My calculations were incorrect because linear ripple rejection (mV) is multiplied in series, not added. And thus logarithmic ripple rejection (dB) is added in series. This means that series configuration should trounce parallel, with the only downfall being series current capacity is half of parallel.
Keep in mind that even 1dB ripple rejection is a significant amount because it's 26% ripple reduction, and -3dB is half the ripple.
For the following examples, lets assume we're starting with a fairly clean 10mV ripple.
Going by the LT3045 data sheet, a 1V voltage differential generates 70dB ripple rejection...
10mV ripple @ -70dB = 0.00316mV ripple
If we again pass that through another LT3045 at 1V voltage differential and another 70dB ripple rejection...
0.00316mV ripple @ -70dB = 0.000001mV ripple
Likewise, -70dB + -70dB = -140dB which is also 0.000001mV ripple from 10mV
Now lets start with a single LT3045 at 3V voltage differential and 83dB ripple rejection...
10mV ripple @ -83dB = 0.00070795mV ripple
So a single LT3045 at 3V voltage differential (-83dB ripple) is 4.5 times more effective at ripple rejection than a single LT3045 at 1V voltage differential (-70dB ripple)
But two series LT3045 at 1V voltage differentials is 708 times more effective at ripple rejection than a single LT3045 at 3V voltage differential, and it's 50% more efficient (net -2V versus -3V).
Also two series LT3045 at 1V voltage differentials is 3160 times more effective at ripple rejection than a single LT3045 at 1V voltage differential, but it's 50% less efficient (net -2V versus -1V).
So definitely series LT3045 has a distinct advantage over single. But if you can't wing two, try one at 3V differential.
Cornan, you must have a REALLY good ear to be able to distinguish these differences in your testing as they seem to reflect on paper. That's quite an ability. I'm very impressed!
Keep in mind that even 1dB ripple rejection is a significant amount because it's 26% ripple reduction, and -3dB is half the ripple.
For the following examples, lets assume we're starting with a fairly clean 10mV ripple.
Going by the LT3045 data sheet, a 1V voltage differential generates 70dB ripple rejection...
10mV ripple @ -70dB = 0.00316mV ripple
If we again pass that through another LT3045 at 1V voltage differential and another 70dB ripple rejection...
0.00316mV ripple @ -70dB = 0.000001mV ripple
Likewise, -70dB + -70dB = -140dB which is also 0.000001mV ripple from 10mV
Now lets start with a single LT3045 at 3V voltage differential and 83dB ripple rejection...
10mV ripple @ -83dB = 0.00070795mV ripple
So a single LT3045 at 3V voltage differential (-83dB ripple) is 4.5 times more effective at ripple rejection than a single LT3045 at 1V voltage differential (-70dB ripple)
But two series LT3045 at 1V voltage differentials is 708 times more effective at ripple rejection than a single LT3045 at 3V voltage differential, and it's 50% more efficient (net -2V versus -3V).
Also two series LT3045 at 1V voltage differentials is 3160 times more effective at ripple rejection than a single LT3045 at 1V voltage differential, but it's 50% less efficient (net -2V versus -1V).
So definitely series LT3045 has a distinct advantage over single. But if you can't wing two, try one at 3V differential.
Cornan, you must have a REALLY good ear to be able to distinguish these differences in your testing as they seem to reflect on paper. That's quite an ability. I'm very impressed!
Very interesting! Thanks for clearing that out. I actually suspected that your first calculations was wrong and Alexey (that produce LT3045 boards) confirmed my doubts at CA.
Thanks! I must admit that it feels especially good when calculations and ears match!
I recently got my balanced & floating AC-AC ps, but realized with help from Michael and others at CA that I made a misstake with the 15VAC output which would give 20,81 VDC to the first HC-HPULN. Luckily Larry (@lmitche) at CA provided a great solution to my problem. The Sigma 12 discrete regulator board S12 Super regulated linear power supply board 5-30V LPS 10000uF 50V Versio PSU | eBay
which is now on order. This will unfortunately delay my “uber-LPSU”, but hopfully just by a couple of weeks or so. Anyway, I will post my progress here and on my CA tweak thread during the build.
Micael
Last edited:
S12 + HC-HPULN is also the topology I'm trying out. That seller of the S12 is very helpful, especially with custom voltages and transformer matching. I'm very interested in how much of a difference a balanced isolated transformer makes with the S12 + HC-HPULN.
My system is an 8 channel HTPC with an Asus STX II 7.1 sound card. The specs (124dB SNR, 0.0003% THD+N) seem better than dedicated receivers I can locate. I'm using Hypex nCore NC400 amps for each of 7 "non-sub" channels and it sounds amazing. Right now, the sound card is powered by an ordinary switching PC power supply, and I can't hear any noise floor with my ear on the speaker at 100% at volume. Yes it's scary and dumb to attempt. But the complete lack of any audible noise floor with just a PC switching power supply is interesting. I've been using this configuration for several years.
I was inspired to see if it could be further improved. Today I upgraded all 12 op amps to Burson V6 vivid, which I'm currently burning in. I saw reports of how much better the sound card is when fed a linear power supply rather than the PC power supply. I'm interested in seeing how much better it could get compared to already having no audible noise floor at 100% volume. So I'm giving it a shot.
I'm also moving the sound card to a dedicated shielded enclosure, using a PCI riser which connects the sound card to the PCI with a USB cable.
Off topic, but all of the components, including the PC and nCore amps are in individual rack mount enclosures. I like the visual consistency and discreetness of the black wall of rack rather than a hodgepodge of various interconnnected components.
The sound card calls for 12vdc 5vdc, and 3.3vdc lines. The 3.3V line is used only for digital processing, not for any analog work, so I'll just pipe this off the 5v line with LT3045 rather a dedicated LPS line. I'm also upping the 12v line to 16v (soldered directly to the out pads of the sound card 78xx regulators) to get the most out of the Burson op amps.
Because there's so much involved with multiple voltages, I'd like the versatility of changing components up as desired. So I don't like those balanced isolated power supplies being locked to a single voltage. They also don't seem to come in 100VA. I'm considering a single balanced isolated transformer (120VAC out) to feed the LPS lines. A 500VA for just over a hundred bucks could provide balanced isolation for all downstream lines. Each LPS line then having an R-Core transformer + S12 + HC-HPULN + HC-HPULN*maybe.
My system is an 8 channel HTPC with an Asus STX II 7.1 sound card. The specs (124dB SNR, 0.0003% THD+N) seem better than dedicated receivers I can locate. I'm using Hypex nCore NC400 amps for each of 7 "non-sub" channels and it sounds amazing. Right now, the sound card is powered by an ordinary switching PC power supply, and I can't hear any noise floor with my ear on the speaker at 100% at volume. Yes it's scary and dumb to attempt. But the complete lack of any audible noise floor with just a PC switching power supply is interesting. I've been using this configuration for several years.
I was inspired to see if it could be further improved. Today I upgraded all 12 op amps to Burson V6 vivid, which I'm currently burning in. I saw reports of how much better the sound card is when fed a linear power supply rather than the PC power supply. I'm interested in seeing how much better it could get compared to already having no audible noise floor at 100% volume. So I'm giving it a shot.
I'm also moving the sound card to a dedicated shielded enclosure, using a PCI riser which connects the sound card to the PCI with a USB cable.
Off topic, but all of the components, including the PC and nCore amps are in individual rack mount enclosures. I like the visual consistency and discreetness of the black wall of rack rather than a hodgepodge of various interconnnected components.
The sound card calls for 12vdc 5vdc, and 3.3vdc lines. The 3.3V line is used only for digital processing, not for any analog work, so I'll just pipe this off the 5v line with LT3045 rather a dedicated LPS line. I'm also upping the 12v line to 16v (soldered directly to the out pads of the sound card 78xx regulators) to get the most out of the Burson op amps.
Because there's so much involved with multiple voltages, I'd like the versatility of changing components up as desired. So I don't like those balanced isolated power supplies being locked to a single voltage. They also don't seem to come in 100VA. I'm considering a single balanced isolated transformer (120VAC out) to feed the LPS lines. A 500VA for just over a hundred bucks could provide balanced isolation for all downstream lines. Each LPS line then having an R-Core transformer + S12 + HC-HPULN + HC-HPULN*maybe.
Awesome!
It would be very interesting to know your experiences with S12 & HC-HPULN. The seller is really great and since he is the seller for the 50VA balanced & floating AC-AC ps as well he is very helpful getting it all optimized. Right now he is suggesting me to rewire the transformer secondary to 18VAC output to be able to get a 14VDC (or possibly 16VDC) output from the S12.All my audio gears are already connected to a balanced isolation transformer with floating secondary (center tap connected to chassi) It is actually one of my best audio purchases ever. I use a DC blocker trap filter pre that IT and a starquad wired and star earthed power dustributor (no filters) post it.
I agree with the hassle of balanced transformer having fixed outputs. I use Gophert csp-3205II for that very reason (five of them).
I heard a lot of good things of the Burson V5 so the V6 is surely a great upgrade! Otherwise you have a really nice and well thought of setup. Personally I have moved away from PC, Server & NAS due to the complex power solutions to get it optimal. My setup is nowhere near KISS though!
Micael
Unfortunately I have run into a wall with the S12 since it was not possible to get 13,5VDC output from 15VAC input that I was hoping for. I had to make a choise and choosed a more universal solution which was to change the S12 discrete voltage reg to 9VDC out and to order another 12VAC out balanced AC-AC ps to feed the HC-HPULNs in series directly. This will mean ~16,568VDC after the rectifiers of the first HC-HPULN with 14VDC out>HC-HPULN 12VDC out.
This means that I will not use the S12 with the HC-HPULN I am afraid, but I will try it with the LT3045 1A>HPULN (3A) in series probably (My other HPULN is occupied with other work). This combo will power either ISO Regen, Aqvox switch-8 or Luckt BluWave USB to Spdif board depending on where it will make the best improvement.
/Micael
Last edited:
Using the LTs in series has been verified by myself and others as well (some of whom Cornan is familiar with), just not on this forum. Glad to see some numbers to support what we hear.
As everything in this hobby it is a huge community effort. Without the great support of helpful audiophile friends true music magic could never happen in this hobby. I wasn’t the first to try LT3045s in series, even if I was the one trying the crazy idea with HPULN and HC-HPULN in series. I am truly grateful for all the help from others. Without them I would still be enjoying my Pioneer A-656, Pioneer CD-player and Dali 404 speakers without knowing what I missed out!
Last edited:
Hello
Is it possible to know how you can calculate these numbers based on the data sheet ? After check this data sheet I did not understand how to do it.
In my case ,I have got one single rail (6 x LT3045) and the dropout is 0V, by using r-Core transformer 12V AC
... I thought it was a good idea, but it's not the case based on your analysis.Thanks for your help.
Hello
Is it possible to know how you can calculate these numbers based on the data sheet ? After check this data sheet I did not understand how to do it.
In my case ,I have got one single rail (6 x LT3045) and the dropout is 0V, by using r-Core transformer 12V AC
Thanks for your help.
Sure.. See page 10 of the LT3045 data sheet, the upper right graph "Power Supply Ripple Rejection" shows that ripple rejection is a function of voltage differential. There's minimal ripple rejection at .26V differential. I'm confused by your 0V dropout. From what I gather, 0.26V is minimum. You may want to see if you notice an improvement with a higher voltage R-core.
I tend to agree, data isn't everything. But I am pleased how well it matches Cornan's experience.
I'm going with a middle of the road method. I'm going with two in series with 1.5V differential between each, rather than 3V differential. Basically it's net 3V differential divided among two in series. It should be a good compromise for efficiency.
What I'll do to analyze the difference is first run it with just the S12 with single HC-HPULN with 3V differential, and then add the other HC-HPULN in series to divide up that differential into two 1.5V increments. But it would just be a subjective test, which I'm really not the best at. I'd like to see it with an oscilloscope but I'm comfortable enough just going this route after following Cornan's experience in another thread.
Last edited:
That's too bad you can't use the S12 + HC-HPULN + HC-HPULN. I'm sure the two series HC-HPULN + HC-HPULN would already cover the vast majority of filtering, but the S12 is so inexpensive, it's like free filtering.
I just ordered a balanced isolated PS + S12 for each line. The seller was helpful in selecting the correct voltage configurations for each component to arrive at the final voltages I need for each line (18V and 9V) before the first HC-HPULN.
As far as how to wire the balanced isolated power supply, I was just planning on going with how it comes. You can see a picture of it here. I ordered the less expensive one with the smaller enclosures and without EMI filters because I'll be placing those enclosures within my rack mount with it's own own EMI filter, along with the S12s and HC-HPULNs. I'll remove the AC socket from those enclosures and wire directly through grommets/cord grips. You have any other suggestions for improvement?
Yeah it was a real bummer! I think I will be fine with just the two HC-HPULNs though. The balanced isolated ps will be plugged into a balanced isolation transformer with floating secondary (center tap connected to chassi) so it will be balanced and isolated all the way to the end!
The only thing that comes to mind is to use a Kemet A750 series aluminum Polymer cap at the output of the S12 (and possibly also at the input of the first HC-HPULN) which have improved things a lot in my setup lately. Otherwise just mind the cables. I personally use starquad AC & DC cables. Adding a JSSG around the starquad cables will protect them against airborn RFI/EMI. JSSG is a shield mesh tube around the cable that have a single drain wire (~28awg) soldered to each end of the tube acting as a Faraday cage for the cables. A great little tweak by John Swenson.
Ghent Audio knows how to make them to a fair price.
I will be very interested to know how your LPSU with S12 and HC-HPULNs will turn out. I will surely report back here with my own findings of my two versions as well!
Micael
Last edited:
Thanks for all.
Modification done and the sound sounds better I think (input 12.1V AC ; ouput 11,84V DC)...And I'm limited based of the input voltage of my DAC. I'll try to found an another transformer in 15V AC (it will not be easy) to have 3V as dropout and check if it's better.
Last edited:
Very interesting! Thanks for clearing that out. I actually suspected that your first calculations was wrong and Alexey (that produce LT3045 boards) confirmed my doubts at CA.
Thanks! I must admit that it feels especially good when calculations and ears match!
Hi,
Which type (amperage and mark, because the voltage is 14V AC if I'm right) of transformer you have choosen for your Linear PS ?
Here it is HIFI Audio 50W AC 12V Balanced isolation power supply 50VA EMI AC version PSU | eBay
Available in 12VAC, 15VAC and 18VAC. I have ordered both 15VAC and 12VAC. To calculate AC to DC with Stammheims PS use this formula AC x 1.414 -0,4= DC output voltage after the rectifiers.
Here it is HIFI Audio 50W AC 12V Balanced isolation power supply 50VA EMI AC version PSU | eBay
Available in 12VAC, 15VAC and 18VAC. I have ordered both 15VAC and 12VAC. To calculate AC to DC with Stammheims PS use this formula AC x 1.414 -0,4= DC output voltage after the rectifiers.
Thanks a lot.
Are you sure about the formula, because when I apply it : 12V AC in input, I've got 16,5V DC for the output, it's strange because usually we need to have a voltage lower than the input. It's the opposite no ?
You’re welcome!
I have got the formula from Michael (Stammheim). AC-DC= 1.414 Drop down of rectifiers= 0.4v
The figures you’ve got is probably BEFORE the rectifiers. Remember that the LT3045 can only manage <20VDC.
12VAC x 1,414= 16,968 -0,4= 16,568VDC so it is actually correct!
Last edited:
You’re welcome!
I have got the formula from Michael (Stammheim). AC-DC= 1.414 Drop down of rectifiers= 0.4v
The figures you’ve got is probably BEFORE the rectifiers. Remember that the LT3045 can only manage <20VDC.
12VAC x 1,414= 16,968 -0,4= 16,568VDC so it is actually correct!
To be honest I'm totally lost
sorry ... based on this example :