I use two HC-HPULN in series since it improves SQ compared to one. This is not something specific for the HC-HPULN but with LT3045 voltage regs. I have series LT3045s in all of the audio devices and ethernet devices in my setup and they improve in series everywhere. The HC-HPULN in series replaced the HPULN in series since they sounded considerable better powering my Mytek Brooklyn DAC. I could’nt be happier!
I happened across another forum in which you refer to configuring two HC-HPULN in series with 1V voltage differential between intervals. A key consideration is that for the LT3045, Power Supply Ripple Rejection increases with voltage differential.
On page 10 of the LT3045 data sheet, "Power Supply Ripple Rejection" is around 60dB@.5V, 70dB@1V, 78dB@2V, 83dB@3V voltage differential (averaged over the charted frequencies)
So there would be a 13dB ripple rejection improvement between stages if the voltage differential were changed from 1V to 3V.
Let's compare two configurations:
14V -> LT3045 -> 13V -> LT3045 -> 12V
In this configuration, cumulative ripple rejection is 70dB + 70dB = 73dB over two LT3045 in series.
15V -> HC-HPULN -> 12V
Here we have cumulative ripple rejection 83dB with one LT3045.
So in theory, one LT3045 with 3V differential should have considerably more ripple rejection than multiple series LT3045 having 1V differential. Of course there are many other factors affecting ripple rejection, such additional capacitance. But this illustrates a significant consideration for when planning a configuration.
Probably the ideal configuration would be...
18V -> LT3045 -> 15V -> LT3045 -> 12V
This configuration would have cumulative ripple rejection 83dB + 83dB = 86dB over two LT3045 in series.
Or only slightly better...
21V -> LT3045 -> 18V -> LT3045 -> 15V -> LT3045 -> 12V
This configuration, would have cumulative ripple rejection 83dB + 83dB +83dB = 87.8dB over three LT3045 in series.
This also illustrates the diminishing returns with additional series LT3045.
Last edited:
@get-lit
Many thanks for your explanation. That was my To-Do for the next days, too
Who ever wants to reach the most possible ripple rejection, please notice, that by increasing voltage difference (Vin - Vout) in combination with a huge power (ampere) consumption massive heat is generated by the reg's . Getting very hot has also negative influence on life time of the components.
So your implementation should be dependent on your application and environment.
Best regards
Stammheim
Many thanks for your explanation. That was my To-Do for the next days, too
Who ever wants to reach the most possible ripple rejection, please notice, that by increasing voltage difference (Vin - Vout) in combination with a huge power (ampere) consumption massive heat is generated by the reg's . Getting very hot has also negative influence on life time of the components.
So your implementation should be dependent on your application and environment.
Best regards
Stammheim
Very interesting! The thing is that I have always preferred the SQ with LT3045s with 0.7-1v drop-down rather than 2v. I have tried that many times. Therefore 2v drop-down or more was’nt really an option for me when trying out HPULNs or HC-HPULNs in series. I have not tried any LT3045 with 3v drop-down to keep them as cool as possible.
I will definitely try out the HC-HPULNs in series with 3v drop-down to see if your theory holds true IRL. I will report back here and on CA as soon as I have tried it out.
I have a comming project with a balanced and floating AC-AC PS coming up as well (15VAC output) to see if that would concur my Gophert csp-3205II (floating SMPS) as a feeder solution for my series HC-HPULNs powering my Brooklyn DAC.
Cornan, as always your experience is very welcome to me!
Guys, I am working on a webshop for offering the HPULN's from a central website. Feel free to subscribe for an info of launching the website (until the end of May 2018 hopefully) and new products, which I work for right now...
Here we go:
MPAudio
Guys, I am working on a webshop for offering the HPULN's from a central website. Feel free to subscribe for an info of launching the website (until the end of May 2018 hopefully) and new products, which I work for right now...
Here we go:
MPAudio
I will be more than happy to share my impressions!
Awesome, I will go ahead and subscribe. Good luck with the launch!
Any chance you're working on a single board with regs in parallel and series?
@getlit: great to see someone has actually read the sheets rather than just making fantasy claims about adding multiple regs. ;-)
The lt3045 is capable of stellar results, but under most operating conditions its no better than anything else.
Seems like getting the 5amp version and dropping 3v is the way forward.
The lt3045 is capable of stellar results, but under most operating conditions its no better than anything else.
Seems like getting the 5amp version and dropping 3v is the way forward.
Last edited:
Yes, but no fixed results yet.
Please notice that with a current of 5A and a voltage difference of 3V the heatsink and with that the pcb could get to uncomfortable dimensions.
So what would be very helpful to all of us is your experience, guys, testing and listening with several voltage differences to get an overview using many devices, as anyone's hifi chain is limited.
I would work on a knowledge database with all experiences, which could be helpful for all users.
Best regards
Stammheim
Your initial subjective tests are interesting. There are other variables such as noise and transient repsonse, but these should have much less weight on performance than ripple reduction. I'd be interested in seeing your further results. Also try parallel, especially if you're over 300mA per LT3045.
However you put it 86dB ripple rejection is better than 83dB looking at a data sheet, so putting two in series will always be better than one. The law of diminishing results is something you´ll have to ignore in this hobby since it is the end result in terms of SQ that really counts. A lot of other things will affect the result.
We could probably argue all night long about fantasy claims. Interpret data sheets or personal experiences?
I will surely report back how a 3v drop-down will turn out in my specific setup.
There is other things like caps that affects the SQ quite a lot with the LT3045s. Actually that much that I am beginning to realize that it is the combinations of output and input caps that could be one of the reasons that for example a LT3045 1A board into a HPULN actually sounds better than HPULNs in series in my setup. Might be a clue that tantalum caps or Elna Silsmic II caps and aluminium polymer caps is a really great combo with LT3045s in series? This is however something that I have just started to experiment with and it is way too early to give any firm conclusions from my part.
The HPULN and HC-HPULN is already parallel, so in fact putting them in series is both parallel and series.
Yes, that is why I was thinking that this would require mounting to the enclosure rather than using a heatsink which would in turn require moving the protruding components to the other side of the board.
The LT3045 data sheet has specific instruction for the types of caps and values (ceramic/electrolytic/tantalum/µF/ESR/ESL) used on input vs output for optimal ripple rejection, noise reduction and transients. I have no idea what some of these figures mean but I'm sure Stammheim has already taken all of this into account with his assemblies.
One of the first versions was a bottom-mount-cooled version, which was not flexible and not efficient (regarding cooling) enough. Cooling would be very dependent on on material (copper, Alu...) and it's thickness. Enclosures of most hifi equipment have a material thickness of 1mm, so cooling would be very limited.
Heat is now conducted by the LT's exposed pad to a whole copper layer, so don't underestimate the cooling capacity.
On this experience the decision was made to go the actual way.
BUT: there is something in work for your mentioned kind of application, too
@get-lit
Best regards
Stammheim
Last edited:
Sure, in the end this is just a matter of changing the caps and fine-tune it using my ears to guide me. Right now I am putting them just about anywhere to see how they affect SQ in just about all of my audio devices incl router and network switch. I am even planning to do a crazy experiment and add them to the USB path (ISO Regen output into BluWave board). Not sure how that one will turn out, but it will be fun to try!
I`m totally sure that Michael have taken all of this into account. It is quite easy to hear since both the HPULN and HC-HPULN (even more so) sounds much better than the standard LT3045 in singles (using tantalum output caps) in my setup.
Last edited:
