Thanks for the reply. Actually this is easily testable, as others have mentioned.
Instead of analog scopes with persistence, we could just use a DAC whose input is connected at the output of an ADC.
The ADC and DAC could be synchronized, i.e. use the same sample clock. Actually this is not really needed, it just makes the results easier to interpret.
Obviously, the ADC should have at the same or higher ENOB, sample rate and linearity than the DAC.
We could then compute the total error of the ADC and DAC. If they have the same ENOB and sample rate and are synchronized, the total error should be smaller than 2 LSB.
This test could be done with a PC sound card by connecting one of the outputs to one of the inputs.
I would do it but my sound card is internal, therefore the computer noise will be much greater than the error we are trying to measure.
Will someone with an external sound card like to perform the test ?
47 (More than one answer!), 072, 83, 107, 300, 555, 703, 709, 722, 923, 7400
Wrong 83, but a valid answer! (Hint 83 is a partial part number!)
107 and 300 good,
There still are some antiques!
If you went into a parts store and asked for a #47, the guy would have reached under the counter as it was one of the most common replacement parts!
Wrong 83, but a valid answer! (Hint 83 is a partial part number!)
107 and 300 good,
There still are some antiques!
If you went into a parts store and asked for a #47, the guy would have reached under the counter as it was one of the most common replacement parts!
Last edited:
I don't consider my laptop portable in this context. No need to add confusion that sound card is more than acceptable.
Sy,
Well, NOT QUITE.
If they are 10,000 anecdotes that return a 10,000 in 10,000 preference, non-blind, we can actually use classic statistic analysis (assuming we took the module at Uni, I did, twice in the space of 10 years) to establish a confidence interval to see how likely it is that the results are just a mass delusion.
It is quite illuminating, should one do so.
Well, I have not particularly evaluated or experienced these events, so I could not possibly comment.
I do however note that I commonly encounter UFO's...
That is, things that fly that I cannot reliably identify as air-plane, helicopters, alien spaceship, bird, insect, heat inversion, fata morgana, or sundogs.
We may even agree on that. I would use the term "adequate" in the sense an auditor might, that is, improvements are possible and desirable, but at least it's not Enron...
Ciao T
Well, NOT QUITE.
If they are 10,000 anecdotes that return a 10,000 in 10,000 preference, non-blind, we can actually use classic statistic analysis (assuming we took the module at Uni, I did, twice in the space of 10 years) to establish a confidence interval to see how likely it is that the results are just a mass delusion.
It is quite illuminating, should one do so.
Well, I have not particularly evaluated or experienced these events, so I could not possibly comment.
I do however note that I commonly encounter UFO's...
That is, things that fly that I cannot reliably identify as air-plane, helicopters, alien spaceship, bird, insect, heat inversion, fata morgana, or sundogs.
We may even agree on that. I would use the term "adequate" in the sense an auditor might, that is, improvements are possible and desirable, but at least it's not Enron...
Ciao T
Hi,
I agree.
One COULD do so, however, that would be kind of self defeating, unless the AD operates without "fuzzy distortion".
It will be, if you use sufficient averaging to remove the effects of "fuzzy distortion" (n.b. I have borrowed the term "fuzzy distortion" from Malcolm Hawkesford)
It could be, but it would require a way of analysing the data that does not rely on averaging large amounts of data (which does the precise and exact opposite of "infinite persistence").
Ciao T
I agree.
One COULD do so, however, that would be kind of self defeating, unless the AD operates without "fuzzy distortion".
It will be, if you use sufficient averaging to remove the effects of "fuzzy distortion" (n.b. I have borrowed the term "fuzzy distortion" from Malcolm Hawkesford)
It could be, but it would require a way of analysing the data that does not rely on averaging large amounts of data (which does the precise and exact opposite of "infinite persistence").
Ciao T
Last edited:
PIM Testing Growing in Importance as 4G Rolls Out in Europe**|**October 7, 2011**|**Microwave Journal
FYI as per JC
FYI as per JC
Last edited:
PMA, you already did the ADC-DAC test.
Does "FS" mean full scale, i.e. the max. amplitude for the output of the DAC ?
If that's true then from your data the max. error is 0.07% of the full scale which is 10.48 bits ENOB.
Can you post more info about the ADC and DAC, like ENOB, linearity ?
Anyway it seems way less than the 19-20 bits ENOB of current ADCs/DACs.
It's also possible that the circuits around the ADC and DAC introduce their own non linearities and/or noise which could explain the bad results.
Can you post more info about those circuits ?
Scott,
I do not know YOUR Laptop.
Mine has 8 Hours real use battery life, is around 20% larger than an A4 page and weights well under 2kg. It is most definitely "portable in this context" and with the 1616m one could easily record more than two channels too...
They do make a smaller and lighter weight version as well, with a 13" screen and without the blue ray drive. A bit weedy for my taste.
Ciao T
I do not know YOUR Laptop.
Mine has 8 Hours real use battery life, is around 20% larger than an A4 page and weights well under 2kg. It is most definitely "portable in this context" and with the 1616m one could easily record more than two channels too...
They do make a smaller and lighter weight version as well, with a 13" screen and without the blue ray drive. A bit weedy for my taste.
Ciao T
Just to be clear, I agree that the error should be computed sample by sample with no averaging.
If the ADC also generates "fuzzy distortion" then the maximum error will be even bigger, thus easier to measure.
For example if both the ADC and DAC have an error of 10*LSB instead of 1*LSB then the total error will be 20*LSB instead of 2*LSB.
If you do the sawtooth test with the external EMU sound card, please let us know what you find. From the specs of the sound card, it seems to be up to the task, so you should be able to expose any "fuzzy distortion" that might be present either in its DAC or in its ADC.
#47 is bayonet!
But I thought this would be easy. Apparently I am the old geezer now.
Is this really the same context, MH is the king of fuzzy thinking? BTW you guys must really hate class D.
I have no choice, I need to use a fire breathing workstation laptop that burns through its battery in 2hr and weighs a ton. That's beside the point I don't know of anyone doing field recording, i.e. walking around in the jungle, rain forest, etc. with a laptop based system. I'm sure you can find one, don't bother it only adds to the confusion.
Part Numbers
OK Ed,
Without Googling, I can only ID a few:
#47 - Most common pilot lamp in 6-volt tube filament equipment...oft found behind beautiful glass jewels, as in Fender amps, ad nauseum.
TL072 - FET Op amp.
WE300 - Western Electric triode. It is apparently associated with religeous rituals nowadays.
NE555 - Timer.
uA709 - Op amp.
CK722 - Germanium transistor, my first projects utilized these unstable gems.
SN7400 - TTL quad NAND gate.
Pretty poor showing, I'm afraid: looking forward to the other answers!
Howie
Howard Hoyt
CE - WXYC-FM
UNC Chapel Hill, NC
www.wxyc.org
1st on the internet
OK Ed,
Without Googling, I can only ID a few:
#47 - Most common pilot lamp in 6-volt tube filament equipment...oft found behind beautiful glass jewels, as in Fender amps, ad nauseum.
TL072 - FET Op amp.
WE300 - Western Electric triode. It is apparently associated with religeous rituals nowadays.
NE555 - Timer.
uA709 - Op amp.
CK722 - Germanium transistor, my first projects utilized these unstable gems.
SN7400 - TTL quad NAND gate.
Pretty poor showing, I'm afraid: looking forward to the other answers!
Howie
Howard Hoyt
CE - WXYC-FM
UNC Chapel Hill, NC
www.wxyc.org
1st on the internet
I for one don't hate the little giants at all. There is no reason why an amplifier topology that is nothing but institutionalized crossover - the exact opposite of class A - should sound so well, but I have heard some truly amazing applications. The same sort of dryness (I need to resort to wine terms) that I hear in SET's, but with much less distortion. Plus tremendous command over the driver in the low end. I fully concur with Stereophile's review of the Hypex amp. With the addition that the same amazing sound can be had from much smaller packages, like the TA2021B on 12V. Dead silent on a simple wall wart, even on a pair of highly efficient Tannoy K3838's.
But when you look at the output signal on a scope, it looks much like the plot Pavel (I think) showed of a DAC output signal plus dithering. The analogy falls short, though, because in class D the fuziness is much higher in frequency.
vac
Cool, I agree but it's all one bit dither nontheless. The noise shaped dither is supra-audio and there is denial that the underlying theory holds up. The nay sayers don't have enough of a grasp on the theory to usefully critique it so we will remain at an impass.
Last edited:
- Status
- Not open for further replies.