| cjd |
I've gotten to that point where I realize I only understand enough to know that this is not something I'm going to be able to figure out on my own - I've spent the last few weeks digging through information and reading.
I would like to try regulated power on the SuperSymmetry chip-amp I'm working up. I'm using two LM4780 chips per channel which should have a 10-15A capable supply. These will be monoblocks.
Worst case, I'll build it without regulation. :) However, I would very much like to try a regulated supply.
The LM338 datasheet has a couple outlined, but they use obsolete chips in the circuit or are not recommended for more than 25V at best (though, I am not entirely clear where that voltage limit is being derived).
One of the things that has crossed my mind would be to supply each chip with its own regulator and not have a common power source (though I would still use one transformer and one rectifier bridge, and thus one capacitor bank behind the regulators). But I don't know if this is a good idea.
Other ideas and options are welcome. :)
Thanks!
C |
|
|
| neutron7 |
You can parrallel LT1083s without using an opamp, just 2 feet of 18g wire as ballast on each one (how do you make that look nice, i do not know :) )
http://www.linear.com/pc/downloadDo...055,P1281,D3741
on page 12
that resistor with the arrow pointing to the chips output means that the resistor must be physically connected as close as possible to that pin.
that chip is a bit harder to find and more expensive than a 338. |
|
|
| slackman |
| Because it's easyer to get the LM338 regulator chip, here's the schematic of paralleling three items to get a 15A voltage supply from the original datasheet :LM138/338 Datasheet |
|
|
| h_andree |
Are you bridging the chips?
I'm having the same dilemma. If you are bridging
you should be able to use one set of lm338 per chip.
If you are paralleling them then the V of the power supplies should
be close to each other.
I'm planning to do this by using 0.1% resistor and using
a trim pot to get the voltages within a few mV.
I don't know if this works but I don't see why not.
Harry |
|
|
| jackinnj |
if you must, use an external pass transistor with the LM338 (National has the details on their site) -- much less expensive than paralleling multiple discrete regulators -- you can also use a much less expensive regulator -- like an LM317 with an external pass device -- the approximate current that the "regulator" will have to bear is just the base drive (something approximating the output current by the beta of the pass transistor). the LM317 is less expensive than the LM338 and may be less noisy (although each device can be "cleaned up" if you read the application note by Wenzel Associates.)
be advised that you are adding another feedback loop into your circuitry when you "regulate" -- "not necessarily a good thing" to paraphrase Martha Stewart with her orange jumper and ankle bracelet. |
|
|
| janneman |
| quote: | Originally posted by slackman
Because it's easyer to get the LM338 regulator chip, here's the schematic of paralleling three items to get a 15A voltage supply from the original datasheet :LM138/338 Datasheet |
Do you really want 15A capacity? Remember that an unregulated supply has basically zero Amps capacity except for the capacitors at the supply output. The amp runs on those caps except for the 20% of the time that the diodes conduct to recharge the caps and supply the amp if necessary.
So a very efficient regulated setup would be say one 338 with 5A capacity with say 10.000 uF of capacitance at the output. The amp will run 95% of the time on the regulated supply, and when you really need very high output pulses for bass drums etc the caps will provide that. A full 15A or more supply is a prime example of overkill.
Jan Didden |
|
|
| carlosfm |
| quote: | Originally posted by janneman
So a very efficient regulated setup would be say one 338 with 5A capacity with say 10.000 uF of capacitance at the output. |
Jan, I'm using 2,200uf on the LM3886 chips (after the LM338 regs).
This works very well.
But 10,000uf can kill a regulator at power-on, no?
Anyway, I think that the regs are fast enough to keep up with the demands, and 2,200uf seems ok.
They need some capacitance before them and around 10V difference between input and output so that they can regulate in the most demanding conditions, even accounting with some voltage sag (in this case they can give 12A peak current).
cjd, the LM338 regs can regulate at much higher voltages.
The limitation is the input to output voltage differential, not the regulated voltage.
PS: I would not parallel regulators. Better use a pass transistor. |
|
|
| cjd |
| quote: | Originally posted by carlosfm
cjd, the LM338 regs can regulate at much higher voltages.
The limitation is the input to output voltage differential, not the regulated voltage.
PS: I would not parallel regulators. Better use a pass transistor. |
I knew the 338's could regulate high voltages - the one example from the data-sheet that was posted here suggests 25V max, which I assumed was not due to the 338 but something else in the circuit. But, being unsure of WHAT is causing the 25V limit (if it's even a fair limit), and not really enjoying the thought of blowing something up in discovery, I decided it was better to admit my lack of knowledge and ask. :)
Pass transistor gives me something to look into. A couple of you have mentioned this and it seems to make sense if I understand the concept correctly. Now I just get to figure out how to actually implement it. :)
janneman: I don't need 15A probably, though it never hurts to have more capacity than needed. 5A seems to me to be too little though, even though I may never use it. The fact that the chips *could* draw more than 5A though...
C |
|
|
| carlosfm |
| quote: | Originally posted by cjd
5A seems to me to be too little though, even though I may never use it. The fact that the chips *could* draw more than 5A though...
C |
5A continuous, 12A peak "for short periods of time", 7A "guaranteed peak output current".
in
LM138/LM338 datasheet
National Semiconductor
May 1998
:D |
|
|
| cjd |
Is 5A/7A/12A going to never have problems supplying power to two LM4780 chips?
(do you now have to ask "how loud do you play"? :P )
I'm actually having trouble finding documentation that explains anything enough (for me, given my level of understanding) regarding an external pass transistor. I can probably put the basic circuit together, just can't tell what parts go where. As if that's importnant. :clown:
C |
|
|
| carlosfm |
| quote: | Originally posted by cjd
Is 5A/7A/12A going to never have problems supplying power to two LM4780 chips? |
No, I can't recommend this to you.
You need another solution than a single pair of LM338 regs.
If I was you I would go unregulated, high capacitance.
Snubberized.:att'n: |
|
|
| cjd |
| quote: | Originally posted by carlosfm
No, I can't recommend this to you.
You need another solution than a single pair of LM338 regs.
If I was you I would go unregulated, high capacitance.
Snubberized.:att'n: |
That's my fall-back solution. :) This post was exploring the regulated option.
C |
|
|
| janneman |
| quote: | Originally posted by cjd
[snip]janneman: I don't need 15A probably, though it never hurts to have more capacity than needed. 5A seems to me to be too little though, even though I may never use it. The fact that the chips *could* draw more than 5A though...
C |
We are powering an amp, not an industrial welder! Music has an average to peak ratio of, what 10%? 2%? The vast majority of the time you are listening to just a few watts.
The 338 seems to be able to supply up to 7 amps short term. Assuming dual supplies, that's 5A RMS. Assuming, again, 4 ohms load, that means that a simple 338 without anything else can supply this amp up to a level of 100 watts. ONLY with peak loudness above 100W equivalent, the capacitors would need to come in to help the reg. I would be very surprised if that is 1 sec per day. If your amp is able to output 100W or more, which it probably isn't.
And yes, it can hurt to have more then necessary. Paralleling 338's or using pass transistors will degrade the performance of the total reg package in terms of either dynamic Zout or stability or both. And that will influence the sound 100% of the time.
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by carlosfm
Jan, I'm using 2,200uf on the LM3886 chips (after the LM338 regs).
This works very well.
But 10,000uf can kill a regulator at power-on, no?
[snip] |
Yes, I would think 2200uF after the reg would work OK, 10.000uF or more seems overkill. I don't know if it kills the reg, aren't these current limit protected?
Jan Didden |
|
|
| Elso Kwak |
In my experience a regulated supply for a poweramp kills the bass slam. I prefer a Pi-filter for lowering the ripple.
For those who believe in beefing up caps: you can add caps at will after the inductor!:bigeyes: |
|
|
| carlosfm |
| quote: | Originally posted by Elso Kwak
In my experience a regulated supply for a poweramp kills the bass slam. |
Elso, that is your experience with your amps and PSUs.
In my experience with regulated PSU and these chips, it's the most impressive bass slam I have heard out of them.
But notice that I use 2,200uf caps after the regs, and the PSU is not so conventional.
It is fully snubberized, which makes a whole of a difference.:yikes:
:µphone: :snare:
:sing: |
|
|
| Elso Kwak |
| quote: | Originally posted by carlosfm
Elso, that is your experience with your amps and PSUs.
In my experience with regulated PSU and these chips, it's the most impressive bass slam I have heard out of them.
But notice that I use 2,200uf caps after the regs, and the PSU is not so conventional.
It is fully snubberized, which makes a whole of a difference.:yikes:
:µphone: :snare:
:sing: |
Please don't snubber to me!
:clown: |
|
|
| Lars Clausen |
cjd: Do you need a single
V+ GND supply or do you need a
V+ GND V- supply?
Lars |
|
|
| carlosfm |
| quote: | Originally posted by Elso Kwak
Please don't snubber to me!
:clown: |
I couldn't hurt that cute banana.:D |
|
|
| Upupa Epops |
| Look at 338 datasheet, there is parallel connection of three regulators, which give 15 A. Or use power transistor. All is easy, you must read datasheets and not only ask friends here :cool: . |
|
|
| janneman |
| quote: | Originally posted by carlosfm
Elso, that is your experience with your amps and PSUs.
In my experience with regulated PSU and these chips, it's the most impressive bass slam I have heard out of them.
But notice that I use 2,200uf caps after the regs, and the PSU is not so conventional.
It is fully snubberized, which makes a whole of a difference.:yikes:
:µphone: :snare:
:sing: |
... but it is still YOUR experience, and as such has exactly the same value (if any) as Elso's. Let's keep this straight.
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by Upupa Epops
Look at 338 datasheet, there is parallel connection of three regulators, which give 15 A. Or use power transistor. All is easy, you must read datasheets and not only ask friends here :cool: . |
Data sheet applications contain suggestions for possible use with the general aim to sell as many as possible. Where the hell does this 15A requirement come from?? Don't tell me you guys spec you power supply according to the max current the chip can deliver before selfdestruct? What nonsense is that?
Jan Didden |
|
|
| carlosfm |
| quote: | Originally posted by janneman
... but it is still YOUR experience, and as such has exactly the same value (if any) as Elso's. Let's keep this straight.
Jan Didden |
Jan,
I suppose we are talking about chip amps here.
And this PSU.
One member has tried this complete PSU as I recommend and he is very impressed.
If that's worth something for you, of course.
http://www.diyaudio.com/forums/show...1625#post631625
PS: good to see you back on form.:) |
|
|
| janneman |
| quote: | Originally posted by carlosfm
[snip]One member has tried this complete PSU as I recommend and he is very impressed.
If that's worth something for you, of course.
[snip] |
No it isn't. But you knew that already, no?;)
Edit: Carlos, I'm not poking at you, honestly. But you say one thing, Elso another. Just give me one reason why I should believe you instead of Elso. Or maybe believe Elso instaed of you. Ok, you come up with someone who agrees with you. Is that it? 2 to 1, so you're right? There must be something more!
Jan Didden |
|
|
| cjd |
janneman:
Unless I'm totally out of it, two LM4780's per amp on a 4ohm nominal speaker could deliver over 100W. ;) Theoretically, I think over 200W? Though not probable all said and done. Again, thats not saying I'm ever going to actually demand that much, but it could come close depending on the configuration I end up with speaker-wise. We're talking a somewhat high sensitivity dipole, and there can be a lot of power demands by the woofers in this case. Higher sensitivity should mitigate power needs, but I may end up having to give up on that particular design goal to keep the size of the final project within the realm of what will fit in my house reasonably! This is a pair of LM4780's being used in the SyperSymmetry layout proposed by Nelson Pass. 4780's because low impedance is entirely probable, and a simple pair of 3886's may run into trouble.
oh yeah - on "who to believe" - if it's a simple circuit, why not try it? ;)
Upupa:
If you read my initial post, you'll note that I *have* read the datasheet - many times. The layouts they propose use obsolete chips OR are suggesting a max voltage a bit less than I would like (though not so much so that I couldn't go that route if I felt it was the best choice). Besides that, datasheets tend to give you circuits that work, not circuits that are ideal or the best. The conversation generated here has been quite clear (for the most part) on discussing the merits of regulation, and whether I should bother trying it. :)
At any rate, thank you all for the help and comments so far!
C |
|
|
| janneman |
cjd,
I can follow your post and I am sure you are well aware of all the issues; in the end you have to make the decision. My numbers were just examples and it may well be that your numbers change the picture as far as max current requirements. Maybe 15A IS required for 2 channels low impedance bridge etc. But continuously, DC, at 100% duty cycle?? Even in that case, you probably are comfortable with just a few amps capability (DC) plus a couple of capacitors. That was my point, as you know.
Good luck wading through all this and to get to some decision!
Jan Didden |
|
|
| carlosfm |
| quote: | Originally posted by janneman
Edit: Carlos, I'm not poking at you, honestly. But you say one thing, Elso another. Just give me one reason why I should believe you instead of Elso. Or maybe believe Elso instaed of you. Ok, you come up with someone who agrees with you. Is that it? 2 to 1, so you're right? There must be something more!
Jan Didden |
Something more you can ask everybody on this forum.
Seriously, Elso has good intentions and I understand his point of view.
But as I said millions of times here, high capacitance doesn't sound good with these amps, unless it's snubberized.
So, Elso´s choke recommendation will not work well. He also refuses to snubberize, and doesn't have experience with these amps, so who to believe is everyone's choice.
If there's a will to go regulated again, I'm guilty and I'm the one to shoot.
So shoot me.:D
I've been guilty of plenty of things happening here in the last years, so hang me.:cheerful:
PS: ask those (too many to count now) who tried the snubberized PSU if I'm crazy, or if that is a REAL improvement.
I don't play with gold-plated resistors and their polarity. |
|
|
| Upupa Epops |
| 15 A regulator : NS LM 138 , page 14 :cool: . |
|
|
| janneman |
| quote: | Originally posted by carlosfm
[snip]But as I said millions of times here, high capacitance doesn't sound good with these amps, unless it's snubberized.
[snip] |
... and your point? Maybe you start to find out that the value of a statement doesn't lie in the number of repetitions?
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by Upupa Epops
15 A regulator : NS LM 138 , page 14 :cool: . |
up,
That isn't the point. You pay me enough money, I'll design a 50 amps +/- 50V supply for you. The point is that your music system needs this only 1 sec in a forthnight. All the other time it is idling at less than an amp. So my point was, take a standard 5 amps supply and hang a couple of caps on it, and be done with it.
Jan Didden |
|
|
| carlosfm |
| quote: | Originally posted by cjd
This is a pair of LM4780's being used in the SyperSymmetry layout proposed by Nelson Pass. 4780's because low impedance is entirely probable, and a simple pair of 3886's may run into trouble. |
So, it's one LM4780 per channel, right?
Then a complete LM338 PSU per channel will do.
http://www.diyaudio.com/forums/show...6970#post636970
No need to parallel regs, no need for pass transistors.
Just independent rectifiers/caps/regs per channel. |
|
|
| carlosfm |
| quote: | Originally posted by janneman
... and your point? Maybe you start to find out that the value of a statement doesn't lie in the number of repetitions?
Jan Didden |
Forget it, Jan.
I don't need to say anything.
Just hundreds (thousands?) of people around the globe to testify.
And I don't make one penny out of this.
So why bother?:rolleyes:
This has already gave me too much work just trying to explain, and it's not worth it.
While you guys talk, I DO, and there's a time gap between us.
We are alrealdy discussing a PSU that, to my knowledge, only one person besides me have tried it.
But there's too many people lurking here, so you never know.:cool: |
|
|
| PKS |
Enybody have a good tip for were i can buy a good transformator dual 24V? I need one for an psu to an sub-amp.
Thanks-pk! |
|
|
| Upupa Epops |
| To Jan : I agree with you, one regulator + batery of caps can make the same result. But probably better result will be with discrete regulator without feedback, LM ... can oscilate in this case. This oscilations probably are suppresed by " Carlos's snubber " ( it is the same case as " Boucherot " at output of PA, not the same case as " Carlos's snubber " connected beside rectifier ). |
|
|
| janneman |
ue,
There are many ways to skin a cat, but the disadvantage of a discrete reg without feedback for me is that it really isn't a regulator, it's a stabilizer or super-power-zener or whatever. So, you put in all those power parts, but stop at the regulation. It is surely better than just bridge & caps, but I would just close the regulation loop. Stability can be done by the means you say or any other accepted compensation method. This is not different from the compensation of the amp itself.
Much also depends on the amp, its PSRR and other things. It is difficult to give a standard solution without knowing where it will be used.
But one thing is sure: amps are designed with the assumption that the supply is stable and for AC is the same as ground. If that is not the case (and it never is), the supply will influence the sound, in principle.
Jan Didden |
|
|
| Upupa Epops |
| I agree again ;) . |
|
|
| cjd |
carlos: Two LM4780's per channel. Four for a two-channel amp. Hence my desire to have a little more juice available than a single LM338 provides. I can always just go high-cap snubbered or whatever.
Upupa: I'm aware of that one. :) In fact, it's the one I was alluding to when I stated in my first post I haven't figured out where they get their listed voltage range (4.5V-25V) to know if my target 30V was do-able or not. 25V would not be a problem regardless. It's on the table.
C |
|
|
| mauropenasa |
The audio sector is the only ( in electronics ) when select technical are bound by the "opinions".
Here is am touched 3 diverge problems: The power supply dimensioning, the correct value of accumulated energy for audio amplifiers and the differences among a stabilized and non stabilizeded PSU.
How to in all the projects, be necessary tackles a problem for time;
The power supply dimensioning: if not is regulated, is chosen the proper diodes bridge and a battery of capacitors of "levelling". The diodes bridge has to be able to manage the " charge current peaks" of the capacitors and this is the only reason for which are used very strong seas. The middle current (on a diodes) of management of an amplifier in "real"conditions are too low of that theory ( after 2 to 30 % to second of the musical son-in-laws ).
Accumulated energy: Is diffused opinion that a widens has to dispose of a good energetic reserve (joule) from "Discharge" on the load when maids. Affair remembers that the capacitors of levelling works of continue to 100Hz with proportional current goes the instant discharges on the load, and not all the "theory" energy is in hand. To resolve this technical problem are used as the LC circuits and the stabilization. When are used active circuits of stabilization is has to hold account that to work has to be a voltage loss ( 10-30 % ) and that the regulator, for his nature, doesn't dispose of the speed necessary to current "follower" with the load in "real time", and a "cache" energetics reserves is that relative to the capacitors on output.
P.S. Using the protection diodes on LM338 and the slow start-up circuit, are able use cap. in exit => 10.000 UF without problems
Mauro |
|
|
|
