hey all,
i have a switching PSU in a digi 002 rack recording interface. It's pretty noisy (power wise). I'm trying to decide whether to try to clean it up by some simple method - replacing the caps with better, for example - or to replacing the whole thing with a linear psu.
it's powering only low power stuff - +/- 12v for opamps and a 48v circuit (for phantom mic power), + 5v and + 3.3v for digital and logic components.
the caps (post voltage regulators) are Hermei LE caps (mostly 2200uf and 1000uf) and i thought about swapping those with same value nichicon kz or panasonic FM perhaps.
another thought i had was just to drop a 2200uf cap on each lead (power to ground) to filter there rather than messing with the psu itself. there's already chokes on the board for the 12v, but it's still pretty noisy.
my other option is adding a psu for the analog 12v rails - which i tried unsuccessfully... has a bad ground loop noise and i'm not clever enough to source it. think it needs to be redesigned to isolate the AC side. soooo, thought about just building a whole new linear psu for everything with 3.3v, 5v and the +/-12v lines as well. would rather the simplicity of keeping the switching psu and cleaning it up though.
any advice is appreciated.
thanks,
Adam
i have a switching PSU in a digi 002 rack recording interface. It's pretty noisy (power wise). I'm trying to decide whether to try to clean it up by some simple method - replacing the caps with better, for example - or to replacing the whole thing with a linear psu.
it's powering only low power stuff - +/- 12v for opamps and a 48v circuit (for phantom mic power), + 5v and + 3.3v for digital and logic components.
the caps (post voltage regulators) are Hermei LE caps (mostly 2200uf and 1000uf) and i thought about swapping those with same value nichicon kz or panasonic FM perhaps.
another thought i had was just to drop a 2200uf cap on each lead (power to ground) to filter there rather than messing with the psu itself. there's already chokes on the board for the 12v, but it's still pretty noisy.
my other option is adding a psu for the analog 12v rails - which i tried unsuccessfully... has a bad ground loop noise and i'm not clever enough to source it. think it needs to be redesigned to isolate the AC side. soooo, thought about just building a whole new linear psu for everything with 3.3v, 5v and the +/-12v lines as well. would rather the simplicity of keeping the switching psu and cleaning it up though.
any advice is appreciated.
thanks,
Adam
Quieting a switch mode power supply is much akin to tuning a banjo -- when you throw it in the garbage and it hits the "Ozark Harp" it's tuned.
Same could be said of the accordian.
But seriously -- an LM317/LM337 as a post regulator can rid many nasties but you will burn a couple of watts in the process.
Same could be said of the accordian.
But seriously -- an LM317/LM337 as a post regulator can rid many nasties but you will burn a couple of watts in the process.
Thank you both.
I have to do more research about the LC filters, star882. So many variables that I'm not familiar with (I'm just a hobbyist, not an engineer unforunately).
perhaps the LM3** is an ideal simple solution for me. : )
thanks for the help. may post design ideas once i have some.
cheers,
Adam
I have to do more research about the LC filters, star882. So many variables that I'm not familiar with (I'm just a hobbyist, not an engineer unforunately).
perhaps the LM3** is an ideal simple solution for me. : )
thanks for the help. may post design ideas once i have some.
cheers,
Adam
How bout this?
hey there,
so, after a little more research, i discerned that I didn't know:
A) whether it should be a band pass, band cut, or low pass LC filter
B) what frequency I should be aiming for/against
C) the appropriate values of the components given those things.
however, i did discover that i've made LC filters before, only i called them 'crossover networks' ; ) and used them in speakers.
I also built and use Elso's Kwak-Clock which has a little LC filter in it. It's a 2.2mh inductor with a 2200µf cap. From my research i discerned that it is a low pass PI format taking advantage of the cap at the psu.
My question is this:
without getting too deep into the math, could I just use the same simple circuit and values for my ripple/noise filter?
I know it's a bit of a trick getting an inductor with a high enough amperage rating. my positive 12v supply line (as rated on the psu) is 3.5amps, negative is 0.5amps. But from a little bit of very rough calculation, it seems that my board is only really drawing 1.5amps on the positive rail max.
Other question is this:
saw an article on boosting 12v to 19v in the 'favorites' of psu articles here. thought about boosting the 12v rails to 18v to take full advantage of the opamp headroom (they're all opa2134/4134 or 4228 - rated for +-18v). I'm not sure if i'll run into trouble though starving for amperage once i boost the voltage. any thoughts to point me in the right direction?
(i had built, as i said, a separate psu for the analog rails that was +/-18v that worked EXCEPT it had a bad 60hz hum. because i don't know enough, i haven't figured how to resolve it, so I just temporarily gave that up. point being, the board happily works with the voltage boost).
Thanks for any input.
Adam
ps edit:
thought about, with the voltage boost, doing a simple voltage doubling circuit from 12v to 24v and then regulating back to 18v with the lm317/337 - basically trying to do the boost and clean up in one fell swoop. thoughts?
hey there,
so, after a little more research, i discerned that I didn't know:
A) whether it should be a band pass, band cut, or low pass LC filter
B) what frequency I should be aiming for/against
C) the appropriate values of the components given those things.
however, i did discover that i've made LC filters before, only i called them 'crossover networks' ; ) and used them in speakers.
I also built and use Elso's Kwak-Clock which has a little LC filter in it. It's a 2.2mh inductor with a 2200µf cap. From my research i discerned that it is a low pass PI format taking advantage of the cap at the psu.
My question is this:
without getting too deep into the math, could I just use the same simple circuit and values for my ripple/noise filter?
I know it's a bit of a trick getting an inductor with a high enough amperage rating. my positive 12v supply line (as rated on the psu) is 3.5amps, negative is 0.5amps. But from a little bit of very rough calculation, it seems that my board is only really drawing 1.5amps on the positive rail max.
Other question is this:
saw an article on boosting 12v to 19v in the 'favorites' of psu articles here. thought about boosting the 12v rails to 18v to take full advantage of the opamp headroom (they're all opa2134/4134 or 4228 - rated for +-18v). I'm not sure if i'll run into trouble though starving for amperage once i boost the voltage. any thoughts to point me in the right direction?
(i had built, as i said, a separate psu for the analog rails that was +/-18v that worked EXCEPT it had a bad 60hz hum. because i don't know enough, i haven't figured how to resolve it, so I just temporarily gave that up. point being, the board happily works with the voltage boost).
Thanks for any input.
Adam
ps edit:
thought about, with the voltage boost, doing a simple voltage doubling circuit from 12v to 24v and then regulating back to 18v with the lm317/337 - basically trying to do the boost and clean up in one fell swoop. thoughts?
Last edited:
Hey! I resemble that remark!
Are the 3 footed regs like the LM317 fast enough to clean up the HF hash that most switchers have? I'll have to go back and look at impedance plots.
Was thinkin' they are too slow.
hey gents, this will probably show my lack of knowledge, but...
i was reading the lm317 datasheet and thought, perhaps besides regulating, i could use it to also boost the voltage to 18v since it just runs on the principle of set resistance compared to the 1.25v signal it generates.
My first assumption was that it is like most regulators and needs supply voltage higher than its output voltage, but perhaps this isn't the case.
a word from the wise please?
Adam
i was reading the lm317 datasheet and thought, perhaps besides regulating, i could use it to also boost the voltage to 18v since it just runs on the principle of set resistance compared to the 1.25v signal it generates.
My first assumption was that it is like most regulators and needs supply voltage higher than its output voltage, but perhaps this isn't the case.
a word from the wise please?
Adam
You can get the LM317 to act as a boost regulator -- it's in the data sheet -- but you are back to noise generation. TANSTAAFL as we used to say at the UofC.
I go with the work that WJ and Jan did for Audio Amateur in 1995 --
How about the Wenzel "clean-up" circuit?
The HF noise from SMPS can pass right through 3-pin linear regs, which even a little RC in front can help prevent. Here's a little example I did some time:
Using 3-pin regulators off-piste: part 2
Using 3-pin regulators off-piste: part 2
here's a link to the wenzel circuit... looks like a slightly more complex adaptation of a gyrator circuit (like in Martin's article) to me:
Finesse Voltage Regulator Noise!
also, for other's benefit, a link about gyrator circuit (a possible substitute for a large inductor):
Gyrator - Wikipedia, the free encyclopedia
Finesse Voltage Regulator Noise!
also, for other's benefit, a link about gyrator circuit (a possible substitute for a large inductor):
Gyrator - Wikipedia, the free encyclopedia
Hey guys,
so, after looking into this for a bit, I've come up with a draft plan.
I am going to do a combination of things cause i'm trying not only to clean up the power, but hoping to get from +/-12v rails to +/-18v to get more headroom for the opa2134 opamps.
I've attached a schematic of the first phase, which is an excerpt from the lm 117 datasheet.
First is a 'lower noise' adjustable voltage boost/regulator with an LM350 or 338 for the positive rail because the psu is rated at 3.5amps on the positive rail. I'll also build a parallel circuit out of the negative counter part (the lm 337), which is only 0.5 amp. The datasheet recommends using solid tantalum caps, so i'll probably stick with that. Only change I'll make is to take Martin's advice to put a small choke before the regulator (probably this 10µh from mouser 542-PM124SH-100M-RC) and maybe after?
So, those two will boost my +/-12v rails to +/-18v (should be about that if I set the pot to around 3.3k). Not sure which is the best kind of pot to use for this application though, if you have any thoughts.
Then, I intend to put some sort of low pass filter (around 75 hz) - a pi filter on each line cap-choke-cap, with 2200uf panasonic FM or FC caps and 2.2mh chokes would work, but those chokes are a touch pricey and hard to get given the amperage i'm looking for. Alternately, I could use the gyrator circuit suggested by Martin - the Wenzel one is just a bit involved for me. Another alternate is swiping some chokes from some dead mother boards or old power supplies. : )
Martin (or anyone), could you recommend - for the gyrator circuit - a different NPN than the BC547B for higher amperage? ie 3.5amp or so. Also, would a gyrator circuit for the negative voltage rail look the same or be with a different transistor? I didn't do the frequency calculations (cause i don't know how) but I assume your values would be fine for my purpose: 10k resistor, 3.3µf cap (what kind of cap do you suggest?)
Thank you in advance for any thoughts,
Adam
so, after looking into this for a bit, I've come up with a draft plan.
I am going to do a combination of things cause i'm trying not only to clean up the power, but hoping to get from +/-12v rails to +/-18v to get more headroom for the opa2134 opamps.
I've attached a schematic of the first phase, which is an excerpt from the lm 117 datasheet.
First is a 'lower noise' adjustable voltage boost/regulator with an LM350 or 338 for the positive rail because the psu is rated at 3.5amps on the positive rail. I'll also build a parallel circuit out of the negative counter part (the lm 337), which is only 0.5 amp. The datasheet recommends using solid tantalum caps, so i'll probably stick with that. Only change I'll make is to take Martin's advice to put a small choke before the regulator (probably this 10µh from mouser 542-PM124SH-100M-RC) and maybe after?
So, those two will boost my +/-12v rails to +/-18v (should be about that if I set the pot to around 3.3k). Not sure which is the best kind of pot to use for this application though, if you have any thoughts.
Then, I intend to put some sort of low pass filter (around 75 hz) - a pi filter on each line cap-choke-cap, with 2200uf panasonic FM or FC caps and 2.2mh chokes would work, but those chokes are a touch pricey and hard to get given the amperage i'm looking for. Alternately, I could use the gyrator circuit suggested by Martin - the Wenzel one is just a bit involved for me. Another alternate is swiping some chokes from some dead mother boards or old power supplies. : )
Martin (or anyone), could you recommend - for the gyrator circuit - a different NPN than the BC547B for higher amperage? ie 3.5amp or so. Also, would a gyrator circuit for the negative voltage rail look the same or be with a different transistor? I didn't do the frequency calculations (cause i don't know how) but I assume your values would be fine for my purpose: 10k resistor, 3.3µf cap (what kind of cap do you suggest?)
Thank you in advance for any thoughts,
Adam
AT77 - what i call a 'gyrator' isn't really: its simply an RC filter with a pass transistor (emitter-follower) added
Scaling the idea up or higher currents is easy enough - use a darlington, Sziklai pair or FET for the pass element. Here's a good article by Rod Elliott that outlines some of the considerations:
Capacitance Multiplier Power Supply Filter
Scaling the idea up or higher currents is easy enough - use a darlington, Sziklai pair or FET for the pass element. Here's a good article by Rod Elliott that outlines some of the considerations:
Capacitance Multiplier Power Supply Filter
Hey Martin,
his final design looks cool and to be a fairly easy build - nice. I couldn't gather anything about voltage drop in the circuit as it is (besides changing the value of the 12k resistor). I assume there is less than 0.5v drop if any. Just concerned because my application of it is post regulator for opamps, not in a non-regulated ps for a power amp. Also considering using the same circuit for the 3.3v and 5v rails (digital and logic power).
Aside from that, how about doing your simple circuit but using a KSD1691 for the positive rail and KSB1151 for the negative rail? or TIP110/TIP115?
And perhaps, for greater capacitance, changing the values of the resistor and cap.
his final design looks cool and to be a fairly easy build - nice. I couldn't gather anything about voltage drop in the circuit as it is (besides changing the value of the 12k resistor). I assume there is less than 0.5v drop if any. Just concerned because my application of it is post regulator for opamps, not in a non-regulated ps for a power amp. Also considering using the same circuit for the 3.3v and 5v rails (digital and logic power).
Aside from that, how about doing your simple circuit but using a KSD1691 for the positive rail and KSB1151 for the negative rail? or TIP110/TIP115?
And perhaps, for greater capacitance, changing the values of the resistor and cap.
Last edited:
Ok, this probably has a simple explanation but I can't see it! I'm trying to get about 200v dc for a valve preamp. I have 2 small transformers back to back giving me 177vac. This goes accross a diode rectifier bridge and should give me about 238vdc (177x1.4). So why do I end up with 160vdc. I've never seen this before!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.