Hi,
I am rebuilding my LM317/LM337 +/-15VDC regulators for my line stages.
I used to use PSUII to model PSU. In my previous regulators, after the transformer, I used CRC (3,300uF/35V + 6.8ohms + 4,700uF/25V) to significantly reduce the ripples before feeding into the LM317/LM337. The result looks much better from PSUII and I did not mind the additional costs of the RC.
But from reading all the schematics, I found nobody else did the same thing. Everything I read has only a C after the transformer and before the LM317/LM337 regulators.
So now I am reviewing what I have done.
Adding the RC will most certainly reduce the ripples significantly (would this make the LM317/LM337 to work much happily?) but I am now worrying about increasing the PSU impedance, because at low frequencies the cap would have a higher impedance than the transformer.
So should I ditch the RC in my new work or keep it? If using C only, what value do you find the best? 4,700uF? 8,200uF?
Thanks.
Bill
I am rebuilding my LM317/LM337 +/-15VDC regulators for my line stages.
I used to use PSUII to model PSU. In my previous regulators, after the transformer, I used CRC (3,300uF/35V + 6.8ohms + 4,700uF/25V) to significantly reduce the ripples before feeding into the LM317/LM337. The result looks much better from PSUII and I did not mind the additional costs of the RC.
But from reading all the schematics, I found nobody else did the same thing. Everything I read has only a C after the transformer and before the LM317/LM337 regulators.
So now I am reviewing what I have done.
Adding the RC will most certainly reduce the ripples significantly (would this make the LM317/LM337 to work much happily?) but I am now worrying about increasing the PSU impedance, because at low frequencies the cap would have a higher impedance than the transformer.
So should I ditch the RC in my new work or keep it? If using C only, what value do you find the best? 4,700uF? 8,200uF?
Thanks.
Bill
Hi
I've seen the use of Pi RC filters but with less than 10 ohms due to I losses. Have you thought of cascading 2 linear regs?
1) What are your specs just after the linear reg. ie Vo, Io , Vac output max, noise out? etc
2) how much power loss can you put up with all series elements.
3) standard AC transformer?
4) Do you model esr and esl or your bulk C?
I've seen the use of Pi RC filters but with less than 10 ohms due to I losses. Have you thought of cascading 2 linear regs?
1) What are your specs just after the linear reg. ie Vo, Io , Vac output max, noise out? etc
2) how much power loss can you put up with all series elements.
3) standard AC transformer?
4) Do you model esr and esl or your bulk C?
HiFiNutNut said:
Your approach is sound. The only thing you need to be sure of is that you have enough voltage at the LM317 etc input so that the reg keeps working at max load. You would want at least 3 or 4 V more at the 317 input than its output, at max load. Remember that as load increases, there is more loss across the R in the CRC. For example, with the 6.8 ohms you cite, an additional 1A load gives an additional 6.8V drop between rectifier and 317.
But the idea is good. The R not only lowers ripple but especially lowers the higher harmonics in the rectified waveform resulting from the sharp charging pulses. It is the higher harmonics that are more harmfull to the final reguklated voltage than the basic 100 or 120 Hz ripple.
Jan Didden
Re: Re: Re: Additional RC after C in front of LM317/LM337?
FYI:
http://www.maxim-ic.com/appnotes.cfm/appnote_number/883/
tonyptony said:
FYI:
http://www.maxim-ic.com/appnotes.cfm/appnote_number/883/
Re: Re: Re: Additional RC after C in front of LM317/LM337?
That depends on several factors, but you can generally say that the dual regulator gives double rejection over the full frequency band, whereas the additional CRC or CLC filtering doesn't help at (very) low frequencies. However, the PSRR of the main regulator is normally very high at low frequencies so there is no problem to solve there anyway.
An advantage of the dual reg solution is that it is smaller and cheaper to implement.
It's partly a philosophical thing. People that like active circuits will tend to chose the dual reg, while others that like passives may be inclined to choose the CRC or CLC route. But if you really want to be sure you must do some calculations. The Maxim App Note posted by QSerraTico-Tico (where does anyone get a name like that ;-) ?) is a good introduction.
Jan Didden
tonyptony said:
That depends on several factors, but you can generally say that the dual regulator gives double rejection over the full frequency band, whereas the additional CRC or CLC filtering doesn't help at (very) low frequencies. However, the PSRR of the main regulator is normally very high at low frequencies so there is no problem to solve there anyway.
An advantage of the dual reg solution is that it is smaller and cheaper to implement.
It's partly a philosophical thing. People that like active circuits will tend to chose the dual reg, while others that like passives may be inclined to choose the CRC or CLC route. But if you really want to be sure you must do some calculations. The Maxim App Note posted by QSerraTico-Tico (where does anyone get a name like that ;-) ?) is a good introduction.
Jan Didden
If you can tolerate the additional voltage drop, an RC filter makes good sense on noisy supplies or where the powered stage is sensitive to noise. HF noise passes straight through a 3-pin reg - some quick'n'dirty comparisons here:
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
(Note this was aimed at a slightly different solution - using a capacitance multiplier on a very low-current supply)
http://www.acoustica.org.uk/t/3pin_reg_notes2.html
(Note this was aimed at a slightly different solution - using a capacitance multiplier on a very low-current supply)
Great info, all.
QST_T, Jan, I like the idea of dual regulators as an easy way to get improved full band rejection, but what if one were to go one step further for additional higherF rejection and still use a CRC? Maybe still use one before the first regulator.
How would the use of 4-pole capacitors at the regulator input modulate this approach?
Martin, read through that web page. Interesting. In my case I'm not sure how much of a problem this will be for me, as I am working on a linear reg.
QST_T, Jan, I like the idea of dual regulators as an easy way to get improved full band rejection, but what if one were to go one step further for additional higherF rejection and still use a CRC? Maybe still use one before the first regulator.
How would the use of 4-pole capacitors at the regulator input modulate this approach?
Martin, read through that web page. Interesting. In my case I'm not sure how much of a problem this will be for me, as I am working on a linear reg.
If you are looking to try two regs in series, I recommend trying a 'tracking preregulator'. The first reg after the raw PSU is not set up to maintain a regulated supply which the second reg is fed from, but rather, set-up to maintain a constant voltage drop across the second reg. There's a diagram in the National LM117 datasheet.
It's easier to rig up a 317 if it is physically close to the second reg, but you can do it simply by lifting the 0v end of the (first reg's) lower voltage set resistor and returning it to the output of the second reg. Replacing both the pre-regs voltage set resistors with 1k items will set the voltage across the second reg at 2.5v, quite enough for most purposes. Add the Cadj cap across the lower voltage-set resistor - it still makes an audible difference (i.e this cap is connected between the 1st regs ADJ pin and the OUTPUT of the second reg.)
This gives a quite noticeable (ie audible) improvement over two regs in series. Effectively it relieves the second reg of all line regulation duties - and the second reg benefits from its own (low) output impedance. Try it...
It's easier to rig up a 317 if it is physically close to the second reg, but you can do it simply by lifting the 0v end of the (first reg's) lower voltage set resistor and returning it to the output of the second reg. Replacing both the pre-regs voltage set resistors with 1k items will set the voltage across the second reg at 2.5v, quite enough for most purposes. Add the Cadj cap across the lower voltage-set resistor - it still makes an audible difference (i.e this cap is connected between the 1st regs ADJ pin and the OUTPUT of the second reg.)
This gives a quite noticeable (ie audible) improvement over two regs in series. Effectively it relieves the second reg of all line regulation duties - and the second reg benefits from its own (low) output impedance. Try it...
Problem with Pi RC is size and picking the right caps.
Highest performance PS ripple and noise rejection is obtained by... Bulk C > 3 term reg > discrete reg shunt/series > RLC highly damped LPF > DUT
Bulk C - knocks down fund. ripple
3term reg - cleans up majority of low freq ripple
Discrete reg - provide HF PSRR and cleans up 3 Term reg reference noise
damped LC - placed close to DUT filters out conducted and radiated pickup and further HF noise rejection.
Highest performance PS ripple and noise rejection is obtained by... Bulk C > 3 term reg > discrete reg shunt/series > RLC highly damped LPF > DUT
Bulk C - knocks down fund. ripple
3term reg - cleans up majority of low freq ripple
Discrete reg - provide HF PSRR and cleans up 3 Term reg reference noise
damped LC - placed close to DUT filters out conducted and radiated pickup and further HF noise rejection.
Hi,
if you have sufficient voltage then I would recommend RCRC filtering prior to the regulator.
Next best may be cascaded but with different regulators using pre-reg referenced to second output.
I fear that two cascaded but similar regs may allow the same frequencies to pass through with little attenuation.
if you have sufficient voltage then I would recommend RCRC filtering prior to the regulator.
Next best may be cascaded but with different regulators using pre-reg referenced to second output.
I fear that two cascaded but similar regs may allow the same frequencies to pass through with little attenuation.
Andrew,
Let's say one reg has a input ripple etc rejection characteristic that runs at -40dB from DC to say 5kHz and then drops at 6dB/octave. Putting two in series will give rejection from DC to 5kHz of -80dB and then drop at 12dB/octave. You just add the rejection curves together. Same as you would do with RCR or CLC followed by reg. So there is no risk of any 'funny' frequencies to be let through.
Jan Didden
Let's say one reg has a input ripple etc rejection characteristic that runs at -40dB from DC to say 5kHz and then drops at 6dB/octave. Putting two in series will give rejection from DC to 5kHz of -80dB and then drop at 12dB/octave. You just add the rejection curves together. Same as you would do with RCR or CLC followed by reg. So there is no risk of any 'funny' frequencies to be let through.
Jan Didden
Hi Janneman,
thanks for that clarification.
I have a suspicion for some bits of theory that I don't fully understand (and resonance is one of them) and then I resort to Thomas and want to poke the finger. If my knowledge and resources prevent me from examining the result I'm afraid my suspicion continues until I can find the relevant proof.
Now, going back to -40db and then rising at 6db/oct. What if there was a narrow band that did not follow the curve and had a blip that showed less attenuation of a narrow band of frequencies? If that were the case, is there a chance that both of the cascaded regulators have the same reduced attenuation at the same frequency band?
thanks for that clarification.
I have a suspicion for some bits of theory that I don't fully understand (and resonance is one of them) and then I resort to Thomas and want to poke the finger. If my knowledge and resources prevent me from examining the result I'm afraid my suspicion continues until I can find the relevant proof.
Now, going back to -40db and then rising at 6db/oct. What if there was a narrow band that did not follow the curve and had a blip that showed less attenuation of a narrow band of frequencies? If that were the case, is there a chance that both of the cascaded regulators have the same reduced attenuation at the same frequency band?
The reason that the regs lose attenuation is because the error amps inside are over-compensated for stability. Only moderate open-loop gain, and a low gain 'corner' means diminishing correction available much above 10Khz - but in a monotonic way.
All 3-pin regs behave like this, so not too much point worrying about it. If you need much better regulation or noise control there are other approaches available, but IMO 3-pin regs are pretty amazing in what they doaccomplish for the price!
One last thing - IIRC, most of them also require low source impedance (as in 10ohms or less preferably) for guaranteed stability. This is why datasheets show small input caps. It just maybe something else to consider before going overboard on RC decoupling ahead of the regs though.
All 3-pin regs behave like this, so not too much point worrying about it. If you need much better regulation or noise control there are other approaches available, but IMO 3-pin regs are pretty amazing in what they doaccomplish for the price!
One last thing - IIRC, most of them also require low source impedance (as in 10ohms or less preferably) for guaranteed stability. This is why datasheets show small input caps. It just maybe something else to consider before going overboard on RC decoupling ahead of the regs though.
... and after the regs. Big large caps after the regs are useless (except for the manufacturers ;-) ) because they are in parallel with a sub-omh output Z of the reg, so have no effect to mention.
At higher frequencies, where the reg Zout increases, a foil cap at the reg output may help to limit the Zout rise, BUT that cap should not be too good: it must have a non-zero ERS to prevent resonance with the (inductive) Zout of the reg. That is why on many reg data sheets you find a min ESR for the output capacitors. Many audio people use here very high quality and expensive caps, that cause oscillations or resonances.
Jan Didden
Good point, and there's a nice demonstartion here:
http://www.tnt-audio.com/clinica/regulators2_impedance1_e.html
http://www.tnt-audio.com/clinica/regulators2_impedance1_e.html
Martin, Jan, given that even with cascaded regulators (or with two regs in a tracking pregreg design) the higher frequency rejection is still not as great as with a Jung design, where would be the best spot to implement a CRC or RCRC arrangment to improve the higher range rejection, and still insure lowest output impedance across the widest bandwidth (within the limits of what this sort of thing can get you)?
bypass the resistor from the adjust pin to ground with a 1uF cap. SY pointed out (and I followed up with some pictures) an article written in one of Bob Pease's books that this is most helpful in combatting the random noise in these devices:
here -- I found them -- they are on the borderlineof the "happiness" region of load currents for the LM317:
here -- I found them -- they are on the borderlineof the "happiness" region of load currents for the LM317:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.