So is this thread about interference or noise? Or parasitic oscillation - there are various sources of what can sound like noise. In any case, surely all these topics are fully explored in other threads?filthyone said:
It is good that the OP has now learnt one of the basics of noise: it comes mainly from the first stage. That may help us know at what level to pitch the thread.
Thank you very much guys for the inputs, especially Bespav for the read material, very interesting.
Well, actually more about noise like the usual constant hissing and the mechanisms that contribute to it. But, why not, more broadly, anything that comes from the output of an amp that it shouldn't be there. I found some sporadic info on other threads, but i was hoping to collect all of the wisdom and experiences on the subject in this thread.
I clearly stated in post #1 that i am an amateur with basic knowledge, willing to learn more about the subject. Like, i suppose, everyone of you was at some time
Truth is that i already learned a lot from you and now i know where to look for more in-depth information.
In other news, i see everyone mentioning the input stage, but what about the VAS? I always thought that the VAS, being a transimpedance amplifier with high voltage gain, was a major contributor to total noise? What about it and it's current source?
Well, actually more about noise like the usual constant hissing and the mechanisms that contribute to it. But, why not, more broadly, anything that comes from the output of an amp that it shouldn't be there. I found some sporadic info on other threads, but i was hoping to collect all of the wisdom and experiences on the subject in this thread.
I clearly stated in post #1 that i am an amateur with basic knowledge, willing to learn more about the subject. Like, i suppose, everyone of you was at some time
Truth is that i already learned a lot from you and now i know where to look for more in-depth information.
In other news, i see everyone mentioning the input stage, but what about the VAS? I always thought that the VAS, being a transimpedance amplifier with high voltage gain, was a major contributor to total noise? What about it and it's current source?
Hi, I find it interesting how and why people ask basic questions here, whether they have looked elsewhere etc, but not been able to find answers. Do you mind telling me what led you here? Yet another basic question
The reason the input stage is the most critical is because that's where all the gain is available. Noise in the base of an input transistor gets amplified by that stage, noise in a VAS stage doesn't see the input stage gain. So for noise to be comparable it would have to be larger than the input noise by the gain factor of that first stage.
Short answer: usually negligible compared to the input stage.
Longer, but still simplified answer: in an overall feedback amplifier, you can calculate the noise of stage n back to the input by dividing it by the total open-loop gain of stages 1 ... n-1. That is, the noise of the first stage occurs directly at the input, the noise of the second stage is suppressed by the open-loop gain of the first stage, the noise of the third stage is suppressed by the open-loop gain of the first and second stages and so on.
Reason why it's simplified: to be completely correct, you have to consider voltage and current noise and four different types of gain: voltage, current, transimpedance and transadmittance gain. That doesn't change the basic idea, but just makes the calculations more complicated.
Shouldn't questions that YOU find basic (and probably are) be asked in this forum? Or are you accussing me for being a troll wasting your time?
Yes, i did look elsewhere for several days, and the reason i am here is my because i want to learn more about noise directly from guys that share the same interest as me and have some knowledge to spare. Also, i was hoping that this thread could potentially help others like me some day.
Thanks MarcelvdG and bwaslo for your answers. I think i get it now!
They are, often the best way to find something here is to do a Google search and add diyaudio at the end. There are quite a few good suggestions on the thread I linked to about getting better search results. People often say, Google is your friend, but it's not really it can be quite restrictive.
https://leachlegacy.ece.gatech.edu/papers/MultiStage.pdf
Little dense but goes through the multistage noise analysis.
Little dense but goes through the multistage noise analysis.
Anyone who's a Hifi enthusiast, whether consumer or DIY, will do odd things like listening for faults. The seeds of doubt are sown by posts or article references here, comments on other sites etc. and this feeds into paranoia - the fear that your system may not be as perfect as the other guy's and you may be missing something 
Yes, we all started knowing nothing. We learnt by studying, listening, thinking, experimenting. Most of us didn't start a thread asking for everyone else to tell us all they know about some subject.filthyone said:
On noise itself it appears that you do not have "basic knowledge", but almost no knowledge. About the most basic possible knowledge of noise would be to know that it arises mainly from input stages, which you didn't know. As I said, knowing where we are starting from helps us pitch answers at the appropriate level. You and Einstein could ask the same question "Where does noise come from?" but I might give quite different answers because those same words would have quite different meanings due to the different context of who is asking the question.
Classic newbie mistake to think that all knowledge on some subject can be gathered together in one place by asking a few apparently simple questions. I have already highlighted that we need to be clear whether we are talking about noise or interference or parasitics - each of those would need separate threads, which may then need to separate further. We would end up with the online equivalent of chapters of a book, but unlike a proper textbook half of what is written will be untrue because the internet is a wonderful place for people to share their confusions with each other.
Basic questions are often asked here. The usual answer is to point someone to where they can find the answer - which in many cases they could have found for themselves. However, "tell me all you know about noise" is not a basic question and the most helpful answer may have to be "I can't tell you because it would take too long and you would not understand most of the answer".
So, helpful questions might have been:
"Where can I learn about noise in audio circuits?"
"How can I reduce the thermal noise of a resistor?"
"Why is there no partition noise from junctions in a circuit, even though the current splits into two paths?"
Yes, that we can agree on. We are told that the new forum software (coming soon) will be much better in that respect. In the meantime, use google.
OK DF96, i get your point. It seems to be a very broad subject indeed, so let me restate my question as follows:
"I want to learn more about noise in conventional amplifiers, it's causes and ways to eliminate it. Can you suggest material to read, or share some practical knowledge? "
I don't know much (well, anything) about noise, because i never had insterest on the matter before. Now i do, and i'm already starting to learn (thanks for the application notes)
Ian, i totally agree with you. I have become fixated on a non issue, that it never bothered me before. One day, or actually one very silent night, i noticed some hissing comming out of the tweeters and i wondered what can be done to reduce it. It's not nearly as bad as i may have made it sound like, it's probably totally normal, and i admit i am being kind of paranoid.
For the record, my main amp that i use all the time is based on these boards NEW LJM L20 V9 Two Channels Power Amplifier Kit | eBay
which i think is a copy of the blameless topology with some changes in the VAS and driver stages. Also, to answer some previous questions, i have installed rc filtering right at the input RCAs (1k and 470pF), i have installed a ground loop breaker (which did a great job in my current setup and was totally needed). I can assure you that the amp is totally immune to rfi, at least while testing with a cellphone calling directly around the amp.
"I want to learn more about noise in conventional amplifiers, it's causes and ways to eliminate it. Can you suggest material to read, or share some practical knowledge? "
I don't know much (well, anything) about noise, because i never had insterest on the matter before. Now i do, and i'm already starting to learn (thanks for the application notes)
Ian, i totally agree with you. I have become fixated on a non issue, that it never bothered me before. One day, or actually one very silent night, i noticed some hissing comming out of the tweeters and i wondered what can be done to reduce it. It's not nearly as bad as i may have made it sound like, it's probably totally normal, and i admit i am being kind of paranoid.
For the record, my main amp that i use all the time is based on these boards NEW LJM L20 V9 Two Channels Power Amplifier Kit | eBay
which i think is a copy of the blameless topology with some changes in the VAS and driver stages. Also, to answer some previous questions, i have installed rc filtering right at the input RCAs (1k and 470pF), i have installed a ground loop breaker (which did a great job in my current setup and was totally needed). I can assure you that the amp is totally immune to rfi, at least while testing with a cellphone calling directly around the amp.
Not clearly stated yet--- "noise" is like bugs in your house. There are many kinds. They get in many ways. Termites come up from the ground. Bees come in the attic. Fruitflies come in with the groceries. Stinkbugs creep through windowsill gaps. You usually must do several changes to get a real reduction. And then you find more bugs not noticed before, and must work on them.
Specifically: hisss, hum/buzz, radio, and elevator clunks are very different things, and may get-in different ways.
Everything hisses. In a "Well Designed" low-hiss amplifier chain, *usually* the first stage has significant gain. Therefore if the second stage has similar intrinsic hiss, the first stage's amplified hiss overwhelms the second stage hiss.
(There is a point here: The input stage of a conventional power amp actually gets little voltage gain in the top of the audio band, limited by the compensation capacitor. I've never really looked at that.)
SPICE can compute the noise contribution of each part toward the final output (or input-reference) hiss. The mess of numbers can be bewildering. Hiss of active devices is often not well-modeled.
Now look at the WHOLE system. I don't listen to shorts for fun. I have ANOTHER amplifier in front of the power amp, a Line Amp or PreAmp. In most useful operation, THAT amplifier will dominate the power amp's hiss. Taking the very long view, the playback hiss "should" be dominated by the microphone amp in the recording studio (and in many cases, is). Gains should be trimmed so all later amps have lower hiss (at their high working signal level) and add little to the studio/mike/preamp hiss. Sometimes a later bottleneck sets the hiss. High-level music through vinyl or tape, the media sets a hiss level higher than the original performance. Some over-produced recordings have so many sub-optimal gain choices that the mix-desk and sidecar hiss dominates.
If the hiss rises when you remove your shorts and connect to an actual playback system, whacking hiss out of the power amp is a low/no-profit labor. The stage after the volume control is more likely what you hear when you USE the system. Vinyl cleanliness (both your dust and the crap fillers in commercial vinyl) is well known.
The "1Meg" at a tube amp input is NOT a hiss source. In the test above it is bypassed with a short and clearly out of the picture. It is generally true that the input grid(whatever) resistor is bypassed by whatever is driving it. It may appear for "open" input, but again that is not a common listening condition. And open-input begs all the electrical buzz and hash in the room to come in at high amplitude.
Hiss is POWER, not just voltage or resistance. The 10K notional resistance is lower hiss V but higher hiss I than the 1meg. All the resistances around a node split the same power by the usual network formulas. If we had ideal transformers, mucking with impedance would be pointless.
Douglas Self's writings have some examples of power amp input stages with extra-satisfactory low hiss. But even following his low-hiss line-amps I'm dubious such designs are needed.
Power-supply crap is another notorious source of "noise", though rarely hiss. Since we quit batteries, we are plagued with 50/60Hz artifacts, and stray pops from large loads switching on/off. This is usually solved by a several-layer system. The Main Caps smooth the DC. The power amp topology is chosen to reject supply crap. NFB around the amp helps. Of course like house insects, thick walls may have cracks which let bugs/crap in. The classic '301/'741/etc opamp has good PSRR at DC but near zero at 1MHz because the compensation cap injects from rail to hi-gain node. This may not be a problem if the main rail caps hold good to past the audio band.
Radio troubles are getting worse. In my youth I only had to deal with a 50,000 Watt transmitter a mile behind my house. Now I have 0.1W transmitters EVERYwhere, often inches from my audio. Metal boxes help but the devil is in the small cracks at joints and jacks. Low-pFd bypass caps may actually inject radio into the "ground" system.
Tubes do have a higher hiss level than transistors because they work 3X hotter. Hiss is random motion and mostly due to thermal agitation of atoms (charges). Hotter is more excited and higher hiss power. Tubes also have high 1/f (bass) hiss. In real life this is due to "dirt in the cathode". Transistors used to be as bad as or worse than tubes, but for other reasons semiconductor processors got obsessed about "lattice defects" and super-purification, so a 1970 transistor has 1/f corner at the bottom of the audio band instead of midrange.
The "number of transistors" in a circuit block does NOT imply more hiss. I know many too-simple amplifiers which hiss like a steam-leak because too many compromises were needed. Several of the lowest-hiss amplifiers made are just filthy with transistors. As mentioned, ideally only the first few devices really matter for hiss. In later stages added transistors may *reduce* hiss by isolating side-paths and biases. Adding a part in the 2nd stage can sometimes reduce overall hiss by allowing the 1st stage to work at higher gain making all later stages' hiss less important.
In all of this, the output stage (where "AB" is used) should be the least of our worries. I do know one non-Hi-Fi case where output AB affects idle hiss. The Fender 5F6a runs the output "cool" for tolerable life with monster power in hand. Enough so the gain at idle is several dB down from the high-power gain, borderline crossover distortion. Insufficient NFB to mask this completely. It idles relatively low-hiss for a BIG amp, yet BARKS the big chords. This of course is not re-production but music Production, a dynamic tool for the musician to work with.
Specifically: hisss, hum/buzz, radio, and elevator clunks are very different things, and may get-in different ways.
Everything hisses. In a "Well Designed" low-hiss amplifier chain, *usually* the first stage has significant gain. Therefore if the second stage has similar intrinsic hiss, the first stage's amplified hiss overwhelms the second stage hiss.
(There is a point here: The input stage of a conventional power amp actually gets little voltage gain in the top of the audio band, limited by the compensation capacitor. I've never really looked at that.)
SPICE can compute the noise contribution of each part toward the final output (or input-reference) hiss. The mess of numbers can be bewildering. Hiss of active devices is often not well-modeled.
Now look at the WHOLE system. I don't listen to shorts for fun. I have ANOTHER amplifier in front of the power amp, a Line Amp or PreAmp. In most useful operation, THAT amplifier will dominate the power amp's hiss. Taking the very long view, the playback hiss "should" be dominated by the microphone amp in the recording studio (and in many cases, is). Gains should be trimmed so all later amps have lower hiss (at their high working signal level) and add little to the studio/mike/preamp hiss. Sometimes a later bottleneck sets the hiss. High-level music through vinyl or tape, the media sets a hiss level higher than the original performance. Some over-produced recordings have so many sub-optimal gain choices that the mix-desk and sidecar hiss dominates.
If the hiss rises when you remove your shorts and connect to an actual playback system, whacking hiss out of the power amp is a low/no-profit labor. The stage after the volume control is more likely what you hear when you USE the system. Vinyl cleanliness (both your dust and the crap fillers in commercial vinyl) is well known.
The "1Meg" at a tube amp input is NOT a hiss source. In the test above it is bypassed with a short and clearly out of the picture. It is generally true that the input grid(whatever) resistor is bypassed by whatever is driving it. It may appear for "open" input, but again that is not a common listening condition. And open-input begs all the electrical buzz and hash in the room to come in at high amplitude.
Hiss is POWER, not just voltage or resistance. The 10K notional resistance is lower hiss V but higher hiss I than the 1meg. All the resistances around a node split the same power by the usual network formulas. If we had ideal transformers, mucking with impedance would be pointless.
Douglas Self's writings have some examples of power amp input stages with extra-satisfactory low hiss. But even following his low-hiss line-amps I'm dubious such designs are needed.
Power-supply crap is another notorious source of "noise", though rarely hiss. Since we quit batteries, we are plagued with 50/60Hz artifacts, and stray pops from large loads switching on/off. This is usually solved by a several-layer system. The Main Caps smooth the DC. The power amp topology is chosen to reject supply crap. NFB around the amp helps. Of course like house insects, thick walls may have cracks which let bugs/crap in. The classic '301/'741/etc opamp has good PSRR at DC but near zero at 1MHz because the compensation cap injects from rail to hi-gain node. This may not be a problem if the main rail caps hold good to past the audio band.
Radio troubles are getting worse. In my youth I only had to deal with a 50,000 Watt transmitter a mile behind my house. Now I have 0.1W transmitters EVERYwhere, often inches from my audio. Metal boxes help but the devil is in the small cracks at joints and jacks. Low-pFd bypass caps may actually inject radio into the "ground" system.
Tubes do have a higher hiss level than transistors because they work 3X hotter. Hiss is random motion and mostly due to thermal agitation of atoms (charges). Hotter is more excited and higher hiss power. Tubes also have high 1/f (bass) hiss. In real life this is due to "dirt in the cathode". Transistors used to be as bad as or worse than tubes, but for other reasons semiconductor processors got obsessed about "lattice defects" and super-purification, so a 1970 transistor has 1/f corner at the bottom of the audio band instead of midrange.
The "number of transistors" in a circuit block does NOT imply more hiss. I know many too-simple amplifiers which hiss like a steam-leak because too many compromises were needed. Several of the lowest-hiss amplifiers made are just filthy with transistors. As mentioned, ideally only the first few devices really matter for hiss. In later stages added transistors may *reduce* hiss by isolating side-paths and biases. Adding a part in the 2nd stage can sometimes reduce overall hiss by allowing the 1st stage to work at higher gain making all later stages' hiss less important.
In all of this, the output stage (where "AB" is used) should be the least of our worries. I do know one non-Hi-Fi case where output AB affects idle hiss. The Fender 5F6a runs the output "cool" for tolerable life with monster power in hand. Enough so the gain at idle is several dB down from the high-power gain, borderline crossover distortion. Insufficient NFB to mask this completely. It idles relatively low-hiss for a BIG amp, yet BARKS the big chords. This of course is not re-production but music Production, a dynamic tool for the musician to work with.
Last edited:
As said, hiss is little problem in decently designed amplifiers.
What should be paid attention is the first stage of amps and préamps.
The place where hiss and hum is of a paramount importance is microphone amplifiers.
Hiss and hum is a serious issue where signal levels ar very low. There, active parts must be very low noise and impedances kept as low as possible. These noises will be largely amplified so the Signal/ Noise ratio should be optimized. This is what makes a cheapo mike preamp, cheap and a pro quality mic preamp a challenge not many engineers are able to master.
What should be paid attention is the first stage of amps and préamps.
The place where hiss and hum is of a paramount importance is microphone amplifiers.
Hiss and hum is a serious issue where signal levels ar very low. There, active parts must be very low noise and impedances kept as low as possible. These noises will be largely amplified so the Signal/ Noise ratio should be optimized. This is what makes a cheapo mike preamp, cheap and a pro quality mic preamp a challenge not many engineers are able to master.
^ and phono, if that's your thing. Although I think you implied phono preamps. Unfortunately few of us have control over the mic preamps being used (nor the subsequent processing) for the music we enjoy.
PRR covers, in pretty tractable language, what the concerns are -- input stage (which, for clarity *includes* the noise injection from the current source), power supply hash into mediocre PSRR (esp around the VAS), RF/environmental noise at vulnerable connections, etc.
Unfortunately few of us have control over the mic preamps being used (nor the subsequent processing) for the music we enjoy.PRR covers, in pretty tractable language, what the concerns are -- input stage (which, for clarity *includes* the noise injection from the current source), power supply hash into mediocre PSRR (esp around the VAS), RF/environmental noise at vulnerable connections, etc.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.