Tír Na HiFi

rickmcinnis wrote:Each battery will supply one "Vref" - plus and minus for each channel, hence four.

What I intended to ask was whether battery power would negate the NEED to leave the gear on all of the time. My question was along the lines of what takes the most time to stabilize: The power supply or the circuit itself. Thinking the batteries are as ON as they ever will be at turn on maybe one does not need to leave the thing on? And then I thought about the clock ...

THANKS for the iron recommendation and your comments in general.

Take care,

jkeny wrote:Rick,
If running +/- 4V from 4 batteries (2 in series on each side) you are running them at below their minimum recommended voltage (2.7V) so it's a very sub-optimal setup & I would suspect that this will be a problem.

Charging at these voltages really means that you are not trickle charging the batteries - they will act as fully discharged & draw the max current that the charger can provide (these batteries can recharge at 20Amps if the charger can do this). This may or may not be a problem for the charger, depending on its thermal build?

Furthermore, putting 4V into 2 (3.3V) batteries in series is a recipe for failure as it is not balancing the voltage to each battery so one battery. It may not be such a big problem as this is close enough to the battery's max voltage of 3.8V to not kill one of the batteries but I could see one dead battery in the series being the result of this scheme over a longer period of time. This is a difficult one to predict as when a battery imbalance starts, it becomes a cascading effect - it will take up less current & the other battery more until one or both die.

I've said all this to Nige & he disagrees with me on some of this as his experience suggests that a batteries in series work fine in this sort of config. That may well be the case because the current drawn (& therefore current fed from charger) is so small that a battery imbalance isn't happening to a large extent or it may be that he is lucky & his batteries are well matched to start with? Or it may be that he has a dead cell in there already? A dead cell in a series will not be seen when the series is being constantly charged.

Sorry, Nige, hope you don't mind me writing this (I've said most of it to you already) but I just feel that batteries in series with trickle charging is a wrong config when no balancing circuitry is used & my experience with balancing circuitry (at least the ones I've tried) have a detrimental effect on the clean power that these batteries provide.

That's why I suggested to Nige that using +/- 3.3V for DAM vref is probably better & to suck it & see. I too predicted a small drop in volume but it appears this isn't audible. Neither is it audible going up to 5V (he has tried this) so I don't know what's happening here. I know he reported 6V as audibly a bit strained but this may be for other reasons.

jkeny wrote:

rickmcinnis wrote:No, I am running two pairs of batteries (or will be) for each channel, each polarity, and sticking with Nigel's recommendation that 3.3 volts is sufficient. Only one battery per rail.

I'm not with you here - maybe I'm dense? You have two pairs, that's 4 batteries on each channel? So you are putting 2 batteries in parallel for +3.3V & two in parallel for -3.3V, right? BTW, you should share the vref between both L & R channel

No question running 26650s at 2 volts would be a bad idea!

I was talking about 2 batteries in series (on each polarity) & using them for getting the original +/-4V vref of the DAm DAC - this was Nige's config originally which I was talking about.

I am tempted, after all is actually running well at 3.3, to creep up the voltage and see what happens.

With the negligible current draw one wonders what would happen as you near 4 volts on one cell. After all you can charge them with 4 volts.

Do you think this is asking for trouble?

Thanks for your concern and your counsel.

I think it's pointless (according to Nige's reports of 5V sound) & asking for trouble - who said you can run these batteries at 4V? 3.8V is considered max. OK, I know datasheet maximum voltages can be exceeded (DDDAC's PCM1794 DAC for instance) but there are audible gains realised with these overvoltages - in this case there doesn't appear to be any upside to this, only downsides.

jkeny wrote:

rickmcinnis wrote:

nige2000 wrote:you no power outages over there?

volume seems the same i didn't measure it

It took me awhile to get what you are saying. We do not have power outages very often.

Yes, I do plan on keeping the thing on all of the time so if there was one while at work and I left it on there would be the problem of the battery running the DAM and the regulator. I wonder how long it would take to bring the battery down with an extended power loss? Ours are usually less than fifteen minutes and rarely anywhere near that long. Compromise is always with us.

I wonder if battery power might negate the benefits of running the thing all of the time?

I don't get what you're saying here - how would battery power negate the benefits? Oh, 15 minutes power down will not cause any problem with the batteries - if you are doubling them for each polarity then you should have many hours before running low

I wonder if shift registers work better after "warming up"? I guess the clock would work better after being on awhile.

Yes, leaving on all the time allows ICs & clocks to reach stable operating conditions. What the DAM is doing is reproducing the very low level detail in the digital signal - the LSBs. These are at very low voltage levels & the stability of these very low voltages is what is the secret here - it's what gives the DAC it's realistic sound, I believe. The stability & low noise of the vref is the most crucial node in the whole DAC - it's this reference voltage that is used directly by the resistor ladder to generate the output voltages corresponding to the digital bits of the digital byte. These batteries provide a number of advantages for use in digital audio - they provide 3.3V fairly steady output for most of their charge - perfect V for most digital circuitry so don't need a voltage reg - they are so fast, low noise & stable that using them with a cap or voltage reg is usually detrimental to their operation. These 3 factors low noise, stability & speed are exactly what's needed for digital ICs & it's difficult to find regulators that combine these attributes - I believe the Paul Hynes come closest if not the equal of these batteries.

If you do not detect a volume change then no need to measure it. One more thing to make one think that 3.3 is just fine. Makes one wonder why SOEKRIS picked such an odd value for the voltage ... One thing for sure the batteries have infinitely better current delivery than that opamp based reg/reference which could well explain the ability to get by with less. Has anyone ever measure just how close those SOEKRIS regs track each other's voltages?

The current requirements are probably not very great in this part of the DAm but you are correct, the batteries are over engineered for the task of supplying such small currents - they are 2.3A rated but can provide 20Amps continuously or something like 120Amps instantaneously - which accounts for the voltage stability needed in this circuit.

jkeny, what do you think the real world capability of those reg/refs to supply "equal" voltages would be over time?

I haven't looked at them but if you are worried about the potentiometer, then find the setting that gives you 3.3V output, measure the potentiometer's resistance & substitute a fixed R of this value for the pot

What soldering iron do you use for your SMD work? What size tip?

Metcal (OKI 9000) - I recommended it to Nige & he finds it great too. Various size tips depending on what I'm doing but the most useful one for smd work is this one OKI Soldr Tip SFV-CH15A Tapr Chisl 1.5mm

I am thinking my setup is just too large and I might be better off getting something more suitable before soldering anything to the board.

I worry that the more I obsess these details the more certainly there will be that the thing will blow up upon first turn-on.

Hope this helps you avoid the magic smoke?

nige2000 wrote: everything sounds better warmed up :)
this build will definetly best left on for a while for voltage stability

using caps with lifepo4 is kinda like putting a sponge on the end of garden hose it slows everything down, so dynamics/bass are effected along with slowing down its attack and affects fast transients adding linear regs will slow it to but will remain reasonably low noise
the key is direct if at all possible

the 4 v tracks bang on +/-3.9998
i challenge anyone to notice say a 0.3v skew with the batteries
cause i cant

yes john i think the fixed r in my current audiowind regs will be the way ill go
i can get them to 0.0002 off target with the std bourns pots

yea the Metcal (OKI 9000) i love it
its a joy to use
i turn it on i sit down and its ready to go straight away
i used to waste a heap of time waiting for heat up of old iron
also its very fast to react and compensate for tip heat loss
i was lucky sourced 2 s/h out of the uk for Fran and myself
fran likes his very much too
great selection of tips too

oh btw i have vref at 3.4v but it also powers the clock, fpga and ucontroller i didnt check the upper limits but theres hardly that much room for movement

jkeny wrote:

nige2000 wrote: everything sounds better warmed up :)
this build will definetly best left on for a while for voltage stability

Yep, no harm at all

using caps with lifepo4 is kinda like putting a sponge on the end of garden hose it slows everything down, so dynamics/bass are effected along with slowing down its attack and affects fast transients adding linear regs will slow it to but will remain reasonably low noise
the key is direct if at all possible

The sponge is a good analogy :)

the 4 v tracks bang on +/-3.9998
i challenge anyone to notice say a 0.3v skew with the batteries
cause i cant

Yea, I think you tried 3.3V on + & 4V on - at one point & didn't notice any audible difference. When you asked Soekris about this imbalance he said that 2nd harmonic would be -14dB down or so. 2nd harmonic is fairly innocuous, AFAIK?

Didn't you have to adjust the regs in Tony's before we listened to the DAM? When you measured the two regs, one was away from 4V by a bit, no?

yes john i think the fixed r in my current audiowind regs will be the way ill go
i can get them to 0.0002 off target with the std bourns pots

Yea, it makes sense

yea the Metcal (OKI 9000) i love it
its a joy to use
i turn it on i sit down and its ready to go straight away
i used to waste a heap of time waiting for heat up of old iron
also its very fast to react and compensate for tip heat loss
i was lucky sourced 2 s/h out of the uk for Fran and myself
fran likes his very much too
great selection of tips too

Yep

oh btw i have vref at 3.4v but it also powers the clock, fpga and ucontroller i didnt check the upper limits but theres hardly that much room for movement

Do you not have vref separated from powering these other ICs - I thought your vref was also just powering the 595 chips?

nige2000 wrote: yes thats right i was pre testing 3.3v on vref theory but i realised the sm dc-dc 1.2v reg wouldnt function under 4v input so it was +4 -3.3v goin to tonys
adjusted it back to +/-4v before we started playing it

the latest config is much improved since then
lucky that reg wouldnt work @ 3.3v as i wouldn't have realised that orginial 1.2v reg was adding so much noise

i tend to say separated but what i actually mean is that there is a cell in parallel for each vref and then another for clk,fpga,ucontroller,1.2v ldo reg,usb to i2s
a cell in parallel is as good as an additional seperate rail

jkeny wrote:

nige2000 wrote: yes thats right i was pre testing 3.3v on vref theory but i realised the sm dc-dc 1.2v reg wouldnt function under 4v input so it was +4 -3.3v goin to tonys
adjusted it back to +/-4v before we started playing it

Right, I understand

the latest config is much improved since then
lucky that reg wouldnt work @ 3.3v as i wouldn't have realised that orginial 1.2v reg was adding so much noise

Ah the attention would have fallen on that reg eventually :)
eh yea but it was well down the to do list
there is always ni-mh batteries

i tend to say separated but what i actually mean is that there is a cell in parallel for each vref and then another for clk,fpga,ucontroller,1.2v ldo reg,usb to i2s
a cell in parallel is as good as an additional seperate rail

Right, yea, because all cells connect to common ground & only their positives drive different parts of the board, it can be considered that these cells are kinda in parallel. But maybe not - did you tell me that running the FPGA, ucontroller, clock off of the same cell as was providing the +3.3V Vref, sounded slightly worse than using a separate cell for these ICs?

randytsuch wrote: Nigel,
quick question.
I noticed you're isolating the i2s between your usb to i2s converter and the dac. The dac already has an isolator, so I was wondering about adding another isolator.

Randy