Dip your op-amp into liquid nitrogen. That will lower noise.
I think, as you surmise, if in the application of digital electronics... shot noise could be misinterpreted as a bit of data. Especially with computers going gigahertz.
I think beyond a cetain point it is moot anyway. Even if an amp has some noise, even a tube amp, the transient noise we hear all around us in a room is louder than that coming from a well designed amp through most speakers anyway. It is S/N ratio one needs to worry about. I think -60dB to -70dB is sufficiently low so as not to worry about levels lower.
IMHO
Gabe
I think, as you surmise, if in the application of digital electronics... shot noise could be misinterpreted as a bit of data. Especially with computers going gigahertz.
I think beyond a cetain point it is moot anyway. Even if an amp has some noise, even a tube amp, the transient noise we hear all around us in a room is louder than that coming from a well designed amp through most speakers anyway. It is S/N ratio one needs to worry about. I think -60dB to -70dB is sufficiently low so as not to worry about levels lower.
IMHO
Gabe
Nelson Pass said:The article is simply mistaken.
Thank you for the support. But why did the author ever write this article? I sounds very good but the real life experince tell you different. I would also say the basic theory tells you also different.
I think that the author has forgotten the feedback thing. The conlusion he's making is
Opamps (at that time) were noisy AND also had high open-loop gain
-> High open-loop gain devices = noisy (which is obviuosly wrong)
OpAmp noise
I think the article had a few good points - for it's time.
If you try to design an op-amp to maximise open-loop gain , then you run can end up with a noisy design I think.
as has been mentioned before , it's tricky to design resisters accurately on a substrate ( and 10 years ago it was harder ) , let alone match them. capacitors are even harder etc...
and then all your components are running at the same temperature - great for transistors , not so good for resistors.
now are 'silicon' resistors inherently noisier than good metal film?
I would think so.
I don't design chips so my opinion is just that , but I would think that you get lower noise by carefully selecting discrete components than an using an opamp , even if you follow exactly the same topology.
modern IC design has come a long,long way in recent years so the difference may be so small as to be not worth worrying about.....
ray
I think the article had a few good points - for it's time.
If you try to design an op-amp to maximise open-loop gain , then you run can end up with a noisy design I think.
as has been mentioned before , it's tricky to design resisters accurately on a substrate ( and 10 years ago it was harder ) , let alone match them. capacitors are even harder etc...
and then all your components are running at the same temperature - great for transistors , not so good for resistors.
now are 'silicon' resistors inherently noisier than good metal film?
I would think so.
I don't design chips so my opinion is just that , but I would think that you get lower noise by carefully selecting discrete components than an using an opamp , even if you follow exactly the same topology.
modern IC design has come a long,long way in recent years so the difference may be so small as to be not worth worrying about.....
ray
sonnya said:I do not think his phono amp is that much overkill! or extreme...
It's only two opamps with correction! per channel.
Sonny
Yes, a little bit overkill is it but everything is pure basics, nothing weird or odd and plain components. My pcb's are like a evaluation board. It's very easy to omit parts and make some sections simplier. If you don't like cascodes don't use them. If you don't like current sources use plain resitors instead. If you want super transistors, use them etc. You can realy taylor the performance in almost any way.
mad as a fish....
Peranders , please don't think I'm having a dig at you - I'm not.
I did regard the phono stage as a little over the top , simply from my point of view , not from a high-end audiophile aspect.
to be able to get that sort of performance out of a phono preamp is quite simply out of my league - I wish I could design something as well as that, but I can't.
I have a great deal of respect for those that can.
I've been messing around with electronics for about 20 years or so and I recognise that some folk take a lot more care and effort to extract the last n-th degree of performance from their designs.
Me , I'm happy if mine stay alive long enough to test
Sorry, this is hijacking the thread , so enough now from me on this.
back to the topic of op-amps though , are fet-input op-amps noisier than bipolar op-amps?
the reason I ask is that I've found that the last few pre-amps I've built ( discrete class-a j-fets ) have seemed significantly noisier than their npn equivalents.
is this just bad transistor choice on my part or are fets noisier? and the op-amps?
ray
Peranders , please don't think I'm having a dig at you - I'm not.
I did regard the phono stage as a little over the top , simply from my point of view , not from a high-end audiophile aspect.
to be able to get that sort of performance out of a phono preamp is quite simply out of my league - I wish I could design something as well as that, but I can't.
I have a great deal of respect for those that can.
I've been messing around with electronics for about 20 years or so and I recognise that some folk take a lot more care and effort to extract the last n-th degree of performance from their designs.
Me , I'm happy if mine stay alive long enough to test
Sorry, this is hijacking the thread , so enough now from me on this.
back to the topic of op-amps though , are fet-input op-amps noisier than bipolar op-amps?
the reason I ask is that I've found that the last few pre-amps I've built ( discrete class-a j-fets ) have seemed significantly noisier than their npn equivalents.
is this just bad transistor choice on my part or are fets noisier? and the op-amps?
ray
Re: mad as a fish....
This week I will send my normal phono amp, normal headphone amp and my monster headphone amp to my pcb manufacturer. Maybe the normal ones can be of any interest? You will get extreme performance despite the "normalness". They are very easy to build and will look very nice, see the pictures of my QSX monster amp, same type of pcb..
Yes, JFET opamps are noisier (at the moment at must add). Discrete JFET's can be very quite, 2SK170 for example. You can make a discrete amp quite both with BJT's or JFET's (not MOSFET's, just for RF). You must have seen good MC amps with JFET's?
Here is one: http://www.passdiy.com/projects/pearlono1.htm
TheFettler said:
This week I will send my normal phono amp, normal headphone amp and my monster headphone amp to my pcb manufacturer. Maybe the normal ones can be of any interest? You will get extreme performance despite the "normalness". They are very easy to build and will look very nice, see the pictures of my QSX monster amp, same type of pcb..
TheFettler said:
Yes, JFET opamps are noisier (at the moment at must add). Discrete JFET's can be very quite, 2SK170 for example. You can make a discrete amp quite both with BJT's or JFET's (not MOSFET's, just for RF). You must have seen good MC amps with JFET's?
Here is one: http://www.passdiy.com/projects/pearlono1.htm
j-fet opamps
perhaps it's just the transistors I chose then.
I was using BF994S , because I needed 2 gates for the design ( a very simple 2 input class a mixer for a guitar preamp ).
I found the input noise was significant especially at the very low current that I was using , 0.7mA )
I also tried a similar design with 2N3819 , not quite as noisy especially running at slightly higher currents , but still definately noisier than a typical NPN design.
I was using fets because of the very high impedance I needed ( >6 MOhm ) to match a particular pickup without affecting it's tonal character. It was needed more as a buffer than a preamp, so the low gain wasn't an issue, simplicity was.
I couldn't easily use an opamp due to the small physical size constraints of the design ( hence the 2-gate smd device ).
I will look for your simpler phono preamp with interest, I could do with a better design than the TL071-based thing I'm currently using.
ray
perhaps it's just the transistors I chose then.
I was using BF994S , because I needed 2 gates for the design ( a very simple 2 input class a mixer for a guitar preamp ).
I found the input noise was significant especially at the very low current that I was using , 0.7mA )
I also tried a similar design with 2N3819 , not quite as noisy especially running at slightly higher currents , but still definately noisier than a typical NPN design.
I was using fets because of the very high impedance I needed ( >6 MOhm ) to match a particular pickup without affecting it's tonal character. It was needed more as a buffer than a preamp, so the low gain wasn't an issue, simplicity was.
I couldn't easily use an opamp due to the small physical size constraints of the design ( hence the 2-gate smd device ).
I will look for your simpler phono preamp with interest, I could do with a better design than the TL071-based thing I'm currently using.
ray
noisy fets
... of course I realise that the BF994 is a MosFet not a Jfet... that's probably part of the reason for the noise...
I was thinking of redesigning the cicuit with 2 BFR30/31's instead, perhaps that will reduce the noise.
and you are quite right, there are some very nice fet-based preamps , I have been looking with interest at some of Erno Borbaly 's designs - but even he slips an op-amp in there from time to time for error correction
ray
... of course I realise that the BF994 is a MosFet not a Jfet... that's probably part of the reason for the noise...
I was thinking of redesigning the cicuit with 2 BFR30/31's instead, perhaps that will reduce the noise.
and you are quite right, there are some very nice fet-based preamps , I have been looking with interest at some of Erno Borbaly 's designs - but even he slips an op-amp in there from time to time for error correction
ray
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.