Discussion of LiFePO4 battery mods of REGEN and other out-of-warranty adventures

[Superdad]

Man you guys are all over the map! I am not sure JS is even going to want to wade into this mess.

[Gonzbull]

I find it quite amusing. Love the level of obsession. Often how great discoveries are made or not.

[Qhwoeprktiyns]

Yes, I expect the batteries will maintain a constant 3.2V (this voltage will not be detrimental to the Regen's USB hub chip or clock) over 6 hours powering a load which only requires max of 100mA or 0.03C current.

 

But, yes, there's probably no need to run solely off battery - just use the 3.3V regulated supply as recommended

 

I tried it and actually prefer the stock Regen fed by a high quality power supply.

[Qhwoeprktiyns]

I tried it and actually prefer the stock Regen fed by a high quality power supply.

 

And yes the regulator on the Regen are very good. I have in fact recycled my Regen green to use it as a 5 to 3.3v converter to feed an Audio gd I2S card.

[4est]

I imagine this to be true to an extent as I have previously noted. The TI part is current limited and has a thermal shut down for starters. If one were to supply 5v(or a bit less) in, it would then be dissipating 1.7 watts @ 1A. Remember, that same part is used for the 5v supply, and that section is spec'd for .5A with no heat sink. Somewhere JS gives some guidelines on dissipation levels IIRC. Using the stock supply @ 7.5V we are dissipating 1.25 W with .5A. Lowering that voltage to 4V (a 1 ohm 1W resister in series would do fine here) and we are at .7 W. Alternately, a 3.5ish ohm 5 W resister puts the stock Meanwell safely in play.

 

Once again, I have no skin in this game, and am not recommending anything...

You won't be able to obtain anywhere near that amount of current from the 3.3V regulator due to it's lack of a suitable heatsink. I wouldn't want to try and charge a fully , or almost fully discharged, 3.3V battery using this internal 3.3V regulator, but I may be a little too conservative here ?

Floating the 3.3V battery when it is already close to it's nominal output voltage is an entirely different matter.

 

Perhaps John Swenson could chime in here ?

[4est]

Is this what your asking about? The onboard Regen regulator seems to output 3.3V 1amp, should work well charging the LiFePO4 battery.

 

Yes, I also looked up that part for giggles. See my other post.

[4est]

I tried it and actually prefer the stock Regen fed by a high quality power supply.

 

Out of curiosity, can you describe that power supply? I can hear mild differences between transformer PSs and am looking for a reference.

[Qhwoeprktiyns]

Out of curiosity, can you describe that power supply? I can hear mild differences between transformer PSs and am looking for a reference.

 

I use a Daitron HFS30.

[zilch0md]

Man you guys are all over the map! I am not sure JS is even going to want to wade into this mess.

 

Alex,

 

I really admire your patience. You guys are so "accommodating" for lack of a better word. :-)

 

Again, I have no intention of trying the mod myself and have not recommended it, but one take-away that I find appealing is John Kenny's assertion that a battery pack can filter noise coming in from a PSU. I wouldn't blame John Swenson for staying well clear of the mod discussion, but I would really love to get his opinion on whether or not a continuously "trickle-charged" LiFePO4 battery pack can be as noise-free as a state-of-the-art voltage regulator.

 

Keep in mind that even though some people have suggested using a 3.3V SMPS, we could improve things by using a less efficient but quieter linear regulated power supply. I also think the larger the pack (mAH), the better it could perform as a "noise eater."

 

I have no interest in the mod, but John Kenny has opened up the possibility that we can force a battery pack to stay at a constant voltage of our choice - anywhere within that battery pack's minimum and maximum permissible voltages, instead of suffering the usual decay in voltage that occurs between charge cycles, just by "trickle-charging" it with a regulated supply at the desired voltage. That alone has a lot of applications, but it's only practical if the battery pack actually reduces noise. Otherwise, why put a battery between the regulated power supply and the load?

 

Thanks,

 

Mike

[4est]

I generally agree with you (especially about the accommodating part), but there is more to PSs than just noise. That battery has tons of current headroom for starters. I do not know much about them, but this has piqued my curiosity enough to order one up to try.

Alex,

 

I really admire your patience. You guys are so "accommodating" for lack of a better word. :-)

 

Again, I have no intention of trying the mod myself and have not recommended it, but one take-away that I find appealing is John Kenny's assertion that a battery pack can filter noise coming in from a PSU. I wouldn't blame John Swenson for staying well clear of the mod discussion, but I would really love to get his opinion on whether or not a continuously "trickle-charged" LiFePO4 battery pack can be as noise-free as a state-of-the-art voltage regulator.

 

Keep in mind that even though some people have suggested using a 3.3V SMPS, we could improve things by using a less efficient but quieter linear regulated power supply. I also think the larger the pack (mAH), the better it could perform as a "noise eater."

 

I have no interest in the mod, but John Kenny has opened up the possibility that we can force a battery pack to stay at a constant voltage of our choice - anywhere within that battery pack's minimum and maximum permissible voltages, instead of suffering the usual decay in voltage that occurs between charge cycles, just by "trickle-charging" it with a regulated supply at the desired voltage. That alone has a lot of applications, but it's only practical if the battery pack actually reduces noise. Otherwise, why put a battery between the regulated power supply and the load?

 

Thanks,

 

Mike

[fmak]

Buy a 5200 mAh 7.4v LiPO with a smart charger and an explosion proof bag (for charging),

connect it to the Regen, and you have loads of play hours. Play the dac thru the Regen into

an iFi usb power, and the sound is even better!

 

Try it.

 

Using a battery and connecting it with a mediocre charger to power the 3.3V input of the

usb chip in the Regen makes little immediate sense for the general audiophile. First you

are powering the chip with a high current device with no protection. Second, you are

assuming that a battery supply (with a charger) doesn't need bypassing with the

coupling. It seems that Jkeny doesn't even do it with his mainstream modified M2Tech dac

 

However, I shall this against a 3.3V superegulator when the bits come in. I must say though,

that the Regen has removed the last bits of digititis from my audio PC, a 'minimised'

Zotac CI320 and made the system better.

[JohnSwenson]

DO NOT under ANY circumstances use the internal 3.3V regulator to charge a lithium battery!!!!

 

 

These batteries have a very low impedance, hooking one up to the regulator in the REGEN will try and pull 50A or so out of it, which will current limit down to 1.2A or so. When it does this the voltage will be very low (there is a very low resistance path between the regulator and where the recommendation is to put the battery), which means almost the entire raw supply voltage will be converted to heat by the regulator. For a 7.5V input that is over 8 Watts. This will heat up the reg chip VERY fast, it will hit its thermal limit and drop the voltage even further. The reg chip is designed to live with this temperature for a short time, but it will fry if left in this state for long.

 

Even worse the whole board will quickly get into this temperature range and many of the other parts on the board are not designed to withstand this temperature, thy will fry within a few minutes of this condition.

 

So using the internal regulator as a charger for a lithium cell is almost guaranteed to well and truly fry the REGEN.

 

BTW the above scenario is NOT covered by the warrantee.

 

John S.

[Distinctive]

And even worse, wouldn't this result in a fire hazard?

[JohnSwenson]

And even worse, wouldn't this result in a fire hazard?

 

I don't think so. None of the parts will get hot enough to actually start combustion (unless you are doing this in a high pressure oxygen atmosphere).

 

The plastic will char and probably start smoking, but it is doubtful there will be actual flame.

 

John S.

[JohnSwenson]

On using a 3.6V charger with the HUB and clock chips.

 

This will probably work, BUT it places very high stress on the transistors inside the chips, which will radically shorten their lifetime.

 

So instead of a lifetime measured in decades, you get a lifetime measured in months. It is that long because the chips are done in an older technology which can withstand this for a longer period. With the newer chips it would be weeks.

 

So yes it will probably work, but not for long.

 

John S.

[fmak]

No, you keep missing the point.Some have already powered a higher quality Xtal Oscillator using a good linear PSU and reported improved results, but the present idea from Forrest has yet to be tested by C.A. members , as far as I am aware.

There is also the risk of damage to the Regen when modifications are performed by those without good SMD device soldering skills. Damage the Regen and you will have to get on a queue to purchase a new one.

The charger should only be connected as indicated by automatic voltage supervision , perhaps using an opto coupler for isolation, to operate a relay to connect the linear PSU as required. With the charger connected permanently via a Linear PSU there will be a small earth loop as previously mentioned by John Swenson, due to capacitive coupling via the Primary and Secondary windings of the transformer via AC mains Neutral which is connected to mains Earth.

 

There are loads of ill informed and downright ignorant posts being made about the subject in AA.

 

It is being said that a LiFePo4 battery, charged by a smps charger, provides the 'best' SQ and that

the battery acts as a 'capcitative reservoir' to provide low noise (my words to describe what I think is said).

 

Clearly the power to invent exceeds the power to understand what is going on clearly (when a battery's

chemicals are subjected to simultaneous discharging and charging).

 

I recently spent time in a dealer's showroom with 3 expensive LiFe usb to I2S/spdif units lying

around. This has a supervisory charging system and all of them have failed batteries.

 

So much for long life battery chemistry!

[nige2000]

There are loads of ill informed and downright ignorant posts being made about the subject in AA.

 

i saw

It is being said that a LiFePo4 battery, charged by a smps charger, provides the 'best' SQ and that

the battery acts as a 'capcitative reservoir' to provide low noise (my words to describe what I think is said).

think of it this way

you have an external 3.3v ps (it doesnt need to exceptional but a cheap smps might be pushing it)

route this through a switch onto your 3.3v battery(massive capacitor) through another switch on to bypass the onboard reg on the regen

Clearly the power to invent exceeds the power to understand what is going on clearly (when a battery's

chemicals are subjected to simultaneous discharging and charging).

this is how car alternators and alarm system backups work and they seem to be reliable enough?

I recently spent time in a dealer's showroom with 3 expensive LiFe usb to I2S/spdif units lying

around. This has a supervisory charging system and all of them have failed batteries.

 

So much for long life battery chemistry!

over draining cells is the most likely cause of failure

 

it only takes a couple of minutes to solder on a couple of wires and a battery to try

which is a lot less time than it takes at least me to put a single post up on a forum

 

the usb hub is clocking the usb signal so results are similar to clocking/reclocking i2s signals (like ians fifo reclocker or on usb to i2s converters), clocking pcie usb cards on audio pcs

Clock ps is of paramount importance and no reg ive come across can compare to a direct supply from lifepo4 but i have heard that a Paul Hynes is in a similar category after a bit of a-b testing

 

anyone going to try it

i would advise using a linear 3.3v supply with 2amp capability properly heatsinked

the onboard reg wont work for two reasons its position in the circuit will affect performance and it will probably overheat and fail as the cell could potentially draw well over 500ma

Do it at your own risk

you cant expect warranty if you tamper with it, however if you know the difference between + and - you should be ok

 

After ab testing regular regen with lifepo4 modded one performance doubled in regards resolution and noise floor

no-ones house burnt down there was no explosions etc

[pl_svn]

I recently spent time in a dealer's showroom with 3 expensive LiFe usb to I2S/spdif units lying

around. This has a supervisory charging system and all of them have failed batteries.

 

So much for long life battery chemistry!

 

been running this Audiophilleo and a RWA Signature 15 for close to 3 years now (RWA sold to a friend 1 year ago)

Both use LiFePO4 batteries and both are still running way more than fine

[Superdad]

Several members suggested that I move recent discussions of people modifying their REGENs (to power the hub chip directly--with a 3.2V lithium battery--bypassing the REGEN's already ultra-low-noise regulator) to a new thread so as not to confuse or divert the majority, who are non-DIY hackers.

 

So here are all the past few day's posts on the topic, moved from the thread http://www.computeraudiophile.com/f27-uptone-audio-sponsored/uptone-audio-regen-22803/ .

 

Cheers,

 

--Alex C.

[zilch0md]

Thanks Alex!

[Theobetley]

Seems like nobody wants to comment here after JS warned of the risks. If you are good at diy and you heed JS comments (don't use the onboard reg to trickle charge, keep the charge to a max of 3.3V) you can get a fantastic sq boost with this mod. It was as a great a mod as I've done in the past year or so.

[4est]

I am waiting for the battery. It will be awhile-I opted for the slow boat from China. Obviously I need to brush up on battery charging, but I will be trying this. To begin with I will simply use a power supply that does have a current limiter. I must admit I am confused by that TI part in the Regen.

Seems like nobody wants to comment here after JS warned of the risks. If you are good at diy and you heed JS comments (don't use the onboard reg to trickle charge, keep the charge to a max of 3.3V) you can get a fantastic sq boost with this mod. It was as a great a mod as I've done in the past year or so.

[tims]

From post 28

"Yes, that's a great idea as long as the USB Regen can perform well at voltages between mmerrill99's stated 3.0V and 3.65V (having bypassed its 3.3V regulator), and you somehow ensure that the battery pack never falls below mmerrill99's stated 3.0V minimum between charge cycles"

 

What happens if the battery's voltage drops below 3V - will it damage the Regen in some way or will the Regen simply stop working?

Also, will the LiFePO4 battery's life be compromised (reduced?) or damaged in some way If the voltage drops below 3V?

Thanks

[JohnSwenson]

Since it looks like there are several people here that really want to try this, here are some warnings and theory behind battery charging etc.

 

First off lets get clear what this is all about. The purpose here is to bypass the 3.3V regulator already in the REGEN. The 3.3V LiFePO4 battery directly drives the hub chip and oscillator.

 

There are two general ways to do this:

 

1) Battery externally charged, then disconnected from charger and connected to the REGEN

 

2) Battery and charger always connected to regen

 

Number 1 is easy to understand and implement and presents little risk.

 

Number 2 is complicated, difficult to get right and could fry parts of the REGEN or just plain not work if done wrong.

 

I'll go over number 1 first since it is the easiest. For this just use a commercial battery charger for the type of battery you are using. One easy way to do this is a charger with jack, plug on the battery, same jack on the REGEN. Plug the battery into the charger until charged, unplug from the charger and plug into the REGEN. This guarantees that the charger is NEVER connected to the REGEN, it can't fry the REGEN and it can't form ground loops.

 

Note this is NOT plugging into the power jack already on the REGEN, this is all about wiring up a separate jack at the output of the internal regulator so it will bypass that regulator.

 

The hub chip and oscillator take about 50mA, so with a 2.5Ah battery that is 50hours, I'm not sure what the cutoff voltage is for that so lets say 30 hours of continuous use from a charge. Most of the discharge time is at about 3.2 V which will be fine for the REGEN.

 

Unfortunately that gives the problem that you can't run it continuously. You have to charge, then connect to REGEN. A simple solution for this is that when the battery is charging plug a normal supply into the normal REGEN power jack. Just make absolutely SURE this is not connected while the battery is connected. A multiple pole switch should be able to do all these connections.

 

Number 2 is a much more complex system. An external source is continuously charging the battery while it is connected to the REGEN. There are some issues, the voltage across the battery should not get above 3.3V, nor below about 3.1V. The charging circuit has to be able to handle the situation of a discharged battery, which in this case is almost a dead short. When the battery is charged it has to provide the 50mA to the hub and oscillator as well as the battery.

 

So what does this charging circuit look like? Some people have talked about using a commercial charger, but it has been noted that these charge at 3.65V, which is too high for the REGEN. See a previous post of mine as to why.

 

The simplest charger circuit is a voltage regulator followed by a resistor. The resistor value is set so the charge current into a discharged battery will not exceed the current and power capabilities of the regulator. For example, lets take a 10 ohm resistor, at 3.3V out of the regulator and 0V on the battery, that is 330mA, that is a reasonable charging current for the battery and most 3.3V reg circuits should be able to handle that. The resistor is going to be dissipating 0.33A X 3.3V or a little over a watt, so it should be at least a 2 watt rating. The regulator will be dissipating 0.33A x raw supply - 3.3V, so if using the 7.5V that is 4.2V x 0.33A which is about 1.4 watts. As long as it is heatsinked properly that should be all right.

 

BUT what happens when the battery is charged and it is supplying the 50mA to the REGEN? A 50mA current through the 10 ohm resistor is 0.5V drop, that puts the voltage at the REGEN of 2.8V, WAY too low for proper operation.

 

OK how about a 5 ohm resistor? That gives 0.25V drop which gives 3.05 V, which may work, but is right on the edge and may cause degradation in performance of the regulator built into the hub chip. Not only that but the maximum charging current has now doubled to 660mA, which is now over 2W on the resistor and 2.8W on the regulator. Again doable if you have the right regulator and resistor.

 

A 4 ohm resistor is probably the best way to go, the voltage on the REGEN is 3.1V which is where I would like it to be, the max current is 825mA. So a resistor rated for 5W and a regulator designed to handle 5W would probably work well for this.

 

Under normal operation when the battery is charged the current will be much less so it won't be getting hot at all. But you need to design the circuit so it won't fry if you put in a discharged battery.

 

One thing to note is that this will take a long time to charge the battery. As the battery charges the voltage across the resistor falls so the charging current decreases, expect to spend at least a day charging one of these cells even with the 4 ohm resistor.

 

I would not plug the REGEN into the computer or DAC until the voltage across the battery gets to above 3V. Anything below that and it will not be functioning properly.

 

Don't even think about trying to do this with the regulator built in to the REGEN. There is no 4 ohm resistor to limit the current. It will max out at 1.2A, with over 8 watts trying to be dissipated, you WILL fry something with that. If you try and add the 4 ohm resistor you will need to cut a trace on the REGEN board, this could be done but is VERY difficult, I could do it with an Xacto knife and a microscope, but I have a LOT of experience doing such things. And even then it probably wouldn't work, with the 4 ohm resistor and 7.5 raw voltage the reg will be dissipating 3.5W, which will really push things above their limits. If you have a fan blowing on the board you could probably get it to work, but that kind of defeats the "silent running" philosophy.

 

As to why having the battery across the charging supply makes the characteristics of the charging supply not matter, it is a simple voltage divider. The battery has less than 10 milli-ohm impedance across a wide frequency range, you are feeding it with a 4 ohm resistior, that is 4/.01 which is a reduction of 400 times, that is a 52db reduction in noise from the charging supply.

 

I know some of you want to try this no matter what, but take these things in advisement when trying so you don't fry something or run the REGEN in such a way it won't be operating up to full potential.

 

John S.

[sandyk]

John

A simpler way is to have 2 batteries , with the original being recharged while the other is swapped in.

I use a 12V 4,000mAh Li Ion battery regulated down to +9V for the Regen supply.

The total cost including the plugpacks is only around $30 for 2 of those.

You could similar to what you are suggesting using a 7.2V Li Ion battery into a LM317T set for +3.3V, (or a LM3940T-3.3) with a suitably sized heatsink inside a small box with a 3.9 ohm PW5 series resistor to float the LiFe P04 battery ?

The additional battery would ensure no down time between listening sessions, and there would be no mains earth loop issues.

Regards

Alex