Active summing bus/output section for mixing console

So I'm tired of the sound of my IC based summing bus/output section on my console. I'm trying to build one purely for sound quality.  I am not an electrical engineer or designer, but can read schematics, and am a bit crafty.  Right now I'm not worried about the meters or solo working, just getting the sound through proper 1st....

...so I've taken John Hardy's summing schematic from his 990 application notes & married it to Jensen's high quality output section schematic, but something is wrong.  The sound gets distorted after 5 seconds (maybe an oscillation?) and there is a lacking of low end.  Sounds like a high pass filter rolling off at around 150Hz or so.  I don't have a sweep oscillator.  I believe the high pass filter problem is in the last 990 schematic, as I've ran an insert in here and am losing the low end.

Here's the schematic of one channel of the bus:



I put the fader between the two schematics.  Anyone have any idea what is going wrong just by looking at this schematic? 

I'm running the 990C's off of a 24v dedicated power supply just for the bus.

I searched for a summing schematic with proper output section but I only found IC designs.

Regards,
Scott

Put a scope on the output of the summing amp to start with....

I don't know if it'll help but I'd put a capacitor in front of the 10k pot to get dc off the fader.

I've never seen an OLI used in an interstage position... looks a lot like a tube amp power supply filter when you add that variable resistor after the choke. 
I'd have to look it up to see for sure, (cause I'm not one of these terribly clever old hands here who know this stuff like you and I know our ABC's) but if I'm right the rolloff should change as you turn the pot... it would seem to be hard to measure coz your levels are going to change at the same time.  LCR filter or choke filter- it's called something like that.
Kelly

What do you need the line driver portion for? The 990 used for the summing can drive whatever. IMHO.

i am planning on doing something like this with my soundcraft 400b. I am still doing some research and am interested to see how this works out for you.

990 for the summing > 10k fader > ~+10db opamp buffer (maybe a 5534 or a DOA) > 600ohm O/P transformer > master out

I was emailing with john hardy about this and he mentioned that the output isolator is only needed when interfacing with the outside world (insert points & outputs). in your case, it appears to just be going straight to the fader, so I would wager that you can get away without that 1st output isolator.

i was also considering ditching the fader buffer and taking the output straight from the 990 to the O/P transformer, but in my case, the fader buffer flips the polarity of the signal. this may be something to consider, depending on your topology.

there are a few posts around here about replacing the summing of a ACN type circuit (soundcraft 200b, allen & heath) with a DOA

outoftune said:

...i was also considering ditching the fader buffer and taking the output straight from the 990 to the O/P transformer, but in my case, the fader buffer flips the polarity of the signal. this may be something to consider, depending on your topology.

Click to expand...

Just swap pins 2 and 3 after the transformer. The 2nd stage would need the polarity flipped.

I have a similar mod I'm adding to my API desk. I am adding 4 inverting ACA's (no trafo) that sum 8 channels each, per stereo side. Those 4 outputs, and a few others, will then be summed through the stock original ACA's. No fader in this case. All I have to change on the stock ACA's is the polarity on the final output since I'm adding the inverting stage in front of it.

This fader could easily be 600 ohms instead of the 10K. Plenty of balls there!

Thanks for all of yr input!  I'll break out my scope this weekend and test after the 1st summing 990...

Originally I tried this circuit without the isolators..so i don't think they are the problem,  but I will remove them where they are not needed in the middle of the circuit.  I'd love to be able to just use one 990, but I think I'll need some make up gain.  Pretty sure John Hardy's summing schematic provides zero gain...but then again I could be wrong!  Unfortunately my knowledge is limited!

Fred Forssell's summing schematic shows a DC servo to remove DC offset.  I might try this in the future as well.  I'll try testing with a electrolytic coupling cap but would rather use a DC servo so there are no coupling caps in the circuit.

so...I'll run some tests & even try the circuit without the output make-up stage, and report back. 

BUT if someone sees something obviously wrong with the schematic that I posted let me know!

Thanks so much,
Scott

The Jensen application notes that involve the 990 use a schematic symbol for the 990 that can be misleading (in my opinion). Specifically, look at the box labeled as "A1 JT-990-C" in the image that you provided:

To properly visualize what is inside the box, you should draw a standard op-amp "triangle" with the inverting and non-inverting inputs connected to the "-IN" and "+IN" points, the output of the triangle should go to the "FB" point (feedback loop connects here), and you should then draw an output isolator from the output of the triangle (or the "FB" point) to the "OUT" point. The "COM" point is the ground connection for the internal bypass capacitors and a constant current source inside the 990C. The external 470uF bypass capacitors also connect to this point.

Deane Jensen included the output isolator in his original design to show the importance of it, but it is NOT normally a part of an op-amp, whether it is a 10-cent op-amp or a $100 op-amp. It is an external part and (in my strong opinion) should be drawn externally. An op-amp should be drawn as a triangle, and any external parts should be drawn separately. The folks at Jensen continue to represent the 990 as having the output isolator built in, but you would not necessarily realize it unless you read Deane's paper. So you need to connect your feedback loop components before the isolator, and the final output after the isolator.

Perhaps that is part of your problem. Just look at my half of the image for the way I think the 990 should be drawn.

John Hardy

Hi John & All,

I've been too busy to perform any testing yet, but I'll have a couple hours free tonight before I leave town for the holiday.

The Jensen schematic did cause me some confusion.  Currently I have the "Load"(going to JT-11-BM OT) & "Feedback"  tied together, and have  "Com" & "Ground" tied together on the 990.  So following the Jensen notes, (looks like I'm following them wrong according to John)  currently I have the output isolator going to ground from the red wire of the JT-11-BM OT. I'll change the output isolator so its between the "load"/"FB"/(output) of the 990 to the brown wire of the JT-11-BM OT.

John, should I remove the the output isolator that is between the summing 990 and my fader?  Does this schematic look okay to you otherwise?  Thank you so much for finding this thread & replying, I've had you in the back of my head as the person to contact with this problem.  I just didn't want to waste yr time.  If I still have problems I'll redraw the Jensen part of the schematic and repost it here.

I'll let everyone know if I get this running.  I know I'll have to tie my meters/solo in and provide trim-gain adjustments to each side of the bus for calibration purposes.

Thanks Again!
Scott

So I'm tired of the sound of my IC based summing bus/output section on my console.

Click to expand...

SSL builds their console summing with about 15$ in ICs.. 

But if you must, be very very careful messing around with the summing network and amps.  Read this:  http://www.forsselltech.com/downloads/design_discussions/summing_buss.pdf  before you continue.

I've been there and have learned the hard way.

I've been there and have learned the hard way.

..and now I'm there too....  ...but haven't learned yet....


yeah I knew that off-handed IC comment might get some criticism.  I shouldn't just blame the IC's... There's an electrolytic coupling cap between each IC in my summing/master output stage.  I think most of my IC gear shares this in common.  Circuit-wise there probably is a better way to work with them.  I should checkout what SSL is doing in their summing/output circuits, and compare them to my stock IC circuit.

Some history why I'm doing this:
I have a spare stereo bus, so its easy for me to experiment with one.  My console is 24 years old, so I replaced/upgraded all my electrolytic coupling caps and that helped.  I also went through the IC opamp swap obsession..that helped as well, but I think I can still get a better sound.

The majority of my discrete gear doesn't have electrolytic coupling caps, which makes me think the circuits are designed more with audio quality in mind, so I shouldn't just slam the IC but the circuit.  My discrete stuff (with probably better circuits) tend to have a clear but muscular sound to them, and an excellent mid-range.  I'll use the unscientific...it just feels good to listen to.  That is what I'm hoping to achieve with this mod.  If it doesn't happen I can always turn these parts into mic pres.

********************** 

Here's where I'm at:

I just had an hour to test before running out to a gig and splitting town for the holiday

I got rid of the isolator before the fader.  I removed the isolator from between the red wire OT > ground and placed between the load/FB > brown wire going to the Jensen OT.  Same problem.  The sound is distorted and then drops out...I move the fader a bit and the sound comes back...then disappears again.  When I run an insert signal just after the fader the sound is there BUT has the low end roll off, like a high pass filter.  However the sound is excellent except for the unacceptable low end roll off.  The gain does seem a bit low...could I be creating an impedance problem? 

What does give me hope is that using a SM58 > UA610 mic pre > bus insert (its going just into the Jensen post fader/output part of the circuit) the sound is what I'm looking for (except the low end roll off).  My voice is so clear and pure and right on the edge of the speaker(though I am leary of the low end roll off creating an illusion of better clarity).  When I put the unmodded IC/coupling cap stereo bus in, inserting the mic pre into the same insert the sound of my voice is much more distant & veiled, but hey I've got the proper low end.  So that's where I'm at...  right now I'm 400 miles from my project and won't be able to get back to it for 5 days.

Thanks for the link for Fred's PDF, I read it a while back, but just printed it for me to read again while on my trip.

Yes it *could* be an impedance problem but it almost seems like a DC offset problem.  The input of the second 990 shouldn't care that you are changing resistance to ground at all.  Stick a bipolar cap between the output of the first 990 and the fader so that no DC is on that fader at all.  Let us know how that works.

Caps can be a really really good thing in a console.  They can do things with one part that would otherwise take 15 parts and still not solve all of your problems(like servos).

The trick is to use a bipolar or biased unipolar caps for best sound.

BTW, SSL uses 2 AD829s and a 5534 in a instrumentation amp setup for their summing amp..

When I get back home I'll place a cap where you are suggesting and see what happens.  Any idea what value?  120uF? I can just reference my stock schematic to see what the manufacture was using.  If this works I might A/B test by putting a servo on the other channel. 

Thanks for the non-polar tip, I've used them before but only as a guess that they might sound better.  Never A/B'd them against polarized ones. 

The API input channel schematics show a 250uf here. I have changed all of mine to 470uf for a little more bottom. Plus, the 250 is a bastard size. I think some would say anything to 1000uf would be fine.

I'd go for 220uf.  That seems to be a good value for most things without going through the trouble of figuring out what you should have.  Unless the impedances are really strange I don't see the need for going as big as 470uf or higher.

Bipolars will exhibit much lower distortion numbers than polar caps.  I really dig the Muse ES series.

Here is a crude drawing of what is actually inside the symbol that the folks at Jensen use in their schematics:

Thanks for the drawing John!  I now have the isolator matching your schematic.

...worked on this quite a bit yesterday...and a little frustrated.

I didn't have the ground of the the power rails hooked up.  Stupid mistake as I should have started with my voltages.  so...its now hooked up and the summing 990's are no longer distorted and I have a somewhat proper signal...well...undistorted audio (to my ear) is passing through the circuit.

I now have 3 obvious problems.

1) There is still a low frequency roll off.  Sounds like a high pass filter starting to roll off below 200Hz.  At 100Hz can barely hear the oscillator and at 50Hz unable to hear the oscillator at all.  This is happening in the output section...somewhere from the fader to the output transformer.  Any ideas? 

There is a bus line in section right before the master fader.  This section I'm not showing on my schematic.  It can be switched in if you want to monitor a 2 channel source, or if you want to hear the summing from the console then you switch it out.  I hear the low end roll off when summing and when having the bus line-in switched in.

2) I have a phase problem.  I can probably figure this one out and believe it is in the summing section.  When a channel is panned in the center the signal cancels itself out when I pan either left or right the signal gets stronger.  I'll test with a scope.

When I run the signal into the bus line-in then this phase problem goes away, hense I believe it's in the summing section.  Its odd the problem is not obvious to me.  BUT then again I had my power voltages wrong!

3) Noise.  I can probably figure this one out as well.  Improper grounding and routing of wires the culprit?  That's where I'm looking.

The biggest problem for me is problem #1, the low end roll-off.  The sound is definitely headed in the direction that I want.  I can hear more depth to the audio signal, either summing or the bus line-in.  I really want this to work. 

okay...  headway

Problems 2 & 3 solved due to improper grounding. 

Without re-reading the entire thread (I realize you've done at least some of this): Have you separated the two circuits and tried each one individually (particularly now that you have fixed a major grounding issue) to see how each one works? What is your signal source and analysis method? A 1kHz square wave and a scope? Check the generator itself to see if it is working properly. An audio track? If an audio track, where is it coming from? Which input of the summing schematic are you using? How much gain are you getting after the summing amp but before the output stage? This is a good way to see what is going on. Try shorting the non-inverting input of the summing amp directly to ground, if you haven't already.

Break the connection where RL1 (on "my" schematic) connects to the fader. That way you can test the summing amp by itself by applying a signal to one of the inputs of the summing amp and listening to the output of the summing amp by using the output of RL1 and ground as your connection points for listening.

Then test the output stage by itself by applying a signal to the fader input and listening to the output of the output stage using pin #2 of the XLR as the "HIGH" connection and pin #3 as the "LOW" connection. You cannot use pin #2 and pin #1 as your connections (nor can you use pin #3 and pin #1). You can listen to the output signal BEFORE the output transformer using the LOAD terminal of the "Jensen" schematic and the COM terminal as your connection points. This is a point after the internal isolator and before the output transformer. You could even disconnect the output transformer completely to see if it has any effect.

In both cases (summing amp and output stage) you could try connecting to the point BEFORE the isolator (the actual output terminal of the 990) to see if there is a problem with the isolator, although your cables or probes should be short so they don't add a lot of capacitance to the connection point (that is why the isolator is there). This would be the "FB" (feedback) point on the Jensen schematic, and the output terminal of the 990 on my schematic.

The signal generator must have a SIGNAL and GROUND connection that connect to the appropriate points too (don't be forgetting any more ground connections).

This is basic troubleshooting, just as checking your power supply connections! Break things down into the smallest workable sections and test them individually.

John