The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Optimal resistor values for passive summing Audio Interfaces
Old 20th August 2008
  #1
Lives for gear
 
gainreduction's Avatar
 

Optimal resistor values for passive summing

A couple of years ago I built a simple, passive summing network. 8 channels to 2, or 4 to 1 per side if you like. I´ve made tons of mixes with it and all is good.

I consulted a gear designer friend of mine who recommended 2k2, 1% metal film resistors and that´s what I used.

The mistake I´ve made now is to start reading up on impedance matching and optimal values for summing.

What would higher vs lower values do to the overall performance of the setup and the make-up gain amp at the end especially ?

The chain is: D/A - various analog processing - passive summer - Crane Song Flamingo @ 12-18 dB of gain depending on what I´m doing.

I´d love to know more about impedance and how different units interact the best.

Please be gentle, I´m a musician, not a tech. heh
Old 20th August 2008
  #2
Lives for gear
 
JohnRoberts's Avatar
 

Optimal for what?

Higher impedance will be less load on preceding stage for lower distortion but higher noise.

Final output impedance should be good match for following gain stage (yeah, passive my a__ ... gain is gain).

Optimal depends on where it's coming from and where its going.

Perhaps figure out what optimal source impedance is for the crane whatever at that gain, and work backwards from there.

JR
Old 27th August 2008
  #3
Lives for gear
 
ulysses's Avatar
A higher-value summing resistor will increase the input impedance of the channels.
The input impedance needs to be high enough that the source feeding it doesn't complain. 10-20k is about right in most cases. If you're only going to feed it from a particular piece of gear (your DAC) you may be able to customize it.

If you're feeding one source into both the Left and Right busses, the effective load on the source will be half the impedance of a single bus. So if you want the load to be no less than 10k, you'll want to make each input 20k.

(If you only have "Right inputs" and "Left inputs" you don't have to worry about it. If you have both, you should consider some way of switching each channel in and out of the two busses without messing up the remaining summing network impedance).

A higher-value summing resistance will increase the insertion loss (attenuation). This will improve the crosstalk performance. It will also, of course, require more makeup gain which means more noise and more "flavor" from your makeup gain.

A higher-value summing resistor will also increase the output impedance. There is an ideal source impedance for whatever you use for makeup gain (your Flamingo, for example). For most mike preamps, the ideal will be fairly low - 150 ohms or even less. Some preamps are different. If you can find out what's ideal for your preamps, you can make that your output impedance. If you want it to be nominally correct for a wide variety of mike preamps, make it 150 ohms.

When you take all of those factors into consideration and try to come up with a "correct" summing resistance, you'll realize it seems like it won't work. Summing resistances high enough to present a reasonable load to your DACs will still not present sufficient attenuation to reduce crosstalk to reasonable levels; and the output impedance will still be too high. The solution is very simple though: A shunt resistor on the output will increase the attenuation, reduce crosstalk, and set the output impedance wherever you want it.
Old 29th August 2008
  #4
Lives for gear
 

Hi
When I was at AUDIX in the 1980's the mix resistors were 15K and the load resistor was added to bring the 'loss' to 35dB, This was on mixers with around 24 channels contributing to the mix bus. This was followed by a transformer with '10dB gain' so the folowing amplifier needed to provide 25dB. I was not involved in the original 'RnD' that proposed this gain structure but it certainly worked pretty well. The 15K was a suitably light load for the simple drive amplifiers following the pan pots.
Yes you can make the resistors smaller but you then have to provide amplifiers with enough 'welly' to drive them. If mixing signals from a variety of units it may get a bit unhappy.
Crosstalk should not be an issue at all if each channel is only feeding one mix resistor. Crosstalk occurs when the (-35dB) signal on the bus goes back to a non zero source resistance from the unit driving it, which, if feeding another mix resistor would present some signal to the second (or more) busses.
Matt S
Old 30th August 2008
  #5
Lives for gear
 
ulysses's Avatar
Quote:
Originally Posted by Matt Syson View Post
Crosstalk should not be an issue at all if each channel is only feeding one mix resistor. Crosstalk occurs when the (-35dB) signal on the bus goes back to a non zero source resistance from the unit driving it, which, if feeding another mix resistor would present some signal to the second (or more) busses.
Right. So, in the case under discussion, only channels that are feeding both the left and right busses (the mono inputs) have the potential to induce crosstalk. And we're only talking about a reduction in separation between the left and right sides of a mix. Since they'll eventually be played through speakers that are in the same room together, you can see that the crosstalk performance is clearly not the most critical parameter to optimize. You don't want it to be terrible, but it doesn't have to be perfect. Getting good crosstalk performance has more to do with good circuit layout than selecting ideal summing resistors.
Old 30th August 2008
  #6
Lives for gear
 

Hi
A gramophone record (remember those) with a good cartrigde and well set up could manage around 25db separation at 1KHz and worse at other frequencies and can still sound as pleasing as an 'all digital' high separation system. Multitrack tapre machines are rather better but around 70 dB in some cases.
Crosstalk is far more of an issue in broadcast micers where there must be NO leakage from channels that are not 'requested'.
Matt S
Old 30th August 2008
  #7
Lives for gear
 
JohnRoberts's Avatar
 

Not to complicate this but crosstalk is a little more complicated with several possible vectors for corrupting nearby audio paths. In a simple x2 mixer this is not a serious concern and may show up as poor fader or pan kill.

Minute errors to pan or mix are not likely to be perceptable.

JR
Old 2nd January 2009
  #8
Lives for gear
 
gainreduction's Avatar
 

Quote:
Originally Posted by ulysses View Post
The input impedance needs to be high enough that the source feeding it doesn't complain. 10-20k is about right in most cases.....

....Summing resistances high enough to present a reasonable load to your DACs will still not present sufficient attenuation to reduce crosstalk to reasonable levels; and the output impedance will still be too high. The solution is very simple though: A shunt resistor on the output will increase the attenuation, reduce crosstalk, and set the output impedance wherever you want it.
Coming back to this topic after some months. If I understand ulysses post right I could go with 10k resistors to keep the input impedance of the summer high enough. However this would result in an output impedance too high for the amplifying mic pre to be driven optimally.

So the big questions for me now are:

- The "shunt resistor" on the output, what value should it be to get the output impedance down to 150ohms if using 10k resistors in the summing network ? 4 channels combined to one, balanced all the way

- How is the shunt resistor to be connected ?

- Can I do this with a transformer instead ?


PS. I want to thank everybody sharing their knowledge. I´d love to buy all of you a big round of beers.
Old 3rd January 2009
  #9
Lives for gear
 

HI
The shunt resistor to get 150 Ohms would be about 160 Ohms as four 10K resistors in parallel are 2K5. (2K5 in parallel with 160 is about 150)
Since you were talking balanced you would actually use 5K0 resistors (one in 'hot' leg, the other in 'cold').
The shunt resistor would go from pin 2 (hot) and pin 3 (cold) if using an XLR to plug it into the following amplifier.
Using a transformer is easily done by just inserting it in line with your resistor arrangement, but in this case you would increase the value of the shunt resistor slightly if you are getting 'anal' about the calculations.
Having a 'step up' transformer can decrease the gain required by the following amplifier which is probably a good thing.
Matt S
Old 15th April 2010
  #10
Here for the gear
 

Sorry for dropping into the discussion, I have been experimenting quite a bit with passive summing in the past, but having just got an 8 extra outs from my adat, I decided to expand the mixert to 16 ins. What happened next was that the DAC (Aphex A141) had a much louder signal than the Motu 828, when both drive the passive mixer's 10K input resistors.

Now I came up with the decision to put 3K resistors on the 828 dedicated channels instead the 10K, is that "healthy" to have a summing network with different load resistors?

The volumes almost match now, but I wonder hot to calculat the right shunt value for the outputs that will load my mic preamp. Obviously the same value I used before (202 ohm I think it was) it's too small.
Old 15th April 2010
  #11
Lives for gear
 

Hi
I would tend to keep the resistances 'up' a bit as 3K while OK is beginning to make the outputs actually work.
I would think more like 5K (when balanced making it a 10K load for each DAC output).
To match the levels better send a reference level, say -3dBFS to the convertors and measure the voltage coming out of each. Using the lowest available level, treat this as the 'minimum which would be attributed to a 2X5 (10K) mix resistor then using your skill and dexterity with maths, work out what INCREASED resistance you need to get the same current that you have from the lowerst level outputs. (if it is 6dB more, double the resistance to 10K each (total 20K) for example). Note from +4dbu to -10dBV is a factor of 4 as it is actually 12 and a small bit dB voltage difference.
Having different 'feed' resistors is fine although you recalibrate the audio out levels from the DACs to make them equal.
Matt S
Old 15th April 2010
  #12
Lives for gear
 
dcollins's Avatar
5k. Not too high, not too low.


DC
Old 21st April 2010
  #13
Lives for gear
 
ulysses's Avatar
A balanced summing network has two summing resistors in series on each input to feed a bus. If an input is feeding both the left and right busses, then there are two of these pairs in parallel. The net result is a load impedance equal to one of the resistors. Choose the minimum load impedance you'd like to present to the source, and make all four resistors that value.

The two summing resistors in series present a source impedance to the bus. Divide that total series resistance (2xR) by the number of channels feeding the bus, and you'll have the impedance of the bus. If your channels are assignable, it's wise to maintain the source impedances regardless of whether or not the channel is assigned - that way your bus impedance doesn't change when you unassign channels. Bus impedance can be lowered further with a parallel resistor across the output. This is only meaningful with a voltage summing amplifier.
Old 14th January 2011
  #14
Gear Maniac
 

All of this discussion, yet no one has posted a schematic. (-:
Old 14th January 2011
  #15
Lives for gear
 
dcollins's Avatar
Quote:
Originally Posted by mysteryman View Post
All of this discussion, yet no one has posted a schematic. (-:
http://www.forsselltech.com/download...s/8chsum_2.pdf


DC
Old 19th October 2011
  #16
Gear Nut
 
eman kayker's Avatar
 

arg...
as I was reading I couldn't dream of a schematic, then saw that DC posted one, clicked on it and it was GONE!
Could you re-post it and make this thread alive again?
cheers, thanks to all, this is realy captivating.
Old 20th October 2011
  #17
Lives for gear
 
dcollins's Avatar
Quote:
Originally Posted by eman kayker View Post
arg...
as I was reading I couldn't dream of a schematic, then saw that DC posted one, clicked on it and it was GONE!
Could you re-post it and make this thread alive again?
cheers, thanks to all, this is realy captivating.
http://www.forsselltech.com/media/at...s/8chsum_2.pdf
http://www.forsselltech.com/media/at...s/8chsum_1.pdf

DC
Old 30th October 2014
  #18
Gear Maniac
 
Kan Kaban's Avatar
 

Ok, let´s make this zombi thread walk a few blocks...

Output impedance RME FF800 = 75 Ohm
Input impedance VP28 = Mic 1K2 / Mic+Pad 1K5 / Line 10K
Input impedance RME = 10K

Channels 1-8 out from RME - passive network - RME´s 9/10 pair input.
An 8ch (4 st) summing.

For 6.5k input resistors, balanced (2x6.5=13k), shunt is near 230 ohms.

But, Is there any advantage on adapting the network values to the 10k line input stage of the Capi VP28, instead of using the common 200ohm mic-like value?

Cheers.
Old 30th October 2014
  #19
Gear Maniac
 
Kan Kaban's Avatar
 

Thanks a lot.
Old 1st November 2014
  #20
Lives for gear
 
ruffrecords's Avatar
It is interesting that only 18 to 20dB of gain make up is needed. If it is a balanced mixer and the input arms are 2K2 each then the input Z is 4K4 - perfectly fine. For four into one the bus impedance is 4k4/4 = 1K1 which is a little high for feeding into a regular mic pre and you would only need 12dB of gain make up The OP makes no mention of slugging the bus to fix this. Maybe his tech thought it was 8 into 1 in which case the bus Z would be 550 ohms which is just about OK for feeding a mic pre and would need 18dB of gai make up.

Cheers

Ian
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
geoff004 / Gearslutz Secondhand Gear Classifieds
8
vintagelove / Geekslutz forum
5
ramjet / Geekslutz forum
8

Forum Jump
Forum Jump