[Tomo]

Hey,

Ok I am posting the calculation for ordinary LM317 regulated PSU.

Assumption: You have adequate post regulator stage.



Lets say 500mA draw

Lets have 4700uF preregulator stage.

2.4 X 500 / 4700 = 0.255V

I get 0.255 V ripple at the end of preregulator stage.

LM317 gives about -80dB rejection.

-80dB = 20 X Log10(V/Vo)

Vo = 0.255 V

V = 25.5 uV after regulation.

For my opamp based amps, current draw is probably 50mA ~ 100mA. Then,

2.4 X 100 / 4700 = 51 mV
2.4 X 50 / 4700 = 25.5 mV

I get 25.5~51 mV at the end of preregulator stage. Then the resulting ripple at the end of regulator stage is:

V = 2.55 uV ~ 5.1 uV

I think it is perfectly nice. However, note that LM317 is not particularly good at regulation at high frequency. I kill this with low pass filter, but if you use Jung's then you should get -80 dB at even at high frequency. (100kHz)

With -100dB which Jung's offer, For 50~100mA draw,

V = 0.25~0.5 uV

Which is even better. That is why I say Jung's full fledged regulator is better.

T

 

[Tomo]

The calculations came from "Building Power Supplies" by Davind Lines. (Published by Radish Shack!)

 

[Tomo]

Ok ... Whatever.

I really care less if YOU make money or not. I am more interested in discussing about the amplifier design or the designs associated with amplifiers.

Let's get back to the issues of power supplies. I wanna see calculations and measurements.

Jung's regulator can give you ripple rejection of 100~110dB, ripple of about 0.5~1uV and noise figure of 10uV. According to my calculation, LM317 is producing about 50~100uV of noise on top of 5~10uV ripple. (But remember to have low pass filters to kill high freq.)

This one order of magnitude difference may not cause much differences. It is especially so if you are using opamps and buffers which has fairly high power supply rejection.

Tomo

P.S. Morsel and Erix please do fill your differences in personal email or something. I know it is really hard. I was bad once and got banned too. Let me tell you, it felt stupid after all the heat is gone.

And, umm ... I don't consider DIYaudio people that way. I think whole a lot of them are extremely experienced more so than many of us here. Stop arguing before you make anymore stupid statements you will deeply regret.

 

[Tomo]

Oh, speaking of which,

Morsel, can you give me some figures so I can make calculations? Despite the fact that I express doubts on cap multi design, I really could use measurements. I have given up on that design and hearing you talk about it makes me think I may have missed something.

Besides, if you got DATA, you can really shut us up real quick.

Tomo

 

[morsel]

Tomo, I don't want to get into it with you atm. Sorry. Maybe in a few weeks after Kurt and I have had a chance to do our tests. We will have plenty of data then.

 

[Tim]

So how 'bout them power supplies?

 

[Tomo]

Thanks

I will be waiting.

It is one of many things I keep in Tomo's X-Files, the unREsolved cases.

Tomo

 

[Voodoochile]

Quote:

It is one of many things I keep in Tomo's X-Files, the unREsolved cases.

Do you keep the Mets in that file, too?
(Not that the Red Sox are anything to brag about

)

 

[Tomo]

Hey!

That reminds me. Darn ... did I miss it? I knew I missed something in last few weeks. One more file unREsolved.

Tomo

 

[ppl]

Hi Tomo. Glad to Hear from you you have been quiet lately. First let's consider the Noise issue. voltage regulators have Voltage gain and also use feedback. this fact alone makes the Device more prone to Limited bandwidth and transient response ringing. more over most regulators being a feedback system do not like large capacitive loads you just can't hang a 2000uF capacitor on the Output of a conventionally configured LM-317 and worse on the LM-337 unless you add a small series resistor on the output and within the feedback loop so as to keep the output impedance low. you can compare a typical voltage regulator as a Closed Loop Amplifier with all the Benefits
and problems one encounters with these systems.

Next consider the capacitance Multiplier this is comparable to an open loop buffer since it is basically an emitter follower. hear you get all the attributes of emitter followers like an extremely wide Bandwidth and a output impedance that is constant over an extremely wide frequency range. In addition to the Ability to drive what ever capacitive load you want with large capacitors only reducing the bandwidth. transient response of an open loop system is also better than a closed loop system. Now you also, get all the disadvantages of an open loop system like higher output impedance and if a noisy transistor is used the output noise could be higher than in a closed loop system. this is not really a problem because the output impedance is a function of the capacitor on the Base of the Pass transistor and the transistors Hfe. so with proper component selection the attributes can be maintained and the drawbacks greatly reduced.

Hay **** I just got a wild idea why not use a capacitance multiplier and place it within the feedback loop of a voltage regulator. humm must work on this.

regarding my contribution to the PPA when tangent approached me about doing this I told him on the phone I did not want any compensation and I was engaging in this project so as to allow the Head-Fi/Headwize community an easy way to obtain one of my amp designs and not have to make boards them self's. I also have no problem with those that wish to clone this Amp on there own board as is often done on the META42. my primary Concern on the PPA is that the Amp dose not get compromised to the point of not performing as well as I think it should. now I understand that compromises are needed to get a practical Project however this is not the way I am used to doing things in Audio. I believe that if your are going the spend the time and expense of the DIY thing that cost of parts nor complexity should not enter into the equation. otherwise you might as well go buy a ready made amp because if you consider your time as valuable then it would be cheaper to buy rather than Build. the Exception is students who are still in the Process of learning and are probably strapped for funds anyway. moreover it is a lot less painful to have $25.00 in Parts go up in smoke than to have $300.00 in parts become a pile of carbon this I can definitely relate to, Been there Done that.

Those that express concerns of the team losing direction on this project from the Posts on this forum it is starting to confuse me also. So I will state what I envision this amp to be. First and formats it is to be portable and the best Portable Available anywhere from anybody. This requires that the Internal circuitry be somewhat complex not because I like complexity for it's own sake but because a cretin amount of complexity is absolutely required in any high performance circuit. Considering the performance and sound quality potential of this Amp it is relatively simple. remember my previous high performance portable Amp used capacitance multipliers in addition to Fet op amp rail isolation. due to space limitations we decided to go with a topology similar to my pocket amp that replaced the capacitance Multipliers on the Op amp rails with virtual ground drivers at the op amps. this was to allow the voltage divider action created by the Jfets and the Virtual ground driver to provide isolation of the sensitive op amp rails from contamination by noise and any Audio riding on the Output rails caused by the Load . while the amount of isolation is not as good as provided by the more complex capacitor multiplier it is nonetheless is about 100 DB down rather than 120-130 Db the cap multiplier would do. To further augment this The PPA uses a dedicated amplifier to serve as the output Ground this removes the contamination of the ground from the load currents. in addition since the ground channel being an Amplifier it also has power supply rejection and this further removes load current transients from the Output supply rails.

considering all the effort done inside the amp to deal with power supply and load current noise the outboard power supply can be less demanding in its performance than otherwise would be required. this is why the outboard supply can be considered as a separate issue and why I am willing to allow Morsel and Kurt to test different outboard power supply topologies and not complain to much. this amp is very immune as it is to bad wall warts. wall warts that would make a META42 hum are dead quiet on the PPA.

OK that's my 2 cents worth

 

[morsel]

Hi PPL. Nice detailed description of the dynamic performance of capacitance multipliers .vs. voltage regulators.

"I believe that if you are going to spend the time and expense of the DIY thing that cost of parts nor complexity should not enter into the equation."

Except that we really do need to be practical. If we design an amp that is 100% your ideal, but uses parts most people can't obtain, has features no one needs, and costs more than people are willing to pay, not enough people would want it to justify making a large run of boards. If the amp is not popular then the service to the community was not desired, which takes away from the whole point of the gesture. That is also why it is so important to listen to what the community wants. (Listening does not equate to adopting all suggestions.)

Team PPA is not losing direction. Neither do we agree with each other 100% on everything, but that is to be expected. One of the things I am trying to do is change our paradigms from "overengineer first, ask questions later" and "continue evolving indefinitely" to "avoid complexity where possible, test what we have now, fix what is not good enough". PPL and Tangent are busy, so Kurt and I are doing most of the testing.

~

Then we go on to SNEE. And the SNEE is for Sneedle
A terrible kind of ferocious mos-keedle
Whose hum-dinger stinger is sharp as a needle.
The Sneedle's too tough to be killed with a smack
So he has to be hunted on elephant back
And your eyes and the elephant's have to be keen
And you have to aim fast and you have to hit clean
And the bullet you shoot is a stale navy bean
That you've dunked for three weeks in old sour kerosene
Which is awfully hard work. So it's easy to see
Why most people stop at the Z. But not me!

-from On Beyond Zebra by Dr. Seuss

 

[Tomo]

Hey,

Brah Brah Brah ... Team ... Brah Brah ... Now that is done lets talk PSUs.

So PPL, are you saying full-fledge regulators does not have enough bandwidth?

Jung' measurements indicates 317's rejection is flat at -80dB up to 10K and rises to -40dB by 100KHz. 337 only goes up 10 1~2KHz. His regulator scores -100dB well above 10KHz and -90dB at 100KHz. I think it is perfect. Noise level you already know is like 0.1~10uV which is way low.

Anyways, PPL, I would like numbers now. Not that I care less about theory, but if there isn't a real data, it is simply a theory and not a fact.

Lastly, topic slightly shift but I would like to Quote Mr Chester Simpson at the Power Supply Division of National Semiconductor:

Quote:

Remember, the headphone amplifier is fed from a line output which is 775mVrms @ 3dB, which means a few microvoluts of noise are too low to be heard (10uV is -100dB from this level) - "A DC coupled, Selectable Gain Headphone Amplifier" by Chester Simpson at Headwize

More thing to think about. I wanna use a sharp cooking knife to cut onions; I don't want to use a circular saw. There is a difference between overkill and meaninglessly-wasteful.

Tomo

 

[Audio-Me]

morsel,
you've stated before that the mETA42 should be better than some $1000+ amps when configured well, I gather do you think this PPA is going to be better than the gilmore by your standards? I'm interested in tangent's response on this matter as well. If you think it will, I really think you should try making a gilmore to compare it to, otherwise you'll have no ground to base its sound on besides the meta42.

 

[morsel]

I have listened to the Gilmore and think it's great. I don't know how the PPA will compare but I hope it will be at least as good. I believe it is possible, otherwise I would not have started the project. I don't own a Gilmore and can't afford one. Kurt has a Gilmore board which we will be able to compare after he builds it. We will post test results when the tests are done. My guess is the differences will be subtle.

 

[ppl]

Tomo You should Know that Numbers have very litle to Due with performance, this has been demonstrated several times. However i am not going to get into a senceless debate over sound vs performance because this is one of those never ending questions like the tube vs Solid state.

The Limited bandwidth and Poor transient response of the LM-317 is well documented in the data sheet and the nurmerious Application notes.

My intention is to provide Yes Tomo theroy as to why things are the way thay are. I leave it up to you to do your own testing and investigating. I no longer have the time nor the inclination to do sutch testing on a subject I have allready proved to myself manny time over the years and redoing such just to generate spacific numbers. If this is what you need for you to sleep at night then test these things yourself.

As far as Walt's regulator is concerned It is good and alot better than the Typical LM-3XX types. This again illistrates my point that if the typical IC Regulator is good enought for Audio and anything else is overkill as you sugest then why do lots of people go throught the trouble of building one vs the more simple LM-3xx solution? IMHO the difference between so so Audio Equipment and the Truly great Grear is in the details PPM if you wish. As far as overKill is concerned. I like the security Overkill gives me, Kind of like not having to worry if it is going to fail. This same logic i apply to sound quality, I dont want to worry if the Equipment is doing dramatic harm to the Music or not.

If Audio design was so cut and dry as you sugest then there would be no reason for the High end to exsist and the typical $300.00 A/V receiver would sound as good as a equv. power and spec Multikilobuck Audiophile Amplifire.