Well,
How about a UPS battery monitoring system. One board running off 4 batteries in series... about 54 Volts. Superimposed on this 54 Vdc was 150-250 Volt spikes. To get this board to measure accurately we had to keep board noise well below 200 uV. This board had 2 stage regulation... TL783 followed by LM2951's... and all the lytic, ceramic, and film you could beg for. All the capacitance in world hardly do jack. The ultimate cure was to gather 32 samples... discard the upper and lower 12, then average the remaining 8. Audio? no, but neither was the question.
Now... line spikes... you get somewhere around 1000, 1000 Volt spikes per day... depending on where you are. I would give you the link but I can't find it, this was a study done by Harris before they got sucked up... at a time when they were hawking their MOVs. Now, being an ironhead... I didn't believe this garbage. Not having a scope with a million septabytes of storage, I built a simple fast comparator that would latch at one 1000 Volts. You couldn't walk away for more than a minute or two before the LED would light. And yes... if you unplugged the source it would quit latching.
My point being... is 2 reg's better... yes. Is it necessary... I don't know... it depends. I'd rather see a guy, who probably doesn't have the big measuring toys, to double regulate, rather than buy silly power cords or Black Mamba Heaven's Gate caps and all that nonsense.
Oh, and I wasn't talking about preregulators. I was talking about C(L||L)C-C filters. They just block line line hash... yes the diodes make more hash... but that's under your control.
And yes... let's not be silly... I know you're a common sense guy... so am I.
How about a UPS battery monitoring system. One board running off 4 batteries in series... about 54 Volts. Superimposed on this 54 Vdc was 150-250 Volt spikes. To get this board to measure accurately we had to keep board noise well below 200 uV. This board had 2 stage regulation... TL783 followed by LM2951's... and all the lytic, ceramic, and film you could beg for. All the capacitance in world hardly do jack. The ultimate cure was to gather 32 samples... discard the upper and lower 12, then average the remaining 8. Audio? no, but neither was the question.
Now... line spikes... you get somewhere around 1000, 1000 Volt spikes per day... depending on where you are. I would give you the link but I can't find it, this was a study done by Harris before they got sucked up... at a time when they were hawking their MOVs. Now, being an ironhead... I didn't believe this garbage. Not having a scope with a million septabytes of storage, I built a simple fast comparator that would latch at one 1000 Volts. You couldn't walk away for more than a minute or two before the LED would light. And yes... if you unplugged the source it would quit latching.
My point being... is 2 reg's better... yes. Is it necessary... I don't know... it depends. I'd rather see a guy, who probably doesn't have the big measuring toys, to double regulate, rather than buy silly power cords or Black Mamba Heaven's Gate caps and all that nonsense.
Oh, and I wasn't talking about preregulators. I was talking about C(L||L)C-C filters. They just block line line hash... yes the diodes make more hash... but that's under your control.
And yes... let's not be silly... I know you're a common sense guy... so am I.
For the amount of money that can/would be spent on using two 3 terminal devices in series along with all the C to go with it. A decent regulator can be built that will probably out perform the previous mentioned.
I'm not knocking using 2 devices in series, there are many benifits that are known.
I'm not knocking using 2 devices in series, there are many benifits that are known.
I want to resurrect this:
I like batteries. I like jung regulators. But I also like three pin regs. I
78xx
Now, is there happy medium? I don't want to go to the trouble of powering all my line level gear from batteries or superregulators, say nothing of massive capacitors and inductors. In what cases will using two 78/79xx type regulators in series drop noise appreciably? I suppose I would have to drop a few volts between them, like using a 7815 and then a 7812, wouldn't I? In diy audio gear, are there places where this won't accomplish anything? (how about those very high frequency spikes from smps) I've seen plenty of measured noise graphs for different regulators, but never for a pair of vregs in series - are there any?
Also, a second question: Up this thread, and the blowtorch thread, people were talking about mixing line and load regulators. Does anyone want to walk me through the difference between line and load regulation? Does a line regulator cut noise? And a load regulator keep a steady current? I've seen the schematics, I just don't know how to apply them intelligently.
I like batteries. I like jung regulators. But I also like three pin regs. I
78xxNow, is there happy medium? I don't want to go to the trouble of powering all my line level gear from batteries or superregulators, say nothing of massive capacitors and inductors. In what cases will using two 78/79xx type regulators in series drop noise appreciably? I suppose I would have to drop a few volts between them, like using a 7815 and then a 7812, wouldn't I? In diy audio gear, are there places where this won't accomplish anything? (how about those very high frequency spikes from smps) I've seen plenty of measured noise graphs for different regulators, but never for a pair of vregs in series - are there any?
Also, a second question: Up this thread, and the blowtorch thread, people were talking about mixing line and load regulators. Does anyone want to walk me through the difference between line and load regulation? Does a line regulator cut noise? And a load regulator keep a steady current? I've seen the schematics, I just don't know how to apply them intelligently.
For your reading pleasure:
50kHz spikes getting through a three pin regulator from a smps:
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
some nice noise measurements:
http://www.tnt-audio.com/clinica/regulators_noise2_e.html
50kHz spikes getting through a three pin regulator from a smps:
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
some nice noise measurements:
http://www.tnt-audio.com/clinica/regulators_noise2_e.html
Hi cuibono,
In my experience, line operated supplies are as quiet as battery operated with regulators. Line operated supplies can be less noisy than straight batteries.
The best supplies I have seen do not use massive capacitors, more often it is the opposite. Resistive input before the first cap. R-C-R-C style followed with whatever regulator system you want to use, then possibly filtered again. The nice thing about this scheme is that it attenuates the high frequency noise and hash before it gets into your regulator. The diode peak currents are greatly reduced so that you don't generate as much noise to begin with.
This does not work for supplies that deliver great changes in currents, but for small signal equipment it works very well. This is also a good supply for the front end of an amplifier.
-Chris
In my experience, line operated supplies are as quiet as battery operated with regulators. Line operated supplies can be less noisy than straight batteries.
The best supplies I have seen do not use massive capacitors, more often it is the opposite. Resistive input before the first cap. R-C-R-C style followed with whatever regulator system you want to use, then possibly filtered again. The nice thing about this scheme is that it attenuates the high frequency noise and hash before it gets into your regulator. The diode peak currents are greatly reduced so that you don't generate as much noise to begin with.
This does not work for supplies that deliver great changes in currents, but for small signal equipment it works very well. This is also a good supply for the front end of an amplifier.
-Chris
cuibono said:I want to resurrect this:
I like batteries. I like jung regulators. But I also like three pin regs. I
Now, is there happy medium? .
A "cleanup" circuit after the regulator will get the noise down into the tens of nanovolts per root Hertz region. Full credit to Wenzel Associates for this fine article:
http://www.wenzel.com/documents/finesse.html
An externally hosted image should be here but it was not working when we last tested it.
jackinnj:
That circuit above is almost chip independant and will work as well with '317 variables and any and all '78xx 3-pin regs as long as the power required is low ... (adding a 100 nF plastic cap to the regulated shunt output (and ground), close coupled, very near the "device to be powered", will darn near eliminate any and all popcorn noise and stray EMF.)
Good stuff for the output op-amps of a DAC, etc. ... Good show
That circuit above is almost chip independant and will work as well with '317 variables and any and all '78xx 3-pin regs as long as the power required is low ... (adding a 100 nF plastic cap to the regulated shunt output (and ground), close coupled, very near the "device to be powered", will darn near eliminate any and all popcorn noise and stray EMF.)
Good stuff for the output op-amps of a DAC, etc. ... Good show
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.