Super Regulator, collecting the facts

[peranders]

Jung Super Regulator, some facts

Hi!

I wonder about a few things regarding the Jung Super Regulator:

How fast does it start with the original component values?

Does it work for negative voltage using AD797, assuming all signs have been changed, NPN->PNP and vice versa?

What happens when you have current limiting and the voltage is below min voltage for the opamp?

 

[PMA]

Per,

could you kindly link to the schematics?

Thanks,
Pavel

 

[peranders]

http://www.e-insite.net/ednmag/archives/1997/010297/01di_03.htm

 

[dimitri]

May I ask a question without concern to copyright issues?

After building super regulator we connect it to several opamp via RC filters to prevent possible loops. The individual supply to each opamp will have source resistance R in parallel with C, now C value and it type again become significant. How should one choose R and C values? Should R be replaced by inductor? I feel that using super regulator for each opamp is too much.

 

[peranders]

jwb, you have a point here and I don't mind also good info collectors.

dimitri, I would have chosen 10-100 ohms + 47-470 uF// 100nF/63 polyster/ceramic. This I think is a good compromise for some isolation between the opamps. This is also good thinking.

Use 100 ohms to start with. Lower the value if you feel that you get to much voltage drops. Less than 10 ohms the isolation effect become rather small.

 

[dimitri]

Peranders, as we use 10-100 ohms + 47-470 uF// 100nF (corner frequency 3-300Hz) why we need super regulator with milliohms in wide frequency range? We will lose the most of it strength.

 

[johnferrier]

Linear Technologies LT1963

Peranders,

Isn't the SuperRegulator idea incorporated into currently available low noise, low dropout regulators??? For example the LT1963 1.5 A regulator has less than 40uV noise (and the LT1763 0.5A regulator <20uV noise). A good voltage reference such as a REF02 has a typical noise of 4uV to start with.

LT1963 Datasheet:
http://www.linear.com/prod/datasheet.html?datasheet=886

Here is the TI datasheet: http://focus.ti.com/lit/ds/symlink/tps78601.pdf

Here is a good LT app note: http://www.linear.com/pdf/an83f.pdf

I've wondered about why people aren't using/talking about these.

(To be frank, my interest is to build a headphone amplifier so 1.5A is more than enough. For higher current capability, can't the LT1963 be used with a power FET?)

Comments anyone?


JF

 

[randytsuch]

Hi Guys,
I just bought a pair of kits from Andy, he sent them out this weekend, but they have to go from England to LA, so not sure when I will get them.

I plan to use them in a simple FET preamp, to generate the +/15VDC I need.

The original design called out for 3 term regulators, but I wanted to upgrade .

I will let you guys know how it turns out, but as with all my projects, I am sure this one will take a while to complete.

Randy

 

[jwb]

Jung (and Sulzer, and other error-amp-with-pass-transistor designs) have extremely low output impedance and very wide bandwidth. The output impedance is much lower than three-terminal jobs like LT1962. That's why my headphone amp, with idle current only a few hundred milliamps, uses four Sulzer-type regulators instead of the commercial parts. The distortion reduction is significant.

What actually sold me on the Sulzer regulator was its amazing ability to regulate the load. Stepping from 0 to 1A output barely perturbs the voltage level. I was truly impressed.

Also three-terminal regulators tend to use noisier voltage references. In a superregulator you can use subsurface zeners or other high-quality refs.

 

[peranders]

You have many good integrated regulators but this thread was about the Jung regulator in particular. The examples which you have mentioned is way over the 78xx/79xx but in some application nothing (at the moment!) beats a discrete regulator.

 

[peranders]

Also three-terminal regulators tend to use noisier voltage references. In a superregulator you can use subsurface zeners or other high-quality refs.

...and you can use any LP-filter you like to filter out the noise.

 

[johnferrier]

Jung (and Sulzer, and other error-amp-with-pass-transistor designs) have extremely low output impedance and very wide bandwidth.
...
That's why my headphone amp, with idle current only a few hundred milliamps, uses four Sulzer-type regulators instead of the commercial parts.

JWB,

Thanks, that is information I wasn't aware of and the feedback I looked for. (Sorry Peranders, at least I'm not on the IP tangent...)

Are you happy with your Aleph headphone amp design? Do you recommend this topology? Right now, I'm interested in a differential design and am wondering about SuperSymmetry. The leading candidate today is a pair of crosscoupled OPA637/discrete output buffers. I don't mind going to the effort of making one headphone amplifier, but I don't plan to wish I had constructed something else. (Like everyone, I would like to make The best that I can.)

I will reconsider the SuperRegulator idea based on your feedback.


JF

 

[grataku]

Going back to the subject matter I think the most refined Jung style SR at the moment is the ALW one. I am very intersted in the subject since I want to use it in my new, for personal use, preamp.
Unfortunately for reasons of PCB size and number of boards needed I need to design my own reg.

 

[fmak]

Re: Linear Technologies LT1963

I've wondered about why people aren't using/talking about these.

(To be frank, my interest is to build a headphone amplifier so 1.5A is more than enough. For higher current capability, can't the LT1963 be used with a power FET?)

Comments anyone?
------------------------------------------------
I did , but no one was interested and went back to 20 yr old designs. I have been using these with good results.

Indeed the TI chips are good but they are spoilt by the need for a large ground plane to combat 50Hz noise. Why I don't know but all of that series is susceptable. They are also spoilt by the packaging not being a drop in replacement for 78xx. Neg regulators are also not extensive.

Perhaps more to the point is that 78xx and 79xx regs can be very good and much better than spec. Thre JRM ones especially so but one does need to select. The LT and LM series arev really not that good but look expensive!

 

[johnferrier]

Re: Re: Linear Technologies LT1963

I've wondered about why people aren't using/talking about these.

(To be frank, my interest is to build a headphone amplifier so 1.5A is more than enough. For higher current capability, can't the LT1963 be used with a power FET?)

Comments anyone?
------------------------------------------------
I did , but no one was interested and went back to 20 yr old designs. I have been using these with good results.

Indeed the TI chips are good but they are spoilt by the need for a large ground plane to combat 50Hz noise. Why I don't know but all of that series is susceptable. They are also spoilt by the packaging not being a drop in replacement for 78xx. Neg regulators are also not extensive.

Perhaps more to the point is that 78xx and 79xx regs can be very good and much better than spec. Thre JRM ones especially so but one does need to select. The LT and LM series arev really not that good but look expensive!

Combat 50hz noise:
LT1973--Low Noise: 40uVRMS (10Hz to 100kHz)

Neg regulator not extensive:
But positive regulators can be used as negative regulators, though with some degradation of performance.

The LT and LM series arev really not that good but look expensive!:
Though cheaper than discrete SuperRegulator circuitry.

If one wants a Jung-like SuperRegulator, take your favorite low noise, low PSRR, DC power amplifier circuit and bootstrap a reference voltage output to input. And use a pair of those to power the target amplifier circuitry. (Although I risk not making a lot of sense, I am trying to stay on topic.)


JF

 

[andy_c]

Re: Jung Super Regulator, some facts

(...)What happens when you have current limiting and the voltage is below min voltage for the opamp?

Well, I think this can be determined by looking at the schematic. I think you're asking what would happen If one added current limiting to the circuit and it was enabled due, say, to a short at the output? From what I can see, it would be the following. The feedback would try to drive the pass transistor to as high a current as possible due to the shorted output. Maximum base current to the pass transistor occurs when the PNP emitter follower connected to the op-amp output is in cutoff, allowing all of the PNP current source's current to be delivered to the base of the pass transistor. Since the PNP emitter follower's emitter current would be zero, there would be zero volts across the zener and its series resistor. This would put the emitter of the PNP emitter follower at about +0.7 Volts. But the op-amp might only be able to swing within about 2 Volts from its negative rail (ground). So the base of the PNP emitter follower would be sitting at about 2 Volts. Looks like you'd end up with a reverse biased base-emitter junction of the PNP emitter follower, but not enough to break it down. There would also be a significant voltage difference across the differential inputs of the op-amp in this state, but if you've read ALW's site, you'll see that the diodes he puts across the differential input will protect against that.

The startup process is described in detail at ALW's site.

 

[jewilson]

S-J-D Regulators

There are regulators designs similar to the Jung-Sulzer-Didden around before their articles were published. If you have an old National Semiconductor application data books you will find regulators that look to be the foundation for this implementation.

While the implementation is different it may or may not be new for the time “it is still very well done” however I believe audio performance claims made in the articles for the circuit is new. The design offers great low noise performance and the ability to track dynamic changes in at the load with excellent stability. The keys are precision voltage reference, the high-speed transistor, the ultra low noise opamp, and the added ground sense implementation for audio.

My understanding is the only copyright protect belongs to the author work and what he is done in the development of the documentation for Audio Amateur-Audio Express magazine. They are no patents or trademarks other than the author name and no intellectual property issues.

To have intellectual property issues, as I understand it, this would have to have information i.e. a process, a method, or a design secret one has acquired at a company. Now, the person is at another company and if he implements the information gained at his previous employer that could be an intellectual issue. Therefore, the act of publishing the article for users, and users paying for the article seems to cancel out the intellectual property rights.

So using this circuit is ok for your home project. However, it would make good sense to alter and improve the design for commercial use.

 

[fmak]

Re: Re: Re: Linear Technologies LT1963

Combat 50hz noise:
LT1973--Low Noise: 40uVRMS (10Hz to 100kHz)

----------------------------------------------------------------
If you try them out ( and I have) you will find that only the grounded tab models are free from 50Hz pickup. All others require ground planes which make their use a hassle.

 

[johnferrier]

Re: Re: Re: Re: Linear Technologies LT1963

Combat 50hz noise:
LT1973--Low Noise: 40uVRMS (10Hz to 100kHz)

----------------------------------------------------------------
If you try them out ( and I have) you will find that only the grounded tab models are free from 50Hz pickup. All others require ground planes which make their use a hassle.

I plan to give them a try. Here in the US, I'm sure that 50hz won't be a problem (obviously, 60hz might be a different story). Thanks for the info about the tab v no-tab.


JF

 

[Fred Dieckmann]

Duly noted

Despite my feelings that a circuit development for 20 years and with input from the readers was "public domain," Jan does not. I have to respect his feelings since he put a lot work into the project. I am very pleased that he approves the results of ALW who has put a tremendous amount of work into his design, that look to me to be real improvements. The Jung regulator was not typical most hobbyist magazine projects in terms of input from others and length of development. I view the work by Mr. Pass in a much different manner. It is the product of his work alone, and commercial copies compete in marketplace were he earns his living. He has weighed in on the issue of the Aleph PCBs and I will let him speak on that matter. I live in mortal of misrepresenting his feelings to even the smallest degree. The fact is that the Jung regulator was based very heavily on the Sulzer article, which addressed the issues important in regulator design in a very concise and straight forward manner. I think the original Sulzer article was one of the best pieces ever published in TAA and is the Rosetta Stone for audio regulator circuits. I like to reread it from time to time to see the forest in addition to examining the bark of one particle tree with a magnifying glass. It is a great overview and is still very relevant after 20 something years.

The AD797 is not a very good op amp for the circuit due to possible stability issues when used with remote sensing. I also believe its phase margin is not as good as several other op amps often used for this circuit. Some of the jfet input op amps seem to work very well in this circuit. I would not even use remote sensing, but put the regulator very close to the circuit it is providing the regulation for. Unless you are careful you can wind up with stability issues and noise pick up when using remote sensing. Line rejection, good transient response, and linearity seem to be more sonically important than lowest absolute output impedance or noise. The noise level has as much to do with the voltage reference, filtering of the reference, and the PCB layout, as the input noise of the op amp.