^ Dang... you have a knack for this! As a matter of fact, I haven't installed the snubber network on the LT4320 boards yet.
Once I get it fully up and running with the snubbers installed (already did the Quasimodo measurements), I'll measure again.
Edited to add - Just in case also... I read earlier in the thread that the 0R16 series resistance of the 159ZJ was enough to quell the ringing of the L in the CLCRC filter... just in case I misinterpreted your meaning. Still learning.
Once I get it fully up and running with the snubbers installed (already did the Quasimodo measurements), I'll measure again.
Edited to add - Just in case also... I read earlier in the thread that the 0R16 series resistance of the 159ZJ was enough to quell the ringing of the L in the CLCRC filter... just in case I misinterpreted your meaning. Still learning.
load - best way to load it as it is going to be in amp itself
so, speaking of one rail
if Iq is going to be (for instance) 1A8, with "nominally" (for instance) 24Vdc , rough calc is R=U/R
so, R=24V/1A8=13R3
now, we all know that Vdc will going to be lower from "nominal" (resulting from our golden standard 18Vac secs), somewhere around 22V5dc
thus slightly corrected R value should be 22V5/1A8= 12R5
heat in resistor is going to be P=I^2*R=1A8^2*12R5= 40.5W
that calls for at least 100W resistor(s group), well heatsinked
one can cheat, if situation calls for that/being more convenient with resistors at hand - doubling resistance if having bipolar supply, and connecting it between positive and negative rail
of course wattage of resistor(s group) must be doubled
so, speaking of one rail
if Iq is going to be (for instance) 1A8, with "nominally" (for instance) 24Vdc , rough calc is R=U/R
so, R=24V/1A8=13R3
now, we all know that Vdc will going to be lower from "nominal" (resulting from our golden standard 18Vac secs), somewhere around 22V5dc
thus slightly corrected R value should be 22V5/1A8= 12R5
heat in resistor is going to be P=I^2*R=1A8^2*12R5= 40.5W
that calls for at least 100W resistor(s group), well heatsinked
one can cheat, if situation calls for that/being more convenient with resistors at hand - doubling resistance if having bipolar supply, and connecting it between positive and negative rail
of course wattage of resistor(s group) must be doubled
If you find that you have some ringing with your CLC, you can try a resistor in parallel with the L, try a resistance value of 10 to 20 times the Rdc of the inductor. Keeps the supply impedance low also. A CLCC can also help, rather than just CLC.
@Zen Mod - It's good to know that even a blind dodo finds a nut every now and again... 
I am testing and setting up the amp that the PSU will initially be paired with. So, I set the initial Iq to ~2A per channel for the amp. The loaded supply was very close to +- 32V.
Backing that out... my method / madness for PSU testing alone was ... that 8ohms would be ~128W... too hot for even my 250W rated dummies. 16 ohm would still be likely pretty toasty at 64W... So, I tripled them up. 43W for a quick test may not be equivalent (1A333) to the realistic load (~2A2 will be final target) in the amp, but I thought it was OK for a fun check. Still pretty revealing (and fun). I wasn't willing to lug the amp around too... Lazy dodo.
Thank you!
I am testing and setting up the amp that the PSU will initially be paired with. So, I set the initial Iq to ~2A per channel for the amp. The loaded supply was very close to +- 32V.
Backing that out... my method / madness for PSU testing alone was ... that 8ohms would be ~128W... too hot for even my 250W rated dummies. 16 ohm would still be likely pretty toasty at 64W... So, I tripled them up. 43W for a quick test may not be equivalent (1A333) to the realistic load (~2A2 will be final target) in the amp, but I thought it was OK for a fun check. Still pretty revealing (and fun). I wasn't willing to lug the amp around too... Lazy dodo.
Thank you!
You can check resonant freq in clc on a an online calc. It will be in the audible range. Resistor in parallell can help.
Edit: parallell=across
10-20R 3 watts should do it, but pa has talked about this in a 2002 thread i think.
Edit: parallell=across
10-20R 3 watts should do it, but pa has talked about this in a 2002 thread i think.
Apologies, I need to unpack / digest that a bit.
I may need to learn how to check the ringing properly on the scope. I saw some things (when triggered properly) that showed what looked to be a decaying ring like on the quasimodo, but capturing it... yeah... nope. I can't figure that out.
Short of that, I might just implement the "known-working-good" solution you're suggesting and call it a day. I may have completely misinterpreted what I read earlier in the thread (highly likely). This seems to be a known phenomena with a known solution. I am very happy that I made the initial measurements that showed (to experts with keen eyes) what was going on.
So.... Two related, but separate things are happening, we think.
1) Toroid ringing from the toroid / rectifier combination. This is quelled hopefully by a snubber network on the 4320 boards with the snubber R value selected by our friend Quasimodo.
2) Hammond inductors ringing also from ... something. This can be quelled hopefully by placing a resistor of approximately 1R6 to 3R2 in parallel with each inductor. I'd assume, we can use some parallel resistors etc, and that we'd need some relatively speaking / chonky resistors b/c of the current flow etc. Same as pi resistors?
Am I kinda sorta getting it correct?
God damn you write insanely fast
Re point 2: Pa sez something about that. He sez put the resistor across the inductors. A bit of a hassle that, probably will need extensions. I don’t remember the exact values. My resonant freq is 4hz and so I haven’t explored that bit yet.
Re point 2: Pa sez something about that. He sez put the resistor across the inductors. A bit of a hassle that, probably will need extensions. I don’t remember the exact values. My resonant freq is 4hz and so I haven’t explored that bit yet.
^ Fast fingers and tongue are great for some things.... not so good when paired with a slow brain. My mouth outruns my brain far too often.
As always... more reading to do for the dodo. Across... parallel ... for my simple mind, I'll need to draw something up before implementation. If not... I may shed a tear if this PSU has a badaboom. Too much time and effort.
As always... more reading to do for the dodo. Across... parallel ... for my simple mind, I'll need to draw something up before implementation. If not... I may shed a tear if this PSU has a badaboom. Too much time and effort.
Last edited:
There it was!!
https://www.diyaudio.com/community/threads/parallel-thermistors.8882/
Thermistors work great too.
https://www.diyaudio.com/community/threads/parallel-thermistors.8882/
Thermistors work great too.
Resonance 8,7 hz. Not much to worry about but still, can’t hurt to tweak.
https://www.omnicalculator.com/physics/resonant-frequency-lc
https://www.omnicalculator.com/physics/resonant-frequency-lc
Also try with std rectifiers, make sure it's not the LT4320 / your extremely fast FETs. I found a 22uF 40V MLCC across the legs of the 4320 to be effective and the recommended (spec sheet) 10uF lytic / 1uF polyprop to not do much. YMMV depends on the FETs/ load.
Guys, don’t worry over issues that don’t exist. LT4320 rectifiers are perfect even without any snubbers and their switching speed doesn’t produce noise. I’ve measured enough around those questions during usual audiophool dispute over power cords and their effect on power supply and sound.
Point is that active rectifiers ringing without any snubbers is orders of magnitude lower than that with diode bridges. Another important point is that nothing reaches even first reservoir capacitor. Here is attached measurement from my LT4320 board with 33.000 uF first reservoir capacitor. You can put snubbers for a peace of mind but there won’t be any difference you can hear or measure.
@ItsAllInMyHead
Notice that there is always some 2-3 mVpp residual noise which is not from rectifier switching but rather environment noise pickup by oscilloscope probes and small portion of noise from the mains. Check with oscilloscope probe where ground is clipped to probe tip. There will be similar minimum noise level.
Point is that active rectifiers ringing without any snubbers is orders of magnitude lower than that with diode bridges. Another important point is that nothing reaches even first reservoir capacitor. Here is attached measurement from my LT4320 board with 33.000 uF first reservoir capacitor. You can put snubbers for a peace of mind but there won’t be any difference you can hear or measure.
@ItsAllInMyHead
Notice that there is always some 2-3 mVpp residual noise which is not from rectifier switching but rather environment noise pickup by oscilloscope probes and small portion of noise from the mains. Check with oscilloscope probe where ground is clipped to probe tip. There will be similar minimum noise level.
Ah, my fingers are as well, faster than my brain. 😀
Picture above is with an ordinary EI transformer. Here is proper picture with toroidal audio grade transformer.
You can observer that ringing is much smaller. That is because ringing is caused by parasitic leakage inductance of transformer and toroids are much better in that regard.
Picture above is with an ordinary EI transformer. Here is proper picture with toroidal audio grade transformer.
You can observer that ringing is much smaller. That is because ringing is caused by parasitic leakage inductance of transformer and toroids are much better in that regard.
I am so thankful for the help from everyone. I have a bit more information than I can put intro practice / or not. Such is the peril of the uninformed.
I've taken some time to look at the 'segments making up the whole'... and appreciating each chunk of the potential issue individually, while also keeping what I hope to be the overall goal in mind. I certainly don't have the skills / talent / knowledge to know which things may be:
1) Generally necessary. They make the circuit(s) work properly / in harmony with each other, even if the improvement isn't audible.
2) Not necessary. Don't hunt for problems where there are none / chase red herrings.
3) Bad. It can be or likely will be a detriment to circuit performance and even worse, sonic performance in an audio system.
I am confident that a lot of my confusion comes from overthinking and not truly understanding who's snubbing whom...
So, I will try to break things down into chunks.
a) Snubber network for the power toroid / rectifier combination. My very limited understanding is that the switching of the rectifiers is what 'smacks the bell' of the toroid and causes it to ring. I have also read that some 'soft recovery' type diodes / rectifiers are better for not ringing the bell in the first place. It sounds like from above, that the LT4320's don't ring the bell, or just give it a soft tap, if I interpreted correctly. Either way... I have measured all the toroids that will be used (for now) in the PSU, and I have the snubber R's determined. I am waiting on some new LT4320 boards with TH snubber components, so I can socket the resistor for easier changes. The boards are pretty known in the community, but a schematic and link are attached and below. The snubber networks are very easy to install, and while they may not help... would likely not cause any harm... I don't think. Unless bad things could happen, I'd toss them in. I have the parts. I think that's what @tombo56 is saying above.
b) The ringing of the L in the CLCRC. Even after reading this thread again and a few other resources, I'm not sure what causes the inductors to ring. Perhaps it's the rectifiers also, but just from the opposite side of the circuit and through the first C? I have much less of a grasp on this. I am more than willing to do the work, but I admit to being at a loss still on this one. Putting a resistor or two in parallel with the chokes is no big deal, but if implemented, I'd like to do it properly. Is there a measurement I could take to more properly indicate if ringing is present in the first place?
I'll try to capture a few more good scope images. This is a good way for me to learn more about the scope. I'm trying to figure out if I can trigger the scope off of one of the high peaks I see, then only capture a small amount of time after. I got it to show something that looked like ringing earlier, but I could not get a photo.
Thanks again to all!
https://www.diyaudio.com/community/threads/lt4320-based-active-rectifier.336572/
I've taken some time to look at the 'segments making up the whole'... and appreciating each chunk of the potential issue individually, while also keeping what I hope to be the overall goal in mind. I certainly don't have the skills / talent / knowledge to know which things may be:
1) Generally necessary. They make the circuit(s) work properly / in harmony with each other, even if the improvement isn't audible.
2) Not necessary. Don't hunt for problems where there are none / chase red herrings.
3) Bad. It can be or likely will be a detriment to circuit performance and even worse, sonic performance in an audio system.
I am confident that a lot of my confusion comes from overthinking and not truly understanding who's snubbing whom...
So, I will try to break things down into chunks.
a) Snubber network for the power toroid / rectifier combination. My very limited understanding is that the switching of the rectifiers is what 'smacks the bell' of the toroid and causes it to ring. I have also read that some 'soft recovery' type diodes / rectifiers are better for not ringing the bell in the first place. It sounds like from above, that the LT4320's don't ring the bell, or just give it a soft tap, if I interpreted correctly. Either way... I have measured all the toroids that will be used (for now) in the PSU, and I have the snubber R's determined. I am waiting on some new LT4320 boards with TH snubber components, so I can socket the resistor for easier changes. The boards are pretty known in the community, but a schematic and link are attached and below. The snubber networks are very easy to install, and while they may not help... would likely not cause any harm... I don't think. Unless bad things could happen, I'd toss them in. I have the parts. I think that's what @tombo56 is saying above.
b) The ringing of the L in the CLCRC. Even after reading this thread again and a few other resources, I'm not sure what causes the inductors to ring. Perhaps it's the rectifiers also, but just from the opposite side of the circuit and through the first C? I have much less of a grasp on this. I am more than willing to do the work, but I admit to being at a loss still on this one. Putting a resistor or two in parallel with the chokes is no big deal, but if implemented, I'd like to do it properly. Is there a measurement I could take to more properly indicate if ringing is present in the first place?
I'll try to capture a few more good scope images. This is a good way for me to learn more about the scope. I'm trying to figure out if I can trigger the scope off of one of the high peaks I see, then only capture a small amount of time after. I got it to show something that looked like ringing earlier, but I could not get a photo.
Thanks again to all!
https://www.diyaudio.com/community/threads/lt4320-based-active-rectifier.336572/
Attachments
Last edited:
My 'lessons' are perhaps moving into more of the how to use a scope for this application for PSUs... vs. the benefits of CLC vs. CRC. In an effort to keep it on topic... I'll measure a CRC also... 🙂
Below are a few pics after (I think) I've learned some better triggering techniques. Goes to show that giving a novice a nice tool can be a fool's errand...
Laughter and levity aside, I made an effort to understand what some of this could be other than pretty pictures on a screen, which they are (to me). I took pictures of the screen b/c saving to floppy and transferring... LOL! IF I learn how to properly use 1/1000th of the capabilities of this scope, I may treat myself to a newer model that makes capture and a few other things easier.
Again, this is the CLCRC PSU. I don't see anything in the 60Hz and other multiples of 'ripple', but because the 'noise' voltage peaks are prevalent, and my techniques are limited, I couldn't capture anything too relevant.
Starting with the lowest frequency pattern. 30kHz, I think. No idea if this is a concern or native to the circuit. I think it's either native to the circuit or EMI b/c the pattern persisted even after disconnecting from the mains / mains earth.
And moving slowly toward what I thought was the 'ringing'. It may very well, be, but it's at a relatively very high frequency (I think).
I saw a nice sine wave ... and I thought, well that must be the ripple with noise overlaid... but the sine is 20Mhz with 1 GHz intertwined w/in it... I think. I don't know diddly squat about signal theory or practice, but is this a 20Mhz carrier of a 1GHz signal? I haven't the absolute foggiest idea. My guess is pretty noise. That's all I've got.
A bit tighter zoom of same...
And finally what I thought was the ringing I saw earlier... and it may be some sort of ringing, but it's at an awfully high (I think) frequency to what a rectifier would produce.
While you all have a chuckle to yourselves... If you're willing to share any plausible explanations for what I'm seeing (or not seeing) ... I'd again be grateful.
It's like my first time looking at a paramecium under a microscope. Wildly fascinating, but WTF!?
Edited to add - Forgive the m vs. M for Megahertz in some of the annotated pictures. I'll learn... eventually.
Below are a few pics after (I think) I've learned some better triggering techniques. Goes to show that giving a novice a nice tool can be a fool's errand...
Laughter and levity aside, I made an effort to understand what some of this could be other than pretty pictures on a screen, which they are (to me). I took pictures of the screen b/c saving to floppy and transferring... LOL! IF I learn how to properly use 1/1000th of the capabilities of this scope, I may treat myself to a newer model that makes capture and a few other things easier.
Again, this is the CLCRC PSU. I don't see anything in the 60Hz and other multiples of 'ripple', but because the 'noise' voltage peaks are prevalent, and my techniques are limited, I couldn't capture anything too relevant.
Starting with the lowest frequency pattern. 30kHz, I think. No idea if this is a concern or native to the circuit. I think it's either native to the circuit or EMI b/c the pattern persisted even after disconnecting from the mains / mains earth.
And moving slowly toward what I thought was the 'ringing'. It may very well, be, but it's at a relatively very high frequency (I think).
I saw a nice sine wave ... and I thought, well that must be the ripple with noise overlaid... but the sine is 20Mhz with 1 GHz intertwined w/in it... I think. I don't know diddly squat about signal theory or practice, but is this a 20Mhz carrier of a 1GHz signal? I haven't the absolute foggiest idea. My guess is pretty noise. That's all I've got.
A bit tighter zoom of same...
And finally what I thought was the ringing I saw earlier... and it may be some sort of ringing, but it's at an awfully high (I think) frequency to what a rectifier would produce.
While you all have a chuckle to yourselves... If you're willing to share any plausible explanations for what I'm seeing (or not seeing) ... I'd again be grateful.
It's like my first time looking at a paramecium under a microscope. Wildly fascinating, but WTF!?
Edited to add - Forgive the m vs. M for Megahertz in some of the annotated pictures. I'll learn... eventually.
Last edited:
I'm no authority/ just a bodger, but a fair guess might be interference from a SM PSU. Something that 'snaps' open /closed quickly to impersonate a signwave is a Switch Mode PS...
Unplug that phone charger (maybe)
Unplug that phone charger (maybe)
^ Thank you! It could be so many things...
This is truly fascinating for a newcomer. I spent 30 mins using search engines to try and see what things were roughly in that 1 GHz / 20 MHz range. The 30kHz has me more interested, but it's still not totally relevant. At one point, I thought the higher frequency stuff might be my WiFi, or my microwave ... or ... or ... or... It's fun to chase that rabbit down the hole, but I want music. 🙂
Either way, what I'll likely do is see if I can limit the bandwidth further on the scope below 20MHz. 100kHz seems reasonable, I think. I'm also trying to see how I can have the bandwidth limiting 'hold' during a measurement session. It seems to pop on / off when I change other settings. That way, I won't get distracted by high-frequency noise that I'm not really looking for. It's fascinating, sure... but it's a distraction from what I'm really trying to measure. SillyScope for DoDos 101... at least I haven't blown anything up.
Thanks so much again! I kinda high-jacked your thing. Sorry for that.
This is truly fascinating for a newcomer. I spent 30 mins using search engines to try and see what things were roughly in that 1 GHz / 20 MHz range. The 30kHz has me more interested, but it's still not totally relevant. At one point, I thought the higher frequency stuff might be my WiFi, or my microwave ... or ... or ... or... It's fun to chase that rabbit down the hole, but I want music. 🙂
Either way, what I'll likely do is see if I can limit the bandwidth further on the scope below 20MHz. 100kHz seems reasonable, I think. I'm also trying to see how I can have the bandwidth limiting 'hold' during a measurement session. It seems to pop on / off when I change other settings. That way, I won't get distracted by high-frequency noise that I'm not really looking for. It's fascinating, sure... but it's a distraction from what I'm really trying to measure. SillyScope for DoDos 101... at least I haven't blown anything up.
Thanks so much again! I kinda high-jacked your thing. Sorry for that.
Well, thanx… Darnit! 🙂
…Now i will have to snitch and add the Ls assigned for my upcoming Singing Tokins and Lazy Singing Tokin Birds. And arrange CRCLC in my twin ❤️ Aleph Js.
🎷🙂🎸
…Now i will have to snitch and add the Ls assigned for my upcoming Singing Tokins and Lazy Singing Tokin Birds. And arrange CRCLC in my twin ❤️ Aleph Js.
🎷🙂🎸
^ FWIW, plus a bag of chips from a total neophyte...
CLCRC both simmed and measured better than CRCLC (for me with my stuff etc. and all the caveats).
I doubt it's much of a practical difference... but... it's something to consider.
CLCRC both simmed and measured better than CRCLC (for me with my stuff etc. and all the caveats).
I doubt it's much of a practical difference... but... it's something to consider.
I like simple. My typical CLC of 44mF - 159ZJ - 44mF measures well and there is no audible noise from my 103dB sensitive speakers.
With an external power supply though, it is good to add the RC section to the amplifier chassis, with the R being the umbilical.
With an external power supply though, it is good to add the RC section to the amplifier chassis, with the R being the umbilical.