Summing amp best performance

I want to build a new summing box for mixing a small number of channels (min 4-max 16). I have the CAPI ACA-BO and a Neumann V475-2C. What would be the best option for this. I saw on a build that Holger did, a switch where he would set different resistors depending on the number of that channels that was being used. I liked that idea.
Essentially I would need the best performance and I don't mind using a fixed number if that's best. I will be using 4 channels more often than 16.


Zander

  For up to 16 channels you shouldn't have much to worry about, in performance, I'd try to lift the unused resistors from the bus if it's not a problem to do so in the design. Bigger mixers get trickier, a few channel summing box shouldn't.

  I don't know the V475 in detail, so I can't take that decision for you.

  If you will be using only 4 channels most of the time and you want the design optimized for that you could do so, it could probably still handle 16 channels, but if you want to switch between two summing networks I'd use separate bus with separate inputs, let's say a DB25 input for the 4 channels and another two connectors for the 16 inputs at the 16 bus, then you need a switch at the bus it self which should be pretty carefully since it's a critical place. Please don't run a shielded cable all the way across the case back and forth to do this switching.

  While that could be done I guess I'd go for an optimization for half way, let's say 8 channels and live with 4 or 16 at a more than reasonable performance. Unused resistors should not be connected to the bus or connected to ground at the other end. Floating resistors connected to the bus is usually not considered good practice. Also the end of the resistor which usually goes to the bus should go to ground, that's good practice, if you don't do that because you don't care or it's not convenient don't worry, your boat won't sink.

JS

joaquins said:

  While that could be done I guess I'd go for an optimization for half way, let's say 8 channels and live with 4 or 16 at a more than reasonable performance. Unused resistors should not be connected to the bus or connected to ground at the other end. Floating resistors connected to the bus is usually not considered good practice. Also the end of the resistor which usually goes to the bus should go to ground, that's good practice, if you don't do that because you don't care or it's not convenient don't worry, your boat won't sink.

Click to expand...

I like the idea. I will use push button switches, 2 per channel. So you can take it out of the circuit, L, R or L+R (cfr. Folcrom).

Now the question really is. Which summing amp is most "sterile"/clean? I know people have build both here. Any comments on that?

Zander

API is known for a specific sound while Neumann is clean and neural imho.
The V475 can handle up to 100 channels easily without the change of any resistor.
I doubt there will be any noticeable or measurable difference between 4 and 16 channels.

Zander said:

I want to build a new summing box for mixing a small number of channels (min 4-max 16). I have the CAPI ACA-BO and a Neumann V475-2C. What would be the best option for this.

Click to expand...

A very simple question that triggers a lengthy answer! First, it depends what is your opinion on "best". Some like spanking clean, others like a dose of "dirt" or "warmth". Both alternatives are definitely on the clean side. However, the CAPI can be fitted with different DOA's, and that may start and endless quest for the ultimate DOA. The Neumann has both input and output xfmrs, which may impart a slightly different colour, although the input xfmr, being under heavy NFB, is essentially transparent. The CAPI output stage, with the 2503 xfmr has a reputation of adding "colour".
You are likely to get as many opinions as there are members in this group about the respective sonic signatures of the proponents.
On a different level, that of purely objective performance, there are mainly two aspects:
Headroom: In this area, there is no doubt the CAPI is the winner, with its bipolar 16v rails, compred to the single 20V rail of the Neumann.
Noise: The Neumann, with its input xfmr, is optimized for the low bus impedance (typically about 200-500 ohms) resulting from a large number of channels being routed via ca. 10k resistors. Otherwise, the CAPI (with the standard 2520) is optimized for much higher bus impedance (typically about 5-10 kohms), which is consistent with the recommended use of 47k bus resistors.
Depending on the bus resistor value and number of channels, the Neumann may be more noisy than the CAPI, if the resulting bus impedance was much higher than the optimum value. Unfortunately, the noise figures (input noise voltage and input noise current) for the 475 seem unavailable.
Please note that there are DOA's that are optimized for lower source impedance than the 2520; they are low noise only if the source impedance is low enough; they may actually be noisier in a low channel-count application.

Thank you abbey. Very insightful explanation. I must add that I had in mind to use app992  DOA's with the aca-bo.  Now I'm thinking that maybe I could just build a circuit with those, configured for 4->2?

Zander said:

Thank you abbey. Very insightful explanation. I must add that I had in mind to use app992  DOA's with the aca-bo.  Now I'm thinking that maybe I could just build a circuit with those, configured for 4->2?

Click to expand...

992 is not a good choice since it's optimized for low source impedance.

abbey road d enfer said:

Zander said:

Thank you abbey. Very insightful explanation. I must add that I had in mind to use app992  DOA's with the aca-bo.  Now I'm thinking that maybe I could just build a circuit with those, configured for 4->2?

Click to expand...

992 is not a good choice since it's optimized for low source impedance.

Click to expand...

I have designed the APP992 with a topology that is much different from JE-990.
APP992 is a FET input DOA and it uses a matched pair of 2sk170 fets in its input stage, so it can be used with low, medium and high impedance sources.
BTW it is  very used on Igor Sontec eq's and on several line applications, as on several preamps.

ppa said:

abbey road d enfer said:

Zander said:

Thank you abbey. Very insightful explanation. I must add that I had in mind to use app992  DOA's with the aca-bo.  Now I'm thinking that maybe I could just build a circuit with those, configured for 4->2?

Click to expand...

992 is not a good choice since it's optimized for low source impedance.

Click to expand...

I have designed the APP992 with a topology that is much different from JE-990.
APP992 is a FET input DOA and it uses a matched pair of 2sk170 fets in its input stage, so it can be used with low, medium and high impedance sources.

Click to expand...

That is interesting; I assumed (wrongly) that the input was bipolar.
I have noticed that VLN FET's noise performance seemed to degrade with high source impedance. Have you noticed that? Have you documented the EIN current?

abbey road d enfer said:

ppa said:

I have designed the APP992 with a topology that is much different from JE-990.
APP992 is a FET input DOA and it uses a matched pair of 2sk170 fets in its input stage, so it can be used with low, medium and high impedance sources.

Click to expand...

That is interesting; I assumed (wrongly) that the input was bipolar.
I have noticed that VLN FET's noise performance seemed to degrade with high source impedance. Have you noticed that? Have you documented the EIN current?

Click to expand...

we speak of 1 db of NF at 100k of input source impedance and 2dB at 1M ,
2SK170 is used also on condenser mics just to be clearer, so it is very quiet also with very high impedances.
And however in a summing amp is very often very very quiet, a lot more than many opamps.
Toshiba semiconductor designed a very  good one.

and BTW seems that Forsell DOA's use 2SK170's and that they are well knowed to be quiet a lot also in line stages. 

ppa said:

2SK170 is used also on condenser mics just to be clearer, so it is very quiet also with very high impedances.

Click to expand...

I know this, however the high g-s capacitance makes it a poor performer with SDC's; LSK489 is a definite improvement in that respect, in spite of having a higher EIN voltage.

ppa said:

and BTW seems that Forsell DOA's use 2SK170's and that they are well knowed to be quiet a lot also in line stages.

Click to expand...

Line inputs, which are commonly mated with lowish source impedances, present no challenge at all in terms of noise.