Missing the hills, rain, more rain, black puddings and Manchester Tarts (oh the M6 is still the same, static)
I tried all of that, this May
Yes, rugged is how I would describe the country side around the Pennines, of course commuting from Blackburn to Accrington every day I don't see the best of it.
And ever lower cost --- and for portability in products.
But not for this DIY'er. I'm sticking to the TH til the end.
I'll do the SM transistor on a header, though.
BTW - I find it faster and easier and more flexible to do bread-boarding with 'flying' parts crossing all over the area.
THx-RNMarsh
Great help to us all. especially the conclusion of your work. Makes life so much easier to get best results from battery powered audio circuits.
Any Z and noise data near end-of life? Or, approching full discharge? -- other than fully charged batteries?
THx-RNMarsh
How did your measurements correlate to those of NIST? One of the problems is that the battery manufacturers change chemistry and even manufacturers all the time. I visited one 9V factory in China that displayed batteries with every major brand that were made there. Further there are not many vendors of the chemistries and everyone evolves in sync.
is there a correlation between ESR at a given frequency and noise at that frequency? ESR would be much easier to measure.
If you are using different FFT resolutions for each segment of the measurement does the software correct for the different noise bandwidth in each segment? This issue would be a real challenge to sort out manually.
That analyzer looks pretty capable but at $2K+ is out of my budget.
SMD is here to stay, most assemblies these days are surface mount based for the many advantages it brings, size reduction, better EMC, better thermal management, smaller signal loops etc. I have seen technicians replace 0201 resistors (and some engineers), and on some designs we have placed 0201s between the balls on a BGA device on the same side.
The main advantage is size, the picture below shows a traditional 8 leg dual in line op-amp footprint with a SMD circuit (Audio) consisting of two dual op-amp packages and associated resistors and caps (HDI PCB construction, microvias laser drilled).
The one advantage of designing the PCBs is you can zoom in, so size becomes irrelevant when laying out, and most assemblies are automated so manual dexterity is only a concern when re-work is required. A big magnifying glass and steady hands
Unfortunately that construction is only meaningful in larger quantities. The MOQ to justify that stuff is usually larger than the commercial life of a high end audio product. If you need to spin the board a few times its even more painful. Laser drilled blind vias in pad are very much state of the art and only the most premium board shops can even do that stuff. It makes sense for a million piece cell phone order.
With the tiny resistors tempco and modulation with signal will be a much bigger issue,there isn't much thermal mass to absorb changes.
I am adapting to surface mount. I acquired a good stereo microscope and a handful of other specialized tools. I can deal with 0201's if necessary but not because I want to. Coming from through hole 0604 looks tiny. In the world of SMT they are gigantic.
No, I cannot explain that. For some of the NIST batteries, there seems to exist
no 1/f corner. And it is claimed that the noise mechanism is Johnson noise over
the generator resistance.
In my loaded cases, there is a 24 Ohms resistor (47 par. 47 wire wound) in parallel
to the battery. If there was only Johnson at work, that 24 Ohms alone would keep
the results below the 0 dB line (1 nV ~ 60R), no matter what is inside the battery.
But the load makes the result plots soar high above thermal noise.
Do you think that the 1/f noise of the amplifier itself is OK? There are 3 hi pass poles at
slightly less than 0.1 Hz, so the 1/f should not continue forever. But that would be to left,
outside of the picture.
Last edited:
Do you think that the 1/f noise of the amplifier itself is OK? There are 3 hi pass poles at
slightly less than 0.1 Hz, so the 1/f should not continue forever. But that would be to left,
outside of the picture.
Your plots are 20-30dB/dec in some cases, it might be worthwhile to find the root cause.
Yes, the software corrects for the 10 dB steps at each decade boundary, one
would get a staircase otherwise. My 89441A has the generator option also, so
I could do Bode sweeps. But then this correction has to be switched off since
the sine wave fits the whole frequency bin, and the Hann window would also
not be appropriate.
There are lots of other things that must be set up. One can put that into
a PRESET, but under remote control, I can write it into the source code,
why I did this & that.
Yes, the name of that man was Julius von Hann. He is not a twin brother
of Mr. Hamming, who has a different window all for himself. There has never
been a guy called Hanning in signal processing, at least not such a prominent
one.
Last edited:
Just to be safe I would try a white noise source into the instrument. Getting one that is uniform to .1 Hz may be a challenge. However sometimes the arithmetic may not actually be what it seems. Agilent has been known to make mistakes even.
The specs on the vector analyzer don't mention residual noise below 1 KHz and it drops ABOVE 40 KHz. Its quite possible its much higher at 10 Hz and 1 Hz. Its phase noise residual is pretty high as well. However it can demodulate a CDMA signal which is quite the undertaking.
An alternative would be HPW's HpW-Works cross correlation software + a good stereo sound card and preamp. 1 Hz may not be possible but 10 Hz and even 5 Hz is. I think you can try the demo version to see if it works.
I am always cross checking measurements when they seem unusual, which is often for me.
The specs on the vector analyzer don't mention residual noise below 1 KHz and it drops ABOVE 40 KHz. Its quite possible its much higher at 10 Hz and 1 Hz. Its phase noise residual is pretty high as well. However it can demodulate a CDMA signal which is quite the undertaking.
An alternative would be HPW's HpW-Works cross correlation software + a good stereo sound card and preamp. 1 Hz may not be possible but 10 Hz and even 5 Hz is. I think you can try the demo version to see if it works.
I am always cross checking measurements when they seem unusual, which is often for me.
That is a real challenge. It seems, nature abhors DC also, not only vacuum.
The AD4898 with its 30 Hz or so corner frequency and 60 Ohms thin film
is probably as close as I can get, and this displays cleanly, I think.
It IS much higher, so much that it takes 60 to 80 dB of external preamp gain to mask.
I already must be careful not to compress the noise.
The preamp masks the noise of the 89441A just as the 60 Ohm resistor masks
the preamp's 1/f corner from 30 down to 5 Hz.
No way. I would rather modify one of my own RF digitizers. I'm after clean oscillators
and a clean dc supply and reference is only a precondition, not the goal.
Maybe it was a not so good idea to solder the 22 Wima 10u-Capacitors into the
preamp input. Living with the 10 Hz lower corner and computing the rolloff away
would actually give MORE dynamic range in the 1/f region. The corner must
stay constant, then. Electrolytics and thick film resistors on a sound card won't do.
Thank you for the measurements.
> Sizes does matter .....
I wonder if it has anything to do with the internal resistance.
NiMH has high capacity but internal resistance is high compared to NiCd.
There are special high discharge current NiMH types used in RC models or power tools.
But they are not AA, but Sub-C or D instead.
Master Instruments - High Current NiMH
The other type suitable for high discharge (and most likely very low internal R) is :
Lithium Iron Phosphate Battery Technology | A123 Systems
http://www.a123systems.com/Collateral/Documents/English-US/A123 Systems ANR26650 Data Sheet.pdf
Patrick
> Sizes does matter .....
I wonder if it has anything to do with the internal resistance.
NiMH has high capacity but internal resistance is high compared to NiCd.
There are special high discharge current NiMH types used in RC models or power tools.
But they are not AA, but Sub-C or D instead.
Master Instruments - High Current NiMH
The other type suitable for high discharge (and most likely very low internal R) is :
Lithium Iron Phosphate Battery Technology | A123 Systems
http://www.a123systems.com/Collateral/Documents/English-US/A123 Systems ANR26650 Data Sheet.pdf
Patrick
One trick is to mix down RF noise to DC, Quantec sold a box that did that. I have one.
EDIT - Model 420, can't find a free schematic.
Last edited:
That is known as direct conversion and was used in a Ten-Tec receiver late 60's early 70's if I recall correctly.
The inverse which I do use is to mix LF with HF to move stuff up a bit.
The last trick is to use two RF oscillators to beat and create an audio band signal. As the distortion harmonics are nowhere near the audio beat you can get very low distortion.
But everything is DSP today as digital rules and is always poifect.
- Status
- Not open for further replies.