An ideal line input stage

[Monte McGuire]

I have veroboard with copper on one side.  I guess it's time to at least consider PCB.

ps. Ever use LC passes on the audio input for RFI filtering?

You're welcome!

A sleazy trick, which can avoid complex PCBs, is to tape a single sided un-etched board to the back of your circuit board to make a fake "ground plane". You have to somehow meaningfully connect your ground to this add-on board, hopefully at a number of places so that your circuit has low impedance to the board. Maybe you could drill holes from many ground nodes through the Vero board to the un-etched ground board and use short jumpers to stitch it together?

Barring that, I've found that even placing a board over a grounded un-etched board will provide a plane that will tend to eat up a bunch of RF. I've done this with single sided SMD - I didn't have component leads sticking out that could short out the real circuit board, so I can keep the fake ground plane facing up, making it easy to stitch to the real PCB, but you get the idea. My bench is on the 2nd floor of a metro Boston building, only 5 miles from an antenna farm, but with a lot of copper planes in and near my circuits, I have zero problems with RF and hum, despite having no enclosure for the circuits I test.

PCBs are great, but it'll take you a while to learn the software and built a library of footprints that work. For now, that might be tough, but it's worth considering. I use Eagle along with PCB Library Expert to make footprints, since I mostly use SMD, and it works well now. Still, it took me literally a couple of years to get fully up to speed. But, once you get SMD down, it's beautiful to be able to use a stencil to print solder paste, place the parts, and then reflow the board and watch everything magically flow and align perfectly.

As for LC RF filters, I've experimented with this a bunch, and found that simple RC filters are safer and often more effective. The basic problem I've found is that ferrites or lossy inductors can 'eat up' RF without conducting it to ground, which is great, but these parts, when used as a series element, tend to add distortion. The parts that add less distortion tend to be the ones that have low 'real component of loss'. In other words, they're basically clean inductors. So, if you use lossy ferrites, then you risk distortion from signal currents - the loss that appears to be resistive at RF is actually from nonlinearities that can bite you in the audio band.

So, if you then decide to use clean ferrites or wound inductors, you find that you need external resistors to add loss - a clean LC circuit simply rings, and does not get rid of RF. You can actually get RF "gain" out of them if you're unlucky. So, you're adding R to the LC to get loss, and you're worrying about inductor linearity, and at the end of the day, a simple RC filter can do just as well as an RLC. RC filters are much cheaper than RLC, much more linear, they won't ring at unfortunate frequencies, and they're physically smaller. So, I've given up on ferrites and wound inductors for RFI filters, and haven't been unable to do whatever I needed to do without them.

 

[Newmarket]

> bare bones differential ....few discrete

That's sort of simple and 'cool' as the Fonz would say 

 

[Newmarket]

Am I right thinking I could take the output off of the collector of T1, and the output off of the collector of T2, feed both to a double throw switch and use it for unbalanced phase reversal?

I've noticed a need for that function and decided to incorporate it.  Ideally I'd like to avoid additional circuitry.

Adam

No. That won't work.  The circuit isn't 'symmetrical' as the summing is done at the lower transistor only. The top transistor is simply acting as an inverter.
Drawing it out as an opamp circuit (2 x inverting stages with the summing link as shown) should make it easier to think about what happens.

 

[Newmarket]

Yes I'd like to understand the circuit better.  I'll mull it over again later today and see if I can get some clarity on it.

Adam

I'd suggest to simulate to see what you're dealing with.
LT Spice seems a favourite around here. I find it a bit difficult to get into. I'm sure others will have a different opinion  ;D
For an easier learning curve I'd suggest the free TI version of TINA (ti.com) and/or the free version of ICAPS. Also the simulation in Altium Designer if you happen to have access to it though I'm guessing that's unlikely.

 

[PRR]

I'll disagree. You should be able to visualize what this does in your head. If not, you need to review basic amplifier and resistor action.

SPICE is sometimes useful to give a second opinion on your analysis, and to give 8-place "precision" numbers to impress your boss.

 

[Newmarket]

I'll disagree. You should be able to visualize what this does in your head. If not, you need to review basic amplifier and resistor action.

SPICE is sometimes useful to give a second opinion on your analysis, and to give 8-place "precision" numbers to impress your boss.

Well I sort of agree with you. I was going to suggest a quick catch up with Horowitz and Hill Art of Electronics and other texts - but I thought my post might sound a bit too 'prescriptive' then 
Different people best understand things in different ways so looking at simulation results can reinforce the understanding and get stuff in focus.
I was really thinking simulation simulation to see the basic voltages / polarities / signals etc. rather than extreme precision.
Hardware building and bench instrumentation arguably gives better experiential learning but depends on having the 'scope channels etc...

 

[Newmarket]

I drew the circuit by PRR a different way and started tracing what I think is the flow... I think I'm getting a little more clarity.

Both negative - copies show up on the collector of T1, while meeting at the base of T2 as well.  The 100K/100K is some kind of magical 0V due to the cap and resistor divide.  They mix happily there, and show up on the collector of T2 as a positive signal.  If there's different signals inputted they grow/morph together, but if it's the same balanced mono signal with + and - going in the level stays the same.

Hopefully I've got this. 

I tried putting a L and a R stereo mix in, a The Guess Who song, to see what would happen.  So obviously the L and R of the mix were different waveforms.  Burton Cummings' vocal got canceled out and the rest of the band went through.  His vocal must have been mono.  Maybe it's not that the vocal got canceled but the rest got amplified because it was different?  The vocal did sound very low though.

I realize this isn't meant for taking L and R of a mix, I'm just trying to understand what the heck goes on. 

Also, why is there no resistor from emitters to ground?

Adam

The emitters are at 0V. (No resistor from emitter to 0V as you point out so there is no emitter degeneration). That defines the voltage at the base. Take it from there...