Yes, I get why they should be earthed. Even metal windows should be earthed, per code. For example, in case of a fault that energizes any of the metal, a breaker or RCD will trip (if all is done correctly).
In my own home I never had the "tingles", thankfully. I don't think the copper pipes are earthed properly, and they are actually well insulated from the brick walls/cement with thermal insulation (old stuff), but it hasn't been a concern so far.
Thanks for the good advice, I am proceding carefully.
Last edited:
🙂 your windows don't need a 0V reference..
If your floor is at some deviation from 0V, and this can be a thing with cables imbedded in metal conduit cemented into concrete (dont ask me how, but it does happen) - so for example your tooth brushing tap is a 0V and the bare-foot person standing still gets a tingle because there is a voltage at the floor. (Yep, I know).
If your floor is at some deviation from 0V, and this can be a thing with cables imbedded in metal conduit cemented into concrete (dont ask me how, but it does happen) - so for example your tooth brushing tap is a 0V and the bare-foot person standing still gets a tingle because there is a voltage at the floor. (Yep, I know).
Last edited:
That's very nice. AI will not replace you anytime soon. Actually, AI needs you 😉
Maybe. (I would write what I really think, but through fear of deletion in the absolute sense).
Where/when I grew up, we all had a trade, if we wanted it enough = not easy (and many did not).
Where/when I grew up, we all had a trade, if we wanted it enough = not easy (and many did not).
All this interesting story about reference ground and stories, windows as window ligthning rods made me wonder about this new TDA1541A design and the super-caps iterations talked few pages ago.
Half-illiterate's question assumed off topic for you aeropagus : could the full BJT boards could be switched off from the main with its bad EMI and noise during playback ? (Assuming the pcb with supercaps from Thorsten is chosen)
How much time could it last till being repluged to the main for feeding the Super caps from Thorsten or at least be charged by slow cells like dry ones for home environment as it doesn't matter here they have inductance and capacitance for that reloading Super caps job ?
Don't want to slide into a project with expensive and explosive LiFEPo or alike pcb project story to add stories (stacking) ,I just wonder if even a simple switch on-off from the main could make the DAC sounding better (often people talk about darker blackground, or more "analog" blah sound if those words make sense - take the words just as : it sounds better when depluged from the main). Btw those supercaps made me remember old tweaks on the CD723 with TDA1545A where we putted back then (end of 90s' ?... my memory is falling like a manga on the head of a teen in a mangas store). It was a flat memory cell cap used as an inertia caps to filter/isolate the crystal from the main PS, just putting a decoupling with a fast lythic just before the crystal... was the first use I remember of the super caps but those one had of course a very bad ESR, not as the ones discussed here.
Off topic. In my question I assume there still is few EMI from the main not totally filtered whatever the supercaps ? Bad idea ?
Half-illiterate's question assumed off topic for you aeropagus : could the full BJT boards could be switched off from the main with its bad EMI and noise during playback ? (Assuming the pcb with supercaps from Thorsten is chosen)
How much time could it last till being repluged to the main for feeding the Super caps from Thorsten or at least be charged by slow cells like dry ones for home environment as it doesn't matter here they have inductance and capacitance for that reloading Super caps job ?
Don't want to slide into a project with expensive and explosive LiFEPo or alike pcb project story to add stories (stacking) ,I just wonder if even a simple switch on-off from the main could make the DAC sounding better (often people talk about darker blackground, or more "analog" blah sound if those words make sense - take the words just as : it sounds better when depluged from the main). Btw those supercaps made me remember old tweaks on the CD723 with TDA1545A where we putted back then (end of 90s' ?... my memory is falling like a manga on the head of a teen in a mangas store). It was a flat memory cell cap used as an inertia caps to filter/isolate the crystal from the main PS, just putting a decoupling with a fast lythic just before the crystal... was the first use I remember of the super caps but those one had of course a very bad ESR, not as the ones discussed here.
Off topic. In my question I assume there still is few EMI from the main not totally filtered whatever the supercaps ? Bad idea ?
If I temember correctly, John @ecdesigns proposed a PSU solution with complete isolation. Read this thread from post #1 on.
As I did , using li-ion battery ( they are totaly save ) , one for each supply solve these issues 😉
.
Search back for "charge transfer power supply"
https://www.diyaudio.com/community/...ac-using-tda1541a.79452/page-129#post-1750512
https://www.diyaudio.com/community/...ac-using-tda1541a.79452/page-129#post-1750512
I do not understand the question. EMI on the mains gives rise to EMC and as long as we do EMC (aka EMI mitigation) right, absolutely no.
A separate case is "missing earth", but a local network with multible devices can still function efficiently, without actual mains earth, as long all chassis connect together correctly.
EMI is mostly a common mode problem.
There are calculations for that, but I am not using supercapacitors as holdover device, in this case they should be before a regulator and have extra voltage (say 1.5 times) so the voltage can drop significantly, before the actual supply voltage to the circuit drops.
Depends.
If the many fundamental issues that I seek to address using RC & LC filters plus screens in transformers have not been sufficiently mitigated, yes it will.
I had one of those.
But that supercap tweak would have been pointless, the crystal, as well the 4x oversampling filter were part of CD processor IC (it even had a basic DS DAC on chip).
There a few things that could be done, but the potential was limited.
It's a complex topic. The SC's are for the actual direct supply in my case, so they deal with DM (differential mode) noise, primarily everything in the extended audio band.
If we use 8pcs 10mOhm 25F on series for 20V, we end up with 80mOhm from around 1Hz to above 50kHz with a very real and substantial transient storage.
Prefixed by 3R DCR from the chokes before them we get a 32dB noise attention from the preceding electronics at around 1Hz rising to ~-36dB at 100Hz with a first order slope above that (so ~ -56dB @ 1kHz and ~ - 76dB @ 10kHz)
Can we do better with electronic regulators?
Yes, but to do so in practice is non trivial, SC & Choke are a trivial passive circuit.
I also worked with such supplies.
In the end, unless you have a reason specific to require a "never connected" power supply (e.g. marketing) there is not much merit.
Why? Transistor switches or FET's usually have enough capacitance, to make the dominant source for EMI conduction the rectifier capacitance. So shorting the transistors out doesn't really changed much.
John's "charge transfer" however incorporates many elements that should improve EMC. It is this:
Rectifier capacitance is cut by a factor 3.
Rectifiers are Schottky and thus have no reverse recovery.
Multiple diodes in series and capacitors in series for the first capacitor increase resistance in circuit, causing a wider conduction angle and less dI/dT than standard rectifiers.
Once we add the chokes and resistors (and switching transistor resistance) between the two capacitors we have a symmetric PI filter (lowpass) with a corner somewherebat a few 100Hz and a common mode impedance of around 70 Ohm at 100kHz.
I would suggest perceived improvements (objective/subjective) over a simple rectifier bridge + single Cap supply are down to these factors by far more than the "charge transfer" nature of the powersupply.
Reviewing the power supplies I posted earlier utilise all sorts of such measures, perhaps, if I may be so bold, much mure deliberately and intentionally.
If a CT supply is nevertheless desired, I would suggest to add current control (limit) and threshold based early turn off of the transistors, to create a crude voltage regulation and to isolate drive signals with optoisolators.
For control circuit, first whenever the rectifier diodes are not reverse biased, the transistors must be forced off.
The turn on is ideally "soft" with an RC delay ramping current to the set limit, with the turnoff affected by when the second capacitor reaches a limit voltage. This can be peak or average.
This "regulation" is slow (<< 100Hz) and coarse, but it can help limiting heat wasted, especially in shunts.
If we look at it closely, btw, this regulated CT supply (as well as John's unregulated version) is actually a switched capacitor regulator running at very low switching frequency synchronised to the main frequency and with the rectifier bridge dides forming one pair of the switches used.
Doing it at 100(120) Hz with big capacitors or doing it at 500kHz with small capacitors, is there REALLY a difference?
Thor
Last edited:
There are Schafner EMI filtering plug modules, but some noticed not so good behavior here at diyaudio when checking with a scope (while some others automaticaly use them in their pre or dac)
Thanks for the inputs.
In the question I assumed the Supercaps were reservoir caps powerfull enough to feed the BJT I/V and DAC circuitry w/o plugged to the main wall. Assuming as well as untill it is still plugged to the wall there is still EMI that can spread into the DAC by that wires whatever there are big super caps on each rail.
Indeed it assumes there was no EMC strategy in the power supply of the DAC (common chocke, snubber, else....?).
Off topic question. But you know, back then, everyone putted their little amps on battery (remember Hiraga le monstre on polluting car lead battery in living rooms (a stupid idea for health) and DACs too ! ). OK, sorry for the divertion, let go back to the main topic... I know more or less about pcb and EMC having reading Ott (some will be happy just watching Altium YTs) , but haven't off course the instruments to check.
There are also that good rectifier chips LT4320 (?) that swap the diodes, but I don't think it makes anything from the EMI filtering. Btw, is that important for a 16 bits DAC with that (good enough) noise floor ? Can be heard from your experience ? After reading Ott, you see also EMI problems everywhere as well as bellow the pillow !
Last edited:
Current over time. Hmm. Could add a series resistor between capacitor -ve to common for similar/same effect ?
What does "noticed not so good behavior" mean?
These filters are specifically designed to help equipment that either generates or is sensitive to EMI Electro Magnetic Interference) in specific bands to pass the EMC (Electro Magnetic Compatibility) standards set by the Comité International Spécial des Perturbations Radioélectriques (CISPR) of the IEC. While these international organisation standards are not legally binding, many countries legislate EMC on the basis of and with reference to CISPR.
As a general rule, CISPR Band's A & B (9khz - 30MHz) are not legislated and thus both compliance engineering and compliance solutions tend to ignore sub 30MHz problems.
For competently designed audio gear we expect at the minimum an 80...160kHz lowpass on the input is expected, mains inputs should be filtered at even lower frequencies.
Thus by the time legislated emissions and emission mitigation solutions become effective Audio gear should be substantially resilient to EMI. Where is little or no resilience unless explicitly designed in are band A & B which is for example where a majority of the noise generated by switching mode supplies (including "Inverter" motor drivers in fridges, air conditioners, heating systems and electronic ballasts for LED Lamp's) falls.
What does it all mean?
Standard EMC solutions to support legislated compliance with emissions and/or resistance to emissions limits are not useful in Audio. They may still be employed, but their limitations must be understood.
EMC in Band's A & B is crucial for audio and related gear and cannot be addressed using easy OTS solutions aimed at CISPR Band C and up.
EMI in modern setups is almost exclusively common mode. It occurs at frequencies where balanced audio connections no longer suppress common mode noise. So balanced and SE connected and internal operation gear are affected equally.
EMI and Electrical Safety legislations create conditions in modern homes that are EMI rich and leave many pathways for EMI to affect audio equipment.
ALL power supplies are actually "switching type" and create switching noise, including so called "linear" power supplies. Dealing with this is "Power supply engineering". There are no patent recipe's, power supplies must be matched to the circuit they supply.
The best solution is a dedicated listening room, with EMI shielding on walls and windows and powered by an electromechanics double conversion system (E-motor drives generator with suitable governor system) of at least 3 times the maximum peak draw capacity and thus perfect low impedance, low distortion and stable frequency.
It will, but the capacitors will discharge and they will reach a 10% drop in voltage in 10% of the time needed to completely deplete all stored energy.
They are still just capacitors.
Remember, EMI is (mainly) common mode, between earth/chassis and all the signal/power circuitry.
That would be a very stupid thing to do, in 2024. It was already questionable in 1984, problematic in 1994 and criminally negligent in 2004.
I have done this too. It does work well, actually and can be designed to be unproblematic. But cost / weight is an issue and for commercial gear it is better to use other solutions. Correctly recycled SLA/SLC batteries are not an environmental issue.
If still worried Greta may "How Dare You" (plus battery lifespan) run your house on NiFe batteries with solar/wind generation. These batteries are essetially eternal with a bit of TLC (like a "Smiling Benz").
No, these are "Ideal diodes" but focusing on losses, not other behaviour. Switching off a giant Mosfet will generate a huge spike on switch on and off.
With 2V RMS SE out and a zero data SNR of 110dB and with a system capable of reaching or exceeding 110dB SPL (my milkmaid calculation is 2.83V SPL + 6dB for stereo -9dB for ~3m distance + 20*LOG(Amp max Vout)) we want less than ~2uV noise in the output attributable to the analogue stage and power supplies, or -114dBV. Of course, full output SNR of the amplifier needs to be ~120dB and the speakers should have a sensitivity of > 93dB/2.83V for an amplifier with 28V undistorted output.
My Upstairs (Big) system with a 1989 Marantz PM-75 (100W RMS @ 8R, 105dB(A) SNR at 2.83V, TDA1541 DAC based on the LHH-1000/DA-12) driving Technics SB-E-100 (95dB/2.83V, 3-Way with 12"LF + Horn Mir/Treble, 50Hz-18kHz +/-3dB) is largely compliant with all these requirements, the background noise from the nearby beach probably spoils the in room SNR. I do use mains conditioning for the DAC/Amp.
What is "can be heard"?
The worst EMI causes significant audible birdies, the most subtle one is just a little more grainy sounding than without.
Yes. Adding resistance in the circuit will reduce δI/δT and δV/δT. Adding chokes will do as well.
Define your goals, design to meet them. Avoid "patent medicine" including a lot of snake oil.
Thor
Hi,
I said I would show the "Super Capacitor" version of the power supply that is fully split A/D. including analog stages.
Here you go:
Regulation is simple. The main AC Impedance and Noise specifications are down to the super capacitors. The rest of the circuit is made simple, with common mode noise conduction significantly reduced.
Most TL431 are actually "protection" circuits for start-up.
Only the +5V ones should be active.
Thor
I said I would show the "Super Capacitor" version of the power supply that is fully split A/D. including analog stages.
Here you go:
Regulation is simple. The main AC Impedance and Noise specifications are down to the super capacitors. The rest of the circuit is made simple, with common mode noise conduction significantly reduced.
Most TL431 are actually "protection" circuits for start-up.
Only the +5V ones should be active.
Thor
The soulution (pun intentional) is to use a Camille snubber. That is C parallel to secondary windings, then two resistors (or more depending on lines from TX) and another capacitor.
In effect the R's and C's resolve to a parallel C and snubber across it, like so many recommend "snubbers".
You can use all the "Quasimodo" stuff and adjust and do it that way, or you do it my way.
I take effective secondary DCR and make the resistors equal. So secondary effective DCR is doubled. For capacitors, practical consirations and the need for a => 1:10 ratio between cap in parallel with the secondary winding and cap after resistors means commonly I use 100nF and 1uF. Plus a pair of 1uF to PEN/Chassis to send common mode leakage directly back to the source, before entering.the remaining circuit.
If we have (say) 3R secondary winding DCR (indeterminate leakage inductance) and 1.5R + 1.5R with 100n + 1uF the first stage lowpass (secondary DCR + 100nF) is 500kHz with an unknown inductive component in the mH range.
The second stage with all DCR in the circuit comes out as around 18kHz. We can go up here on capacitance, but I have this habit of using 1206 SMD's in this position which limits values.
So the rectifier now has a 'look forward" impedance of 6R // 1.5uF with a resonnant tank wrapped in of unknown frequency, but likely much higher than 25kHz.
After the rectifier we do the same, 100nF-> resistors -> 1uF + first electrolytic cap. Schottky diodes, multiple on series to reduce reverse capacitance coupling noise.
We now have trippled DCR, so peak currents are reduced by a factor 3, δI/δT and δV/δT are dramatically reduced. Transformer "free wheeling" ringing is dramatically reduced. Of course, losses are WAY UP, TANSTAAFL.
By now using several smaller capacitors with another few ohms between them, we have more filtering, for example 2 Ohm + 2,200uF makes a single 36Hz lowpass.
By the time we get to the final LC cell that has a lower turnover but overall ripple gets attenuated a lot.
Compare a simple Bridge, a big low DCR torroid transformer and a big Nichicon Super Through 15,000 uF 50V to this.
Which will sound better?
Thor
Hi
@ThorstenL and all other members interested in this topic 🙂
.
I would like to have an opinion about:
Since the C on MSB to MSB-7 pins are calculated for say 176.4KHz,
Would it be good to have also constant Fo of 176.4KHz at DEM oscillator (pins 16, 17)
?
Note:
Only in case of 44.1KHz Sampling rate,
Fo for DEM MSB pins
and
Fo in DEM oscilator pins are the same = 176.4KHz
.
In every Sample rate increase, the external DEM oscilator Fo is higher for SR x n times
while DEM MSB pins remain fixed 176.4KHz
@ThorstenL and all other members interested in this topic 🙂
.
I would like to have an opinion about:
Since the C on MSB to MSB-7 pins are calculated for say 176.4KHz,
Would it be good to have also constant Fo of 176.4KHz at DEM oscillator (pins 16, 17)
?
Note:
Only in case of 44.1KHz Sampling rate,
Fo for DEM MSB pins
and
Fo in DEM oscilator pins are the same = 176.4KHz
.
In every Sample rate increase, the external DEM oscilator Fo is higher for SR x n times
while DEM MSB pins remain fixed 176.4KHz
There is a theoretical possibility (yet to be proved) that the DEM oscillator might interfere with the BCK and/or with the LE signal, causing spurious signal in the audio band. This could be be due to on-chip leakage, ground bounce, whatever. I think that syncing the DEM oscillator to LE signal (á la Grundig in 4x OS) or to the divided MCK or BCK would not harm, it does not increase coplexity too much. In any case, I would not go beyond 176.4 kHz, regardless I don't believe in the "4-times sample rate" rule. Even 50 Hz DEM clock worked fine, with properly sized decoupling capacitors, didn't it? Most important is perfect (I mean rigorously perfect) 1:1 symmetry of the DEM oscillator duty cycle. It should be also jitter-free, naturally.