Passive Monitor Controller: First ever design from rank n00b

[spacelabstudio]

I've been meaning to build some gear for my studio. Don't have much experience building gear and have never designed anything. I mostly intend to just borrow tried and true designs and adapt them to my ends, which is pretty much what I'm doing with this guy:

http://spacelabstudio.com/gear/proj/MonitorController.gif

It's really just a passive mixing network followed by a stepped attenuator. It's balanced throughout. I have three stereo input channels, each with a mono switch, and then I send the four busses to a stepped attenuator and then a mute switch in front of the speakers. Although I don't currently have a need for a speaker select switch (only one pair of speakers in my room) I may add one for future proofing.

I'll probably start with just this, but as you can see, I think I can probably take a tap of each bus into a buffer amp, for then using for whatever active electronics I might want to add eventually: a headphone amplifier, a talkback circuit, maybe just an aux output, whatever....

Will this work? Am I doing anything really dumb? Will the attenuator screw with output impedance to an extent to cause problems?

I am a rank n00b! I have no idea if this will actually work or not!

I appreciate any input or feedback you experts can give me.

Thank you!
Chris

 

[solder_city]

for one thing, the 'mono' switch is actually wired as a 'mute and burn up your source' switch :?

 

[NewYorkDave]

Yeah, shorting an output is a no-no.

In my opinion, the Prime Directive for any monitor controller is that it must not disturb or degrade the signals being monitored. Passive controllers may be fashionable at the moment but if you agree with that Prime Directive, it's hard to deny the advantages of high-quality active circuitry--especially if any kind of mixing (i.e., a mono mode) is to take place in the monitor controller.

I'd be inclined to use balanced-input buffer amplifiers at each input, with an unbalanced, passive source select and volume control circuit following them. This arrangement would satisfy the P.D. while also conforming to the other main imperative: minimum circuitry that can add noise, distortion or "coloration." An unbalanced output is fine for a monitor controller in typical control room situations--fairly short cable runs, not many sources of interference in the vicinity, etc.

If you wanted to be hardcore, you'd use bridging transformers on the inputs to provide galvanic isolation between the lines being bridged and the monitor controller. A real bridging transformer incorporates a resistive pad to isolate the line from the transformer's primary reactances. They used to be available as a single unit with resistors built-in but nowadays you'd probably have to make your own using an external pad. Of course, bridging a line in such a way entails make-up gain at the monitor side.

 

[spacelabstudio]

[quote author="NewYorkDave"]Yeah, shorting an output is a no-no.[/quote]

Yeah, what kind of ******* would do that? (Thanks for the catch, guys.)

[quote author="NewYorkDave"]In my opinion, the Prime Directive for any monitor controller is that it must not disturb or degrade the signals being monitored. Passive controllers may be fashionable at the moment but if you agree with that Prime Directive, it's hard to deny the advantages of high-quality active circuitry--especially if any kind of mixing (i.e., a mono mode) is to take place in the monitor controller.[/quote]

So, in my naive view, a purely passive, resistive network would be completely transparent whereas any sort of active circuitry would have some inherent color and be more difficult to do right. Why am I wrong?

I knew this would turn out to be more complicated than I thought . . . .

chris

 

[Kamel]

purely passive resistive network is transparent at audio frequencies, due to the fact that we can avoid capacitive and inductive effect in resistor.

in theorical life, resistance are perfect
in real life they lower the cmr due to their difference of value
they insert a little bit of noise (carbon pot too)
and las but not least, they burn a little bit of signal, so you have to, like dave wrote, add a make up gain (active)

 

[NewYorkDave]

See my "prime directive" above. It's crucial to isolate the lines being monitored from any corruption by the switching, mixing, etc. networks in the monitor controller.

You can get a reasonably good degree of isolation by passive means, it's true, by using transformers and large resistances in the networks. But the loss would be such that you'd have to use substantial amplification to bring the level back up, so you still haven't avoided additional amplification in the monitor path. You've just moved it downstream where the S/N ratio will be worse.

Anyone who's been around here a while knows that I'm not adverse to using passive networks, but sometimes active circuitry presents advantages that are hard to deny.

 

[the scum]

A quick question about your design intent:

Is there a specific reason that you're mixing your sources in this device, rather than just switching between them?

 

[spacelabstudio]

[quote author="the scum"]A quick question about your design intent:

Is there a specific reason that you're mixing your sources in this device, rather than just switching between them?[/quote]

Personal preference. No real reason, other than I'm used to the monitor section on my mackie which lets you assign several sources at once to the CR output. I'm leaning back towards switching, actually, having thought about it some more. Just because I can doesn't mean I should, and what do I really gain? I'll still need to at least sum L+R for the mono switch, though.

I ran across the Coleman Audio stuff today and it looks like they're doing pretty much what I want to do.

chris

 

[NewYorkDave]

Notice that the Coleman uses active circuitry to provide the L+R sum, which is a good idea.

If you insist on not using active circuits in the main monitor path, you should still include some series resistors even if you're selecting between sources and not mixing. The series resistors will help to prevent clicks/pops from coupling into the lines you're monitoring.

 

[spacelabstudio]

[quote author="NewYorkDave"]Notice that the Coleman uses active circuitry to provide the L+R sum, which is a good idea.

If you insist on not using active circuits in the main monitor path, you should still include some series resistors even if you're selecting between sources and not mixing. The series resistors will help to prevent clicks/pops from coupling into the lines you're monitoring.[/quote]

Thanks for your help. I've been busy poring over this thread, here, too:

http://www.soundonsound.com/forum/showflat.php?Cat=&Number=118645&page=0&view=collapsed&sb=5&o=&fpart=3&vc=1

I hope you guys don't mind me peppering you with questions and schematics. It's occurred to me I could probably do some simple, fixed attenuation prototypes just to experiment. See if I hear any problems or not.

chris

 

[NewYorkDave]

Although I still think buffered inputs are the best way to go, I'm drawing up an idea for a "minimally invasive" passive controller with active mono summing. I'll post it sometime soon.

 

[spacelabstudio]

[quote author="NewYorkDave"]Although I still think buffered inputs are the best way to go, I'm drawing up an idea for a "minimally invasive" passive controller with active mono summing. I'll post it sometime soon.[/quote]

Wow. Sweet! I'm not married to passive, unbuffered inputs. I get that cmrr would be better in an active design and I'd be able to use unbalanced topology after the buffer. I was just afraid that in my n00bness I'd mangle the output of my Mytek DAC with a poorly implemented op amp design--passive just seems easier/safer in terms of just getting an attenuated signal into my fixed gain powered monitors.

I was doing some measurements last night to figure out how much attenuation I want in my design. I measured RMS output of a 1k sine wave at -1dbFs from the Mytek (3.6V). Then routed the Mytek through my current monitor chain, turned up a mix louder than I would ever want to listen to it in real life, then ran the same sine wave through that chain and measured the RMS: 0.25V. So, right off the bat I can have an attenuation of 23db at the *loudest* setting and be ok. Maybe make it closer to 1/10 (20db) or 1/8 (18db) for a little extra room at the loud end.

Remember, my monitors (Tannoy Reveals) are powered and fixed gain, and you can see from these numbers that I'm getting way more make up gain than I really even need out of my monitors. You can see why I wasn't worried about the loss from summing, although I am worried about showing wacky impedances to the DAC and/or monitor amps.

Anyway, assuming a passive attenuator and no summing, still, a minimum attenuation of 8 (18db) and a 24Kohm impedance (arbitrary number to make math easy) then I ought to be able to use a 21k series resistor into a 3k stepped attenuator. Does that make sense? Any gross errors in logic?

chris

 

[NewYorkDave]

OK, I finished the sketch of my idea:

21kB PDF

the 2.37K series resistors provide a reasonable degree of isolation of the monitored line from the switching circuitry, while still keeping loss and output impedance within manageable limits for most monitoring setups.

The number of inputs and outputs can be expanded or reduced to any number you like; the main limiting factor is the availability of multi-pole switches. But there's no reason why you couldn't do it with relays. In fact, if DPDT relays were used for the inputs, a 10K resistor could be switched-in in the "non-selected" state to keep the load across the line roughly the same whether it's being monitored or not.

I double-checked but didn't have time to triple-check the drawing, so please let me know if you spot any errors.

 

[spacelabstudio]

Wow Dave. Thank you! This is awesome.

Is there anything special about the 5k number for the attenuator? I probably want to put a resistor in series with that guy to knock the level down a good bit more, per my earlier post. I can try to work it out so that the total resistance would still be 5k, with the attenuator using only the bottom 1k or less, or I could maybe increase the total value to give myself more room at the bottom end, like maybe a 10k resistor in series with a 2k attenuator?

The load in my case, the monitors, is measuring around 33k input resistance on my ohm meter.

So using your 5k, it looks like 7.37 input impedance and a 2.37-7.37 output impedance. So there should be room to go a little higher? I don't know how to figure the impedance for the mono piece.

chris

 

[NewYorkDave]

Sometimes it's easier just to draw a picture:

Big Dave presents "Fun With Equivalent Circuits"

This should clarify things. Beyond that, someone else is gonna have to jump in because that's about all the time I can spare for this.

 

[Gold]

[quote author="NewYorkDave"]
This should clarify things. [/quote]

Wow, thanks. I've been looking for something this concise and focused for years.

 

[ChrioN]

[quote author="NewYorkDave"]OK, I finished the sketch of my idea:

21kB PDF
[/quote]

Why not put a 100k resistor to ground, and then a ~5uF film cap at the input of the opamp? This way you could take away the output electrolyt...

 

[NewYorkDave]

Argggh... I spotted a typo in the "Equivalent Circuits" PDF. In the equation for worst-case output Z, "2/Rhunt" should read "Rshunt/2". I've pulled the PDF and will re-post it after I've had a chance to correct it, probably sometime tonight.

This way you could take away the output electrolyt...

Nope. There'd still be some amount of offset at the output, which could (depending on its severity) make for a jarring experience if you switch to Mono mode with your monitor amps turned up.

 

[spacelabstudio]

[quote author="Gold"][quote author="NewYorkDave"]
This should clarify things. [/quote]

Wow, thanks. I've been looking for something this concise and focused for years.[/quote]

Yeah, that makes sense to me. Thanks!

chris

 

[NewYorkDave]

Corrected "equivalent circuits" PDF has been uploaded