Speaking of noise from batteries ----- The simplest model of a battery is much like a capacitor... a series R and a shunt R across a C. This implies a rising Z at low freqs but also that noise might be computed from the R values if they were measured.
It is also important to know that as the battery discharges the Z increases; Noise would increase as battery discharges. Also, as the battery capacity (Amp-Hrs) increases the z decreases. All-in-All, batteries may not be the 'ideal' dc source of supply. But. the newer lithium-ion types are worth considering for portability apps and may be quieter and for a longer time in use (flatter discharge curve).
see a 9 vdc alkaline battery vs Lith-ion Z from :
View attachment App_Note_Battery_Impedance_V1_1.pdf
THx-RNMarsh
It is also important to know that as the battery discharges the Z increases; Noise would increase as battery discharges. Also, as the battery capacity (Amp-Hrs) increases the z decreases. All-in-All, batteries may not be the 'ideal' dc source of supply. But. the newer lithium-ion types are worth considering for portability apps and may be quieter and for a longer time in use (flatter discharge curve).
see a 9 vdc alkaline battery vs Lith-ion Z from :
View attachment App_Note_Battery_Impedance_V1_1.pdf
THx-RNMarsh
Last edited:
I do NOT think that this model is valid. The rising Z of a capacitor towards lower
frequencies does not generate noise, only real resistance does. Also, in the
app note, the impedance stays in the mOhm range even at 1 Hz and when the battery
is close to discharged.
That also does not explain the huge 1/f noise that I recently measured. If that was
voltage noise from a resistance, it would take MegOhms at low frequencies.
Methinks, low source impedance is necessary for low voltage noise, but it is not enough.
We have discussed my measurements in this thread already at
< http://www.diyaudio.com/forums/loun...ch-preamplifier-part-ii-5630.html#post4049608 >
Just downloaded and read Ron Quan's latest AES paper, convention paper 9197.
Very briefly, Ron has measured the variation in several types of distortion as a function of the momentary (dynamic) output offset voltage, for instance the variations in high frequency signal distortions superimposed on a slow-varying 'offset' signal in the form of a stair-case or a lf sine wave.
His findings are of interest to anyone serious about audio amplifier design. Not necessarily counter-intuitive, but it is good to see some numbers attached.
Recommended read - thanks Ron!
Jan
Very briefly, Ron has measured the variation in several types of distortion as a function of the momentary (dynamic) output offset voltage, for instance the variations in high frequency signal distortions superimposed on a slow-varying 'offset' signal in the form of a stair-case or a lf sine wave.
His findings are of interest to anyone serious about audio amplifier design. Not necessarily counter-intuitive, but it is good to see some numbers attached.
Recommended read - thanks Ron!
Jan
makes sense to me.... I discussed this with JC decades ago right after publishing in Audio magazine a piece about Common-Mode affects and that a music signal has a non-zero average signal level. A sine wave average or dc component is zero. But music has a common-mode component (average) which varies with the waveform asymmetries.
If a circuit is sensitive to cm distortion will increase dynamically. Some direct coupled circuits can also cause a dynamic shift/offset from zero which is an unbalanced condition leading to increased distortion. A servo cannot reduce that affect.
I am glad someone figured out how to measure its affect. I'll have to check it out how he measured such.
THx-RNMarsh
If a circuit is sensitive to cm distortion will increase dynamically. Some direct coupled circuits can also cause a dynamic shift/offset from zero which is an unbalanced condition leading to increased distortion. A servo cannot reduce that affect.
I am glad someone figured out how to measure its affect. I'll have to check it out how he measured such.
THx-RNMarsh
Last edited:
It isnt my model. However, I suspect the Rs of that model would increase at lower freqs (heating perhaps or chemical affects or both).
From the figures shown of a 9 v battery -- "the battery impedance at 1KHz did increase to approx 3.8 Ohms" . the graph shows a further increase with discharge [7.5vdc gives 1KHz z of about 6.5 Ohms]. Below that is another graph of a Lith-ion battery is 0,126 Ohms.
Does the Lith-ion measure as lower or higher noise than alkaline? Does this trend follow from Z of battery and noise with all battery types? You would have to also measure their Z as well as noise to find out. I am suggesting the higher measured Z will also measure noisier. Certainly chemical reactions might add to the noise.... but does it dominate? If so, where.... at low freqs only? Is that an exponential curve rising at low end?
THx-RNMarsh
Last edited:
Such experiment can be quite easily done, with LF sine + HF signal.
The main problem is that we are trying to measure (or to test) better DUT with worse test system. Any A/D and D/A chain has numerous opamps in D/A I/V, LP filter, output stage . Then we have a very linear DUT, regardless it is discrete or best available opamp. And then several opamps + ADC chip in the A/D converter. What are we testing, in fact? Yes, we may use a kind of magnifying lens like notch filter. But not for the test that you are mentioning. I am afraid such test may be used only for mediocre designs.
BTW, I still can access the forum only through downuptime proxy.
Well you have a point there, in some cases Ron used vintage opamps and/or very high loading to make the measurements possible. Also, some results are as I said expected - that with offset = 0, around crossover, distortion is higher that with the output offset from zero. Made me think about Doug Self's XD design. But the paper is worth the $5 it costs from the AES (for members).
Jan
Many PS do not treat all freq in the same way a battery does. That is, the battery provides pretty much the same Z to all audio freqs. Some electronic PS have a rising Z within the audio range. How low a PS Z needs to be for audio depends on the ratio of the PS Z to the circuit Z being as high as possible. Low Z (eg low noise) need a much lower Z from the PS and visa versa. My rule of thumb is minimum of 80dB ratio.
But, then again, PSRR minimizes much of the effects of PS. But if you are on battery power which degrades over time/load..... better to have it buffered with a regulator/C-Multi. Or at least bypass the battery with a C value.
Thats only IME, of course.
THx-RNMarsh
Last edited:
True, BUT ...
Batteries are good for pure class A operation only. Haven't done much experiemneting, but the little I did clearly shows that they definitely donot like being asked for lots of current in one go.
They are impractical for power amps. I remember a Stax power amp from way back with weighed in at around 20 kg/44 lbs on its own, but its battery pack was on a cart you had to lug because that weighed around 60 kb/132 lbs. And it took its sweet time to charge up.
Tue, for preamps you have no weight problem, but you still have the problems of charging them and their real esate requirements needed to obtain any meaningful voltage are no joke either.
Ultmately, as you said yourself John - the CAN be better than PSU's, but that's a comment you can attach to any poorely designed PSU.
Yes, batteries are annoying, but what a difference !
It is not the noise of the batteries itself, but all the AC and HF currents flowing between grounds of the various AC powered requirements witch are the problem, specially when wiring is not symmetrical.
R core AC transformers are better on this point of view, because the capacitances between primary and secondary coils are minimized, but not perfect. Not to talk about the ground loops that occurs with earth connections.
Attachments
Last edited:
Someone kindly sent me a copy, again too many dated op-amps used to exaggerate effects and IMO mostly diff-gain and diff-phase at audio frequencies, not a new concept.
The staircase plus AC waveforms are lifted right from video tests and he is not particularly careful about separating input from output effects. The staircase hardly exercises the input transfer function at all (for any even remotely recent op-amp) but the effects are heavily influenced by the output offset and load referred back to the input.
Using the LM301 in one example seems rather dated, and the results are obvious from looking at the open-loop gain/phase plots.
Ron is a good guy and sure likes his maths, if the audio community needs a refresher (too long buried in digital?) he provides useful input.
I do agreeThis simple 100mA loaded (250mA max) PSU shows (almost) 90dB PSRR at 10Mhz!
I would think at 10MHz the raw source of energy storage has little effect. A while back I suggested 36V locomotive starting batteries, maybe 200mV sag at 0-1000A transient (assuming the wire has enough 0's in gauge).
This is over the top DIY after all.
- Status
- Not open for further replies.