[BUILD]CAPI 2-ACA-Bo~Official Support Thread

[horvitz]

So, good progress is being made.  Fired up the whole thing today.  To pick up a previous topic, we were wondering about how much noise I'd wind up with if I float the Lo signal on my input.  The answer is some..

Just generally looking at peak levels on the meters in PT (I know not really accurate, but it's a good enough measure), I get:

Just 2-ACA-Bo            : -95db
2-ACA-Bo + 4-ACA      : -85db
Input cables hooked up : -65db

As I have it right now, the only path to ground for the audio on the input side is via the 100k on the input to the 4-ACA.  If I run that straight to the chassis, I wind up with about -70db.  That's better, but I still hear a very slight hum if I crank the output.  I guess maybe that's what I live with?  I'm not sure how else I could configure it.

Speaking of that 100k, one thing that I noticed in my checking of grounds is that the 100k on the input to the 2-ACA isn't really doing much.  On the front side, it's connected to the ground on the output of the 4-ACA, which goes to the chassis.  On the back, it's also connected to the chassis.  So in my configuration, that resistor is rendered ineffective.  Nothing to really worry about I would gather, but just a note around the way I'm using it here.

Bug hunting aside, the thing works great!  I'm really happy with how it all turned out.  Did a VERY quick mix on something here and I love it   Thanks Jeff for making all this available!

 

[Johndcx]

I wonder will I have better specs if I use the doa line in??
Those specs don't seem so good for a descrete mixer?

 

[horvitz]

No, it's not great.. yet   I'll keep on with the troubleshooting and I'm sure I'll get the noise down.  Maybe it's the cables?  Maybe it's the DA?  Maybe I missed something?  Don't know, but I'll get it sorted.

 

[jsteiger]

Hey guys,

Been offline for a few days with computer "issues".

Anyhow, the 100K hanging on the non-inverting DOA input of the ACA portion of the card, is a dedicated ground reference resistor for the opamp, if I am not mistaken. With out this resistor, the base on the first input transistor in the 2520, will possibly not have any "bias reference". This depends upon how your ground is connected/brought to the ACA card. I believe API put this resistor there in case the input to the card had the shield or ground floating on one end. If the shield of the input cables was floating on the supply side, the DOA still had a ground reference thru this resistor. FWIW, the shield on the bus cables is about the only shield that is connected at both ends in my console. So, this 100K resistor is effectively useless for me too. But, may be critical for the next guy.

You could try physically floating the PCB so that the only ground connections are thru the Molex PSU connection. All of the mounting holes connect to the ground plane on the card. This is no issue for my console as the frame is all wood. If this helps, maybe nylon screws and standoffs will be a good solution?

The noise specs for a 2520 in this configuration are pretty damned respectable I think. Look at the info for Figure #3, Lossless combining Amplifier on the 2520 spec sheet.

I think there may be some other noise causes that need to be tracked down.

Keep us informed. 

Cheers, Jeff

 

[horvitz]

It's always the grounding!  I'm getting close, though.  Pretty sure I've isolated it to somewhere around the DB25 inputs or the cables to my DA.  I used some pretty cheap stuff there, so maybe I'm paying the price.  With no signal on the input, it's pretty darn close to the noise numbers in that spec sheet, so I know I'm in the ballpark!  I'll get it sorted, I'm sure.

 

[horvitz]

Jeff, a question for you.. You mention that your bus cables have the shield to ground on both ends.  Is the shield in those cases the actual audio ground?  As in, a single conductor cable using the shield to carry the ground.  In my case, I used a two conductor cable for each channel and have tried running the grounds in a number of different configurations, all with roughly the same effect.  The shields however, are connected to nothing on either end.  I figured the run was so short (maybe 12" ea) that it didn't matter.

I basically have the grounds from all 16 channels bussed together at the LCR assignment switches.  Kinda wondering if this is the wrong approach entirely.  You can see what I've done in the attached picture.  The two terminal blocks on the far left receive the ground for each channel and then you can see it bussed across the bottom to the outputs, which go to the 4-ACA.

It's funny, you can get everything working and then it's the audio-101 junk that trips you up!  I'm hoping that some fresh eyes will see something that I'm not.. one of those things where I need to not think about it for a few weeks and then come back to it and all of a sudden the problem is obvious.

 

[horvitz]

So yes, I do realize what an outstandingly stupid question I asked.  BUT, it did get me thinking more about the right things and I'm happy to say that with proper grounding of shields, fixing some bum cables, and some general sorting, I was able to significantly reduce the noise in the system to the point that I can't hear it at all with everything cranked.  Sounds good to me!  Or I guess, it doesn't sound and that's good for me

Final numbers here:

Just 2-ACA-Bo                        : -98db
2-ACA-Bo + 4-ACA + LCR board  : -95db
Input cables hooked up            : -85db

That extra 10db is entirely 60hz.. just a bit of a bump, and I think I could even chase that out although I'm not really too concerned.  This is 20db better than it was before!

  Brian

 

[jsteiger]

Nice job Brian!  8)  Cool little bus board you got there too!  ;D

I remember when I had talked with Paul Wolff a few years ago about modding my desk and refurbing. One of the first things he told me was to replace the cables than run from the motherboard bus tracks to the ACA boards with "that nice, big Mogami instrument cable". This I did. It is some big **** which takes up lots of room in the desk. It has a very nice shield though which is important for this use I think.

Thanks for the positive updates.

Cheers, Jeff

 

[sintech]

Completed the build yesterday, took me a little while to get used to the API sound, as I'd never experienced it b4: silky top, upfront mids and tight bass.

About to drop it into a 1998 40 channel Trident-MTA console, summing in buckets of ten channels.

Was going to come off the consoles bipolar 18v supply, but suspect as it's the consoles largest configuration, the PSU is prolly flat out, so using an external bipolar 16v for the 2-buss.. any grounding tips or suggestions are very welcome?

Thanks goes out to Jeff for a a great project.

 

[horvitz]

Sintech,

Looks like you had as much fun building DOAs as I did!  Although, I only did 8.. gets a bit tedious, I'd say.

At any rate, it took me quite some time to get the grounding sorted.  In the end, it wasn't anything too fancy.  Basically, I did a star ground with the sig input ground, PEM, and PSU.  The sig ground does kind of a trip around the chassis though before it gets there and I think that could be improved.  Things that didn't seem to make a difference were floating the PCBs or allowing them to ground to the chassis via their mounting holes, and the sig ground on the output (I have that to the GND at the output from the BO).

Regarding shields, Jeff mentions that in his console the bus cable shields are grounded both ends.  I kinda did that.  The inputs to my bus board (a few posts up) are grounded just one end but the shorties from the output of that to the 4-ACA are grounded both.  Seems to improve me a few DB.  I really just had to play with this part a little to see what would get me the best results.

I think my biggest noise problem on mine still is the float Lo scenario on my inputs from a balanced source.  I'm currently breadboarding an input section with 5532s to see how that works.  This thing is turning into a bit of a playground for me to try things.  Bound to wind up with the most unnecessarily overbuilt summer around

  Brian

 

[gar381]

VERY NICE!! Sintech


GARY

 

[sintech]

Quick lash up in context, got a feeling the Trident-Mta channel into the API 2 buss is going to be a killer combo.

Real world test, the 47k summing resistor effects the pan law, as the  summing for the 24 groups are staying stock, reverted back to the 22k summing resistor.

The correct pan law is now back, and it sounds cool, as in it doesn't sound badly wrong... the ACA feedback resistor and cap are calculated for the 47k.

What would be the best method to check and adjust this?

 

[jsteiger]

You will probably want to go with a 22K feedback R then as well. You can play around with this to make sure the gain staging is how you want it. This will probably also require a slight tweak on the feedback cap.

You can use Bicycle Bob's handy little calculator to figure your -3db frequency knee. www.bicyclerecordings.com/various/gainchart.xls

Just enter 22000 (or whatever value you end up with) into the Rf box and adjust Cf to a common value. I would gather 68pf or 82pf would be a good value. Typically, from what I have seen, API was between 65K and 105K with that -3db point.

For instance, the original 312 pre was 20K and 120pf. The console pre (VP26) is 20K and 82pf.

Jeff

 

[sintech]

Thanks Jeff, so.. I change just the feedback resistors in the inverting ACA to 22k, then leave the 2-ACA-BO the same? as they're fed with 47k's from the outputs of the Inv-Aca.

 

[jsteiger]

Yes. That should be a solid starting point at the worst. Should be fine though.

 

[Tubemooley]

Veddy interesting developments. Ya' know..... I've been wondering about that 100k resistor on the inverting input of the ACA (I think it's the inverting input; I'm working from memory here). Not to be a ****..... but this reinforces once again how very, very important grounding is. Basically, ya' never connect both ends of a cable shield. Never. But it still gets done occasionally (under certain circumstances). I need to remember to add these new posts to my ACA notes file. Nice work Sintech. Really nice work. DW.

 

[myker]

hi guys, so from what jeff is saying i should be able to sum 16 channust floels from my 192 i/o with just one 2-ACA-Bo? by jating pin 3 and splitting channels with 47k resistors into the ACA side? then route that signal to the 2 gang pot and into the booster amp and monitor/talkback etc.? then i could run the r and l signal back into the 192 and print it? as well as monitor that signal im listening to as well?
sorry if this doesnt make sense
-mike

 

[sintech]

Connecting more channels today, I'm noticing ground issues going into the Inverting ACA's,.

I'm taking the channels earth and the signal after the summing resistor (either left or right) down a screened lead, in to the + and common inputs, all is ok for the first 10 channels.

I then switch to the next ACA input.... and hummm!!! But.. if I remove the common leaving it floating all is good, and the channel 11 is fine also running just with the positive in place.

I'm wondering what's the correct scheme? I'm doing 40 channels in buckets of 10.

 

[jsteiger]

Wow! That is looking killer!

I can tell you what I know from my board. For almost all the cable runs, API would only connect the screen on one end. I would try floating the screen at the source (channels) and only connecting it at the destination (ACA). The ACA boards are probably getting a ground reference from your 3 wire PSU lines right? I can't really tell form the pics. My bus cables have the screen connected at each end but, I think it all plays into the over ground scheme of the desk. Only connecting at one end should be fine. That is why there is the 100K reference R on the non-inverting DOA input of the ACA. Someone mentioned a while back about removing it. That would not be good in this case. You do have the PSU ground connected with the bipolar supply voltages?

Lot's of work there, no doubt!

 

[sintech]

It's getting a lot better for some fooling

I'm using a second power supply for the ACA's, so..  the API ground is connected to it's bipolar supply ground and also the consoles star earth. Without the star earth connection it's hum city.

Can hear a faint mains hum, should I float the ACA supplies center tap and leave it just connected to the console star earth?.. I've been a bit nervous to try this!! heheh