crazy high rail voltage = better????

[buschfsu]

so i was looking at the ADL 600 (600 volt rails!!) and the new neve stuff (90 volt rails) and wondering what componenets could even handle that much voltage????


Seems like these high volt rails are supposed to bring higher headroom to the table. I understand that wave amplitude(transitent) and headroom are related but isn't this a function of current not voltage?

 

[thermionic]

Marketing...

BTW - there are small-signal BJTs that go up to 120v around from firms such as Sanyo.

On a complementary / PP design, you could run them @ +/-120v!! Just think of the dynamic range - the guys in the marketing department would be smiling!

Read Paul Frindle's comments here: http://www.gearslutz.com/board/mastering-forum/169598-headroom-analog-equipment.html

It appears that the rest of the forum aren't too keen on agreeing - but they can't say why! "err..higher rail voltage is better, isn't it?"

:roll:

Why make life difficult for yourself? If you need the headroom, have the voltage; if you don't, why not be content to know that, aside from extra cost, it's easier to keep a lower voltage circuit quiet?

BTW - anode voltage is a different animal entirely - I write the above for discrete / op-amp-based designs.

Justin

 

[JohnRoberts]

All things equal, if you have a constant noise floor and more headroom (signal swing), in principle you have more dynamic range available. In practice it's not quite so simple. You are not going to encounter line sources with silly high voltage swings so you'd have to apply gain to those sources to get the swing up. That gain would also increase source and input noise along with the signal for a wash. Likewise you aren't going to find outboard gear ready to accept huge peak to peak voltages so once again you must pull back on level. This would only preserve S/N if you added a pad to interface with lower rail gear.

Looking into the way back machine, once upon a time the API consoles all discrete design and hotter than IC op-amp rails delivered slightly more bus headroom, but this benefit was more perceived than real, only a couple dB.

Good attention to design detail will do far more than brute force high voltage power supplies. With input stages the noise is characterized by the very first stage. To fill +/- 45V rails will take 3x the gain as +/- 15V rails so while the signal is 3x so is the noise, so S/N is the same.

At the summing bus, you could combine 3x as many 15V signals before clipping, but you will still have to pad that down before sending it to any conventional path again, so you end up with the same S/N as the bus running at appropriate gain to not clip in the first place.

I see some transitory benefit in bus headroom, but this is not better than properly scaled normal rails, just perhaps a little easier. While probably sellable to some recording types, for use before they route their signals to A/D convertors running from a 3.3V rail.

JR

 

[thermionic]

It appears that you and Paul Frindle are in perfect agreement, John - great post btw :guinness:

In the link I provided, Bob Katz suggests that high-end OEMs over-engineer for better sound, with higher rail voltage being one of the attributes.

My attitude would be, if you only need +/-15-or so volts for an application, why not stay at that voltage, but still apply the same standards of overall engineering? If your budget for a project is 'x' and 30v P-P gives you all that you need, instead of building it with +/-60v, why not spend the money elsewhere, keeping it to +/-15v?

I also wonder if some users are unaware that if you double your rail voltage, you only get an extra 6dB!! If a modest voltage keeps you clean to +22dBu - and your convertor is calibrated so that 0dBfs = +4dBu, having a clean signal to +28dBu, with the extra expense that it entails, is hardly going to turn your record from Gold into Platinum, is it?

Elephant, not mouse...

Justin

 

[solder_city]

that paul frindle seems like a smart chap. he should be hanging out over *here*, not at stupid gearslutz

 

[thermionic]

[quote author="solder_city"]that paul frindle seems like a smart chap. he should be hanging out over *here*, not at stupid gearslutz[/quote]

Tell me about it... I sent Paul a link to this place a while back. I do know that he endeavours to keep BS out of the industry, so there could be an angle that, whilst it's more tech-savvy over here, he feels that the spare time he has is better spent curbing GS from putting forth more misinformation into Google...

Aside from being a world-recognised authority, PF is also one of those guys you can't 'buy'... If he doesn't like the look of a project, he won't take it on - period.

Justin

BTW - You do know he's the main man behind the SSL E / G design and Sony Oxford?

 

[JohnRoberts]

This was all thoroughly sussed out decades ago by designers working in this area.

Another data point, most pro gear that provides active balanced (differential) output drivers, enjoy 2x the nominal voltage swing. Again, no more actual headroom or S/N benefit since you must scale up with 6 dB of voltage gain to use that signal swing.

This higher voltage swing only has practical benefit for long sends in noisy environments, and even then the benefits are modest.

Note: there is a practical limit in cost effective design due to standard opamp technology process voltage limits. Further the reliability tradeoffs from running the ICs right up to the edge of their process limits are not justified for low single digit headroom differences.

Modern opamp technologies push this process limit and it's mostly a GOP (good on paper) spec, but a dB here and a dB there with no downside are worth taking advantage of of the price isn't too dear.

JR

 

[tv]

What about gear that was "debalanced", modded? There would be some benefits with more volts in rails. But that wouldn't fit in marketer's speech either.



Snatched this pic from some mic-pre design webpage which I can't remember. Likewise it looks like I can have a badass hi-volt 5534an for cheap ...

 

[Guest]

[quote author="tv"]What about gear that was "debalanced", modded? There would be some benefits with more volts in rails. But that wouldn't fit in marketer's speech either.



Snatched this pic from some mic-pre design webpage which I can't remember. Likewise it looks like I can have a badass hi-volt 5534an for cheap ...[/quote]

I don't think that lost bandwidth and slew rate means "cheap", especially for 741.

 

[Michael Tibes]

Probably totally silly, I'm not good at theory (and perhaps my english isn't good enough): might a higher supply voltage lower distortions because the transfer curve is spread over the whole supply voltage range and therefore the area we use at line levels becomes more linear?

Michael

 

[Samuel Groner]

Might a higher supply voltage lower distortions because the transfer curve is spread over the whole supply voltage range and therefore the area we use at line levels becomes more linear?

Yes. Given a specific amplifier and signal level it will usually have lower distortion with higher supply rails as early effects and voltage-dependent junction capacities are reduced. On the other hand, thermal effects might be more pronounced.

But this is just theory, in practice other design decisions will be more important for low distortion. And I doubt that the gear with high supply voltages which has so far emerged is especially good with respect to distortion--at least the SPL stuff isn't (as far as I know).

There are small-signal BJTs that go up to 120 V around from firms such as Sanyo. On a bipolar design, you could run them @ +/-120v!

No--you could run them on +/- 60 V.

Samuel

 

[12afael]

that`s why fender use a more high voltage power supply and van halen use less voltage.

 

[clintrubber]

[quote author="12afael"]that`s why fender use a more high voltage power supply and van halen use less voltage.[/quote]
Yeah, and AC/DC rectifiers in their Marshalls :roll: :wink: :shock:


But serious, and as mentioned in fact already, there's more going on at the same time than one property will result in in a ceterus paribus experiment.

 

[Guest]

[quote author="12afael"]that`s why fender use a more high voltage power supply and van halen use less voltage.[/quote]

Output toobs were cheap for Fender. And he forced to manufacture toobs that can live longer when abused.

 

[thermionic]

[quote author="Samuel Groner"]

There are small-signal BJTs that go up to 120 V around from firms such as Sanyo. On a bipolar design, you could run them @ +/-120v!

No--you could run them on +/- 60 V.

Samuel[/quote]

Samuel,

As I wrote earlier in the thread in post no. 4, I am fully aware of the concept of P-P voltage. I have been building complementary power amps for many years. I should have worded it as a 'complementary design', in which case you could run them @ +/- 120v.


Justin

 

[mikep]

great topic

one more thing to consider, if you opt to bootstrap low voltage op amps to enable wide output swing, it doesnt buy you any input range! hi-gain is not possible, and no inverting stages. well it will work, but with reduced headroom so what is the point?

IMO some designers must simply do it (high rails) because they can. dynamic range? you lower the noise, you get the same result.

IIRC API did not use more than +/- 16. they have good current driver output amp into an output TX with some step up ratio.

In my opinion, you would go to elevated supply voltages if you need to interface with circuitry/devices in that range: high power audio amplifiers, electrostatic speakers perhaps, and industrial apps. or it could be worthwhile if it simplifies things, like to utilize an existing system rail voltage. say, a mic pre that is powered by +48V single supply.

mike p

 

[JohnRoberts]

I may be misinformed on old API consoles, I never saw schematics or looked under the hood.

JR

 

[PRR]

> Snatched this pic from some mic-pre design webpage

A variant of that scheme is used in a DBX limiter, to run a bifet chip opamp and buffer to "Pro" output swings.

> I don't think that lost bandwidth and slew rate

The trick is universal. Probably goes back to Burr-Brown app-notes for 100KHz GBW modules. Rediscovered by the 741 generation. Works fine for the biFETs used when DBX made my limiter. Should work fine for whatever is good enough for you.

Using heroic rail voltages seems to me to be the "last cheat", -after- you have spent appropriate money/time on good topology and exquisite detailing.

Voltage alone is bogus. S/N is about POWER. Running 100V and 1mA gives the same performance as 10V and 10mA stuff. The real limit is the ~~100mW power output of small chips, ~~300mW for a compact pair of TO92s, or the 100 Watts of a board-full of high-power bus drivers.

And in a complex system, going to high power is self-defeating, because "small leakage" floods your buses and grounds with garbage.

 

[Samuel Groner]

I should have worded it as a 'complementary design', in which case you could run them @ +/- 120 V.

Can you elaborate? With signal just below clipping a transistor in the voltag amplifier stage will see the full power supply voltage, complementary or not. The input stage is another matter for a power amplifier as CM range is limited, but for opamp-like circuitry we want full CM range so the transistors might see the full rail voltage there as well.

If you opt to bootstrap low voltage op amps to enable wide output swing, it doesn't buy you any input range!

LTMag_V04N2_Jun94.pdf. See page 20-22.

Samuel

 

[Guest]

Upon thinking I like the idea. Let's create a new fasiion! It is the only fashion I like! :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb: :thumb:

If you want I can invent a lot of reasons why it is good and needed.

For example, MOSFETs have lower and more linear capacitances when working on higher voltages. This idea may be supported by few engineers from our forum. Take curves, design schematics, start DIY projects, etc...

I have soe more ideas...