The true experimental tweak thread

[JohnSwenson]

If I cannot find anything simple and useful for data isolation I think I will order the USBISO-AA to give it a try USB to RS485 converter shop

 

[ATTACH]32745[/ATTACH]

 

It works at high speed.

 

Since I am just after the data isolation this might be a valuable option for EUR60 +VAT & freight.

 

Sent from my Nexus 7 using Computer Audiophile mobile app

 

This device is NOT high speed, in the data sheet it specifically says it is a Full speed only device.

 

There are a bunch of these isolators on the market, they all use the same Analog Devices chip which has been around for many years. Several of these companies say they are "USB 2.0" trying to make you think they work with high speed when in fact they don't. Strictly speaking it is true since the USB 2.0 spec includes low speed and full speed. Many people are a bit confused about this and think that USB 2.0 means high speed, high speed is part of USB 2.0, but so is low and full speed as well.

 

On the other device up-thread, it almost certainly is the same chip and the transformer symbol on the case is just to give the idea of galvanic isolation not that it actually uses a transformer.

 

The USB bus has to be bidirectional even after enumeration, even though we think of the audio protocol as one direction, it is not. Every so often the host asks the DAC if it is still there, the DAC then responds with a packet letting the host know it is alive. If the host doesn't get this it will drop the connection. So you still need a bidirectional isolation even while the music is playing.

 

The only way to know when to switch the direction of the isolator is to actually follow everything that is going on bus, ie a USB protocol engine. This is not trivial. Getting it right is NOT easy and has been one of the reasons such devices have been very scarce. The Intona is one of the few devices that has actually done this. And they still didn't get it right for audio at first.

 

I'm not trying to disparage you from trying things, but this is just not something you are going to be able to do without a LOT more knowledge of USB and electronics.

 

It is HARD.

 

John S.

[JohnSwenson]

Hmmm, I am not so sure that it will work amymore. My latest four attempts incl. this one...

 

[ATTACH]33579[/ATTACH]

 

...which is based on a LAN/HUB switch (pin1&2 to 3&6 plus 3&6 to 1&2) configuration did'nt work either. I think it is time to re-group and spend some Google time to fully understand the opto-coupler pins/legs before making a final attempt.

 

You are never going to get this to work for ethernet, you need to understand the ethernet signaling scheme and bandwidth issues. The optocouplers you are using will not do this directly.

 

It certainly can be done, but it is NOT trivial, it will take a significantly more complex circuit than just a a pair of off the shelf optocouplers.

 

Some tdbits:

 

the symbol rate for 100Mb is 125 Mega Symbols per Second. Your isolators have to be at least this fast. 100Mb uses three different voltages, NOT the normal two of digital circuits. It has a high, medium and low. Your isolator circuit has to distinguish the three values on the input side and output the three values on the output side. No off the shelf ones do this. You have to figure out how to make this work yourself. (maybe some comparitors ond multiple isolators).

 

Another way is to use two RMII PHYs with isolators between them. The signals between them are standard logic level signals so off the shelf isolators can be used (you need at least 50MHz). This is probably the simplest most straight forward method, but will require a high quality PCB and soldering very small parts onto the board.

 

Please don't even try this on gigabit, it is way more complicated than 100Mbit. The symbol rates are actually the same, but it uses 5 different voltage levels, 4 pairs, and each par can be going in BOTH directions at the SAME time. Getting that right is NOT trivial!! In comparison 100Mbit is two pairs, one for each direction and only three values, this is much easier to deal with (still not trivial but a LOT easier than gigabit).

 

If you really want to pursue this I recommend looking up how ethernet signaling works (the actual voltage levels and timing) and or look up how MII and RMII PHY interfaces work. Then go to one of the big distributors (DigiKey, Farnell etc) and find a 10/100 PHY that looks like it won't be too hard to design a board for and read the datasheet and see if you want to take on the task of designing and laying out a board for it.

 

John S.

[JohnSwenson]

35 minutes ago, lmitche said:

A DAC interfaced with Quantum entanglement would be the ultimate in galvanic isolation.

 

35 minutes ago, lmitche said:

A DAC interfaced with Quantum entanglement would be the ultimate in galvanic isolation.

And wireless to anywhere in the universe, and completely un-hackable. Just a few small matters of engineering to work out.

 

John S.

[JohnSwenson]

On 3/26/2017 at 1:10 PM, Cornan said:

Preparing for my 10 day China trip. I'll need atleast a little something except my portable headphone setup to keep me satisfied in the hotel room after work hours! ?

 

Hi Cornan,

we sold our house and so I am on the road for three months now. Before everything got packed up I did some experimentation to pick the "on the go" system. What I wound up with is a Squeezebox Touch powered by an LPS-1 with Senheiser PX-100 headphones plugged in directly to the SBT. LMS is running on a Fitlet with a 500G mSATA drive. The Fitlet is plugged into a TPLink travel router for a local network.

 

The system sounds amazing. I've been listening to it this morning and really enjoying it. The LPS-1 does amazing things to the SBT, It sounds like something that should cost $5K rather than something that cost $200.

 

Well back to listening!

 

John S.

[JohnSwenson]

3 hours ago, Cornan said:

This little bugger have annoyed me for 6 hours now! ?

 

 

The TP-Link TL-RW902AC have been anything but cooperative! Set to client-mode it hi-jacks my router and completely ruin my whole WiFi connection, giving me hell to get everything working again. For the first time in a very long time I'm forced to give up. I will put it in a drawer for sometime and hopefully it will behave when I'll take it out next time. I'll need a brake. My family need a brake! ?

It took me roughly 1,5 hours to get everything in the same working order after disconnecting it. I hate network problems! ?

Interesting, I have a TL-RW802N, it works extremely well, I'm typing through it right now. I have mine set in hotspot mode, it is a client to the apartments wifi, and then generates its own LAN and wifi network.

 

Are you trying to use it as a wifi to wired Ethernet bridge? I haven't used that mode yet, but I could try it out that way and see how to make it work.

 

It sounds like you have the internal DHCP server turned on.

 

What I found out with mine is to make sure that you do configuration over the hard Ethernet port, half the time I tried to do configuration over wifi I would not get many of the configuration options showing up on the web page. By going hardwired I could then see all the options and get things setup correctly.

 

Let me know and I can go over this with you.

 

John S.

[JohnSwenson]

4 hours ago, Cornan said:

 

Thanks a lot John! Very helpful! ?

I did all my settings using wireless. I couldn't even find my 5Ghz ethernet channel in the beginning. I changed my main router channel from 108 to 60 without luck, but when I changed it again to channel 48 it was suddenly found and all hell broke loose! ?

I am actually a big fan of TP-Links ethernet products and have a wireless extender (RE450) that have a rock solid wifi connection, never failed me and runs like clock works. 

When my family forgive me for this incident I will make a new setup attempt with a hard ethernet connection to my router and try the hotspot mode instead. I did'nt know it was possible to use LAN in that mode (and no info about it in the manual either) sothat was very useful info.

 

I am looking forward to give this 5v pocket router working so I can start to MOD my banana DC terminal for my low budget setup into a starquad solution and power ALL the DC powered devices (Chromecast audio, USB ethernet adapter, network switch, DAC and pocket router) with a floating bench supply (SMPS). The active speakers are powered through a Ultra isolation transformer.

 

/Micael

 

 

One other thing, the setup address is 192.168.0.1 period, no matter what DHCP is telling the computer. So what I do is make sure I turn off wifi on the computer (so it doesn't get confused), set a static IP address for the computer to something like 192.168.0.13, then plug into the wired port on the router. You can then always connect to 192.168.0.1 no matter WHAT the router is already doing.

 

It's then easy to configure it for whatever you want. In many cases you will need to power down and power up the router to get the new configuration to take affect.

 

John S.

[JohnSwenson]

8 hours ago, Cornan said:

BTW. Due to my new setup with Uptone hard adapter between Regen Amber and Brooklyn DAC I ordered a USPCB the other day. It will be fun to see if it will improve things further. Ofcource I am planning for a ISORegen as well asap, but I will 100% power it by the dual floating Gophert. 

 

 

Another thing I have been wondering about is if it would be possible to add a world clock input to the Regen Amber? In my mind I would then connect it to the world clock output of the Brooklyn. Would'nt that give optimal SI? ?

Hi Cornan,

there is no word clock in a USB hub which is what a USB REGEN is. A word clock is a clock at the sample rate of the audio stream. A USB hub has no concept of audio sample rate, there is an oscillator that goes through some PLLS in the hub to generate the USB clocks, but they have no relationship whatsoever to audio sample rates.

 

Now if you want to hack up a USB REGEN, you can replace the 24MHZ oscillator with a 24MHZ Crystek 575, but it is a lot easier to just get an ISO REGEN, which already uses the 575 (plus the isolator and other things that make it significantly better than the original REGEN).

 

John S.

[JohnSwenson]

On 5/28/2018 at 4:52 PM, Hauser said:

The result seems to make sense.  I'm surprised you can run the LPS1.2 on 8V HPULN though, 24W is not enough.

 

Martin.

The 36W is the maximum power when running at 12V output and greater than 0.5A output current. As the output voltage goes down the maximum input power also goes down. In addition there are two charging currents, one used for less than 0.5A and one for greater than 0.5A.

 

So under some circumstances (lower output voltage, lower output current) you can get away with a lower wattage feeder supply. Unfortunately there is no simple formula to compute how much it is going to take for a particular condition, it's a very complex situation, so we just recommend that people use a feeder supply that can handle the maximum and you will never have a problem.

 

John S.

 

[JohnSwenson]

On 4/30/2019 at 4:07 PM, Cornan said:

 

I was only explaining why I personally did’nt choose to buy the Hammond choke at the time. As far as I knew then and still understand it is DC filtrering.

Remember the original post to use the Hammond choke as a power conditioner tweak.

The choke is NOT just a simple filter. It is part of a complex tuned circuit containing the transformer impedance, the value of the two caps and inductor and the parasitics of all of them. The primary purpose is to recharge the final cap as quickly as possible. That particular circuit WITH the choke recharges the final cap 50 times faster than the circuit without the choke. When driving digital devices which contain lots of rapidly changing load current events this can be VERY important!

 

It does other things as well: the load regulation is much better, The ripple is MUCH lower, and the diode conduction is almost the complete cycle, all very good things, but recharge time is the biggy.

 

There is no way to come up with a "one size fits all" design of this. Every set of transformer, voltage and current has to be tweaked independantly. Since appropriate inductors are VERY few and far between, the inductor is chosen first (to match the output current requirement) then everything else is adjusted to get the best results with that inductor. I spend many hours with spice iteratively working out the design, then build it, measure and do some more tweaking.

 

John S.

[JohnSwenson]

6 hours ago, Cornan said:

 

Thanks for the detailed explanation John! So, did I understand you right that the Hammond choke is not suitable as a power conditioner (one size fits all) as the original poster said? Rather a good option for a PSU powering digital sources?

 

 

Just randomly putting the choke in someplace will frequently make things worse. To get the above advantages it has to be very carefully designed INTO the power supply, not tacked on as a "tweak".

 

I originally did all this PS design stuff for power amps where it excels. It also works very well for digital stuff.

 

John S.