Need help for a little mixer

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

MRecords

Well-known member
Joined
Jan 9, 2009
Messages
65
Location
Austria
Hello to all!

First i wanna thank you all for this great forum. I´m a real newbie but i´ve learned alot in reading in this forum. I´ve build 8 of the 9K Preamps and they sound great. Better than anything i had before.

Now i´m planning to do a little mixer where I can put different Preamps and EQs in it (51x) and maybe some AUX sends for different headphone mixes. The direct out´s per channel should be made in the same way like the AUX send with better spec´s but this comes later. I also wanna have the option of analog summing so there will be an input after the preamp. 16 to 24 channels with forsells active circuit. I thougt of 9K2 for the summing resistors for 24 channels.

Can somebody help me with the first step of this design before i go on with it. The fader buffer is from the Midas XL3 but i´m not shure with R11. The Panpot is the one from NewYorkDave.

Thank you all again.

Cheers Martin
 

Attachments

  • Mixer Channel.pdf
    11.3 KB
Thank you.

I know for you it´s a stupid question but i´m learning much with doing that and thinking about how could it work.
 
No it is not a stupid question.

Although it is a very simple circuit, it can turn this into quite a long thread to complete the analysis. But since you seem not to know the opamp basics, the great majority of the advance stuff will go over your head. Therefore I suggest you start reading about operational amplifiers. There is plenty of information on the net.

However, this is called a non inverting amplifier, because the signal is applied to the non inverting (+) input of the opamp. So the signal on the output of the opamp has the same phase to the one applied to the input. The other input (-) is called inverting. Totally opposite. Signal comes out as inverted in phase.

C1's job is to block any possible DC from the previous stage entering the opamp and buggering up the biasing.  Also causing other troubles. R3 provides a DC path to the non inverting input. But also sets the input impedance of the opamp. Therefore, as a rule of thumb it must be at least 10 times the impedance of the signal arriving from the previous stage. Also because it completes a voltage divider network with the signal impedance and attenuates the signal.

Now, because the signal goes through the fader, the resistance of the fader adds up to the impedance of the signal arriving from the previous stage. Although when the wiper of the fader is in full contact with the signal (at maximum volume) there is no added fader resistance, the fader is now full in parallell with R3 (at AC). Their overall value is going to decrease, reducing the input impedance. Also as you pull down the fader, the value of the high side of the fader will start to add onto the signal impedance. So, to be on the safe side you select R3 as 10x fader value, hence 56K is a good one. As the value of the low side of the fader is going to be even smaller, it will not bother anybody.


A little intermediate info here. You can certainly make the R3 higher value to increase the input impedance but you will not gain anything significant in terms of signal attenuation. However, you will be increasing the noise of the amplifier, because the resistors generate a thermal noise (you will also come accross it as Johnson noise). So all the masters here will tell you that one of the first rules of designing good audio is to keep your resistance values as low as possible.

R4 and R6 form the feedback network by means of a voltage divider and set the gain of the amplifier. Because the feedback network puts back some of the output signal into the opamp, again there is the danger of high frequency oscillation. So C2 is used to short the signal at high frequencies beyond audio. It effectively shorts R6 and makes the amplifier unity gain at that particular high frequency.

C4 is called a "compensation capacitor" to optimize the frequency response of the opamp. Acording to the data sheet 5534 is internally compensated equal to or greater than gain of three. Therefore if the gain is set to below this level then it would have to be compensated to run stable. Now, calculate the gain and find out.

Now we have R5. Because the opamp is driving the signal into C3, it is referred as capacitively loaded. Because C3 becomes load on the output. Again there is the danger of instability in the form of oscillation. Therefore, R5 is used to isolate the capacitor so that opamp actually does not see it at AC.  The value does not have to be very precise. Again the data sheets quite often suggest a value or values.

R11 is now obsolete.

You are using R2 in series to increase the load impedance as both gangs of the panpot is in parallell and their values are effectively halved. R7 and R8 alter the law of the pan-pots at which R9 and 10 also have a share.

Pan-pot operation is simple and self explanatory. But there is a cheaper and terrible way of doing it with a single potentiometer.




I have amended the text to better the link between the feedback and the use of compensation capacitor.
 
Thank you very much. That helps alot.

If i´m right the gain of the opamp is 2.2 =6.8dB. So it´s under the internal frequency compensation of the 5534. So i think 3.3pf is a god value. Or would 10pF be better?

Is it okay to change the opamp with an OPA604? It´s unity gain stable so i don´t have to think about oscillation from the opamp.

On the Midas XL3 schematic they use JFET (J112) on the AUX busses(see below). But they have 16 channels. For me 6 channels are enough so i think i don´t have to use it. Like explained in another thread the signal flow should be:  low impedance unbalanced signal --->pot--->AUX bus--->summing amp--->balancing--->output. Am i right?
 

Attachments

  • XL3 AUX.pdf
    15.4 KB
MRecords said:
If i´m right the gain of the opamp is 2.2 =6.8dB. So it´s under the internal frequency compensation of the 5534. So i think 3.3pf is a god value. Or would 10pF be better?
I would use 10pF rather than 3.3 Any parasitic capacitance or reactive load can unbalance the precarious equilibrium of a critically compensated opamp...
Is it okay to change the opamp with an OPA604?
The OP604 is too expensive to use in this application that will never do it justice.
  It´s unity gain stable so i don´t have to think about oscillation from the opamp.
You might as well use a TL072, dirt-cheap, unity stable and quite adequate for the level of performance you can expect to achieve. If you want to spend money on opamps, reserve them for the summing stages, where noise voltage and high bandwidth are critical.
On the Midas XL3 schematic they use JFET (J112) on the AUX busses(see below).
The FET's are not used to amplify or buffer the signal; they are used as global ON/OFF switches.
Like explained in another thread the signal flow should be:   low impedance unbalanced signal --->pot--->AUX bus--->summing amp--->balancing--->output. Am i right?
Yes, correct. If it wasn't someone would have already sneered at you and made you miserable... :D
 
Thank you all. You´re very kind to me. I never learned so much about electronic as in the past few days.

I´m going on with planning this project and also with having questions. No stress. Only if anybody has nothing else to do :)

I changed the fader resistance to 10K because they are laying around at home. The fader loss will be 10dB(?) in normal operating position so the gain of the buffer should be 3.13= 9.9dB. Also R3 should be about 120K as sahib explained.

I´ve added a pan pot buffer to keep the impedance for the summing bus low (see below). The loss at the pan pot is 6dB at full, 9,4dB at center(?) so the gain of the buffer is about 2= 6dB. I don´t know if it´s better to use a panpot buffer or not. But to keep the impedance low it make sense to me.

And i think i don´t need R87 and R88 on the AUX buffer because DC discharge will be done by the Level Pots in the AUX busses. But i´m not really shure.

If somebody could help me out again. Thanks
 

Attachments

  • Mixer.pdf
    28.5 KB
You don't need the resistor right before the pan-pot (R2, R101...), then you have more level on the busses, good for S/N ratio. Pan-pot loss would be 0dB, 3.5dB at center.
But, why stop on the way? By just adding two resistors, you get the Soundcraft Active Pan Pot.
 

Attachments

  • Souncraft Active PanPot.jpg
    Souncraft Active PanPot.jpg
    50.2 KB
Hi Wolfgang!

Yes I´ve done a few changes on the channel and draw the summing stage. The summing section is the one from radiance in this thread: http://www.groupdiy.com/index.php?topic=27992.0

I hope it´s ok if i use this.

Where do you life in Austria?


 

Attachments

  • Mixer channel.pdf
    32.4 KB
The Insert switching is not adequate. You need DC blocking caps if you want it silent. See attached schemo. You don't need two switches. You want to have the SEND active all the time (compressors and time-related effects need it).
I see you have specified tantalum caps for PSU decoupling. I wouldn't; for me tantalum has always been synonymous with disaster in conjunction with bipolar power supplies (even with Schottky reverse-voltage protection diodes).
The headphone amp needs some more thinking. As it is, it will be satisfactory only on a restricted selection of headphones. Just check a recent thread where PRR expained everything in detail.
http://www.groupdiy.com/index.php?topic=34983.0
 

Attachments

  • INS SND&RTN001.jpg
    INS SND&RTN001.jpg
    132.1 KB
Good progress!!!!!!

The mixer is growing!!

I dont understand R14 and R15 at the Panpot. I think they should be connected between Signal from the Faderbuffer and the wiper of the pot????

I am living in Graz and you?


regards,
Wolfgang

 
Martin, the 1646 has 6dB gain, so the operating level is -2dBu inside for +4 outside. The mater buffer needs to have at least 6dB gain. I would suggest you provide some trimmability there.
If you want silent mono/stereo switching you need DC blocking caps there. Again, you don't need to switch off the mono sum circuit, it should be there all the time, just switch the output.
Am I correct you are using Eagle for your schematic dgm?
You should nomenclature the IC's U1, U2,...
 
Quote
Like explained in another thread the signal flow should be:   low impedance unbalanced signal --->pot--->AUX bus--->summing amp--->balancing--->output. Am i right?
Yes, correct. If it wasn't someone would have already sneered at you and made you miserable... Cheesy

I would always think about buffering the input for the sake of always having a known impedance driving your pot.  But that's just me.. and I don't want to sneer..   ;)

EDIT:  nevermind, I continued to read and look at the newer schematics and see that's already taken care of.
 
Hello guys!

Sorry for my late reply but all that christmas and new year stuff...

To abbey road d enfer: I can´t say it enough thank you for your help. It always need some time until i really understand what you mean. You know i wanna learn and not only doing what anybody said.
I tried to do all the things you told me.

For the DC blocking caps on the insert what value should they have. Is the only important thing that i don´t create a high pass or are there another things to take care off? 

I changed the switching on the insert and the mono cicuit so that it´s always active.

The power caps changed to Film/Foil ones or anything else.

The headphone amp is the one from the thread above.

Are the dc blocking caps at the mono circuit at the right place. The value question is the same as above.

I didn´t rename the IC´s but maybe i will next time.

Theres another big question. What is if i change the THAT1646 and 1246  to 1640 and 1240. Do i need all the same a trimmer there or can i leave it out. I don´t really understand what you mean with: The mater buffer needs to have at least 6dB gain. Do you mean the OPA604 or do you mean the 1646. My english isn´t the best.

Thanks and cheers Martin
 

Attachments

  • Mixer Master.pdf
    47.6 KB
MRecords said:
For the DC blocking caps on the insert what value should they have. Is the only important thing that i don´t create a high pass or are there another things to take care off? 
These caps are typically driving 10kohms nominal value loads. So with 10uF, it creates a high-pass at 1.6hz. Which is fine. BUT: When the signal has travelled from the input to the output, it has gone through about 8 of these high-pass filters, the result being a 5.4Hz turnover frequency. Or -0.9dB @ 20Hz. Which is not too bad, but may be considered too high by some purists.
So I would recommend using a larger value; the cost difference is almost irrelevant. Using 47uF or 100uF, you are sure to avoid any LF problem, particularly when the caps value decreases with age.
And anyway, the caps you have put at the outputs of the 1646 are too small. On a 600 ohms load, they would introduce -3dB attenuation at 50 Hz, -9dB at 20Hz, which is unacceptable. Even if you don't plan to drive 600 ohms load, again, the cost of increasing these caps is trivial. Putting 100uF caps there would allow you to drive 600ohms with only 0.3dB attenuation at 20Hz.
As a conclusion, I suggest you normalise all caps to 100uF. Makes for a better buy.
The power caps changed to Film/Foil ones or anything else.
I don't remember where the discussion ended on this subject, but ceramics are ok, just make sure they are of a good quality. And you'll need at least one large value electrolytic cap on each power rail; use one of the numerous 100uF you will have bought in quantities. ;)
Are the dc blocking caps at the mono circuit at the right place. The value question is the same as above.
Yes they are. Same value (47 or 100uF); the unknown res (R34, 35,...) can be anything from 22k to 100k.
Theres another big question. What is if i change the THAT1646 and 1246  to 1640 and 1240. Do i need all the same a trimmer there or can i leave it out.
I don't see any trimmers on the schemo...
I don´t really understand what you mean with: The mater buffer needs to have at least 6dB gain. Do you mean the OPA604 or do you mean the 1646.
Typo error; I meant the METER buffer. The "outside world" operating level is +4dBu; since the 1646 has 6dB gain, the INTERNAL operating level will be -2dBu. If you want to drive properly a moving-coil VU-meter or a professional peak-meter, you need to have 6dB gain on the meter buffer.
 
Hi all,

I came across this thread and found it very instructive. I would like to design a  fader buffer stage with a gain of 1 (or suggested) for main volume or monitor volume. Although i understood more or less the reasons for the different values of some of the resistors and capacitors, i don't understand the values of the 3k9 resistor at the inverting input and 4k7 resistor in the feedback loop. ¿why not 10k or so? Also i would like to know if it would be a good idea to follow this buffer stage with an ssm2142 to balance the output.

¿Can anyone explain me?
Thank you for your help

JAY X

Attached is the original schematic of the fader buffer of the input stage.
 
For some reason, the designer has decided he needed about 7dB gain there; he chose the values of the resistors low enough to not contribute any significant noise and large enough not to load significantly the output.
A unity-gain buffer is a tad different. You could just use a voltage-follower (with the output tied to the negative input), or a a non-inverting stage with little gain (about 6dB) and complementary attenuation at the input. The former has the advantage of simplicity and the lowest noise, the latter has a better transient response.
 
Back
Top