Why a linear power supply?

[tne]

Question:

 

Can't we have a thread on an interesting technical topic such as this without a "subjectivist" / "objectivist" pissing match taking place ?

 

I am a firm believer in free speech and all that, but it just is so jejune and predictable - on both sides. Speaking for myself, but I believe also in agreement with the vast majority of readers who are in neither camp and who have nothing at stake in regard to our internet persona, we just want to use the collective intelligence and experience of members of CA to learn more about both the technical aspects of CA and about good music. Opinions, whether DBTed, sighted or unsighted, etc., are welcome. We are intelligent people who can figure out for ourselves how to evaluate and act on the information that sites such as this offer.

 

There are a few members who are able to express their points without descending into this quagmire, but so many (once again, on both sides) cannot.

 

So there. It's off my chest now. Snarky followup replies will be happily ignored.

[jtwrace]

What is the projected ETA for these products? Obviously testing and completion needs to be done but is there an idea of when that will be? Will it be a mass produced PCB? What are you guys thinking?

 

I know, too many questions that don't have answers yet. :-)

[Superdad]

Question:

 

Can't we have a thread on an interesting technical topic such as this without a "subjectivist" / "objectivist" pissing match taking place ?

 

I am a firm believer in free speech and all that, but it just is so jejune and predictable - on both sides. Speaking for myself, but I believe also in agreement with the vast majority of readers who are in neither camp and who have nothing at stake in regard to our internet persona, we just want to use the collective intelligence and experience of members of CA to learn more about both the technical aspects of CA and about good music. Opinions, whether DBTed, sighted or unsighted, etc., are welcome. We are intelligent people who can figure out for ourselves how to evaluate and act on the information that sites such as this offer.

 

There are a few members who are able to express their points without descending into this quagmire, but so many (once again, on both sides) cannot.

 

So there. It's off my chest now. Snarky followup replies will be happily ignored.

 

+1

 

And on top of all that, I, as an audio enthusiast entrepreneur (not the first time round the block either), now need to worry about what and how I post on my favorite forum (currently the only site where I post). Guess it just comes with the territory. I have made friends here, and despite all the barbs an pointed debate, I don't think I have any foes.

 

Cheers,

Alex C.

[Superdad]

What is the projected ETA for these products? Obviously testing and completion needs to be done but is there an idea of when that will be? Will it be a mass produced PCB? What are you guys thinking?

 

I know, too many questions that don't have answers yet. :-)

 

Shooting for production before mid-year. Mass produced? If you can call initial 50-100 unit runs "mass"… But yes, we use a board house for PCB fabrication and stuffing (both surface mount and through-hole parts), and the Japanese chassis maker that I picked will do the required machining/drilling/tapping of the heatsinks and bottom/front/rear panels and silk screening and ship that to me unassembled. My teenage sons and I will do all assembly/testing in my converted garage. Everyone will get paid per unit including Mr. Swenson. I do already operate as a business (UpTone Audio MusiCap film-and-foil capacitors). That, combined with my years at Hovland Company, plus really talented friends, hopefully take some of the "amateur" out of the "cottage" nature of it.

 

Thanks for your interest. Hope the above answers your "direct questions." If anybody wants to know more (like voltages and amperage ratings), you will have to ask. I don't want to get in trouble...

[Daudio]

First I didn't see Eloise's question as an attempt to pull the discussion anywhere. That does not seem to be the way she operates.

 

No, I don't think it is her intent, but I have seen it happen more then a few times

[Daudio]

Shooting for production before mid-year...

 

If anybody wants to know more (like voltages and amperage ratings), you will have to ask. I don't want to get in trouble...

 

Alex,

 

Nice !!

 

Can you give us a some kind of range for a price point ?

 

Thanks,

[Jud]

So there. It's off my chest now. Snarky followup replies will be happily ignored.

 

I am lost in admiration of your use of both "jejune" and "snarky" in the same reply.

[sandyk]

John Swenson

Before you put it all together, did you draw a concept diagram of your PSU ? I imagine that quite a few would be interested to see how you used Kelvin Sensing to achieve such great results in voltage stability. I do realise that a few other well known PSUs do similar.

 

Alex

[acg]

I believe that most members are interested, even those Aussie banana benders who love some weird device named a Killerdac.

 

Would you believe that last weekend I went and listened to a Killerdac in its natural environment (i.e. surrounded by valves and heavily modified cd spinners)? It was delicious.

 

Anthony

[One and a half]

Question:

 

Can't we have a thread on an interesting technical topic such as this without a "subjectivist" / "objectivist" pissing match taking place ?

 

I am a firm believer in free speech and all that, but it just is so jejune and predictable - on both sides. Speaking for myself, but I believe also in agreement with the vast majority of readers who are in neither camp and who have nothing at stake in regard to our internet persona, we just want to use the collective intelligence and experience of members of CA to learn more about both the technical aspects of CA and about good music. Opinions, whether DBTed, sighted or unsighted, etc., are welcome. We are intelligent people who can figure out for ourselves how to evaluate and act on the information that sites such as this offer.

 

There are a few members who are able to express their points without descending into this quagmire, but so many (once again, on both sides) cannot.

 

So there. It's off my chest now. Snarky followup replies will be happily ignored.

 

In total agreement, thank you.

[Superdad]

I do realise that a few other well known PSUs do similar.

 

Really? Others that use a remote sense line for voltage feedback? Examples please.

Of course the principle and practice of Kelvin-sense circuits is not new, I am just unaware of any commercially available low-voltage DC power supplies that offer it. At least not in the voltage and current range of interest for audiophiles.

[sandyk]

Alex C

I believe you will see a few examples discussed in DIY Audio, but they aren't commercial power supplies. I believe that a version of the Salas Regulator does this too, using remote sensing. I have never delved deeply into this area , which is why I asked John the question.

Perhaps I should have said well known to the DIY community ?

Alex

[jtwrace]

What voltages/amps will the complete PS be offered in? Also, what is the board capable of?

[hochopeper]

At the moment acg and I are working through details for a linear psu for part/all of his PC so I've done some research recently. Since the OP was asking a more general question and thread is not specific to the John S Mac mini supply I'll add a few comments of my own.

 

I have done almost no research on smps as my supplies for acg will also be used for a few other applications where smps just isn't an option, my comments are focused on linear supplies.

 

Regarding kelvin sense wiring, not sure if it has been mentioned already but the 3.3V rail in ATX standard has option for a 3 terminal version of this config. Some fixed voltage regulator ICs offer this and some of the newer regulator ICs for wide bandwidth and low noise offer this and adjustible output too, though these are mostly limited to 1A. The fixed voltage ICs can be found that offer 3.3V at up to 3A, though become rarer for higher voltages at this sort of current (I havent found any IC 'chip regulators' that do more than 1A and offer this with higher than 3.3V output).

 

Traditionally kelvin sense seems to be used for voltage drop over cable for what would seem to be low freq variations in voltage at the load terminals. The presence of this functionality in wide bandwidth focused ICs does make me wonder about the feedback loop and it's effect on higher freq noise (provided the wiring allows this to be carried reliably). I have no real data to back that claim / idea but it's something I'd like to look into and why I'm building (diy) measurement preamp and measurement focused ADC(s) to better understand the interaction.

 

In terms of filtering before the regulator I will be using a snubber across the bridge rectifier and a pre-reg before each of the final low noise IC regulator. Noise for higher than 100kHz would seem to be dominated by sources on board the PC rather than the power supply so IC regulator PSRR is mostly important for lower freq.

I hope that is more helpful than confusing!

 

Chris

[Superdad]

Nice post Chris. Do keep us apprised of your project.

 

 

Jtwrace: To answer your questions, the supply we will be offering has two separately regulated outputs, independently selectable for 5V,7V,9V,12V.

Only one of the two includes the pseudo-Kelvin-sense circuit. There is a slide switch next to the SMA jack (see pics upthread). When you slide the switch towards the jack, the supply then ignores the output voltage setting (for that output) and gets its voltage set by the value of the pair of resistors at the far end of the thin coax cable (in the Mac mini example, it's 12V set by resistors on our internal adaptor board; we plan on offering little cube modules that can range anywhere between 5V and 12V).

 

The max current is limited by two factors: The 5 amp rating of the choke filter we use, and also the rating and thermal dissipation of the regulators (and heatsinks) we are using. We have not finished stress testing a unit (didn't want to blow up the first one before getting to enjoy it!), but I think we will be quoting around 5A@12V, on down to something less @5V. (Larger the drop, the more the heat.) And while yes, that will be total for the supply (only one set of diodes, one 5A inductor, and 66,000uF caps feeding both regulators), I think it will still be a robust unit capable of handling instantaneous loads.

 

The idea is for a user to be able to power a pair of devices--say a Mac mini or CAPs, etc., and also a small DAC, hard drive, headphone amp, whatever. John's choke-based design (and careful balancing of cap values is part of that) is very clean, but it is a challenge to scale it into a higher-current arrangement. The chokes get very large, heavy, and expensive, and we would likely have to leave behind the regulators we like. So I think that overall we are sticking a useful balance as small computers really don't draw very much at all (so far the heatsinks and regs have not gotten hot).

 

Lastly, 120/240V operation is standard. We are using one of those nice Corcom combo power entry modules (IEC/switch/fuse) where you just flip over the fuse tray to go 240V or back.

[all300b]

Following this thread with interest - would like to try a linear power supply but am also put off by their relatively high cost.

 

From my very linear point of view (no pun intended) - the "noisy" switching power supply of the computer could theoretically harm the sound by injecting noise through the power leg of the USB, injecting noise back into the AC line - and then the other components - or both. (I'm just not ready to believe that the there's an effect on the bulk transfer of data within the computer itself).

 

By this logic should addition of an isolation transformer at the AC side plus something like an iFi USB power generator be equivalent? The linear power supply would be a more elegant one step solution but maybe more expensive.

[Boris75]

Come on folks, must EVERY thread end up at these places? This one is just barely 7 posts in...

 

Your response to Eloise, which I find quite surprising (why flame Eloise when she asked a simple factual question?), suggests that the test was not blind, or was it?

[Superdad]

Your response to Eloise, which I find quite surprising (why flame Eloise when she asked a simple factual question?), suggests that the test was not blind, or was it?

 

Nah, I don't think I was "flaming" Eloise or SandyK (who I simultaneously chided from the other end). But from the way the question was phrased, it seemed clear that the answering that it was not a "blind" test would quickly lead to the implication that what we heard was not valid. That debate has been hashed out a hundred times before, and I thought people would be more interested in talking about power supplies. Again, I was just as uninterested in starting to make comparison rips of CDs with and without the better power supply since that leads into SandyK's realm, and the endless debate of checksum-identical files sounding different.

 

I happen to like and respect both of them, and I enjoy learning from everyone here. I did not mean my gentle chiding to seem uncordial.

 

As for the answer to the listening test method: These differences were large enough--and my room/system/ears good enough--such that "blind" testing was entirely unnecessary. Yet, as I have stated in all my reports on CA, for all the thousands of items (s/w and h/w) that I have tweaked in my system (and over the years being involved in design and manufacture of amps/preamps/cables), I would be happy to submit to any blind test--so long as it is in my room, with my system, and my music.

But to those folks (NOT pointing to Eloise) for whom such methods are the only valid tests, I know that a sample of one or two people getting 98% accuracy is still not acceptable proof of anything. (I can point you to THAT thread where Dennis and others explicitly state so.)

 

Cheers,

ALEX

[esldude]

But to those folks (NOT pointing to Eloise) for whom such methods are the only valid tests, I know that a sample of one or two people getting 98% accuracy is still not acceptable proof of anything. (I can point you to THAT thread where Dennis and others explicitly state so.)

 

Cheers,

ALEX

 

Superdad, If you are going to make claims for me, make them accurately. I didn't say 98% accuracy. I said 3 sigma significant results and with enough samples. Further I would consider with enough samples even one person doing so as showing it possible.

 

Also for all you knew, Eloise wanted to know so she could judge the importance of how you were testing according to her opinion, and would have said little or nothing more.

 

Now we do know sandyk hasn't stopped there. Further he even admitted he only posted to complicate matters related to Eloise's post ("shot across her bows" were his words).

 

I also said I had similar questions. That I didn't think blind testing was done as it is such bother. I did wonder about measurements, but didn't ask the question because I knew this was prototype stuff, early in the process etc. etc. I mentioned this, but still haven't asked that question of you or John as I knew it was too early in the process.. Though he responded my pondering somewhat anyway. In other words my unasked questions were about the design process as were the question Eloise wrote. SandyK on the other hand had no such question he only wanted to stir things up.

[sandyk]

SandyK on the other hand had no such question he only wanted to stir things up.

 

And your purpose in posting further in this thread is ???

In case you haven't noticed, my original remark was 86 posts back.

You already made your point 44 posts back.

[esldude]

And your purpose in posting further in this thread is ???

In case you haven't noticed, my original remark was 86 posts back.

You already made your point 44 posts back.

 

In case you have reading comprehension issues, my purpose was my moniker was used with an inaccurate attribution to me I wanted to correct. What's your excuse?

[tne]

Back to the LPS / Mac Mini issue -

As part of the LH Geek Pulse campaign, I purchased the LPS perk. Mainly I was hoping for the cleanest and most stable possible power for the Geek Pulse through the USB port, but also since it has a separate 12V outlet.

 

Here is my question for the CA LPS and Mac Mini gurus - if I remove the internal switching PS from the Mac Mini, I assume that I can hook up the Mac Mini to the LPS from LH. In this case both the Pulse (through 5V USB power leg) and the Mac Mini would then be powered by the single LH LPS system.

 

Are there any problems with this that I am not currently seeing ?

 

I appreciate your learned input.

[Superdad]

Back to the LPS / Mac Mini issue -

As part of the LH Geek Pulse campaign, I purchased the LPS perk. Mainly I was hoping for the cleanest and most stable possible power for the Geek Pulse through the USB port, but also since it has a separate 12V outlet.

 

Here is my question for the CA LPS and Mac Mini gurus - if I remove the internal switching PS from the Mac Mini, I assume that I can hook up the Mac Mini to the LPS from LH. In this case both the Pulse (through 5V USB power leg) and the Mac Mini would then be powered by the single LH LPS system.

 

Are there any problems with this that I am not currently seeing ?

 

The 12V output of the Geek LPS is rated to only 0.8 amps. That is far too low a current capability to run any of the 12V Mac minis on--or most any computer music server. The LiteOn $35 internal switching supply from the 2010/2011/2012 Mac mini is rated at 7.1 amps, but that is much more than the machine needs for duty as a music server. It think Apple went that big because they could within the form factor and because the fastest quad-core i7 model running some heavy-duty graphics rendering might get the machine to draw a bit more during 100% CPU utilization and they want to err on the conservative side.

 

I just put an ammeter in line with the DC feed from our prototype linear PS (around 5A@12V) to my 2010 2.4GHz Core 2 Duo Mac mini. At idle it draws just over 0.5 amps, while during start-up and operations such as Audirvana Plus upsampling CD tracks to 176.4KHz it draws a little over 2 amps (after A+ finishes converting and loading the track, demand drops down to about 0.8A.

 

So while a 2 amp supply technically could take care of running a mini (and the Geek LPS is much smaller than that), units that small usually don't have very ample storage capacitance for instantaneous demand and can result in a somewhat lightweight sound--usually with with a DAC or preamp, but oddly robustness seems to impact the computer a bit as well.

 

Hope the above info is helpful.

 

Best,

ALEX C.

[tne]

The 12V output of the Geek LPS is rated to only 0.8 amps. That is far too low a current capability to run any of the 12V Mac minis on--or most any computer music server. The LiteOn $35 internal switching supply from the 2010/2011/2012 Mac mini is rated at 7.1 amps, but that is much more than the machine needs for duty as a music server. It think Apple went that big because they could within the form factor and because the fastest quad-core i7 model running some heavy-duty graphics rendering might get the machine to draw a bit more during 100% CPU utilization and they want to err on the conservative side.

 

I just put an ammeter in line with the DC feed from our prototype linear PS (around 5A@12V) to my 2010 2.4GHz Core 2 Duo Mac mini. At idle it draws just over 0.5 amps, while during start-up and operations such as Audirvana Plus upsampling CD tracks to 176.4KHz it draws a little over 2 amps (after A+ finishes converting and loading the track, demand drops down to about 0.8A.

 

So while a 2 amp supply technically could take care of running a mini (and the Geek LPS is much smaller than that), units that small usually don't have very ample storage capacitance for instantaneous demand and can result in a somewhat lightweight sound--usually with with a DAC or preamp, but oddly robustness seems to impact the computer a bit as well.

 

Hope the above info is helpful.

 

Best,

ALEX C.

 

Thanks. Extremely helpful. I will explore other options.

[JJJ]

Forgive me if this sounds a little trite, especially from a newbie to electronics, but I thought a load-sensing regulator is very much "standard fare" in the super regulator game. Jan Didden and Walt Jung et al have made their designs public and their regulator boards are available on DIYAudio. Am I missing something in relation to the design element that's being introduced here?

 

I'm very keen for this discussion to be focused on the PSU design and that it continue. The commercial aspects of the product that may/will be for sale can be presented in other parts of the forum.

 

Does the external enclosure build structure and the resultant long umbilical pose challenges for a load sensing design?