A simple input module

[ramshackles]

Hi
I'm new here but have been browsing through for a few days. I'm quite a novice at electrical engineering - however the combination of having to start designing circuits for uni (geophysics), a healthy obsession with playing music and unhealthy obsession with gear gear gear and finally, lots of time but a small (if steady) input of money inevitably leads to.....DIY!

Anyway, I've been thinking about very simple input & fader modules. Something with say a panorama control, mute, polarity reverse, output to a fader module and *perhaps* an insert. The idea being that with a number of these I could output to something like the folcrom summing mixer and some amplifier and I have a very nice, fairly sophisticated summing setup.
Some sends would be fun, but I think thats getting far ahead of myself

My audio circuit knowledge is fairly limited - in geophysics things a different, circuits are generally current driven and we are concerned with measuring different signals. I have a good knowledge of theory and vaguely related things like computer circuitry/programming and electromagnetics :S

I've looked on this site and around for anything similar to what I want but I find mainly full blown channel strips or people trying to make full consoles. Thats way beyond me. Are there any places I can turn to, threads I missed, schematics to look at?

The Small Signal Audio Design is a very useful resource and I'd quite like to implement the active panning scheme, that uses 2 opamps to achieve a very accurate panning curve and maximum width.

One thing that troubles me is bringing the signal level up to the nominal internal level (I think I'd go for -2dbu maybe even -1)...So the first thing the signal encounters is some kind of opamp (discrete or IC etc) to do so, but the signal going into the opamp is not constant? This is noob crazyness I know....

Is going out from this little channel to a seperate fader module and out from that to whatever summing mixer a good idea, or should the fader be connected differently?

Is it a good idea to put a buffer amp between each circuit block (between the panner section, mute, polarity etc) or is that overkill?

If I used higher spec opamps (such as the 990) for gain stages and cheaper IC opamps for buffer things, would that be beneficial? (I know manufacturers might use the best opamps for critical parts, but cheaper ones for where it doesnt matter as much, but I have trouble to identify the 'critical' places)

I've drawn up some schemes of seperate blocks (pan section, mute section etc) but how to connect them all in a low-noise, working, high spec single circuit is different...

Is such a simple input module more complicated than I think?

Thanks for all help

PS, I cant complete anything without a link to my band and a super-christmassy video

http://www.youtube.com/watch?v=zp2z2JlTzxY

 

[abbey road d enfer]

I've looked on this site and around for anything similar to what I want but I find mainly full blown channel strips or people trying to make full consoles. Thats way beyond me. Are there any places I can turn to, threads I missed, schematics to look at?

Google Steve dove mixing console.

One thing that troubles me is bringing the signal level up to the nominal internal level (I think I'd go for -2dbu maybe even -1)...So the first thing the signal encounters is some kind of opamp (discrete or IC etc) to do so, but the signal going into the opamp is not constant?

There is a concept of nominal level. The signal should hit as close as possible to the nominal level. this should ensure that the maximum instantaneous amplitude don't exceed the maximum level that the circuitry can handle at that point, and also that the signal is sufficiently higher than the noise level existing at this point of the circuitry. The only time when the signal is equal to the nominal mevel is when a tech connects an oscillator for checking/aligning/troubleshooting.

Is going out from this little channel to a seperate fader module and out from that to whatever summing mixer a good idea, or should the fader be connected differently?

Depends on the summing box and your preferred MO. Volume can be controlled in the DAW -some would say that it may generate some degradation, subject to a lot of debate and BS -, it can be controlled by a separate fader, or via a knob on the summing box.

Is it a good idea to put a buffer amp between each circuit block (between the panner section, mute, polarity etc) or is that overkill?

depends on the source. If it's low Z and floating, you can use it directly to go to the phase-reverse, mute and fader. The fader generally requires a buffer to drive the pan, even the active type. You could use a low Z fader (600R or 1k) instead of a more common 10k, then you may drive an active pan directly.

If I used higher spec opamps (such as the 990) for gain stages and cheaper IC opamps for buffer things, would that be beneficial? (I know manufacturers might use the best opamps for critical parts, but cheaper ones for where it doesnt matter as much, but I have trouble to identify the 'critical' places)

Everything is critical, because everything is in the signal path. I wouldn't completely agree with the claim that the 990 is "higher specs". The 990 aims to combine features than can be rarely found in one single opamp, like low-noise and high output capability. The 990 used as a mic preamp is LN and is capable of driving a 600 r load at +26dBu. Using standard ICs, you'll have to use a combo of let's say a 5534 for the LN pre and a 5532 for the output level, if it was necessaryis often used for its mojo, once it's in there, you generally don't need more, so you may use standard opamps for buffers.

Is such a simple input module more complicated than I think?

I would say it's not complicated because the necessary skills are clearly identified (this would be different if you said you want to build a mixer - see http://www.groupdiy.com/index.php?topic=46986.0), but still I detect that today, you don't have the necessary knowledge to embrace all the required competences. The nice thing is that, when you have all the individual parts worked, connecting them together will be easy. So start with a block diagram and a level diagram.

 

[ramshackles]

Thanks a lot, good help. Diagrams...yes, Ill post as soon as I stop doing them on paper. As far as the different blocks go, I think I'll start with:

Input stage - polarity - mute - panorama - output stage.

As far as the fader is concerned, a seperate fader would be fun and I also plan on a summing box like the Folcrom - something that only sums as I can then hook up different pres for the make-up gain (plus a nice little summing amplifier could be a fun little project after this). So I'd need to do some kind of volume control on the channels.

Starting with the input stage block, something like the superbal diff. amplifier looks great. I also saw that using a transformer in place of caps improves CMR but most likely at the expense of more noise - how to decide which is more important?
Something simple with transformers would be like this:
http://www.jensen-transformers.com/as/as069.pdf

What are the benefits of an all discreet design; such as:
http://www.vintagedesign.halmstad.net/diy/Balanced%20discret%20input.pdf
?? In the world I'm coming from, you'd be a pretty crazy geophysicist to go round filling your instruments with discrete opamps when there are perfectly good ICs lying around....

Well, Ill not move on to the other blocks before finishing this one up. BTW, as you can tell from me just researching other designs, Im not interested in coming up with new ideas, rather just to take the best of what I can to make a great spec, channel. As much for learning and understanding what happens to the audio when I mess around with it as anything else.

 

[PRR]

> you'd be a pretty crazy geophysicist to go round filling your instruments with discrete opamps when there are perfectly good ICs

Geophysicists want to play with rocks, not wire.

Many audio people want to play with wires, not audio.

> fairly sophisticated summing setup

Remember that a lot of great music has been tracked through very UN-sophisticated systems. You can go far with some resistors and a couple 5532, and good musical sense.

 

[ramshackles]

Geophysicists want to play with rocks, not wire.

Tease us when you have any idea of what we do

I decided to start very simple with an output stage. The AD797 looks really good, although I'm not sure I got all the specs right.
I put a 10K load at the end as that seems to be a typical input impedance of commercial summing mixers (EG folcrom).

I only have the freeware version of 5spice at home which doesnt allow me to measure noise or distortion - are there any freeware programs where I can do that?

Transformer balancing on the output - I dont really understand the benefits of doing this (so I havent...yet), for one thing I couldnt find inductance measurements for any of the transformers I found which are specified as having 'audio' applications (Jensen, sowter...). Putting in some typical values (from books), naturally leads to big linearity problems...

Another noob question (Im coming from a physics world where the IC reigns supreme...), what are the benefits of putting in a discrete opamp? If there are no benefits, why do respected and expensive models insist on using discrete circuitry (with plenty of transformers )

 

[ramshackles]

A more sophisticated circuit which I think (hope) will work well, with 2 AD797's giving a 1.78db (4dbu) output:

 

[PRR]

> measure noise or distortion

Forget about it. You are at the high end of the signal chain, noise is entirely dominated by earlier stages times all the gain from them to this output stage. A TL072 or even '741 will be quiet enough. Distortion of a complex opamp is not well modeled by any practical simulation; for low-low-THD jobs like AD797 the THD below clipping would be lost in numerical noise.

Your first plan: I do not see what C1 does or why it is there.

Your second plan: if the load is 10K then _ANY_ chip op-amp will drive it authoritatively, even the soft TL072. Why do you need two horses for a pony's job?

You have named these opamps "AD797" and "U1". Be consistent or you will confuse your helpers and customers. And me.

Re-naming to the obvious "IC1" and "IC2": IC1 outputs a signal to output jack "TPv1" via 100r. IC2 gets the same signal, _inverts_ it, multiplies by the strange factor of 4.651, and outputs a signal to output jack "TPv1" via 100r.

Assume that at a specific instant the signal voltage in is 1V. IC1 outputs 1V. IC2 outputs -4.65V. Your two horses are going in OPPosite directions at quite different speeds. Your 100 ohm "hitch" is being pulled apart. Assuming each opamp can pull something as low as 100 ohms, yes you can compute that the output is negative 1.825V, and the two opamps are fighting each other (via 100r) much more than they are pulling a 10K load.

I do not see where the 4dbu number happened.

Also the input impedance of the IC2 stage is nearly 430 ohms, and this must be pulled by IC1. Whereas if you threw-away the IC2 stage, IC1 would only have to pull the 10K load.

BTW: the DC gain of any chip opamp is not 20, but over 10,000. This used to be the #1 boast on the datasheet; I see AD797 omits it. However Fig 16 shows gain of 120db at 100Hz and apparently rising at lower frequencies (like zero).

If you go through the datasheet you may note that the output impedance is not 0.003 ohms when you get to the upper audio band, may exceed 1 ohm at frequencies of interest.

Take Abbey's advice "start with a block diagram and a level diagram". What does "a simple input module" DO? Probably has gain. Probably has a high input impedance. Probably does not rely on an abstract signal generator to get its DC reference bias. May have user-turned loss to handle large inputs.

If that advice fails, try mine: PLAGIARIZE!! Study-study-steal every input module plan you can find. You have to sort the gneiss from the schist. SOme are too simple. Some are far too fancy.

And yes I know geophysicists don't just "play with rocks", in the same way that audio-freaks don't just "play with wires". I'm just a little grumped because the fluid dynamics and electro-dynamics of my land is mostly about the rocks.

 

[ramshackles]

Thanks a lot, I've made a lot of changes based on that - pointed out a lot of daft mistakes!
You mentioned plagarizing - but the circuit is totally taken from Small Signal Audio (fig 15.3c) and the resistances/capacitances altered to suit the AD797. Updated the plan with realistic values now.

You should do geophysics all we do is play with wires...albeit very thick, heavy, annoying wires with electrodes the size of your arm...
No sign of rocks..we solely exist to confuse and contradict geologists with questionable facts based on outrageous assumptions.

Simple input module - like I said in first post; line input, phase, mute and pan control. Before pan would be some out and back again to a fader (which may be able to provide the normal 10db of gain or just attenuate). So no gain in the 'input' module. Input module is a good name for it..but I dont know what is. Stripped-down channel strip?
First diagram is updated version of the small signal audio circuit, second diagram is with the IC2 stage thrown out.

Block diagram, yes, it's so small/simple, I dont know how to make this more of a diagram other than to put boxes round the stages:

                                                                                  fader (seperate module)
                                                                                  |    |
line input -> phase reverse switch -> mute switch ->        -> pan -> line out (stereo)

 

[ramshackles]

and the small signal audio circuit (didnt realise 1 attachment per post :S)

 

[Harpo]

and the small signal audio circuit (didnt realise 1 attachment per post :S)

When you plagiarize, look at all the details. Your schematic differs from small signal audio fig. 15.3 in some respect.
IC2 will be a unity gain inverting stage that should give the same, but inverted signal level as its feeding stage IC1, so R3 will be the same parts value as R4. Measuring level between both outputs, connecting to XLR-pins2/3 will give you a voltage gain of 2 or +6dB. The amount of gain needed in IC1 (actually unity gain voltage follower) to compensate for the loss in your fader and pan circuit would be shown in your level diagram.
D.Self shows this R4 at 1K to keep resistive noise low with a dual, bipolar, unity gain stable NE5532 opamp that can drive this 1K load in parallel with the input impedance of the following stage, connecting to XLR-pin3. Same power requirement goes for IC1 with R3 shunting to a virtual ground node (inverting input of IC2), in parallel to the input impedance of the following stage, connecting to XLR-pin2. Substituting this NE5532 with same surrounding parts values, but a less powerful opamp such as TL072 will give a different result. If your plan -for whatever reason- shows two single AD797s, read this parts complete datasheet and app notes to keep it stable at least on paper/simulator. Pcb layout (not perfboard) will matter with these nerveous and expensive beasts. Page 12 from this datasheet also shows, there are other opamps that might be a better (and probably cheaper) fit for this task.
Your 10pF cap value across this R4 probably is a typo (will be a 1st.order LPF with -3dB cutoff at 15.9MHz).
After removing the short between XLR-pins2/3 from your schematic (TPdv1/TPv1), connect the same assumed load as R2||C2 to this spot, to keep the line impedances balanced with maxxed CMRR.
As already posted, get a block diagram with level diagram first.
Just my 2ct and YMMV.

 

[PRR]

> After removing the short between XLR-pins2/3 from your schematic (TPdv1/TPv1)

I suspect that is a simulator "probe" for "differential voltage".

I don't have Doug's latest but I'm sure there's transcription errors in the plans posted here.

 

[gemini86]

not showing the entire circuit, out of respect, but it's pretty obvious what I've clipped...

 

[Brian Roth]

Downside of that last circuit variation is that the output level will drop 6 dB if driving an unbalanced load versus a balanced load.  Also, if driving an unbalanced load, the poor opamp "driving into ground" will be clipping like crazy as it goes into current limiting, and can dump distorted crap into the ground line.

Digidesign uses that sort of "brain dead" output circuit on their 192 interface, making it tough to happily integrate into a studio's patchbay system with a variety of different outboards.

Best,

Bri

 

[gemini86]

Whatever the topology used, I don't exactly know how we've wandered over to balanced output circuits if he's building an input module... I was just posting the image to make the convo easy to follow.

 

[CJ]

hey shackles, have you seen the Elliott sound pages?

cool stuff, and lots of it,

here is a link that walks you through mixer development,

http://sound.westhost.com/project30.htm

 

[ramshackles]

Thanks for the web link and replies. The ESP site is awesome. Looking at the block diagram from the channel module and borrowing it, what I'm going for would be something more like the modified pic attached.
I was not necessarily interested in the aux sends to keep things simpler, although having the possibility to add them in the future could be cool.
I throw AD797 out of the window as it seems its very difficult to get their best performance. Although they do look great on paper. Funny that the 5534 hasnt really been improved upon after all this time...

 

[ramshackles]

Well, I started this thread so I should have the decency to update/close it...
I decided that while I read much more about the electrics (art of electronics and small signal audio... ), I would start with something much much much simpler - a passive summing module (yes yes...).

There are hundreds of threads and schematics here so it's a good place to start. I'll basically follow NYD's summing mixer schematic, going to 16 channels (L & R) using db25 inputs.

Theirs lots of things about people wanting to put transformers on the outputs of their summing mixer, but I've not seen any schematics that follow through with the idea. I don't want to do that, but I would like to learn the theory behind it? E.G what would the math be for, e.g. adapting NYD's summing mixer for transformer balanced outs? What is the (dis)benefit of doing that??

 

[ramshackles]

Well, I built a couple of kits and I keep coming back to this idea. So here's a ressurection ...
I'm pretty much going to piece this together from the designs in 'Small Signal Audio', which is excellent.

A block diag of the channel is attached to show what I'm thinking about currently. I'm after something simple, low cost, with the basic functions covered, very transparent with excellent headroom and noise.

So, starting with the input stage, I was looking at Self's balanced input (I dont know if I can post the diag here? Fig 14.7 p359)
Those 47uF non-polarized capacitors would be 1. Difficult to come by and 2. Expensive.

I was thinking you could achieve the same result with 2 100uF electrolytic caps in series? At the mid-point between the two would be a very high resistor (1M) connected to -voltage.  The caps connected like:  - + + -

Is that viable? 100uF electrolytics are certainly easier to get a hold of and cost a few pennies rather than a few pounds.

Could it be improved in other ways (i.e. improve noise) with only minimal added expenses?

For the whole channel, I'm looking to keep the cost well below £100, perhaps even more like £50 if possible.