They do run warm - typically 20-25 c above ambient - but I wouldn't worry about it. I have a class A power amp that's been running at 65C for the last few days and last year as well when it was hot because we had a heat wave* - and the same over the last few years with an older version (sx-Amp) and they have never failed. The original amp was built in 2011 and ran perfectly for 8 yrs. I've seen devices in professional applications running at these temps year in an year out.
An opamp running at these temps in a non-critical application like audio is a not an issue
*UK heatwave c. > 30 C. Give a thought for the people in Baghdad where temps have been >52 C
An opamp running at these temps in a non-critical application like audio is a not an issue
*UK heatwave c. > 30 C. Give a thought for the people in Baghdad where temps have been >52 C
@dreamth, It is a waste of my time having a discussion with you about design, I will bow out...
Andrew I respect your input and work, cheers mate
Andrew I respect your input and work, cheers mate
Last edited:
@rsavas
"
@dreamth, It is a waste of my time having a discussion with you about design, I will bow out...
Andrew I respect your input and work, cheers mate"
I don't see any discussionDiscussions are done WITH people, not AT people!
"If lf356 is your choice, that is your option, probably still in production for people like you and legacy designs 🙂 "
I said nothing about buying LF356H ! I just have a lot of them from x-ray CT equipment i used to fix a decade ago and i thought of finally use them if they are good enough for the job.
In my humble experience H cans are better than usual cans especially when linked to thermal management and thermal induced distortions can be significant when the input is overloaded.Otherwise i can't figure why Siemens engineers made their medical equipment using such silly expensive products by the thousands.By the way X-ray imaging needs to be a low noise process too for high resolution ...
"
@dreamth, It is a waste of my time having a discussion with you about design, I will bow out...
Andrew I respect your input and work, cheers mate"
I don't see any discussionDiscussions are done WITH people, not AT people!
"If lf356 is your choice, that is your option, probably still in production for people like you and legacy designs 🙂 "
I said nothing about buying LF356H ! I just have a lot of them from x-ray CT equipment i used to fix a decade ago and i thought of finally use them if they are good enough for the job.
In my humble experience H cans are better than usual cans especially when linked to thermal management and thermal induced distortions can be significant when the input is overloaded.Otherwise i can't figure why Siemens engineers made their medical equipment using such silly expensive products by the thousands.By the way X-ray imaging needs to be a low noise process too for high resolution ...
Attachments
Marcel, I see what Hans is doing as a simple normalization not computing the equivalent input noise. You could simply take the 5534 as the reference and call it 1 so (ignoring input current noise) an amplifier with 7nV noise would be 2. With MC carts this does not show much but with MM you include the current noise contribution. Comparing SNR you take the log ratio so any normalizing falls out.
It might be related to reliability. As far as I know, ceramic and metal packages are better at keeping moisture away from the chip than plastic packages.
I think the newer plastics from the last 20 years or so are very good. At Philips we used to do 3 lots of 78 pieces in 100% humidity with bias for 1000 hrs (IIRC) to get auto qualified - that was 2000/2001. Zero failures allowed and after the test all devices had to pass all parametric specifications 100%.
Its a world away from the early plastic stuff which had all sorts of problems associated with it.
Any way, lets not corrupt a good phono thread!
Its a world away from the early plastic stuff which had all sorts of problems associated with it.
Any way, lets not corrupt a good phono thread!
Last edited:
Using the word “equivalent input noise” is not really apropriate, but in all cases I showed that what I meant with the added calculation: [A-weighted & Riaa processed output noise] / [gain at 1Khz *sqrt(20Khz)] in Volt/rtHz
When Marcel would have better looked at these calculations, he should have suggested a better terminology for what I called EIN instead.
Suggestions are welcome, because to be honest I don’t know a better word, maybe Scott’s suggestion “Normalised Input Noise” ?
Hans
You need to remember the normalized comparison is only for a given cartridge and circuit. Using the total rms noise at the output will include all sources of noise mostly useful for MM where there can be current noise contribution.
I like to use the term ‘spot noise’ to talk about the noise at a specific frequency and it should include all noise mechanisms - voltage and current. If calculating the total noise, then of course it’s integrated over the BW, and we should call it ‘total noise’ and clearly state the measurement BW.
You can only talk about A weighted noise when talking about total noise - it makes no sense in a spot noise context.
We can remove all confusion by always talking about ‘dBV’ for absolute references, and for SNR, of course we refer to the input signal level - for MM I believe 5 mV is the de facto standard, though not enshrined in any formal specification.
Equivalent input noise is IMV a useful way of comparing amplifiers provided both noise voltage and current are considered along with Rsource. The problem with output noise is you don’t have the gain magnitude so it does not allow a meaningful comparison between different gain amplifiers - almost always the casein a practical context.
You can only talk about A weighted noise when talking about total noise - it makes no sense in a spot noise context.
We can remove all confusion by always talking about ‘dBV’ for absolute references, and for SNR, of course we refer to the input signal level - for MM I believe 5 mV is the de facto standard, though not enshrined in any formal specification.
Equivalent input noise is IMV a useful way of comparing amplifiers provided both noise voltage and current are considered along with Rsource. The problem with output noise is you don’t have the gain magnitude so it does not allow a meaningful comparison between different gain amplifiers - almost always the casein a practical context.
Last edited:
I did look at the calculation and concluded it made no sense, as what you were calculating was not the equivalent input noise, neither the spot noise at some frequency or other nor a weighted average across the band. How was I supposed to know that you were not trying to calculate the equivalent input noise in the first place, but some brand new noise metric?
In any case, since to the best of my knowledge it is a new noise metric that you introduced, you can come up with any name you like, preferably one that can't cause confusion. Normalized input noise is vague enough not to cause confusion, as no-one knows what the noise is normalized to until the person using the term explains what he/she/it means.
Still trying to find the right terminology for future use to prevent this kind of miscommunications.
When computing the A-weighted S/N for a specific Cart while connected to an amp, the total A-weighted noise from the amp’s output, including all voltage and current noise elements, has to be normalised for its gain at 1Khz and compared to the Cart’s output at 5cm/sec@1Khz, mostly 5mV for an MM or 0.5mV for an MC unless a more exact figure is known.
Question: How do we call this A-weighted output noise divided by the gain at 1Khz: This is the A-weighted Equivalent Input Noise @ 1Khz, right ?
One could look at the normalised spectrum and read the spot noise at 1Khz. This is what Syn08 does in his reports, and he calls this En@1Khz.
When looking again at the normalised spectra below as given in #105, spot noise at 1Khz is resp. 1.6nV/rtHz and 5.5nV/rtHz for the used MC and MM.
But since the shapes of these noise spectra are quite different, I find this spot noise figure not as representative as the average A-weighted noise over the same 20Khz, in this case 0.63nV/rtHz and 4.4nV/rthz, quite a difference from the above spot noise figures and giving a much better feeling about their to be perceived noise performance.
Of course nobody has to use these averaged figures, but I find them much more useful than a figure for the total noise.
When the normalised A-weighted total noise in Volt be called: A-weighted EIN@1Khz, then the averaged version in V/rtHz should be called: A-weighted AEIN@1Khz (for AveragedEIN), can we agree on that ?
Hans
When computing the A-weighted S/N for a specific Cart while connected to an amp, the total A-weighted noise from the amp’s output, including all voltage and current noise elements, has to be normalised for its gain at 1Khz and compared to the Cart’s output at 5cm/sec@1Khz, mostly 5mV for an MM or 0.5mV for an MC unless a more exact figure is known.
Question: How do we call this A-weighted output noise divided by the gain at 1Khz: This is the A-weighted Equivalent Input Noise @ 1Khz, right ?
One could look at the normalised spectrum and read the spot noise at 1Khz. This is what Syn08 does in his reports, and he calls this En@1Khz.
When looking again at the normalised spectra below as given in #105, spot noise at 1Khz is resp. 1.6nV/rtHz and 5.5nV/rtHz for the used MC and MM.
But since the shapes of these noise spectra are quite different, I find this spot noise figure not as representative as the average A-weighted noise over the same 20Khz, in this case 0.63nV/rtHz and 4.4nV/rthz, quite a difference from the above spot noise figures and giving a much better feeling about their to be perceived noise performance.
Of course nobody has to use these averaged figures, but I find them much more useful than a figure for the total noise.
When the normalised A-weighted total noise in Volt be called: A-weighted EIN@1Khz, then the averaged version in V/rtHz should be called: A-weighted AEIN@1Khz (for AveragedEIN), can we agree on that ?
Hans
Attachments
"How do we call this A-weighted output noise divided by the gain at 1Khz: This is the A-weighted Equivalent Input Noise @ 1Khz, right ?"
I usually call it the RIAA- and A-weighted integrated equivalent input noise, but that's a term no-one understands, so I have to define what I mean before using it.
I usually call it the RIAA- and A-weighted integrated equivalent input noise, but that's a term no-one understands, so I have to define what I mean before using it.
Forgive me, but we seem to be making a mountain out of a molehill here.
1. Spot noise = noise at a specific frequency
2. Total noise - noise integrated over a total BW where you always quote the BW
3. Don't use A weighting on spot noise - its meaningless - only on total noise
4. Always use dBV UNLESS you are looking at SNR, in which case quote the ref signal as well
5. You can weight SNR if required since it makes sense
6. Vinoise is always preferred for amplifiers because it removes system gain and allows direct comparison of noise performance
7. If you must use Vonoise, quote the gain.
8. Use Vonoise for things like PSU noise evaluation where it makes sense since this feeds into an amplifier stage and can be referred back to the input.
1. Spot noise = noise at a specific frequency
2. Total noise - noise integrated over a total BW where you always quote the BW
3. Don't use A weighting on spot noise - its meaningless - only on total noise
4. Always use dBV UNLESS you are looking at SNR, in which case quote the ref signal as well
5. You can weight SNR if required since it makes sense
6. Vinoise is always preferred for amplifiers because it removes system gain and allows direct comparison of noise performance
7. If you must use Vonoise, quote the gain.
8. Use Vonoise for things like PSU noise evaluation where it makes sense since this feeds into an amplifier stage and can be referred back to the input.
Last edited:
Ha, ha Bonsai ,
Good point, this discussion took already much too much time for just a simple figure.
I will simply keep this AEIN for personal use, because its my favourite yardstick.
Hans
P.S1: on top of all this I would like to humbly say that you are using SNR where you actually mean S/N, an error that I made in the past where I was also corrected. SNR stands for power and S/N for voltage.
P.S.2: spot noise En@1khz as syn08 uses is the same for unweighted and A-weigted spectra.
Good point, this discussion took already much too much time for just a simple figure.
I will simply keep this AEIN for personal use, because its my favourite yardstick.
Hans
P.S1: on top of all this I would like to humbly say that you are using SNR where you actually mean S/N, an error that I made in the past where I was also corrected. SNR stands for power and S/N for voltage.
P.S.2: spot noise En@1khz as syn08 uses is the same for unweighted and A-weigted spectra.
Last edited:
X-Altra has something wrong there, there is no way an MC input has a better S/N then MM.
it's at least 10db worse because of the extra gain stage.
it's at least 10db worse because of the extra gain stage.
The numbers are correct.
The devil is in the detail: spot noise floor, total input noise, SNR and the input source Z parameters.
I kinda like this discussion over different ways of looking at the noise issue. Makes one think about the issue, which is always good.