as a function of gain on the op-amp?

I would do it on the input signal unless you need Automatic Gain Control.

- The reference design uses the op-amp as a 'transimpedance amplifier' vs whatever the lm380 is doing. What is the significance of that?

We are accustomed to adding analog signals with summing amplifiers. In this case, it probably doesn't matter whether a summing amplifier is used or just a few resistors in a resistive mixer. I am of course considering that this will not be used and a quiet listening room.

- I can't remember why I chose this op-amp as the adder. It was probably available down the road as well in DIP... Does it look correct? Is the concept at least in the ballpark or is there a better solution for this? I think it at least had a good enough slew rate to do audio signals at a 1x gain.

The TL071 is fine. @Audioguru again may have better insights into this.

You will need either a negative power supply for your op-amps or to bias up your audio signals so that they don't try to go below the input voltage range of the op-amp.
https://www.renesas.com/en/document...7Y4QbxKfMBndEAX8ub2r6OvdUlJN0IaZdJeptLU9Z9lI3

 

The circuit commonly used to combine a number of audio signals from different sources is called an AUDIO MIXER. If you do an on-line search for that you will get at least 2000hits, none of them being what you require.

 

Thanks for the responses!

Only use polarized capacitors if the average DC level (average based the time constant of the capacitor and the load) is relatively constant.

I would think the load for the preamp would be pretty minimal since the output is buffered to the next op-amp in the chain anyway, right? Is that how it works? Something about no current actually flowing through an op-amp or something?
The coupling cap on the output of the output amp would need actual math.

Ceramic capacitors are often microphonic and generally shunned in good quality audio circuits but in a small aircraft I doubt you will l hear the difference.

Thats an interesting concept. Good to know.
Any weight on regular aluminium electrolytic caps vs poly film ones? After some more digging (Writing all this up has made me question why I chose any of these parts...), cheap polys are usually much better than cheap electros, but your comment about not being able to hear the difference given the surrounds is probably valid too. So long as its not just causing a constant buzz.

They determine the low frequency rolloff. Just keep the rolloff below about 20 Hz and you should be fine for voice.

That will give me some specific material to look up and study. Thank you. It looks like I might be 10x too high for the preamp given that 10k resistor. Is the load impedance just that10k resistor thats there?

I would do it on the input signal unless you need Automatic Gain Control.

Added to the feature list for version 10, releasing in 2030 : )

We are accustomed to adding analog signals with summing amplifiers. In this case, it probably doesn't matter whether a summing amplifier is used or just a few resistors in a resistive mixer. I am of course considering that this will not be used and a quiet listening room.

The main problem I was having was audio volume dropping when other channels chimed in, but when I got mad and removed the summing amp it behaved probably good enough as just a regular resistive mixer. Do the actual values of the resistors matter here, or just the ratios? Thinking about impedance for the preamps to drive, if I understand that concept...

The TL071 is fine. @Audioguru again may have better insights into this.

Would appreciate the consult : )
I had a note scrawled in my pad saying that 'TL071 isn't rail-to-rail', so does my +9V PSU limit me when using this chip? If so, how?

You will need either a negative power supply for your op-amps or to bias up your audio signals so that they don't try to go below the input voltage range of the op-amp.

I was trying to avoid creating a negative supply just because of the complexity, but it is starting to look like it might be the easier option. Part of my hesitation as well was due to the fact I wasnt sure how much current I'd need to make it handle. Or how to even guess that.

The circuit commonly used to combine a number of audio signals from different sources is called an AUDIO MIXER. If you do anon-line search for that you will get at least 2000hits, none of them being what you require.

Ran into that issue a lot when researching for this. Not knowing the jargon or what exactly to Google just got me pages of irrelevancy.

Thanks again!

 

So that's my first issue. Audio out of this circuit is seemingly clipping and not-great

I suggest trying this starter circuit. Refinements can added later.
Use the TL071 as the mic preamp to drive the LM380 audio power chip.
Note resistor R79 added to the mic switching circuit otherwise Q6 will be damaged.

 

Welcome to AAC.

The idea of using simple summing of the sources is probably not going to work. Not because there is an issue with it electronically, rather, because there will be collisions that will render transmissions unreadable.

I suggest you work out a gating (muting) and/or attenuating scheme that allows a priority channel to be selected such that when that channel is active, the others are muted/quieted. This can be extended to include levels such that the channels have a "pecking order", but at the very least I would expect you would want the radio comms to override local talk.

By the way, you should also include a manual override for local talk, a PTT that can take the output channel on demand.

 

I suggest trying this starter circuit. Refinements can added later.

I will breadboard it up and have a go, thank you. I also appreciate the photoshop skills.
What is the purpose of replacing the 20k resistor with two 10ks and a decoupling cap between them? Admittedly it is the only 20k resistor in the BoM so I won't be sad to see it go, just curious what function it serves. Thanks again.

The idea of using simple summing of the sources is probably not going to work. Not because there is an issue with it electronically, rather, because there will be collisions that will render transmissions unreadable.

But will it work good enough? I had figured that some music, plus one person keyed up on intercom, plus a radio probably won't create a hot mess on the output if you jam them all into one signal. But I suspect you know a lot more than me...

I suggest you work out a gating (muting) and/or attenuating scheme that allows a priority channel to be selected such that when that channel is active, the others are muted/quieted.

This was kind of originally the plan, but I'd decided to at least get the basics down before adding these kinds of 'nice-to-haves'. Other features still on the list to eventually be added include piping the radio straight to the pilot headset if the intercom loses power, expansion module to include back-seaters on the intercom loop, auto-PTT/voice activation on the intercom loop, auto gain control maybe (as of last night...), BLE for the music loop, and a nicer UX that isn't just like 8 volume/control knobs on the front of the unit.
Just brainstorming your idea though, how would I accomplish it? Comparator dumps the other signals to ground when a priority loop is keyed?

Thanks for your input, and thank you for the welcome.

 

While the circuit in post #5 will work well, consider that the PTT switch is included in the microphone assembly and it will be a big deal to make it separate. And I would think very carefully about having music on a comm system! Missing an important comment because of music could be serious.

The main concept of an actual MIXER is to avoid adjustments for one channel affecting any other channel. That can be very useful.
Now a question: Are the headsets already on hand?? Are they commercially made, or possibly adaptions of telecom headsets? Are the PTT switches a big part? or an add-on?
Looking at the headsets used in helicopters, they tend to block out the vehicle noise. So do the headsets in older TANK comm systems.
So if this vehicle flies or goes fast, the comm system is a big deal package.

 

While the circuit in post #5 will work well, consider that the PTT switch is included in the microphone assembly and it will be a big deal to make it separate. And I would think very carefully about having music on a comm system! Missing an important comment because of music could be serious.

The main concept of an actual MIXER is to avoid adjustments for one channel affecting any other channel. That can be very useful.

Explained below, but the PTT button can be completely separate from the headset/mic.
The system thankfully wont be used in anything so critical or fast as flight. Its an old, loud, and hot ex-army truck. It's going to be driven around the country, but the drivers can't safely be in the cab with that level of heat and noise for extended periods, days on end. My job is to handle the noise issue. Both these problems can be resolved with off-the-shelf parts probably but there's no fun in that, and it won't look period correct!
A mixer would be ideal. That way pilot and copilot could have different volumes for the different loops and not affect the volume of the other occupant. Might have to wait until a future revision, though.

Now a question: Are the headsets already on hand?? Are they commercially made, or possibly adaptions of telecom headsets? Are the PTT switches a big part? or an add-on?
Looking at the headsets used in helicopters, they tend to block out the vehicle noise. So do the headsets in older TANK comm systems.
So if this vehicle flies or goes fast, the comm system is a big deal package.

Headsets already on-hand. They are some Howard Leight Impact Sport electronic earmuffs. They passively have I think 22dB attenuation out of the box. Used to use them for shooting, but I haven't done that in a while, so they were donated to the cause. Originally they were battery powered (2xAAA) to operate the noise cancelling feature (Not really noise cancelling, more like 'don't pipe the louder noises into my eardrums'). But the problem with that was that they needed to be turned on to use the 3.5mm aux input, which was annoying. So I ripped out the innards and wired the 3.5mm aux jack directly to the speakers. Ripping the innards out also left a lot more space for noise-cancelling foam, so the passive attenuation is a smidge better now I reckon.
The microphone is courtesy of a generic boom-mic kit from AliExpress that bolts onto the side of the headset.
PTT button is a separate wire that goes up only as far as the little button handpiece you see in the pic. Doesn't go all the way to the headset. So it can be moved or completely ignored and isn't a big part of the headset itself at all thankfully.

Thank you for your input.

 

I suggest that you try to make the circuit of figure 5 and then you can experiment by adding other features. Get that one working first (that is what sghioto meant by "starter circuit").

The reasons for doing it that way are
1) It is simple and should not take too much to get it running, and
2) You will always have something that works as you refine your project.

R5, R6, and C3 provide a quiet 4.5 volts that will be added to your audio signal, so you don't need a negative power supply.

I started a similar project for a friend who had a two-seater fixed wing airplane, but eventually my day job got in the way. That was about the time such a system was introduced for motorcycles (which are still available).

 

I suggest that you try to make the circuit of figure 5 and then you can experiment by adding other features. Get that one working first (that is what sghioto meant by "starter circuit").

Agreed. Just in time for the weekend.

R5, R6, and C3 provide a quiet 4.5 volts that will be added to your audio signal, so you don't need a negative power supply.

Thank you for clarifying this.

 

But will it work good enough? I had figured that some music, plus one person keyed up on intercom, plus a radio probably won't create a hot mess on the output if you jam them all into one signal. But I suspect you know a lot more than me...

My experience with multi-channel feeds leads me to believe you will find it very hard to use without this functionality. Also, being in a vehicle, you should add a button that mutes everything and maybe even brings up live audio in case of an emergency where you need to hear what is going on around you.

You can try the simple summation it might work for you, but as I said, my experience in dispatch centers and with personal multi-receiver audio feeds is that they quickly become a garbled mess—and as Murphy would have it, just when you need them most.

 

Just brainstorming your idea though, how would I accomplish it? Comparator dumps the other signals to ground when a priority loop is keyed?

To be honest, I would use a small MCU for something like this. I much prefer making revisions in software to hardware.
For detection, I used CD4093 quad Schmidt Trigger NAND gate chips. You just bias the input to one of the gates and then trim the input so it will trigger when audio appears in the channel, then buffer the output with another one of the gates. This can go to a GPIO pin on an MCU, and you can use reed relays or analog switch chips to manage the feeds.

If you set a hard hierarchy, you could eliminate the MCU, but, meh.

And, you are most welcome—I hope you enjoy your time here.

 

My experience with multi-channel feeds leads me to believe you will find it very hard to use without this functionality.

I'll see what I can come up with then. Thank you for this.

To be honest, I would use a small MCU for something like this. I much prefer making revisions in software to hardware.

This is what I like to hear : ) Much more familiar territory.

 

What is the purpose of replacing the 20k resistor with two 10ks and a decoupling cap between them? Admittedly it is the only 20k resistor in the BoM so I won't be sad to see it go, just curious what function it serves. Thanks again.

The process of isolating one stage of an amplifier from another. Decoupling prevents unwanted oscillations and other noises that may occur due to unwanted feedback through common power supply connections.
Same as for the op amp using C4 and R6.

 

The process of isolating one stage of an amplifier from another. Decoupling prevents unwanted oscillations and other noises that may occur due to unwanted feedback through common power supply connections.
Same as for the op amp using C4 and R6.

Thank you.

 

The very simplest scheme unfortunately demands using carbon microphone elements, like in the 1950's operator's telephone headsets. Battery, microphone, switch, and headphone all in series Switch on to talk, and both can talk at once. No amplifier.
But since that is not an option, both an amplifier and a mixer are required. For a mixer circuit, look at the "schematics for free" website under audio / circuits. And they will not try to sell anything. And it all downloads very well. There are also amplifier circuits of whatever power you want.

 

For a mixer circuit, look at the "schematics for free" website under audio / circuits.

Looks like a great resource! Thank you!

 

Mixer circuit suggestion.

 

Mixer circuit suggestion.

Many thanks!
Question: For individual output volume control to each headset, using this circuit as a mixer amp, do I need to buffer the output of this circuit again for each headset, and then add a volume pot before hitting the actual final output amp stage for that headset, or do I just put a coupling cap between the output of the circuit above, and each volume knob?
roughly like:

Will the changing ratios between these two pots essentially cause the volume in each headset to change whenever one person adjusts their headset volume?

Thanks again.