Looking for a neutral DOA for summing

[Rocinante]

I've had really good results with these:
You can pick, class A/B, class A, dc servo and it takes a variety of ic's. I've had best results with 2604.

https://www.jlmaudio.com/shop/hybrid-opamp-kit.html

Also if you can do 24v+- the app992 is a great DOA.

 

[bluebird]

Then you can start attacking the system with modern regulators (not LM317/337).

You mean low drop out regulators? Whats wrong with the LM317/337? Or rather, why are the modern ones better?

Hey living sounds, For shits and giggles you should try a MC33078. I like how they sound when pushed. Different output than most IC opamps.

 

[Monte McGuire]

You mean low drop out regulators? Whats wrong with the LM317/337? Or rather, why are the modern ones better?

The output impedance of the old regulators is not all that low, so while they provide input ripple rejection, they don't establish a 'stiff' rail to drive the amplifiers. For discrete circuits, op amps or otherwise, that do not generally use current sources, and thus have a much higher gain at their power supply terminals (i.e. low PSRR), a better regulator that provides a low impedance output can minimize some faults that these amplifiers and circuits are prone to.

These days, I've been using the Analog ADP7142 and ADP7182 regulators, which by my own measurements, have an output impedance of around 22mΩ around 1kHz when loaded around 20mA, which degrades at higher frequencies due to gain bandwidth limits. This is pretty good performance, and you can do much better with a custom "high gain amp and a pass transistor" regulator, but it's a lot better than a 317/337. Plus, the parts are available in a 3mm x 3mm DFN chip, making them easy to sprinkle everywhere, if you don't mind paying for them.

Power supply regulation can be important, but it's just one tool to allow a designer to structure the transient signal related currents within a design, as part of a complete power and ground system that isolates what needs to be isolated, and combines what needs to be combined. IMHO, simple discrete circuits are most sensitive to power supply rail impedance, since they have very low attenuation from the power supply rail to their output - if this rail 'bounces' with the signal, and there is only 10-20dB of power supply attenuation to the stage's output, that regulator is probably going to be audible, and that's a design failure.

If you have a simple circuit and you want the highest performance, it might make sense to use a custom regulator, probably using a reference, a high gain-bandwidth op amp, and a pass transistor - this will provide unbelievably low output impedance and noise, but it's complex. Still, if the basic circuit is small and you want to maximize it, a super supply (along with a clever ground layout) might be the only way to meaningfully improve it.

The biggest annoyance is that there are very few modern, high performance positive-negative regulator pairs, since the electronics industry generally doesn't use high voltage bipolar analog anymore - it's heading toward 5V or 3.3V unipolar. So, while the 7142/7182 are a nice 200mA regulator pair, there aren't that many other modern pre-made regulators - let me know if I've overlooked any worthy candidates!

 

[volker]

The biggest annoyance is that there are very few modern, high performance positive-negative regulator pairs, since the electronics industry generally doesn't use high voltage bipolar analog anymore - it's heading toward 5V or 3.3V unipolar. So, while the 7142/7182 are a nice 200mA regulator pair, there aren't that many other modern pre-made regulators - let me know if I've overlooked any worthy candidates!

There is also the TPS7A30/TPS7A49 and TPS7A33/TPS7A47 from TI. Or just use a positive regulator for the negative voltage as well.

 

[Monte McGuire]

There is also the TPS7A30/TPS7A49 and TPS7A33/TPS7A47 from TI. Or just use a positive regulator for the negative voltage as well.

Yes, these are also some good regulators, with higher current capacity, but I haven't used them yet. Thanks for listing them here, since there are so few modern positive-negative regulator pairs!

Indeed, you can use a positive regulator for a negative output, but only if you can provide floating drive to each regulator, and that can't always be done. But, if you can do that, the advantage is that the positive and negative regulators will be identical, and will have identical impedance etc., something that is not possible with the positive-negative pairs listed here. The positive and negative versions always behave a little differently, and while it might not be a problem, it's yet another annoyance to deal with. Plus, being able to use only one type of part rather than two is always a win re. assembly and parts cost, even if it provides no electrical benefit.

 

[JohnRoberts]

I do not mean to offend anybody but this is getting a little amusing... quibbling about the sound of voltage regulators.  ;D

Here's a very old VR trick I shared many years ago. The classic 3 terminal regulators were saddled with modest gain bandwidth due to early IC technology. This modest gain bandwidth resulted in a rising output impedance with frequency as the NF loses effectiveness. Adding an electrolytic cap (say 1,000uF) in parallel with the regulator output complements the rising VR impedance with the cap's falling impedance up to the electrolytic's very HF resonance where a ceramic disc takes over. .

BTW if you can hear the VR that suggest the circuit has poor PSRR, or a poorly designed PS. (marginal unregulated voltage can introduce ripple into the regulated rail).

I make a point of not arguing with people on the WWW about what they say they can hear. I know what I can't hear (accurately, reliably). That's why I used bench equipment. It is also why I ran, not walked away from the audiophool infested hifi market.

JR

 

[abbey road d enfer]

The output impedance of the old regulators is not all that low,

That's why they need to be complemented with output capacitors. So far I have found nothing wrong with this. And I don't think the new über-regs can be used without input and output caps.

These days, I've been using the Analog ADP7142 and ADP7182 regulators, which by my own measurements, have an output impedance of around 22mΩ around 1kHz

In order to maintain this low impedance, the source impedance must be kept low. For example, in a mixer with a remote PSU, the source impedance is rather high - this is even truer for mixers where distribution resistors are used for protecting against shorts. So the role of maintaining a low source impedance on the input side is devoted to the input caps. The current flow is thus derived to the local gnd, not the remote ground.

Power supply regulation can be important, but it's just one tool to allow a designer to structure the transient signal related currents within a design, as part of a complete power and ground system that isolates what needs to be isolated, and combines what needs to be combined.

I wholeheartedly agree. The implementation of capacitors, the circulation of signals and current paths are essential. IMO the regulator's dynamic impedance performance is secondary. Many mixers do not have on-board regs; instead they have these 47 or 22r for short-circuit protection. On-board regs in a mixer have often a serious drawback; the noise currents resulting from shunting their output with caps to ground can find their way into the audio ground. I've seen some mixers where replacing the regs with resistors decreased significantly the bus noise.

 

[JohnRoberts]

On-board regs in a mixer have often a serious drawback; the noise currents resulting from shunting their output with caps to ground can find their way into the audio ground. I've seen some mixers where replacing the regs with resistors decreased significantly the bus noise.

I learned that lesson the hard way.... ended up blackballing the the regulator manufacturer when they wouldn't guarantee a noise spec. Changing to different brand (quieter)  regulator solved my immediate drama, but I can confirm that what Abbey describes can happen in the wild on large enough console platforms.

JR

 

[Michael Tibes]

Well, an SSL 4000 is not class A, but its summing bus still sounds very nice to me.

...and if I remember right on every pcb the +/- 18V passed through 10R at least once, entering the board - in addition to the resistance of the cables and connectors. Does this mean something in terms of regulator output impedances?

Michael

 

[abbey road d enfer]

...and if I remember right on every pcb the +/- 18V passed through 10R at least once, entering the board - in addition to the resistance of the cables and connectors. Does this mean something in terms of regulator output impedances?

Yes it does. The line regulation factor expresses how the input variations affect the output voltage. But it's moot, since DC is out of the audible spectrum. The dynamic impedance presented to the circuit is what matters, since all currents are subordinated to it. DC stiffness is almost irrelevant when dealing with transients; most of the energy necessitated to properly reproduce them comes from the reservoir capacitors. Indeed, the supply rail must be able to maintain their correct charge.

 

[Gold]

Abbey, could your reply be restated as 'it doesn't matter how much current the regulator can output if the resivoir capacitor can't supply the necessary current on demand'?

 

[abbey road d enfer]

Abbey, could your reply be restated as 'it doesn't matter how much current the regulator can output if the resivoir capacitor can't supply the necessary current on demand'?

Well, it could be, and it would be as questionable as my voluntarily simplified initial statement  ;D
I guess we could say "the regulator cannot output more than what the source of primary voltage combined with the capacitors can supply".

 

[Gold]

Thanks. I'm cool with some blissful ignorance.

 

[Kingston]

You mean low drop out regulators? Whats wrong with the LM317/337? Or rather, why are the modern ones better?

Stability, noise, output impedance. LT3080 and its variants are quite superior to those two. I once managed to kill all spurious oscillations of a console with a PSU based on super regulators (jung regulators - whatever).

Then repeated the feat. with LT3080. I wouldn't build super regulators anymore when they are basically available on a chip now.

If an opamp is setup correctly it should be totally inaudible?  Is that even possible?

Yes and yes.  Everything else is audiophile cable talk. The claims and apparent differences are always the same. With measurements nowhere to be found.

 

[abbey road d enfer]

So, unregulated supply with 47R or 22R placed where?  Before the B+ reaches each channel amp, have a cap and resistor from B+ to 0V as in the BA283 output amp example?

Adam

Check that (Soundcraft TS24)
https://groupdiy.com/index.php?topic=44786.msg815216#msg815216
page 1 south-east corner, the 10ohms resistors

 

[abbey road d enfer]

I once managed to kill all spurious oscillations of a console with a PSU based on super regulators (jung regulators - whatever).

That suggests a very poorly designed console.
When I made consoles, I tested them on an unregulated supply. Only when I had ironed out all the buzzes and oscillations would I use a regulated supply.

 

[Kingston]

That suggests a very poorly designed console.
When I made consoles, I tested them on an unregulated supply. Only when I had ironed out all the buzzes and oscillations would I use a regulated supply.

Now that's a hardcore designer! I think I will have to borrow that trick sometime.

And yes, said console was poorly designed.

 

[JohnRoberts]

So, unregulated supply with 47R or 22R placed where?  Before the B+ reaches each channel amp, have a cap and resistor from B+ to 0V as in the BA283 output amp example?

Adam

In large console design there is a motivation to isolate single points of failure, so they do not crash the entire console. Back last century it was not uncommon for ICs to release their smoke and put a dead short across the PS rails, basically killing the entire console. The popular remedy for this was modest 10-20 ohm FP (flame proof) resistors in series with each rail to act like a cheaper, more reliable fuse, to allow a channel strip to fail gracefully and not shut down the entire console.

JR

 

[80hinhiding]

.

 

[JohnRoberts]

Now that's a hardcore designer! I think I will have to borrow that trick sometime.

And yes, said console was poorly designed.

Just experience and good practice... when designing a large  product with multiple sub systems operating together, they need to all be robust when working by themself.

Regulating the PS is just adding another zero in front of the hum noise floor. Stability "should" not depend on PS regulation, but there are many less than optimal solutions out in the world.

I like to joke that consoles are one of the hardest simple circuits to design. The individual circuit blocks are pretty simple but connect 10s of channels together and you can have a different experience.

JR