Fully balanced summing box, 16 channels, with level and pan

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Thanks for your input, John.

Just to go back to your previous post:

JohnRoberts said:
if you scale everything down a factor of 1/2... the input Z will be the 4.7K pot in parallel with the wiper load (2.4k in series with perhaps another 4k).  So not quite 2.5k, but perhaps low for consumer gear. I am not a fan of passive mixers so good luck.

JR

So I calculate that if I scale everything down a factor of 1/2  then it'll have an input Z of 2.8k. Will that be ok to be driven by some of my DIY kit that has an output Z of 600r?

Since most passive boxes use variable gain mic preamps for make up gain knowing the precise loss is not a big deal either.
I'd be interested to know the formula to work it out just for the sake of education really! I'd also like to know because I'd like to know if I can use a 1272 type circuit that only has a usable gain of 50dB, or do I have build a 1290 type circuit with the extra card to achieve more gain.
 
letterbeacon said:
What are the formulas to work out the insertion loss and then the return loss? I would like to work out what the cross talk amount would be for NYD's schematic that I posted earlier.

Referring to the schematic you posted earlier, the bus loss is given by Rshunt / (Rshunt +20K)

Rshunt is typically 234 ohms which gives a bus loss of  38.74dB.

Return loss is harder to workout but usually the worst case is with the oan centred. Assuming VR2 is 20K then we need to work out the voltage at the junction of R1 and R2.

From slider to ground is 10K so the voltage on the slider is roughly the bus voltage times 10K / (20K + R2) or ~ 10dB
down

From slider to the top is 1) in parallel with R2 i.e 5K. This forms the top arm of another divider. The bottom arm of this is:

1. R1 in series with whatever the pot VR1 resistance is. assuming the source has a very low impedance then the worst case is with VR1  at 5K + 5K which is 2K5 . So the total is 4K7 +2K5 = 7K2

2. The impedance to ground of the bottom pan pot  and  R3 and R5. The bottom half of this pot is 10K and the top half is 10K in parallel with R3 i.e 5K so the total is 15K. This is in parallel with R5 (20K) which makes the combination about 15 *20 / (15 + 20) which is 8.6K.

1. in parallel with 2. is 7.2 * 8.6 / (7.2 * 8.6) = 3.9K

So the second divider is 3.9 / (3.9 + 5) = 7dB.

So the total return loss is 10dB + 7 dB = 17dB.
Bus loss is 38.74 so total crosstalk is 38.74 + 17 = 55.74 dB.

Not up to real pro standards but definitely good enough for normal stereo.

The reason the return loss is relatively poor is the use of R1. Without it the crosstalk would drop another 4dB to nearly 60dB.

These figures are not 100% accurate because the network is quite complex and I have made some assumptions in order to simplify it. The best way to get the exact answer is with a simple simulation.

Cheers

ian
 
JohnRoberts said:
For a simple stereo mix I wouldn't expect crosstalk to be a huge issue (more of a concern for broadcast mixing).

Since most passive boxes use variable gain mic preamps for make up gain knowing the precise loss is not a big deal either.
Agreed.

The loss will be a function of how many stems are being summed, more stems more loss.

JR
[/quote]

In a simple passive mixer this is true but the other consideration is you need to make the bus impedance low enough for the following mic pre to work with. This the purpose of Rshunt. In simple mixers with few inputs the parallel total of the bus resistors exceeds 200 ohms. In the NYD design Rshunt is added to make the bus impedance equal to 200 ohms. This means in general that no matter how many inputs the design has, the bus loss is pretty constant at 200 / (200 + 20K) = 40dB.

Cheers

Ian
 
Thanks for this Ian - so much useful info here!

I think I follow the maths, but I'm getting stuck on how you calculate the dB.

Referring to the schematic you posted earlier, the bus loss is given by Rshunt / (Rshunt +20K)

Rshunt is typically 234 ohms which gives a bus loss of  38.74dB.

So that's bus loss = 234/ (234 + 20k)

which gives me: 0.01156. I think I'm missing a calculation somewhere!

In a simple passive mixer this is true but the other consideration is you need to make the bus impedance low enough for the following mic pre to work with. This the purpose of Rshunt. In simple mixers with few inputs the parallel total of the bus resistors exceeds 200 ohms. In the NYD design Rshunt is added to make the bus impedance equal to 200 ohms. This means in general that no matter how many inputs the design has, the bus loss is pretty constant at 200 / (200 + 20K) = 40dB.
What happens when you don't use a mic pre for make-up gain, but a 1272 line amp? Does the 1272 expect to be driven by 200r?
 
letterbeacon said:
Thanks for your input, John.

Just to go back to your previous post:

So I calculate that if I scale everything down a factor of 1/2  then it'll have an input Z of 2.8k. Will that be ok to be driven by some of my DIY kit that has an output Z of 600r?
The simple answer is yes it will be OK...

Now to confuse you with TMI, output impedance is not the same as drive capability.  That said the legacy standard of 600 ohm output impedance implied an old school 600 ohm termination (load) so it "should" easily drive a couple k ohms. In fact you could probably drive your mixer with wimpy prosumer outputs, only experiencing any issues from very hot output levels (running out of drive current). There may also be a dB or two loss from high output impedance circuits.
I'd be interested to know the formula to work it out just for the sake of education really! I'd also like to know because I'd like to know if I can use a 1272 type circuit that only has a usable gain of 50dB, or do I have build a 1290 type circuit with the extra card to achieve more gain.

Pretty simple .. resistors in series add resistance... Resistors in parallel are a little harder but Reffective= 1/(1/R1+1/R2+1/Rn...  )

JR
 
letterbeacon said:
Thanks for this Ian - so much useful info here!

I think I follow the maths, but I'm getting stuck on how you calculate the dB.

So that's bus loss = 234/ (234 + 20k)

which gives me: 0.01156. I think I'm missing a calculation somewhere!

And 20*log(0.01156) is 38.74dB
What happens when you don't use a mic pre for make-up gain, but a 1272 line amp? Does the 1272 expect to be driven by 200r?

The 1272 is effectively a mic pre. It has the usual 10468 Neve mic input transformer and a couple of gain stages which together will provide at least 40dB of gain. It will be quite happy with a source of around 200 ohms.

Cheers

Ian
 
Thanks both for your replies.

Bus loss is 38.74 so total crosstalk is 38.74 + 17 = 55.74 dB.

Not up to real pro standards but definitely good enough for normal stereo.

The reason the return loss is relatively poor is the use of R1. Without it the crosstalk would drop another 4dB to nearly 60dB.

Just having a quick look around the net and the Studer A800 specs say their crosstalk was 55dB at 1kHz, so I guess if it's good enough for them, it's probably good enough for me.

What is the purpose of R1 in that schematic? Would it be possible to lose it or would it mess with the input Z?
 
letterbeacon said:
What is the purpose of R1 in that schematic? Would it be possible to lose it or would it mess with the input Z?

Pass, you probably need to read NYD's descriptions of the circuit. His designs are always very well crafted. However, I am pretty sure I have seen another design of his where this resistor was not present and the input pot was much lower in value.

Cheers

Ian
 
I have one more question!

If I do go down the route of halving all the values so I can use a dual 10K pan pot, can I expect the bus and return loss to be halved as well, meaning the crosstalk will double?
 
letterbeacon said:
I have one more question!

If I do go down the route of halving all the values so I can use a dual 10K pan pot, can I expect the bus and return loss to be halved as well, meaning the crosstalk will double?

I think the return loss will remain the same because all the ratios remain the same. However, the bus loss will be halved so the crosstalk will be 6dB worse.

Cheers

Ian
 

Latest posts

Back
Top