Please help fix my error!

[sharks739]

Is there a better place to put this thread?

Thanks!

[Lotus 7]

Originally Posted by sharks739

Im using this box for talkback purposes for remote broadcasts...

The goal was for the engineer to press the button to dim program audio and add the engineers voice to the headphones of the talent. The headphone amps are Rolls PM50's and are powered by 12VDC down the ring of the TRS, signal down the tip (I modified the PS16 to only send power, as the other circuitry would handle the signal).

Most of this works, but the issue I'm having is that when the button to talk to Person A is pressed, you can faintly hear the engineer in Person B's headphones.

What can I do to fix this? Please help!

The drawing of what I have is attached.

When any of your talk buttons are pressed the engineer's mic pre output is being fed back through the resistor (so is attenuated) to the program amplifier. You either need isolation amplifiers on each of the PGM outputs or (ideally) the PGM and the engineer's mic signal should be summed in an op amp. Connecting the PGM level dropping resistor and the engineer's mic pre directly together (instead of summing them through isolated op amp inputs provides a feedback path through the pgm amp.

The mic pre output doesn't just go to the "tip". With your circuit, it can also get back to the pgm source and then be sent to the other channels. Although a pgm signal,is flowing "out" through the resistor, there is nothing to stop the engineer's mic signal from flowing back "in".

See diagram below:

[sharks739]

Lotus,

That's exactly the kind of help I'm looking for! Thanks for your reply.

I have a hard time realizing that current can flow in both directions at the same time. I always think of it like pipes and water flow, and for my normal purposes, that works well for me, but when I get into the nitty-gritty, I guess it doesn't always work the same way.

I understand now that it does indeed flow backwards, and taking your advice with some op-amps, I came up with this circuit. Does this seem like it would work?

I'm a little confused about the resistor values (I took them off of a summing amplifier schematic I found). I understand that the gain is dependent on the ratio between the feedback resistor and the input resistor, so no matter what values I have for those two, I'll get the same voltage, just less current, correct? Since the output is going to be forwarded to a headphone amp, would that current level be sufficient, or should I use less resistance for R4 and R5?

For R1 and R2, I gave them different values because I want the program level to dim when the button is pressed and the engineer speaks. Normally, I just throw a resistor in there and it takes care of the dimming, but with the op-amp needing an input resistor anyway, how do I make sure that the signal will "drop" when the button is pressed, relative to the signal out of the U2?

If I placed a variable resistor in the place of R1, would that allow me to change the level of the ducking? Would I need to adjust the remaining resistors if I went down that route?

Thanks for your help!

[Lotus 7]

Originally Posted by sharks739

Lotus,

That's exactly the kind of help I'm looking for! Thanks for your reply.

I have a hard time realizing that current can flow in both directions at the same time. I always think of it like pipes and water flow, and for my normal purposes, that works well for me, but when I get into the nitty-gritty, I guess it doesn't always work the same way.

I understand now that it does indeed flow backwards, and taking your advice with some op-amps, I came up with this circuit. Does this seem like it would work?

I'm a little confused about the resistor values (I took them off of a summing amplifier schematic I found). I understand that the gain is dependent on the ratio between the feedback resistor and the input resistor, so no matter what values I have for those two, I'll get the same voltage, just less current, correct? Since the output is going to be forwarded to a headphone amp, would that current level be sufficient, or should I use less resistance for R4 and R5?

For R1 and R2, I gave them different values because I want the program level to dim when the button is pressed and the engineer speaks. Normally, I just throw a resistor in there and it takes care of the dimming, but with the op-amp needing an input resistor anyway, how do I make sure that the signal will "drop" when the button is pressed, relative to the signal out of the U2?

If I placed a variable resistor in the place of R1, would that allow me to change the level of the ducking? Would I need to adjust the remaining resistors if I went down that route?

Thanks for your help!

Looks like you've got it. You circuit is fine as it stands. The only concern is that since both op amps are configured as inverters, in the "Normal" feed-through mode the program source is inverted in absolute polarity, but when the "talk" button is pressed and the program is routed through the summing amp, the program is inverted again (back to its original polarity) Normally, absolute polarity is not a concern for most people for headphone monitoring, but I don't like the idea of it being switched between the two modes.

I'd suggest that you use a simple non-inverting voltage follower for the input buffer (see drawing). That will save the use of (2) resistors. If you then use a single unity gain inverter (like your original input amp) after the summing amp (before the switch) the polarity will not change. It is possible to make a single op amp non-inverting summer, but for it to work properly it takes high-precision resistor values and it does not lend itself to easy gain changes with a single pot. It's just simpler to add the second op amp.

You are correct in that the value of R1 will determine the level of the program feed. It's perfectly OK to use a potentiometer (variable resistor) for R1, but you should also use a fixed series resistor to provide an upper gain limit. Put the fixed resistor between the pot and the op amp (1) input, don't connect the pot directly to the (-) terminal (you want to keep any shunt capacitance there as low as possible). The series "limit" resistor will set the loudest level and the max value of the pot will set the amount of attenuation possible.

With the 15k value you have listed for R1, the voltage gain of the summing inverter is 0.6666 so converting that voltage ratio to dB, you will have only a 3.52 dB reduction. I'd limit the max gain to unity (1.000) so would use a 10k resistor in series with a 100k pot. That will give you a possible voltage ratio of 1.000 to 0.091 and a attenuation range of 0 to 21dB.

As far as the resistor values, using 10k as the feedback (R3) is an excellent choice. Any appropriate op amp will easily drive the 1 mA or so current required with plenty left for the output drive.

The above circuit can easily be implemented using anything from the popular TL074 ($ 0.40 each or less) to a much higher performance op amp like the LM4562 (about $4.00 each). Don't forget to use some small bypass caps on each op amp power rail as near to the op amp power pins as possible.

One other suggestion: You can ignore this if you are hard wiring the output to your Rolls headphone amp inputs, but you mentioned that you were powering the headphone amps with 12v. through the rings of the TRS connectors. If that's still the case, if the TRS plugs are inserted or removed with the 12 v, power on, the 12V. supply will be directly connected to the summing op amp output for an instant. Some op amps have internal current limit circuits, but some of the higher performance chips don't and could be damaged by that sort of transient power application.

You can protect the op amps by adding a 150 ohm series resistor to the "TIP" feed after your PB switch. Since you're feeding the relatively high input impedance of the Rolls phone amps, you should not have any major signal loss or other ill effects from that resistor.

Good Luck.

[sharks739]

You are the man...

So here are a few questions for you:

1) How much quality will I lose with the cheaper op-amps compared to the high-performance ones? None of this will be in the air-chain, so if it's negligible, I can live with it, but if it's distinguishably worse, then I'll use the higher-performance amps. (I'll probably get a few of each and try them, but if you can steer me in the right direction, it might not be worth it to try.)

2) So I'm not sure which of the attached drawings is right. Once I re-read your post for the 5th time (A LOT of good info in there!) I realized that I had them as inverting amplifiers, which I don't think I want to do. As I understand, inverting amplifiers are noisier than non-inverting, correct?

So I drew one version with the resistors you specified in the version with non-inverting amps, therefore eliminating the potential of being inverted twice. Would this work?

Otherwise, I drew V1 (as evidenced by the upper left hand corner) as the inverting amp for the talkback "channel" and then I implemented (hopefully correctly) what I thought you meant for the input buffer in the Program chain.

Which of these two is the best way to go? Or did I screw it all up and need to backtrack?

This may sound basic, and I should know this, but how did you come up with the voltages and the dB reduction calculations?

Also, just to make sure I'm on the same page, I'm going to replicate this 4 times, to 4 separate outputs, and they'll all tie into the same program input, and the same mic pre output. I shouldn't have any issues at that point, correct?

Thanks a bunch for your help. I greatly appreciate it!!!

[Lotus 7]

Originally Posted by sharks739

You are the man...

So here are a few questions for you:

1) How much quality will I lose with the cheaper op-amps compared to the high-performance ones? None of this will be in the air-chain, so if it's negligible, I can live with it, but if it's distinguishably worse, then I'll use the higher-performance amps. (I'll probably get a few of each and try them, but if you can steer me in the right direction, it might not be worth it to try.)

2) So I'm not sure which of the attached drawings is right. Once I re-read your post for the 5th time (A LOT of good info in there!) I realized that I had them as inverting amplifiers, which I don't think I want to do. As I understand, inverting amplifiers are noisier than non-inverting, correct?

So I drew one version with the resistors you specified in the version with non-inverting amps, therefore eliminating the potential of being inverted twice. Would this work?

Otherwise, I drew V1 (as evidenced by the upper left hand corner) as the inverting amp for the talkback "channel" and then I implemented (hopefully correctly) what I thought you meant for the input buffer in the Program chain.

Which of these two is the best way to go? Or did I screw it all up and need to backtrack?

This may sound basic, and I should know this, but how did you come up with the voltages and the dB reduction calculations?

Also, just to make sure I'm on the same page, I'm going to replicate this 4 times, to 4 separate outputs, and they'll all tie into the same program input, and the same mic pre output. I shouldn't have any issues at that point, correct?

Thanks a bunch for your help. I greatly appreciate it!!!

Sorry, I probably fed you too much in one gulp!

The issue about the problem with the program signal polarity being reversed is because the path through the summing amp inverts, but the direct path does not. The easy fix is add another op amp after the inverting summer so that the signal is again inverted (so is returned to "normal polarity). That takes an additional op amp (note the polarity of the input pins).

Otherwise, your circuits were close except a few critical connections were left out.

The attached circuit can be replicated to as many channels as are needed. The mic pre path input impedance is 10K ohms, so adding (5) channels will create a 2 k ohm load on your mic pre which is no problem for any mic pre to drive. The program path is fully buffered because of the "U2" voltage-followers. Those will have a high input impedance so won't load your program amp at all.

Note: I've added a 100k "pull-down resistor to prevent DC drift if the input to the program path buffers is disconnected. The resistor keeps the input from developing a DC offset if it's disconnected or connected to an "AC-coupled" output.

Regarding the op amps: As you said it's not in the signal path. For musician's monitoring use with typical studio headphones, it's very doubtful that you would hear any difference between the TL074/TL084 and any more expensive op amps.

If you were using Shure 940's on each channel and were making critical mixing decisions through them, then maybe better op amps would be justified.

The voltages are determined by the ratio of the summing op amp feedback resistor (R3) to the input current source resistor(s) (R1+R6). In your original version: 10K/15K = 0.66666. Note: This is a "practical" approximation of the actual circuit. In the actual circuit configuration, the effective input resistance to the summer is really not just the 15k resistor, but must include the actual output impedance of the amplifier stage that is providing the signal. With modern equipment, that impedance is so low (probably less than 150 ohms) that the effect on the gain is only a fraction of one dB, so would not need to be considered in the basic calculation. The effect of the resistor tolerance is also not considered, but again will have a small effect on the actual gain (or attenuation) realized in the circuit.

dB gain or loss is just another way of expressing voltage ratios. The equation for calculating dB voltage ratios is: dB Voltage = 20 log (V1/V2).

For the signal levels you will have in this circuit, op amp and/or resistor noise will not be a consideration. For a TL084ACN (quad op amp) the noise should be around -90 dB, so unless you use a lot of gain in the Rolls headphone amps, the noise should not be audible at all.

Take a look at this circuit and see if it makes sense to you.

[sharks739]

Let me first start out by saying how incredibly grateful I am of your help. You are the reason that these internet forums are such a useful resource for people like me. Offering personalized advice for no other reason other than you (probably!) enjoy doing it. This world would be a better place with more people like you in it.

I understand the most recent drawing. The program polarity remains the same throughout the bottom leg of the circuit, but when the mic pre leg gets added, the program goes through the amp and gets inverted, so we need to flip it to get it the right way again. That all makes perfect sense to me, as does the pull-down resistor on the program input. Since this will be in a standalone box and the program feed will be disconnected every time the broadcast is over, the resistor should help keep some load on that line if there's nothing plugged into the program input jack on the back of the box.

What I don't quite understand is why I should use a non-inverting amp for program feed, and then an inverting amp for mic pre side. Is that because it's a summing amplifier and you can't have a non-inverting summing amp? Why not?

What I need to figure out now is the best way to supply power to the op-amps. I found out that the Rolls PS16 that I have modified, which usually accepts 12VAC as an input (that’s what the power supply they include with the unit), can also take 24VDC, which is what the RDL mic-pre runs on. So, could the op-amps run on the same 24V, or do I need to convert the voltage to something else?

As far as voltage goes, I know there are a few ways to lower voltages in the instances I had to, but what’s the best way to do so? Just create a voltage divider circuit, or use a voltage regulator?

Assuming I use the same 24V supply for the whole enchilada, the 4 or 5 quad op-amps, the RDL mic pre, the 4 Rolls PM50’s (and a circuit that I haven’t discussed, 2 12V relays in each of the headphone amp boxes), what happens if the supply I use doesn’t give the circuit enough current? Could that damage the components, or would the circuit just not work? If I gave the circuit a lot more current, let’s say 2A, would it take only what is needed, or would it start to blow components because of the current/extra heat?

I’ll be posting another drawing later this afternoon that will reflect the changes you implemented, as well as some other (somewhat) little things that I’d also like to add, but would want to run it by a seasoned veteran before I attempt it!

Thanks again for your wonderful advice!

[Lotus 7]

Short partial answer - on the run! More later.

It is possible to make a non-inverting summing op-amp circuit to mix the two signals, but it's significantly more complicated, needs careful resistor matching to work properly and does not lend itself to easy gain changes because of the required resistor matching. It's just a lot easier to use the second inverter to re-invert the signal.

Re: Power supplys: Virtually all common op amps are designed to run on dual supplys (+ 15v/-15v. or rarely +18v/-18v). They don't have a signal "ground" but may have a ground connection that is a DC offset adjustment. Although it is possible to run the op amps from a single power supply rail (+24 v or +30 v) but is generally not done or recommended because it requires DC blocking coupling capacitor on every input and output. That's a major pain since good caps are physically larger than the op amps and often are more costly. I'd find a inexpensive +/-15v regulated supply and use it exclusively for the op amps. For all (5) channels you will need to power (15) op amps total (4 quad packages with one unused). If you use TL084s the resting current is around 3 mA per op amp and in these simple circuits the peak current per op amp will be below 10 mA on each power supply rail. That means that the +/- 15v. power supplys only need to provide a total of 0.15 Amps. A small supply rated for 250 mA or more be fine. This one by MEAN WELL from Jameco is rated for 650 mA and will work

A separate supply for the Rolls amplifiers and the relays would be best. That should also be a "regulated" supply. The supply just has to be rated for the maximum current that everything can draw at the same time.

For example, if the Rolls amps need 0.2 Amps each X5 = 1 Amp max. and if your relay coils each draw 0.1 Amps then the maximum possible load would be 1.2 Amps. You would pick a supply that had a current rating of 1.5 or 2. Amps. The current used depend on the load. The load will draw as much current as it needs provided it is within the power supply's capability. Having a supply that can provide more current is not a problem. You should run the Rolls amplifiers on DC. Running 12V AC through the same cables/connectors/ground returns is an invitation to have hum in your headphones. DC in the same cables is not a problem.

It might be a good idea to check out a text on basic electronics at some point to get a better understanding of how these things relate to each other.

[sharks739]

The way the box is built right now, I have two supplies going into the box, one for the pre-amp, one for the Rolls headphone amps. That's doable, but I'd like to keep it at 2 if possible, and combine the Rolls and the RDL with the singe 24V supply.

As I mentioned earlier, I'm using a stripped down version of the PS16, which I believe has its own voltage regulator built into it. I tried just using 12V directly from the supply originally and it failed miserably. As you said, lots of hum, probably because there wasn't a regulator. Lesson learned!

If I could use this Mean Well supply and the RDL 24V supply to feed the pre, Headphone Amps and Relays, that would be the bee's knee's.

According to the website, each PM50 headphone amp draws 60mA nominal. The pre draws 25mA, and the relays I'm not quite sure about data sheet here: Allied Elec

So a 1A 24V supply should give me enough current to power all of those devices.

Is the Mean Well supply a "switching" supply? Is that what the +/- means? With that type of supply, is there an external equivalent?

With one output (at least), I'd like to be able to switch between two different program inputs. Does the following drawing look like it would work, or would I run into the same problem I'm currently having with the signals "bleeding" over?

I've also attached the Rolls PS16 schematic.

As far as learning with a text book goes. I attended school for EE for a few semesters, but had to drop out for a number of reasons, and most of the stuff I learned, I've forgotten unfortunately. I never got to work with op-amps unfortunately, so I'm fairly new in that area. I know the very basics (though some stuff gets away from me sometimes!).

As always, thank you.

[Lotus 7]

I. The Rolls power supply is fine although it is a bare minimum for powering the amplifiers. The single 7812 will provide enough current for the amplifiers as long as it's mounted on a good heat sink. Do not use more than 15 volts AC or less than 12 VAC as an input. Even though they say that it can be as low as 9 VAC, that's an error. 9 VAC will give only only 11.29 volts at the input of the regulator. The 7812 needs a minimum of 14 volts to work properly.

2. I don't have the specifications for the Rolls amplifiers, but they (Rolls) seem to want them powered by 12 VDC. Do you have a published specification sheet that says that 24 V is OK? If so then a well filtered and regulated 24 v supply can be used for both the RDL and the Rolls. If not a 7812 can be used to drop the 24 vDC supply down to 12 VDC.

3. With the current loads you've mentioned, a 1 Amp supply should be plenty unless you get some "monster" relays. Most 12 V. relay coils draw 100 mA or less. You will, of course need a dropping resistor if you want to power the 12 V relays from a 24 V power supply. You will have to know the coil current load or will have to measure the coil DC resistance to determine the resistor value.

The little MEAN WELL supply is a switching supply, but that has nothing to do with the (+/-) designation. The (+/-) means it's actually a dual power supply. It has three output terminals one ground, one that provides +15 Volts in reference to the ground terminal, and one that provides -15 Volts in reference to the ground terminal. Measuring between the two 15 Volt terminals will give a voltage reading of 30 Volts.

I haven't seen a "wall wort" type plug-in +/- 15 Volt dual power supply, buy they may exist somewhere. If you don't feel comfortable mounting the open frame MEAN WELL dual supply in a little box, you could always use (2) of these plug-in units (one for +15 V. and one for -15 V. You would have to be careful to always turn them on and off at the same time (plug them into a switched power strip, or just leave them on all the time).

The "switching" designation refers to the way the supply reduces and regulates the voltage. Power supplys can either use "linear" regulators (like the 7812) or "switching" regulators which are typically smaller, more efficient and run cooler.

4. I don't know where you got the design for the Program A/B input circuits (the U2 op amps) on your schematic, but they won't work at all and could damage the op amps if implemented. Why did you think that needed changing?

For those input buffers, PLEASE use the circuit exactly as I drew it on my last schematic. No resistors, the (-) op amp input strapped to the output and the signal applied to the (+) input with a 100K ohm resistor to ground. It's called a "voltage follower" and will work correctly in this design.

With the properly connected input buffers and the summing amps, there is no possibility of any crosstalk unless the circuit ground paths are faulty. Program-A, Program-B and the mic paths are all fully isolated

Finally, (on your drawing) the output of the Program B output inverter (U3) is not connected to the switch, but is mis-drawn as connected to the output of U2. Not only won't that work but it creates a positive feedback loop around U1 & U3 which will cause them to "latch up".

[sharks739]

Whoops!

I switched them to match your drawing originally. Then I reverted back to a file for some reason I can't recall at the moment, and then I forgot to remove those resistors!! That's my bad.

Did I make all the other changes correctly?

I talked to the people at Rolls this morning to try to order the headphone amps without the connectors or the enclosure to try to save some money and time stripping them down. With my small quantity, they couldn't do it, but while on the phone, I asked if I could power the PS16 with 24VDC and they said I could. The regulator would then provide a clean 12VDC to the amps/relays. Am I missing something?

I had the link to the relay datasheet in my previous post, but they are small little things, I can't imagine they'd take that much current.

To be honest, I feel more comfortable using the 2 external power supplies instead of the mean well supply for a few reasons. The first being I don't want to have to deal with the 120V main, and secondly, it would take up a lot of space and I'm trying to keep the box as small as possible. I could just wire up a decently rated DPDT switch built into the box and use that to power them on at the same time, correct? Are these variable voltages? Just at a quick glance, I only saw the output voltage as 15V. How do I get it to -15V? (Very basic, I know... I'm sorry.)

I altered the drawing to include a DPDT pushbutton switch so no matter what program feed you choose for that output, you only have to push one switch instead of trying to figure out which one to press. Does that make sense?

Any other thoughts before I order parts for this thing next week?

Thanks again for all of your help. You are truly a very good person.

[sharks739]

What about this supply? Expensive, but table top, and has both already built in...

At Jameco

[Lotus 7]

Close, but "no cigar".
The DPDT switching is great, but take a close look at how you drew the U2 and U5 input buffers and then take a close look at how I drew U2 in the schematic attached to Post No. 7.

Do you see any input pins on the op amps grounded in my schematic?
Which input node (+/- pin) is connected to the output?
C'mon, you can do it......

I'd re-draw it but it's better if you do it yourself.

The multiple output supply (Model PSU-50A-14E) in your link could be used, but it's really overkill for this project. It has a 5V., 4Amp supply that would be totally wasted, and the ripple/noise is pretty high (it would need some additional filtering).

I think your best bet is to go with the separate, small single supplys. Making them into a dual (+/-) 15 Volt supply is very simple.

1. Cut off the supplied "cylinder" connectors from the output cable.
2. Identify the positive and negative leads with a voltmeter and mark the positive leads.
3. Connect the positive of one to the negative of the other and make that junction the "common" or ground lead.
4. You now have the three terminals of your dual power supply. Solder them to a 3-pin connector of choice, or just run them into your chassis box through a grommet and a strain relief.

Re: Parts ordering: Although they are not usually shown on the schematic, each op amp package needs a 0.1 uF disk ceramic bypass capacitor on its +15 V and on its -15 volt power supply pins (4 packages = 8 caps). I'd also add a single 47 uF / 25 Volt electrolytic capacitor to each 15 V. power supply bus (2 caps total). Remember to watch the polarity of the electrolytic caps.

SEE SCHEMATIC BELOW:

[sharks739]

I'll post more in the morning, but I wanted to put this up tonight to ease my mind!

I can't believe I missed that. I wonder if that was part of the reverting process. I thought I had changed that.

Tomorrow I'll post a more thorough parts list for you to examine, and make sure I'm not missing anything.

I appreciate all of your help!

Did I FINALLY get it right this time?!

Thanks!

[SusanneGilliam]

Thanks for the help!
The link to Lameco really helped me solve my problem. It's not the same but similar.

[Lotus 7]

Originally Posted by sharks739

I'll post more in the morning, but I wanted to put this up tonight to ease my mind!

I can't believe I missed that. I wonder if that was part of the reverting process. I thought I had changed that.

Tomorrow I'll post a more thorough parts list for you to examine, and make sure I'm not missing anything.

I appreciate all of your help!

Did I FINALLY get it right this time?!

Thanks!

Schematic is now perfect. Congrats!

[sharks739]

Awesome!

I'll post a parts list later to look over, but I have some questions first...

As far as the dual power supplies go, I'll probably run them into a 4 pin XLR so I don't confuse them with anything else. In doing so, as long as I plug the transformers into the outlet before plugging the XLR into the box, will that effectively work the same as flipping a power strip switch? My concern is what happens if one pin makes contact before another pin. Will that damage the op-amps, or should I still have a switch for power?

As far as the caps go, I've attached a drawing, which I'm nearly certain isn't correct, but I'm not sure the right way to do it. With electrolytic capacitors, I know that the polarity must be correct, and that the positive side connects to the positive terminal, but I'm not sure if the +15V cap goes + toward the supply, and then for the -15V, the negative side goes toward the supply, or they both go the opposite, or one each direction?

Are the ceramic caps correct? I wasn't sure if they should be connected in series with the supply to the +V and -V of the op-amp, or across the +V and -V of the amps.

With regards to input and feedback resistors of an op-amp, the gain ratio is (Rfeedback+Rinput)/Rfeedback, correct?

What determines the resistor values when you design the circuit? Is it strictly current? What makes 10k ohms an excellent choice for the feedback and input reistors? If I used 5k for both, since the ratio is the same, the gain would be the same, but the current would be higher, correct?

I'm sure I'll have more questions later!

Thanks!

[Lotus 7]

Originally Posted by sharks739

Awesome!

I'll post a parts list later to look over, but I have some questions first...

As far as the dual power supplies go, I'll probably run them into a 4 pin XLR so I don't confuse them with anything else. In doing so, as long as I plug the transformers into the outlet before plugging the XLR into the box, will that effectively work the same as flipping a power strip switch? My concern is what happens if one pin makes contact before another pin. Will that damage the op-amps, or should I still have a switch for power?

A 4-pin XLR is a good choice, however, I'd control the power by switching the line power by putting both of the little power supplys on a switched power strip or plugging them into a simple short multiple outlet extension cord.

Applying power by plugging in the XLR is risky. If both of the powered pins should make contact before the ground pin, it would be possible to overvoltage a bypass cap.

Originally Posted by sharks739

As far as the caps go, I've attached a drawing, which I'm nearly certain isn't correct, but I'm not sure the right way to do it. With electrolytic capacitors, I know that the polarity must be correct, and that the positive side connects to the positive terminal, but I'm not sure if the +15V cap goes + toward the supply, and then for the -15V, the negative side goes toward the supply, or they both go the opposite, or one each direction?

Are the ceramic caps correct? I wasn't sure if they should be connected in series with the supply to the +V and -V of the op-amp, or across the +V and -V of the amps.

Oh My!
The capacitors are bypass caps that are used to provide a very low impedance power source at high frequencies. they all wind up in parallel (with each other and with the op amp power pins), but the small 0.1 uF discs need to be soldered at (or very close to) the op amp package power pins. The 47 uF caps are to provide a current source at low frequencies and can best be located at the power input connector away from the op amps.

See attached schematic. "A picture is worth a couple of hundred kilobits - or something like that"

Originally Posted by sharks739

With regards to input and feedback resistors of an op-amp, the gain ratio is (Rfeedback+Rinput)/Rfeedback, correct?

What determines the resistor values when you design the circuit? Is it strictly current? What makes 10k ohms an excellent choice for the feedback and input reistors? If I used 5k for both, since the ratio is the same, the gain would be the same, but the current would be higher, correct?

For a simple inverting amplifier or inverting summer the gain is simply (Rfeedback/Rinput) * (-1). If the resistors are equal the gain is equal to (-1).

The usable resistor "ball-park" values are determined by the output current capacity of the op amp, the bandwidth of the op amp, the signal bandwidth being processed, the op amp summing junction (- input) input impedance, the acceptable load on the previous stage, and sometimes even absolute power drain requirements (as in battery powered, portable equipment). It's complicated and takes some design experience to pick an optimal value (if one is pushing the envelope to achieve the best possible performance (S/N ratio, very wide bandwith, low current load). Otherwise, for simpler uses like making a low gain summer, it's much less critical and a lot of initial values will have similar performance.

A battery powered, low frequency device using FET op amps might use input and feedback resistors in the megaohm range, while a medium powered device operating at RF frequencies might use feedback components (they're not always resistors) with impedances in the range of a few hundreds of ohms.

Originally Posted by sharks739

I'm sure I'll have more questions later!

Thanks!

No doubt!

[sharks739]

I know I said I'd have a parts list today, but it's been hectic all day so I haven't been able to put it all together, but... I did put this together...

Please tell me if the top section (the only one I added the op-amp power for) is accurate. I also included a switch that will be a slide switch that will kill the program feed to the output that will typically be used to talk to our studio.

Also, because the inputs will be balanced, thanks to you, I now (think) I know how to use a summing amplifier, and then invert the signal back to normal polarity.

Is all of this accurate?

Thanks!!!!!

[sharks739]

That's not right...

I forgot that they are quad op-amps, and they don't all require the connection to the +/- 15V like I've shown in the schematic...

Ignore that part!

I'm only seeing the TL074 at a few places... Is there a reason it's hard to find? One of them is Futurlec, the other is Mouser.

I'll post more later!

[Lotus 7]

Originally Posted by sharks739

I know I said I'd have a parts list today, but it's been hectic all day so I haven't been able to put it all together, but... I did put this together...

Please tell me if the top section (the only one I added the op-amp power for) is accurate. I also included a switch that will be a slide switch that will kill the program feed to the output that will typically be used to talk to our studio.

Also, because the inputs will be balanced, thanks to you, I now (think) I know how to use a summing amplifier, and then invert the signal back to normal polarity.

Is all of this accurate?

Thanks!!!!!

It's time you got a better understanding of how op amps work.

You used a summing amp circuit on the TRS line inputs. As drawn, there will be no output from the first amplifier.

For a summing amp, the output is the sum of the two inputs. The signals on the TRS are balanced, meaning they are equal and opposite in polarity, Suppose that the balanced signal is 600 mV. That means that there will be +300mV on the tip and -300mV on the ring.

If you do the summation: (+300mV) + (-300mV) you get a result of 0mV (no signal at all).

To convert a balanced output to a single-ended signal a differential amplifier is used. An op amp differential amplifier will take the difference between the two inputs: (+300mV) - (-300mV) = 600mV. Any common mode voltage will be the same voltage and polarity on both lines so that will be canceled out and will not appear in the output.

Anyway, you need to make both of the TRS input amplifiers into differential amps. (circuit attached). No inverters are needed. The differential amp will drive the Program buses with the correct polarity.

One thing to note: In all of the other circuits, the resistors just set the absolute gain, and small variations in the resistor value will have little effect on the circuit operation. In those circuits 5% tolerance carbon resistors are perfectly fine. In the differential amplifier, it's very important that the (4) resistors are matched as closely as possible. Use 10k ohm 1% metal film or carbon film parts here.

The "program interrupt" switch is fine, but rather than just interrupt the signal, you should ground the open terminal of the switch so the op amp input is grounded, not "floating" when the program is off.

Otherwise, it's looking good.

[sharks739]

I think I'm starting to get it. Please tell me if this is accurate:

They are op-amps because each configuration performs a different mathematical operation. The summing amplifier does what it says, it adds the two voltages together, then multiplies the output by the gain, set by the resistors. The differential amplifier finds the difference of the two voltages, and multiplies the output by the gain.

For "summing" the balanced output to a single-ended signal, I need to find the difference between the two signals, otherwise I'll end up with nothing because summing both would result in the phases canceling each other out, and not producing a signal.

For a differential amplifier, it's non-inverting, so I don't need the second amp to reverse the polarity to normal.

Is that right?

Updated the drawing to reflect your newly imparted wisdom!

[Lotus 7]

Originally Posted by sharks739

I think I'm starting to get it. Please tell me if this is accurate:

They are op-amps because each configuration performs a different mathematical operation. The summing amplifier does what it says, it adds the two voltages together, then multiplies the output by the gain, set by the resistors. The differential amplifier finds the difference of the two voltages, and multiplies the output by the gain.

For "summing" the balanced output to a single-ended signal, I need to find the difference between the two signals, otherwise I'll end up with nothing because summing both would result in the phases canceling each other out, and not producing a signal.

For a differential amplifier, it's non-inverting, so I don't need the second amp to reverse the polarity to normal.

Is that right?

Updated the drawing to reflect your newly imparted wisdom!

Sharks,

You get a "A" on the exam. Your understanding of the summing and inverting functions is correct. An op amp is a very high gain "building block" and can be used to create all sorts of functions, both linear like a straight "gain block", a summing amplifier, a differential amplifier or many types of frequency filters. With the right additional components, the op amp can also be used in non-linear circuits like compressors or limiters, absolute value circuits and even "sample and hold circuits" for holding on to a signal for the tiny fraction of a second that a ADC needs to digitize the value of a sample.

Also, the differential amplifier configuration can be used to generate either an inverted or non-inverted output (referenced to the balanced signal on the TRS jack) by simply interchanging the wires going to the tip and ring. Normal convention has the initial positive-going signal on the tip, so it's usual to make the single-ended output positive when the tip phase is positive.

You get a "C" on the schematic. Please take another look again at my drawing. Each differential amplifier uses (4) matched resistors. One set seems to have slipped off your implementation.

[sharks739]

DOH!!!

At least I'm learning! That's what I'm here for. The design help is great, but that doesn't really help in the long run. I need to be able to understand what is going on and why.

I fully admit trying to dive into the deep end to see if I can sink or swim, and I'm barely keeping my head above water, but I haven't sunk yet.

The problem I have is that I want to learn how all of this works, but I need a mentor to help guide me along, because there are concepts where I'm just lost. Some of them are very basic too, which really frustrates me because I know it's easy to grasp, but I'm just not able to wrap my head around it! That's why I'm incredibly grateful for your help and advice, even with my dumb questions!

I've attached another updated schematic, this one, hopefully is my building plan!

Am I finally right??????

[sharks739]

Assuming that everything is correct, I put this parts list together. Is there anything else that you think I might be missing?

I didn't include an enclosure or switches, as I may already have an enclosure that might work, and I have plenty of switches laying around.

Thanks again for all of your time and patience!

[Lotus 7]

Originally Posted by sharks739

DOH!!!

At least I'm learning! That's what I'm here for. The design help is great, but that doesn't really help in the long run. I need to be able to understand what is going on and why.

I fully admit trying to dive into the deep end to see if I can sink or swim, and I'm barely keeping my head above water, but I haven't sunk yet.

The problem I have is that I want to learn how all of this works, but I need a mentor to help guide me along, because there are concepts where I'm just lost. Some of them are very basic too, which really frustrates me because I know it's easy to grasp, but I'm just not able to wrap my head around it! That's why I'm incredibly grateful for your help and advice, even with my dumb questions!

I've attached another updated schematic, this one, hopefully is my building plan!

Am I finally right??????

Looks Good! I don't see any remaining problems. Before building the whole thing, you might want to breadboard up one channel to be sure it works to your satisfaction.

Also, since (all but the last) the mic buffer/summer/inverter "blocks" use three op amps, I'd just use one quad package for each channel and forget about the unused one in each channel. That way the channels won't be sharing any packages. To avoid any possible oscillation or excess current draw in the unused op amp you need to connect it as follows:

1. Connect the output pin to the (-) input (just like the voltage follower input buffers). Be absolutely sure it's the (-) input pin. Make no other connections to the wire between the output and the (-) input. It must remain "floating".

2. Ground the (+) input pin.
That will safely "disable that "unused" op amp. (without some kind of a negative feedback path, the op amp will go to full gain and will become unstable)

Unlike a "normal" mic pre or power amp input., when using an op amp you NEVER short the (+) and (-) inputs together.

From the start I've wondered why you need to keep sending the (reduced) program feed while the engineer is talking to the talent. It would seem that the talent would have a better chance of understanding the engineer if the program feed was just cut off while the engineer was communicating with them over a quiet background. Most of the circuitry required is only needed to provide that feature.

[Lotus 7]

Originally Posted by sharks739

Assuming that everything is correct, I put this parts list together. Is there anything else that you think I might be missing?

I didn't include an enclosure or switches, as I may already have an enclosure that might work, and I have plenty of switches laying around.

Thanks again for all of your time and patience!

The parts list looks good. If I were building it, I'd use sockets for the op amps (if one ever fails it will make replacement much simpler, plus it avoids the risk of overheating them during soldering). Be sure to carefully mark the orientation of the socket - identify which end has pin-1.

You will also need a PC board to mount everything on. For one-off projects like this I usually use a plated through hole prototype board like this. There are many variations (different sizes, different numbers of holes. etc.), so you should take a look at the various configurations and pick one that will allow you to wire up everything with a little room to spare. At Jameco, search for "prototype PC board".

Good Luck!

[sharks739]

The reason that the talkback is dimmed, and not totally cut off, is because often times the producers will be talking to the talent while they are still talking themselves. In the current iteration of the box, I have some resistors in there that apparently have been too strong for some talent because they got really annoyed when the producer spoke to them because they could no longer hear themselves. It's almost as if you're off air if you lose your headphone feed.

Unlike in a studio where you aren't live and you can talk to the performers whenever, we need instant communication. Losing a headphone feed could make a person stop talking, and on air, that can sound bad. Most of the people I work with are able to understand what's being said to them when they're still speaking, and won't pause their words. Some can even communicate non-verbally when you are in their ears and they are still speaking!

Also, it helps to keep some program in their ears to listen to the analyst speak for example. If you are talking to the play-by-play guy when the analyst is talking, when the headphone feed gets cut out completely, he no longer knows what the analyst is saying and won't be able to formulate an accurate response, or worse, (if it's a bad producer who doesn't stop on time) when the analyst stops speaking and there's awkward on air silence! With some level still in there, he can hear both at once and know what's being said.

As I went to sleep last night, I had the same thought you did about sockets. I'll be sure to add some of those to the list. There's a guy who builds very nice stuff for the broadcast world who specifies on his website that all ICs are in sockets to make for easy field repair, which is great! But I'm sure that it also helped keep his overhead down a bit from frying IC's with too much heat! I have plenty of protoboards around my workbench from projects that I have planned but didn't get to, but I appreciate your thorough scanning of the parts list! Is there anything else you think I should add?

Now that I've sorted this part out, I have a few other questions about some other parts of the project that range from very basic to maybe a bit more complex. I don't intend to implement them necessarily, but it would be nice to understand how it's accomplished and what's necessary to get it done.

My first question has to do with this attached schematic, which is for the Rolls PM50 headphone amp that this boxes will be connected to. I have a bunch of these laying around, and they are fantastic, which is why I use them for everything, but they aren't flawless and here's why: They have a pot on them for Mic Level, which is a "more me" function. It's to control your own microphone level to your headphones. If you need to hear more of your voice in the mix, you crank up that pot. The problem is, there isn't enough gain on it. So looking at the schematic, I see that there's a 49.9kOhm feedback resistor for the op-amp attached to the mic jack. So, the gain of this op-amp would be (Rfeedback/Rinput)*(-1)=(49.9k/1k)*(-1)=-49.9k. Correct? To get more gain out of the op-amp and more level into the headphones, would I change the value to be a lower resistor, so the output would be more positive, or since it's voltage, the larger absolute value matters, so a bigger resistor would be the right choice?

Thanks!!!

[Lotus 7]

sharks,

The mic amp part of the circuit (U1B) is operating with a voltage gain of 49.9 (approximately 34 dB). And if you have the mic level pot (P1) all the way up, the combination summer and headphone driver adds another 4.5 dB to everything. As an aside, it's interesting that they use one op amp (U1A) to drive the left channel and use two in parallel (U2 A+B) to provide the same summing function on the right channel. I guess they just wanted to do something with the extra available op amp).

Anyway, to get more mic gain, as you've surmised you want to increase the gain in U1B without driving it into clipping. To do that R3 and R4 must increase in resistance while remaining identical to each other OR R1 and R2 must decrease in value (again while remaining identical to each other). (it's a balanced differential amplifier). All the caps are required because of the single-sided power supply which requires that the common reference for the audio signals is a floating bias voltage supplied by R6, R5 and C3.

Because of the current value of R1 and R2 (1k ohms each), there is already some additional loading (2k ohms) of the mic (in addition to the input load from the real mic preamp. I would not like to see it go any lower, so reducing R1, R2 is not a good option. That leaves increasing R3, R4. At 49.9 k each the mic amp has a gain of 34 dB. Increasing the values will slightly increase distortion and may begin to roll off some very high frequencies, but only in the talent headphone mix.

Going to 100k for R3,R4 will give you 6 dB more level on the mic mix. Going to 150k will give you 9.5 dB more mic gain. Any more is probably risky in terms of distortion and the risk of clipping if you plug in a mic with high output. With those higher values, the input capacitance of the op amp starts to become a factor. the combination of the large resistor values in the feedback loop and the op amp input capacitance forms a low pass filter (in the feedback loop )that has enough phase shift that oscillation is a possibility. That means a small compensation cap is needed in parallel with the high value resistors.

For 100k use the same 27 pF that Rolls uses on their other op amps. For a 150 K resistor you will need to go up to 43 pF. "Piggyback" those little disc caps directly on the feedback resistors. Keep the leads as short as possible.

Theory: Increasing the values of R3,R4 decreases the negative feedback current which is balanced against the signal current being fed in through R1,R2. Less negative feedback = more gain (and unforunately, more distortion and less bandwidth).


If the 6 to 9.5 dB is still not enough, you can squeeze out a few more dB (about 3 dB) by increasing the gain in the summer/drivers by changing ALL the 22k resistors (R8, R12, R10 and R11) which must remain matched) to 10k. That will increase the gain of both the mic and the program feed, but then the program feed can be reduced a little with P2A,B and the mic will still be louder.

Hope that helps.