The lowest possible, but not at the expense of better overall sound.
Only going by ear for the moment.
What sort of noise floor can your ear discern!!
If your amp speaker combo can generate 105dB SPL at onset of clipping, and you can measure 40dB SPL ambient in your listening room then that is 65dB span at your listening ear.
You could get up from your couch and put your ear close to the speaker - that would gain at least another 10dB range.
Are you aware of what level a 317 can get down to, and what psrr is likely?
If your amp speaker combo can generate 105dB SPL at onset of clipping, and you can measure 40dB SPL ambient in your listening room then that is 65dB span at your listening ear.
You could get up from your couch and put your ear close to the speaker - that would gain at least another 10dB range.
Are you aware of what level a 317 can get down to, and what psrr is likely?
Not sure what you're getting at: I've already mentioned I am going by ear only for now, so what I am doing is trying to improve on a DIY Linear Regulated PSU I've whipped up quite fast from spare parts in my salvage box. I will try several iterations as little experiments and do listening sessions and using the LM317 because that's one thing I had at hand. It's old and it isn't the best regulator either, but I'll be using it as way to see how others improved on their use of it.
You look like you're getting yourself worked up for no good reason and your pissant tone (twice now) isn't appreciated.
YashN please keep your replies civil, good advice was offered and your response was inappropriate.It's OK, it's just the tone was haughty and dismissive without knowing what I am doing here. I should have mentioned it (it was in one or more threads but I can't expect you to have seen it).
Andrew replied properly to me though.
For Regs followed by additional circuitry to reduce both noise floor and ripple, look for Walt Jung's articles as well as an edn article starting with 'simple circuits reduce...'.
Already seriously improved the sound with my Linear Regulated PSU, with an additional circuit, bringing down both noise and impedance figures (simulated, not measured) way, way down yesterday.
It's quite amazing how all regions expand (bass, mid, high) greatly! The soundstage appears much bigger, with a quieter background.
I'm not even sure I did the layout properly with the additions, nor have I spent time further experimenting with the capacitor circuits for the time being.
I actually have planned 3 more levels of enhancements on this circuit that I am looking forward to, but for the time being, focusing on ripple noise and low, flat impedance has been incredibly rewarding.
Peter Gabriel's "Mercy Street" yesterday was super moving. Been listening the whole day as well today.
I am going to enjoy this for a while, spend some time designing the rest and then I'll continue the builds and the listening impressions. There could be some additional work needed for transient response.
Stellar sound, even the battery charger/pack can't beat this LRP currently.
It's quite amazing how all regions expand (bass, mid, high) greatly! The soundstage appears much bigger, with a quieter background.
I'm not even sure I did the layout properly with the additions, nor have I spent time further experimenting with the capacitor circuits for the time being.
I actually have planned 3 more levels of enhancements on this circuit that I am looking forward to, but for the time being, focusing on ripple noise and low, flat impedance has been incredibly rewarding.
Peter Gabriel's "Mercy Street" yesterday was super moving. Been listening the whole day as well today.
I am going to enjoy this for a while, spend some time designing the rest and then I'll continue the builds and the listening impressions. There could be some additional work needed for transient response.
Stellar sound, even the battery charger/pack can't beat this LRP currently.
No worries, in the end, it's very likely I'll jettison it for audiophile uses and relegate it to a lab PSU for simple circuits.
Great learning platform though, considering all its foibles and how many people have written about it.
In fact, I am still learning a lot with just those circuits built around it.
The interesting initial thing I managed to do was listen for differences in SQ with the output capacitors.
I had great results with an Elna, a Rubycon and really small Vishay metal film caps at the output.
And that was before seeing the Dietz graph... Much fun.
I wonder what the graph represents. Is it the distortion on the rail voltage caused by a 1k sine current draw?
Ah that would make more sense 
Could also be a square wave. I know when I ran a 1Khz squarewave through the DC-B1 that my LM317 based PS was powering that the scope trace on the PS rails post reg was definitely ugly. I don't remember if I tried a sine wave as well. It has me thinking I need to pull out the test gear!! I kinda went silent when I decided I needed the PS to actually power something that reproduces music, and further testing went out the window when I didn't encounter anything I didn't like
Tony.
Could also be a square wave. I know when I ran a 1Khz squarewave through the DC-B1 that my LM317 based PS was powering that the scope trace on the PS rails post reg was definitely ugly. I don't remember if I tried a sine wave as well. It has me thinking I need to pull out the test gear!! I kinda went silent when I decided I needed the PS to actually power something that reproduces music, and further testing went out the window when I didn't encounter anything I didn't like Tony.
Sine wave on the positive supply rail from the AP SYS2722 -- you can see that the LM317 error amplifier is inadequate. Discrete regulators like the Jung, Sjostrom and Sulzer perform much better.
LM317 fared poorly in the regulator bake-off which was written about in Linear Audio -- it made a hash of percussive instruments.
LM317 fared poorly in the regulator bake-off which was written about in Linear Audio -- it made a hash of percussive instruments.
YES! I remember, ok, I'm a bit of an idiot here for forgetting...
Because I'm the guy who read your Linear Audio article on regulators and sent Jan an email saying "hey, why don't you measure THD on the output of your regulators?" and then he told you and you did measure, and the results were damn interesting.
FYI I tested plenty of LDOs and they all have more or less the same behaviour, ie, output impedance decreases with increasing load current, which means high THD on the output.
This is logical, as pass device Gm increases with current. Loop gain is enough to control output impedance, but they're not designed for low THD.
If it supplies anything which draws a distorted current anyway, we don't care. But if it's a class-A stage which draws a current which is an exact replica of the signal, we might as well not have THD on the rails.
Bonus question:
- Load current is a sine
- Regulator output voltage is a sine with harmonics due to regulator THD
- Since this voltage appears across the output capacitor, it means the harmonics create a current which flow into the output cap
- This current also flows into GND and adds harmonics to GND
- Level is probably harmless... should check this.
Because I'm the guy who read your Linear Audio article on regulators and sent Jan an email saying "hey, why don't you measure THD on the output of your regulators?" and then he told you and you did measure, and the results were damn interesting.
FYI I tested plenty of LDOs and they all have more or less the same behaviour, ie, output impedance decreases with increasing load current, which means high THD on the output.
This is logical, as pass device Gm increases with current. Loop gain is enough to control output impedance, but they're not designed for low THD.
If it supplies anything which draws a distorted current anyway, we don't care. But if it's a class-A stage which draws a current which is an exact replica of the signal, we might as well not have THD on the rails.
Bonus question:
- Load current is a sine
- Regulator output voltage is a sine with harmonics due to regulator THD
- Since this voltage appears across the output capacitor, it means the harmonics create a current which flow into the output cap
- This current also flows into GND and adds harmonics to GND
- Level is probably harmless... should check this.
I have a hunch that I might well care - opamps are classAB so fit into this category. And potentially the poor SQ that jackinnj noticed for the LM317 might be down to this. Stages which draw distorted current put lots of distortion on the rail, the amount depending on the output impedance of the reg + associated caps. It seems very likely to me that a high THD regulator also has high IMD and so possibly adds additional HF noise to the rail.
I've only purchased LA 0 through 3 I'll have to get number 4 to check out that article!! (got slack and wasn't reading them so stopped).....
I assume that the injection of the sine wave onto the rail, is to simulate PS modulation by the load. I'm not 100% convinced as yet of the significance of this test. I would have thought that feeding the actual device (which uses the PS) a full magnitude sinewave and then doing the fft on the rails would be more telling...
I also note in the link above to jackinnj's other post that the fundamental on the super-reg is lower than it is on the LM317... It is not clear if this is because the LM317 is amplifying the fundamental, as well as producing harmonics, or whether the fundamental supplied to the LM317 was higher than the one supplied to the super-reg...
In the end, yes the super-reg handles this test better, but is it relevant?
Tony.
I assume that the injection of the sine wave onto the rail, is to simulate PS modulation by the load. I'm not 100% convinced as yet of the significance of this test. I would have thought that feeding the actual device (which uses the PS) a full magnitude sinewave and then doing the fft on the rails would be more telling...
I also note in the link above to jackinnj's other post that the fundamental on the super-reg is lower than it is on the LM317... It is not clear if this is because the LM317 is amplifying the fundamental, as well as producing harmonics, or whether the fundamental supplied to the LM317 was higher than the one supplied to the super-reg...
In the end, yes the super-reg handles this test better, but is it relevant?
Tony.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.