Neve Output Stage Addition to Console

hi:

I have a similar post in "The Lab" discussing some ideas to improve my Allen and Heath Saber, but I think my question/topic might be better suited here.

I have a pair of Neve output stage PCBs (from Fabio) and a pair of Sowter 8751s. I'd like to modify the main output of my Allen and Heath Saber to be balanced by this stage. First, do you think this would be a worthwhile mod? After changing many channel and buss caps, I think focusing on the L/R signal path might have more of an impact.

Here's the schematic:

http://www.lsfsound.com/Public/LRcct.pdf

If you notice, the Saber also has an unbalanced output. If you look at the main out you'll see an inverting gain stage with IC3 to a 300mV unbalanced output. Would you suggest that I first take the output from here and balance with my driver. This could allow me to A/B the original balanced out with my Neve output stage.

One problem: My board runs on +/- 16V, and the Neve needs +24V.

Comment and suggest at will !!! :thumb:

greg-

totally worth your time to do this. I ran my ghost into line amps of all sorts and it really makes a tremendous difference.

dave

Having glanced at the schematic and having no idea what the innards of your A&H master section look like, I would say it's totally doable and you should go for it. I don't know what Fabio's PCB does, but I do think a pair of BA283AV cards could do this. The NV stage would be the first half of IC4, you keep your insert and 10klog fader in between, and the AM stage would be the second half of IC4. If you recall, the AM stage also has unblanced out so this could replace the function that IC3 serves I suppose. Not sure how the metering would figure in, but let's talk about the amps...

First off, notice that both amps on IC4 are running in the inveterting mode (virtual earth aka current summing). The 283 wasn't designed for inverting mode because Neve used passive loss voltage summing, but since I don't know enough about your console or summing in general to say you could just switch, let's see if the 283 will work in inverting mode. Looking at a BA283AV schematic, first the NV section. The emitter of TR4 is the inverting input and the feedback network (both AC and DC) consists of R17, 18, 19 and C17. Now the AM stage, same thing. Emitter of TR1 is the inverting input, with the feedback network consisting of R4,5 and C4, 5. Looks like only AC (audio) feedback this time as C4 is in series with the output tap. I'm not sure about how Neve derived the cap values, but you can calculate the gain set R values like you would with IC4 on your A&H or any other op amp. That's as far as I'm going to go right now, but I guess what I'm getting at is first verify if you can run 283 for current summing to really make it a drop in replacement. As for power, just pop in a little dedicated 24V supply. Seems much simpler than having some sort of switcher hanging of the +16 rail if you ask me, not to mention you shouldn't overlook the big current draw and crappy PSRR that come with a 283 amp.

Thanks for the encouragement guys :thumb: !

First, Fabio's aforementioned boards are the AM stage of the 283. I already have a pair of BA283AV (NV and AM) already stuffed and working in a prototype 1272, but I'd definitely be willing to pull them out and give them a try in my consol-e-o.

But I like the way you think, FH, about replacing the IC4 entirely, not just the 2nd half as I was thinking earlier. I'm still going to have grab some schematics and calculate some gains and such, but let's talk on a "high level" here:

1. Verify the use of the NV stage as the 1st opamp in IC4 and AM stage as 2nd opamp in IC4 with the master fader in between them.
2. I never use master inserts... If I put something on the 2 buss, it'll just go in the signal path after the console but before my converters.
3. Dedicated PSU for the +24... already have this in my 1272 (I shall borrow). Just have properly ground with console.
4. Metering comes later...

Hey there-

I totally gutted and rebuilt by saber with many, many steps of rebuild and testing. Putting in a new output section was a big help to the board. But that is just the beginning. Stock that board has a bit of excess distortion when you engage the eq. That was solved by putting in an 8pin to 16pin IC adapter (2 8pin dips) and putting in better chips. They used a 5532 that was not optimal for the application off either opamp in that chip -split the difference they did. The grounding was also particularly bad on the board - they actuall grounded each channels audio to the chassis (!!!!) check the gain pot.

The power supply distro can be upgraded and the PS bypassing on the board can be improved and upgrading the chips and running them class A improves the soundstage dramatically.

I can give you the name of the guy that teched all of this stuff out. I would expect that you'd have to compensate him for all of the time (lots) experimenting and problem solving he spent on the board but it would be well worth it. My console is astoundingly good and I do not want for something else, except for something with more facilities. Soundwise it is absolutely top level.

Good luck and pm me if you want the info of said tech dude...

Best,

Family Hoof, I see what you are saying about trying to use the BA283AV as a current summing device. I think I'm gonna give it a shot, though.

I looked a little closer at the A&H schematic and made a couple gain calculations (Please let me know if you see anything that might be incorrect):

First half of IC4: A=-213 > 46.5dB gain
Second half of IC4: A=-2.5 > 8dB gain

Isn't this towards the upper gain limit of what the BA283AV can provide? Or is it just the upper gain limit when for a 1272 (w/ BA283AV)?

Here's my reference which I'm sure you're familiar with:
http://www.technicalaudio.com/pdf/neve/0283-allvers.pdf

Assuming the BA283AV can provide the gain, here's what I think I can do.
A 15R resistor between T and V should get me about 43dB gain in the NV stage, and no RC between K and J in the AM stage will give me 15dB. This is close enough for me at first. I can always tweek this a bit later.

In the NV stage, the docs said something about a feedback resistor from S to U. I'm not sure about this one. Is it necessary? Please elaborate.

I'll definitely use some big 2200uF-4700uF caps from N and M as many have suggested before. I can put the fader between P and L, similar to what is done in the JLM Hotrod.

I'll stop here and maybe we can proceed after a little more discussion. :thumb:

PS - bushwick, I PMed you.

Greg - This sounds like an awesome upgrade for the Saber. Thanks for sharing your ideas and results.

Bushwick - Can you elaborate any more on the mods your tech did? I'm most interested in the grounding scheme and PS distro changes.

Thanks
Kevin.

Kevin:

IIRC, you're using a Saber as well? If I can get this mod working and like the results, I'll definitely write something up and maybe you can give it a shot.

I had originally thought about simply changing the L/R DIPs to discrete opamps (Melcor, API, JFET-993) with a transformer output, but I think I'm going to try this first.

I just hope I don't fry something in the process.

Greg - Yeah, I am using a Saber too. I love it but I'm definitly up for messing with the two bus. I have a couple of 283 boards and Sowter xfrmrs, just no time :?

Keep me posted.

Kevin.

GReg,

My calculator is out of batteries right now but your numbers look right. However, 46dB is an AWFUL LOT of gain to demand from the NV stage. In my experience they start to sound constricted past 30dB or so. In any case, you're going to want to enlarge the feedback capcitor quite a lot if you want to maintain goof LF response. As for the feedback resistor from U to S, this is global feedback to set overall stability, max closed loop gain I guess. Never really investigated why, I just know it was always there and somewhere between 15k and 22k depending on context. Maybe make it a little higher if you're going to go for all that extra gain.

In any case, I think maybe you'd ought to find a different amp for the first half of IC4, something that is spec'd to run inverting with so much gain. As for the second half, conside that the output transformer is going give at least 4dB of gain by itself, so you'd want to run the AM stage either at unity, or maybe at its basic gain with some preattenuation if you can't get it stable at unity. In addition, I see no reason to run the AM stage inverting. Just flip the transformer primary to get your polarity correct... although that doesn't allow you to use the unbal out to replace IC3.

I hope you'll get some second opinions and do listening tests the the 283 amps before investing too much time/money into this. Good luck!

Mods... I'll be brief...

-Rechipped the board with super high class chips.
-Upgraded power supply caps
-broke power distribution on cards and in ribbon cable and ran new power distribution
-Put in new master section, based on my techs desgin
-Reconfigured the summing busses
-Modded the board for Class A operation
-Put in a new Power Supply
-Installed Transformers on all of the inputs, custom transformers from Oliver Archut at tab-funkenwerk.com, based on some samples he sent up.
-All old grounding scheme removed and new one installed. There were four(!) ground loops and one to chassis per channel.


I am not kidding when I say that this board is astoundingly good. And has nothing to do with the sound of a Saber. I always thought that the board sounded pretty darn good. The board was modded in steps and in applicable cases was done in ABX comparison. I can attest to the virtues of each of the mods since every one of them helped.

We spent countless hours on this project and lord knows I learned a lot. Its funny, I before I never would have used the eq on the board for anything critical. I now am happy using it along side the stuff in my outboard racks and I have some really cool things. It has its own flavor.

Best

> First half of IC4: A=-213 > 46.5dB gain

Huh? No, the 220 resistor does not set gain. IC4a is a current-to-voltage converter. It accepts currents from the R52..R68 bank of resistors. The gain for one input is (something less than) 22K/47K.

IC4a IS critical to the mix quality, so it would be a logical place to try a change.

But there is nothing like this in the Neve scheme from the vintage you are looking at. In the Neve, you would have the mix resistors and maybe a swamping resistor, then a gain stage. In the A&H, R52..R68 do not present a dead-constant impedance to the mix-bus, but it is close. It would be possible to, say, load the mix bus with 1K and follow with a gain of 47 (34dB) which should be in range of what the 283 input stage can do (put 47 ohms T to V).

Frankly, I think this is putting a Packard carb on a Honda: it is a good part but just does not fit nicely.

So you didn't really mod it, you re-designed it... :thumb: It's more a "Bushwick Saber" as oppose to an "Allen and Heath Saber." But I bet it really does sound great !!!

FH:
I don't really have to get ALL the gain from the NV stage. I could try and get about 30dB, as you suggested, from the NV stage and see if I get the rest from the AM stage.

I have the PSU and the BA283AV cards stuffed and ready to use. I even have the few extra components I'll need for gain setting and filtering. I should be able to at least try out one channel tonight... or least get started on it.

I'm going to take my time.

[quote author="PRR"]Frankly, I think this is putting a Packard carb on a Honda: it is a good part but just does not fit nicely.[/quote]
:thumb: Good to have someone here who knows what they're talking about!

Yea...

[quote author="Greg"]I don't really have to get ALL the gain from the NV stage. I could try and get about 30dB, as you suggested, from the NV stage and see if I get the rest from the AM stage.[/quote]
Well, that would totally change the feel/function/scale of your master fader... if that means anything. Upon further pondering, I really don't think 283 amp for the first half of IC4, the current summing, is a good idea. Apparently PRR agrees, so... BUT I would still try and see if you can use a pair of 283AM with Sowter as the pot fader L+R output, no gain, just balls. That's really one of the best sounding output stages I've ever heard (when you replace the Marinairs w/Sowter, that is).

Current-to-voltage converter... ahh
http://www.phys.ualberta.ca/~gingrich/phys395/notes/node105.html

As usual PRR, thanks for the info !!!

I posted that at the same time PRR was giving us a lesson... :wink: I guess I just glanced at IC4a and thought inverting amp, but after looking at it again, the 220R is not setting gain, nor is the 47K going back to the inverting input. But just to make sure, IC4b is an inverting gain stage, right? A=(22k+8K2)/12k=2.5 ?

Ok let me take a step back for a second:

- Regarding IC4a, I do have a couple MM-990 and a couple Melcor 1731 lying around. Maybe I could replace IC4a with one of those. Hopefully this will help without having to do a total re-design. Thoughts on this?
- I would still like to use the AM stage to balance and add "meat," and as of now I think this is a good idea. This could replace IC4b, right?

[quote author="PRR"]> First half of IC4:

Huh? No, the 220 resistor does not set gain. [/quote]

Just curious: I've been wondering about that 220R (R72). It can't really be for nulling out some DC-offset so the best I could come up with is to isolate the bus-parasitics from the inverting input of IC4-pin#2.

Other ideas ?

Regards,

Peter

Greg are you thinking of making this a switchable option or a straight replacement?