UpTone Audio REGEN Power Supply Add-On

[JohnSwenson]

The development prototype whateveritis is up and running on my bench tonight, working fantastically, measured performance is even better than anticipated. Haven't had time to listen to anything powered by it yet. (slight proud papa syndrome here)

 

I have a big smile on my face right now.

 

John S.

[JohnSwenson]

The biggest problem now is measuring the output, my $13K scope isn't good enough! I'm going to have to build some of my own test equipment to properly see it. (and some special faraday cages to put things in, testing in these realms is tough)

 

John S.

[JohnSwenson]

Will this mystery PS Add-on have a 5v 2-3A output to power a SoC device for NAA?

 

The current crop of 5v LPS are 3+ times the cost of a SoC board.

 

No, it is 5V/7V at 1A max.

 

Which boards are you talking about that take 3A at 5V? I've used several that run at 1A or less. What usually takes the power is a display or hard drive run off the board. If you don't use a display of any type (ie headless) and use Ethernet for storage rather than a board powered drive you should get many boards to run off of 1A.

 

John S.

[JohnSwenson]

It should, but we honestly don't know yet. The boards are still on John's bench and have not made it to the listening room yet.

 

Of course there is a big difference between a little 1A 5/7V supply and a big, flexible, choke-filterer, dual-output, 5-6A linear PS.

 

Scaling up our new tech to build a unit with the current and range of the JS-2 won't be cheap. John and I sketched out one and it ended up being a 2-chassis, $3,000 monster! (Okay, that one would have 4 outputs and be able to do 10 amps, but a wide-range 6A unit would not be much less.) Will be a fun demonstration of the tech and a series of new techniques, but for a practical range of new PS units we will have to reconsider many things.

Truth is, as radical as the new piece will be--and the implications of John's circuits even for audio power supplies within components--I don't want to become just a power supply company. So we are looking at licensing out and/or producing boards (for OEM or DIY) containing the core "brains" of it, allowing others to customize to large or small scale around it. And these things are going to be awesome for bipolar, +/- supplies inside gear.

 

Sorry for the long answer to a simple question. And for having to continue to dance around what the thing actually is. We'll get there…

 

To be honest, it was actually way more than a JS-2, the JS-2 has 5A spread amongst two outputs, so say 3 on one and 2 on the other. The monster we were talking about was 10A per output all at the same time! So that was 40A total. From 5V to 19V.

 

John S.

[JohnSwenson]

Correct,

the whateveritis takes DC in the range of 7.5 to 12V and outputs either 5 or 7V. So you need something else to feed it DC.

 

We are also working on a "feed supply" which takes AC in and outputs 12V to drive the whateveritis. It is NOT a regular linear supply, but I can't tell you exactly what it is. It is optimized to feed the whateveritis, so it may not be terribly useful as a general purpose linear supply. I am working hard at optimizing both devices to keep the cost as low as possible without sacrificing performance. And in addition I have to choose parts that have a good supply chain, one of the parts that was perfect for the design only had 45 available in the world, I had to go with a different part that had many thousands available.

 

I spent the last two days reading over 300 data sheets trying to find a part that would work technically, was inexpensive and had thousands in stock. It wasn't easy but I finally found one that fits the bill. And there are hundreds of parts in this, it has been a LONG process!

 

John S.

[JohnSwenson]

And I am sorry that I got involved. Even in attempting to answer what would normally be easy questions, I fouled it up!

 

There is nothing "normal" about any of this, all the "normal" rules about power supply stuff go out the window. Once Alex gives the go ahead to really say what it is, there is a LOT of technical explanations that have to go on so people can get an idea of how this thing scales and what happens under different scenarios.

 

I will be writing detailed descriptions of what it is and how it works. We won't be keeping this a technical secret like some other companies have.

 

John S.

[JohnSwenson]

It's just amazing how much discussion is being generated around a yet to be prototyped (I believe) device. I have new found respect for the secrecy around most audio gear development. Hats off to Alex for engaging at this early stage!

 

 

Sent from my iPhone using Tapatalk

 

There have been two development prototype boards built so far, I'm in the middle of laying out number three.

 

Once I can actually talk about it I can write up a paper on the development process if anyone is interested.

 

These boards have proven the basic architecture (its actually far exceeding my expectations), but we have run into some parts issues (stuff not working the way the data sheet says they work) and we have found areas that we can make even better, so some detailed implementation is being improved.

 

John S.

[JohnSwenson]

Does that really mean it is going to dump circa 13W of energy as heat then?

Better get some sausages ready to barbecue - bring on the beer - I can't wait.

 

 

Sent from my iPad using Tapatalk

 

I cannot go into details, please don't ask. But remember you cannot assume ANYTHING about this design based on what you know from other supplies.

 

John S.

[JohnSwenson]

So Alex, now you're using an FPGA rather than a CPLD?

 

Yes, I switched from a CPLD to an FPGA, we worked out a "unified control structure" which will work for every power supply variant we could think of, and that needed more horsepower than the low end CPLDs (the FPGA chosen is cheaper than all but the cheapest CPLDs).

 

With the new control scheme there is one control design that will be used for ALL supplies using this architecture, it is a little overkill for our first simple version, but it means I only have to design and layout one controller, period. It makes new versions much easier to design and hence much quicker to market.

 

The controller is doing a lot of different things. It monitors the system, output current, output voltage, input voltage and temperature, it does "the right thing" when any of these parameters are outside of spec. It also deals with power on sequencing and what happens when the power goes off for two seconds.

 

There is also the user interface for selecting parameters, such as voltage and current limit. The first version will have the simplest UI, a switch to select 5V or 7V. But other implementations are going to have significantly more flexibility, we needed a way to get this information into the supply. The new controller design standardizes this.

 

Well that is about all I can say right now.

 

John S.

[JohnSwenson]

There have been some posts about REGEN current consumption recently, I have been rounding off to saying 50mA, I just did the measurement and it is 44.9mA when the DAC is not drawing any current and music is playing. If the REGEN is not connected to a host it draws 2.7mA.

 

So for rough power calculations you can use 50mA as the current draw of a REGEN without the DAC pulling any current. The DAC draw adds to this.

 

John S.

[JohnSwenson]

Hmmm. I've got a 16v linear PS from China hanging about (lets assume it's regulated, haven't checked yet). My Vega doesn't draw power from the USB but I don't know what sort of current it draws through the Regen. Lets assume 110mA but obviously I have to check that.

 

Using Miguel's equation above I get total dissipation of 1.85W. Sounds too much if 1.46 is ok ?

 

Please don't use 16V with the REGEN, the input capacitor is rated at 16V, feeding it 16V continuously will eventually cause it to fail.

 

The specified maximum voltage is 12V. As long as the DAC is not drawing very much current 12V is fine to use with the REGEN.

 

John S.

[JohnSwenson]

Understood, hence my question to John above.

 

However in quick, thoroughly unscientific tests I don't quite hear an improvement running at 9V vs 7V. This is with the same supply, the JS-2, which I can set to either voltage very easily. It does run slightly warmer at 9V.

 

I don't believe in blanket statements either. Hotter is better is nonesense.

 

I ran a whole bunch of tests many years ago and found that most digital circuits involved in digital audio have lower jitter when the chips are hotter. Most of the chips are made with a technology that gets slower as they heat up. This causes the current spikes on the power and ground traces to be less, which leads to lower noise, which leads to lower jitter in the circuit.

 

The slower chips also slow down the ramp time of the USB signals themselves, which can be detrimental to the sound. Whether the decreased jitter is going to be more important than the decreased ramp time is going to be very DAC dependent. Meaning higher voltage may sound better for some and no difference for some and even worse for others.

 

REALLY hot parts lead to drastically reduced lifetime, so it is a balancing act between lower jitter and how long the device will last.

 

So it comes down to "for some it will sound better and for others it won't", kind of like just about everything in audio.

 

John S.

[JohnSwenson]

All known DACs work at full speed or high speed mode. These are supported by both USB2.0 and 3.0. 3.0 adds the super speed mode.

 

The iFi device can supposedly talk high speed to a DAC and convert to super speed to talk to the computer. Nobody really knows whether this is an advantage or not. You force high speed by using a USB2.0 cable between the computer and iFi. IF the port is 3.0 and you use a 3.0 cable the iFi unit should talk super speed to the computer. This test in and of itself is a very interesting test to perform for those that have the iFi unit.

 

Of course if a REGEN is feeding the iFi it will be talking at high speed mode since the REGEN does not support super speed.

 

John S.

[JohnSwenson]

Btw, I do enjoy your music reviews in Stereophile.

 

That is a different John Swenson. I'm the engineer, audio designer.

 

John S.

[JohnSwenson]

Have got this right?

You can’t use USB3 for audio since there are not USB3 audio protocols.

 

You can use USB3 cables/connector for audio, but you are only using the UBS2 part.

 

So: USB3 is not actually relevant for audio (yet), but it’s ok to use the USB3 hardware.

 

OK, let's try and clear this up (again).

 

USB1 has two speeds, slow speed and full speed.

USB2 has three speeds: slow speed, full speed and high speed

USB3 has four speed: slow speed, full speed, high speed and super speed

 

Thus a USB3 implementation can support ALL the 4 speeds. Thus a DAC that uses high speed will work fine plugged into a USB3 jack on a computer.

 

There are some mechanical incompatibilities, super speed uses extra wires, which need extra contacts in the connectors. On the A type connectors they figured out how to add the extra contacts in such a way that a USB2 cable would still fit and connect to the 2.0 contacts and ignore those for super speed. Thus you can plug a 1, 2,or 3 type a plug into a USB3 jack and it will work correctly.

 

Type B on the other hand could NOT have the extra pins added into the existing connector. They had to add an extra shell and contacts to it. The upshot is that with a device that has a 3.0 B plug, you can plug in a 1, 2, or 3 B plug. BUT if you have a device that has a usb 2.0 B jack you cannot plug in a USB3.0 B plug.

 

There have been a few compatibility issues with some USB3 implementations, particularly noticed on some Macs, that a few high speed DACs will not work on some Mac USB3 jacks, this has turned out to be some timing changes, the outputs on the Macs subtly changed their timing from previous generations (but still within spec) and some device implementations couldn't handle it. This is not a problem with the spec, but an implementation issue with some USB chips. It had not been found previously because nothing did the timing exactly this way before.

 

John S.

[JohnSwenson]

Thank you for the clarification. What I was getting at is that there nothing “extra” with USB3 from an audio perspective. It’s the exact same as USB2 – the extra speed (and wires) are not used.

 

Correct, there are no super speed DACs that I know of, so the fact that a hub will not directly affect the DAC, but there MIGHT be some impact because if you have things setup right the connection to the computer might be super speed, I have no idea whether this is a good/bad/indifferent thing.

 

John S.

[JohnSwenson]

Alex and John, can we start the new year with an update on the status of this device and your latest outlook on production timeframe? Thanks.

 

Before Christmas I received the boards from the board house for the New Architecture (NA) version. I got all the parts in and assembled them. I found that one of the parts I had chosen didn't work properly in the NA, I had to find another one that looked like it would work, design a new board, get that built, solder all the parts in. Found out the new part didn't work like it said it did in the spec sheet. Fortunately I was able to hack up the board to get it to work.

 

While waiting for all this I was writing the FPGA code for the controller. I got that all at least to compile. Before Christmas I got one chunk of code working. Then after I got back from the trip to Texas I got the other other chunk working, and plugged everything all together and got nothing. It turns out a very critical part was fried, it is VERY sensitive to soldering temperature, my method of soldering the boards was heating them up too hot frying this part. So now I now I'm waiting for some more parts to arrive from Mouser. I have worked out a way to solder this one part without overheating it. Fortunately the production assembly process can achieve better temperature control than I can. So I don't think there will be any problem from this in production. We are going to test it VERY carefully.

 

In order to make things easier to design, build and debug this R&D version was built on three boards with a bunch of cables running between them. The real production board is going to be on one board, but it will be almost impossible to debug so I have to get everything working before going to the production board.

 

So once the R&D version is working perfectly I can work on the production board. I am already about 40% of the way done with that.

 

With the switch to NA and the almost total re-design things are behind where I wanted them to be. So hitting February may not happen. But I am working very hard on this and making very good progress, it does not seem like it will be too much longer.

 

(It was fun saying that without letting anybody know what it really is! Sorry no pictures at this time)

 

We have decided on the case and Alex is working on the machining and marking for that so it is all ready to go when we get the circuitry working.

 

I think that is about it.

 

John S.

[JohnSwenson]

I just wanted to add that the cable Alex is going to be including with the product is a REALLY nice cable, it is 16 gauge coax, MUCH better than you get with practically anything else. That very special star-quad he is offering IS better, but not by a huge amount. The one everybody gets in the box is a big step above the average. The special one is REALLY special. Just remember you are getting a top notch cable to begin with.

 

John S.

[JohnSwenson]

Squeezebox touch- Rodney Dangerfield of streamers.......

 

Any truth to the rumor that John Swenson uses a

sbt? If so what modifications have been made?

 

Thoughts on whether a sbt/edo->regen, both

powered with UpTone mystery LPSs just might

embarass some high price streamers?

 

pb-

 

Yep,

I have two SBTs, one in the listening room and one in the lab. In the listening room it goes through a REGEN (prototype amber) to a Bottlehead DAC (A DAC I designed). I use SqueezeCommander on an old Android phone for control. The REGEN and DAC are powered by a JS-2 and the SBT is powered by an early choke based PS of mine.

 

In the lab the SBT goes through a REGEN green prototype to a prototype Bottlehead DAC, both powered by a prototype JS-2. No cases, everything screwed to pieces of plywood with cables all over the place, it looks right at home in the lab.

 

I don't have any speakers in the lab, DAC out goes to a Total AirHead amp driving Senheiser PX-100 headphones (original NOT the II), with a $10 cable literally from Radio Shack. This $120 system can EASILY discern subtle details in DACs and cables, etc. So people that say you HAVE to have a megabuck system to hear the differences are wrong.

 

Neither of these SBTs have been modified. All the modifications I have done were to make the internal DAC sound better or improve the S/PDIF out. None of those have any affect on the USB out.

 

That's what it was last week, that is now changing. As some of you know I'm also the designer of the Sonore microRendu, this is now replacing the SBT/REGEN combos. The microRendu is simply an amazing device, when feeding the BH DAC it is like nothing I have ever heard before, major league goosebump time. I knew it was going to sound good, but I was not prepared for the level of transformation this brings.

 

The SBT in the lab is still going to remain, as a controller for the microRendu. The microRendu, has no UI at all, it is in the same case style as the REGEN but about twice as long, so you still need some form of control.

 

John S.

[JohnSwenson]

Will 9V/3.3A work?

 

9V at 3.3A is fine for one unit, but not for two. At 9V one unit takes 2A so 3.3A is not enough for two.

 

John S.

[JohnSwenson]

John

Does it actually take 2A for a 9V output, or is this ONLY when supplying the designed max. 1A at 9V ?

 

Kind Regards

Alex

 

Hi Alex,

yes and no, but I can't tell you what that means at this point, you have to know what the circuit does before I can explain what that means, and we are not ready for full disclosure yet.

 

BTW that max OUTPUT is 7V at 1A.

 

There is definitely power loss inside the unit (ie heat is generated) so there will always be quite a bit more power going in than coming out. I have done the best I can to cut down on that loss, but it will always be there.

 

John S.

[JohnSwenson]

It would be great to have a status update on your pre-production board, even if it's still too early to talk about the design and technology of your power supply.

 

The pre-production board is being layed out right now. I'm taking the lessons learned from the 4 previous boards to make this board very reliable and robust. I've managed to get rid of a lot of parts so it is easily fitting on the board size that fits in the case we chose. The last time I did an all-together-on-one-board it was very tight and took a lot of work to cram it all in, this is going much easier.

 

John S.

[JohnSwenson]

Just replying to get progress updates from this thread on whateveritis

 

I'm still working on the layout. It's getting close, probably done this week. Working on it right now. Well obviously reading forums RIGHT now, taking a break from parts placement optimization, this is where I stare at the screen for hours on end thinking "if I move this here I get shorter routes for this net but longer for that, then if I move this other thing, I could cut down on that net, then move this group over here that would free up room for this part, which makes that better.............."

 

It looks like nothing is happening, just staring at the screen, but it is actually the most important part for getting the placement just right so the routing goes easy.

 

John S.

[JohnSwenson]

I hope your pre-production board is soon in the hands of your PCB fab. It would be so nice if you could start talking about your technology at AXPONA.

 

Well the board is almost done, but I can't go to AXPONA, I'm chained to the chair in the lab, there is SOOO much to do, at least 4 different projects.

 

The second this board goes out I start working on another one...

 

John S.

[JohnSwenson]

Alex,

 

the more I think about the MPS, the more questions I have :-) Sorry, but I can't stop asking - you said earlier that its in essence 1A DC-DC LPS. There are devices that need more than 1A, like the PPA V3/V4 card and it requires 5V/3A according to Paul. So this means that the MPS cannot be used with these devices. Additionally, the microrendu says minimum of 1A, yet their PS is rated 2A. Would using the MPS with microrendu also be a stretch ? Can we make the MPS more versatile by increasing the current limit ?

 

On a different note, what is the use case for 3.3v output ?

 

My tests of the microRendu show it taking about 0.5A at startup and about 0.2A running, so the 1A "mystery" supply will work great with it.

 

Increasing the current of this particular design is not possible. But we are planning additional versions using this technology at much higher currents and voltages. But that also means that they will be larger and cost more money.

 

The 1A is the maximum I can get with the parts used in the first version. Anything larger is going to use considerably more expensive parts.

 

There are several people that wanted 3.3V and it was very easy to add and didn't cost anything extra so I put it in.

 

With "traditional" power supplies a higher than absolutely necessary current rating can significantly increase sound quality, the higher current transformer has significantly lower impedance which produces much faster recovery times which is a major factor in sound quality for digital audio equipment. I've gone into this in excruciating detail in other posts. So for traditional supplies a current rating much higher than seems necessary can result in better sound quality. This is NOT true for the "mystery" architecture. If you need an amp, a 1 amp version will sound just as good as a 2 amp version.

 

John S.