Hello again. Did a quick search and couldn't find this discussion so starting a new thread...
I am planning to build an opamp-based phono pre-amp in a re-cycled full-size chassis/enclosure using an existing transformer with +/- 16.5V centre tapped. As there will be a lot of space left I plan to include a microphone pre-amp running off the same PSU. The transformer also has a +/- 38.5V tap which I will experiment with as a PSU for my TT motor. This last bit may never get beyond taking the AC out via a screened lead (captive lead, no connector) direct to the motor.
The question is: how much better will the noise performance be if I build on a PCB vs. matrix or strip board and _why_ will it be better. My gut feel is that, for the phono pre at least, a sensitive layout on stripboard would be as good as a single-sided PCB. What I mean is that all of the things you need to be careful of when doing a PCB layout can mostly be accommodated on strip board can't they?
Also, what would be the technical justification (physics) for needing a ground plane for an audio frequency circuit in an earthed metal enclosure?
If it all goes horribly wrong then I don't need anything more than a crude metal box around the PCB to exclude induced noise from stray transformer EM fields , right?
I'd welcome your thoughts/advice. Thanks.
PS PSU will be LM317 based with RC LPF on the input with smoothing cap on the voltage divider + diodes.
I am planning to build an opamp-based phono pre-amp in a re-cycled full-size chassis/enclosure using an existing transformer with +/- 16.5V centre tapped. As there will be a lot of space left I plan to include a microphone pre-amp running off the same PSU. The transformer also has a +/- 38.5V tap which I will experiment with as a PSU for my TT motor. This last bit may never get beyond taking the AC out via a screened lead (captive lead, no connector) direct to the motor.
The question is: how much better will the noise performance be if I build on a PCB vs. matrix or strip board and _why_ will it be better. My gut feel is that, for the phono pre at least, a sensitive layout on stripboard would be as good as a single-sided PCB. What I mean is that all of the things you need to be careful of when doing a PCB layout can mostly be accommodated on strip board can't they?
Also, what would be the technical justification (physics) for needing a ground plane for an audio frequency circuit in an earthed metal enclosure?
If it all goes horribly wrong then I don't need anything more than a crude metal box around the PCB to exclude induced noise from stray transformer EM fields , right?
I'd welcome your thoughts/advice. Thanks.
PS PSU will be LM317 based with RC LPF on the input with smoothing cap on the voltage divider + diodes.
RF noise.
For small signal designs a ground plane will always produce a quiteter design.
I would reccomend these sites, regarding EMC.
EMC Information Centre - The EMC Journal (Free in the UK)
And the Godfather of EMC design and noise reduction:
home page
For small signal designs a ground plane will always produce a quiteter design.
I would reccomend these sites, regarding EMC.
EMC Information Centre - The EMC Journal (Free in the UK)
And the Godfather of EMC design and noise reduction:
home page
To keep transformer magnetic fields away from a PCB the main thing is distance. If that is not possible, use thick iron/steel sheet. Personally, I would keep motor supplies well away from a phono preamp.
Ground planes are good for RF, but for audio you may need more control over where the currents flow. Grounding topology is the main thing. Bus, star or plane is secondary; all can work if done properly.
Noise performance is decided mainly by circuit design, secondly by component selection (e.g. low noise opamps).
Ground planes are good for RF, but for audio you may need more control over where the currents flow. Grounding topology is the main thing. Bus, star or plane is secondary; all can work if done properly.
Noise performance is decided mainly by circuit design, secondly by component selection (e.g. low noise opamps).
A crude metal box will make an excellent electrostatic shield. But you must connect it to your circuit correctly.
That box will do almost nothing for electromagnetic shielding.
Marce,
can you jump in here with a set of rules on how to integrate the box with the circuit?
Can you also advise on adding a copper shield plane to the top (or bottom) of a perf board assembly to mimic a ground plane?
That box will do almost nothing for electromagnetic shielding.
Marce,
can you jump in here with a set of rules on how to integrate the box with the circuit?
Can you also advise on adding a copper shield plane to the top (or bottom) of a perf board assembly to mimic a ground plane?
I'll have a dig around tonight, as shielding is mainly (to quote Keith Armstrong and others) is seams and seals!!!
There are notes by Mr Ott that show some advantages of having what they call a mirror plane for emc. I believe that it will help terminate the h and e fields of the signals on the vero-board as well as providing some shielding.
Just been chosen for redundancy, so a bit manic at the moment (its not that bad, I am a lucky one I have a job to go to. And I will get to play with more varid designs as its beureau work, including a couple of pof audio companies He He)
For small signal designs be they any sort of analogue, ground planes are best and produces the quietist circuits, and allow returns to find there own way home. Messing about with stars busses etc is not the best way, especially if you dont understand return currents etc. For POWER designs ie AMPS etc, then yes keeping high currents away from low current is essential. And yes the other schemes work, but for the best results a gnd plane is best. In the early days of layout I would do digital boards on two layers but ALL analogue designs (low signal) I did where a minimum of 4 layers with a contigous ground plane.
Anyway I'll dig out the info and get back.
There are notes by Mr Ott that show some advantages of having what they call a mirror plane for emc. I believe that it will help terminate the h and e fields of the signals on the vero-board as well as providing some shielding.
Just been chosen for redundancy, so a bit manic at the moment (its not that bad, I am a lucky one I have a job to go to. And I will get to play with more varid designs as its beureau work, including a couple of pof audio companies He He)
For small signal designs be they any sort of analogue, ground planes are best and produces the quietist circuits, and allow returns to find there own way home. Messing about with stars busses etc is not the best way, especially if you dont understand return currents etc. For POWER designs ie AMPS etc, then yes keeping high currents away from low current is essential. And yes the other schemes work, but for the best results a gnd plane is best. In the early days of layout I would do digital boards on two layers but ALL analogue designs (low signal) I did where a minimum of 4 layers with a contigous ground plane.
Anyway I'll dig out the info and get back.
Thanks guys this is great. I'll research those links and come back with a few questions I'm sure.
Let's see if I've got it so far:
source - stray EM fields from internal transformer and nearby equipment
distance - keep small signal cct as far away as possible from EM source
signal tracks - as short as possible to minimise induced voltage from EM fields
metal shielding - won't do anything for magnetic fields (how about soft iron?)
PSU ripple - not a problem if I use opamps due to PSRR
OK so here's a question...
If shielding won't do much for me how come a ground plane will? Wouldn't these two techniques work the same way in terms of protection from EMI?
Don't get me wrong: I'm not disagreeing or arguing against what you are telling me, I just want to understand how and why these techniques work. I do DIY in order to learn stuff as much as get decent sounding audio at minimal cost.
Thanks again.
Let's see if I've got it so far:
source - stray EM fields from internal transformer and nearby equipment
distance - keep small signal cct as far away as possible from EM source
signal tracks - as short as possible to minimise induced voltage from EM fields
metal shielding - won't do anything for magnetic fields (how about soft iron?)
PSU ripple - not a problem if I use opamps due to PSRR
OK so here's a question...
If shielding won't do much for me how come a ground plane will? Wouldn't these two techniques work the same way in terms of protection from EMI?
Don't get me wrong: I'm not disagreeing or arguing against what you are telling me, I just want to understand how and why these techniques work. I do DIY in order to learn stuff as much as get decent sounding audio at minimal cost.
Thanks again.
Its not totaly expected, this is the fourth round of them in the last 12 months, defence cut backs here and in the US.
I've got Henry Ott Electromagnetic compatability engineering sat next to me, also what little knowledge I've picked up over the last 25 years doing PCB's, I'll answer in more detail in a bit.
Sheilding does woek, rip open your mobile phone and look at the metal cans in there, I'll dig out some examples of how thats done. The problem with shielding though can be gaps such as slots in the shielding, these can act as waveguides and actually increase interference, whereas having no shielding could be a better solution, strange but true.
With a shielding can on a board you can shield the areas you want 100%, where as if you put a metal box around it you have to have proper terminated cable enty etc to make a complete shield, and that is the key having a complete shield. More in a bit.
I've got Henry Ott Electromagnetic compatability engineering sat next to me, also what little knowledge I've picked up over the last 25 years doing PCB's, I'll answer in more detail in a bit.
Sheilding does woek, rip open your mobile phone and look at the metal cans in there, I'll dig out some examples of how thats done. The problem with shielding though can be gaps such as slots in the shielding, these can act as waveguides and actually increase interference, whereas having no shielding could be a better solution, strange but true.
With a shielding can on a board you can shield the areas you want 100%, where as if you put a metal box around it you have to have proper terminated cable enty etc to make a complete shield, and that is the key having a complete shield. More in a bit.
Yes that is correct. The trouble is there is a lot of VHF rf mush about these days and adding a shield for lower frequencys can work for them but concentrate the VHF mush if it is present. Its another complex and interesting subject (similar to transmission lines at low frequencymore questions on that soon, in another thread of course, the info and reading material has been an education).
The problem wih partial screening is also the cables coming out, acting as antennas, I am probably being over cautious but I think it is worth looking at and discussing.
As to ground planes I am putting some thoughts and pictures together, which I will start a new thread on, we should have some interesting discourse on them, from previos posts youv'e all probably gathered I'm not a fan of excessive star grounding.
As it is a phono pre-amp we are talking about low level signals so any noise is going to have a big effect.
never start a thread on how long a bit of string is!Joking apart, there are so many views that a definitive answer is probably impossible to get, but that is a big part of the education, your beliefs will be challenged, but it is always an education. Dont expect two answers the same, on any point
more questions on that soon, in another thread of course, the info and reading material has been an education).The problem wih partial screening is also the cables coming out, acting as antennas, I am probably being over cautious but I think it is worth looking at and discussing.
As to ground planes I am putting some thoughts and pictures together, which I will start a new thread on, we should have some interesting discourse on them, from previos posts youv'e all probably gathered I'm not a fan of excessive star grounding.
As it is a phono pre-amp we are talking about low level signals so any noise is going to have a big effect.
Wot you doing on DIY Audio then
never start a thread on how long a bit of string is!Joking apart, there are so many views that a definitive answer is probably impossible to get, but that is a big part of the education, your beliefs will be challenged, but it is always an education. Dont expect two answers the same, on any pointHa ha. I never got a definitive answer from a bunch of engineers in my life! I'm gathering information here and then I'll make up my mind. Consult then decide rather than trying to reach consensus if you'll forgive a platitude. That's why I would like to try and understand the advice in the context of the physics that's going on. Physics or data will persuade me - hearsay or superstition won't.
Please, nobody take offense at the engineer comment as I would describe myself as one. There, I said it.
Looks like I'm gonna enjoy this forum.
I'm gathering information here and then I'll make up my mind. Consult then decide rather than trying to reach consensus if you'll forgive a platitude. That's why I would like to try and understand the advice in the context of the physics that's going on. Physics or data will persuade me - hearsay or superstition won't.Please, nobody take offense at the engineer comment as I would describe myself as one. There, I said it.
Looks like I'm gonna enjoy this forum.
I think that you are exactly correct!
And has soon as you start hooking up test equipment or doing a A/B blind test the setup sitution changes and the interference changes.
Rather late, sorry, but other things are taking over my sanity at this time...
Here is a link to some Keith Armstron notes that are worth having a look at. There is a whole series of them (8 now I believe) that I have downloaded as PDF's so dont have all the links handy, this is probably the most pertinant :
EMC Compliance Club – Advanced PCB design and layout for EMC. Part 3 – PCB-to-chassis bonding
Using the follwoing search (changing the number at the end should lead to all the articles) that cover in detail screening, noise etc in both analogue and digital designs.
"Design Techniques for EMC – Part 1"
eg
EMC Compliance Club – Design Techniques for EMC Part 0 Introduction & Part 1 - Circuit Design and Choice of Components (SECOND PART)
Here is a link to some Keith Armstron notes that are worth having a look at. There is a whole series of them (8 now I believe) that I have downloaded as PDF's so dont have all the links handy, this is probably the most pertinant :
EMC Compliance Club – Advanced PCB design and layout for EMC. Part 3 – PCB-to-chassis bonding
Using the follwoing search (changing the number at the end should lead to all the articles) that cover in detail screening, noise etc in both analogue and digital designs.
"Design Techniques for EMC – Part 1"
eg
EMC Compliance Club – Design Techniques for EMC Part 0 Introduction & Part 1 - Circuit Design and Choice of Components (SECOND PART)
Thanks marce. I found the paper on star vs multi-point grounding particularly interesting. I might try some experiments with that when I get round to the build. I am leaning towards a PCB with a decent ground plane; partly so I can experiment with single- vs multi-piont grounding and partly to minimise the problems of coupling VHF+ "mush".
DF96 and Speedskater, thanks for making that point. It makes sense to me to avoid stressing the circuit with any signals outside of the range of interest even if the products of the interaction themselves are inadudible or filtered from the output. That point alone justifies to me a carefully designed PCB. I still reckon that for immunity from audio frequency noise sources a careful layout on matrix/strip board and assembly within the enclosure would be sufficient.
If anybody else is following this here's a link to a top level page with index and access to the papers: EMC Information Centre (you will need to register)
DF96 and Speedskater, thanks for making that point. It makes sense to me to avoid stressing the circuit with any signals outside of the range of interest even if the products of the interaction themselves are inadudible or filtered from the output. That point alone justifies to me a carefully designed PCB. I still reckon that for immunity from audio frequency noise sources a careful layout on matrix/strip board and assembly within the enclosure would be sufficient.
If anybody else is following this here's a link to a top level page with index and access to the papers: EMC Information Centre (you will need to register)
At the risk of being lynched, I am not a fan of star grounds or over complex grounding where too much thought has gone into trying to control return currents and grounds, every time I have seen a system where it has been 'optimised' for ground control, you get problems. Where systems are designed with a scheme more biased to a mesh ground (and where ground currents/returns) can make their own way home, you have less (if any) problems. To solve the problems with the former, you often see nice thick bits of wire tying two ground points together to alleviate the problem (creating a mesh ground). Star grounds can work on simple set ups, but as a system gets more complex, they have limitations and can make fault finding a nightmare. One of my favorite quotes referring to star grounds is:
"Ground leads and shields can cause plenty of trouble, and there is a lot of misunderstanding on this subject. The problem, in a nutshell, is that the currents you forgot about flowing through a ground line can generate a signal seen by another part of the circuit sharing the same ground. The technique of a ground "mecca" (a common point in the circuit to which all ground connections are tied) is often seen, but it's a crutch; with a little understanding of the problem you can handle most situations intelligently."
"Ground leads and shields can cause plenty of trouble, and there is a lot of misunderstanding on this subject. The problem, in a nutshell, is that the currents you forgot about flowing through a ground line can generate a signal seen by another part of the circuit sharing the same ground. The technique of a ground "mecca" (a common point in the circuit to which all ground connections are tied) is often seen, but it's a crutch; with a little understanding of the problem you can handle most situations intelligently."
I disagree with the above. I fought a hum & hiss war with my $15 op amp disco mixer and won a few battles. My system consists of steel case disco mixer as safety grounded hub, with power amp (floating) record player (floating) FM radio (floating) and CD player (floating). The record player tone arm is grounded to the disco mixer through the green wire. This star arrangement minimizes hum. I removed the RCA jacks from the steel case of the disco mixer and isolated them with rubber o-rings.At the risk of being lynched, I am not a fan of star grounds or over complex grounding where too much thought has gone into trying to control return currents and grounds, every time I have seen a system where it has been 'optimised' for ground control, you get problems.
I also removed the AC power transformer from the disco mixer and moved it to the power strip as a wall transformer. This cut a lot of hum. The power switch was right next to the op amps. A separate box in the mixer for AC items might have helped, a la dynakit PAS2 preamp, but the wall transformer solution was cheaper and easier.
Against RF hash the disco mixer had 22 pf ceramic caps from RCA jack center to outside, but was still picking up CB'ers driving by and hash from the lamp dimmer in the next room. I used some chokes salvaged from PCAT power supplies on the power inputs to stop that. The whole hum hiss war is detailed here:www.diyaudio.com/forums/analog-line-level/164102-improving-disco-mixer-mid-fi-performance.html
The disco mixer had its own non-optimum ground plane PC board, but my next op amp project will use these: In the USA Circuit Boards within Project Accessories - MCM Electronics Category
In Europe Prototyping Boards
I do TO92 and leaded component prototyping on undrilled paper 1/16" electronic board I got at a yard sale, also for sale in sheets from mcmaster.com. I drill them myself with a hand crank drill. The glass reinforced stuff is a ***** to drill with steel drills, they really need carbide. I use turret board for tube stuff but they are really too big for transistor projects.
If you build up a prototype and then rebuild it using a custom PC board from a shop, let us know how you fair on the noise front. Most of the people contributing here seem to be rich enough to go right to custom manufactured board and pay for the several revisions. I'm experimenting on a complete shoestring, using garbage mostly, a few key active components and a lot of new electrolytic caps.
Last edited:
Indianajo, i did also state the above. And most of our home systems are going to be quite simple, with only a few components joined together. Both my comments and the Tony Waldren links are to just inject a alternate view to what can be a pain, grounding a system. Every system is different, and every solution different. I was playing devils advocate a bit, but do think the differing views are worth looking at even if they are not
During the day I do PCB's for a variety of systems, most recently communications. We always use ground planes and max copper pours on the top and bottom layers, though where power outputs are involved, copper pours are used to get the current handling, but not full planes. The noise immunity of multilayer boards with planes is better than breadboarded, and due to the complexity we often go straight to PCB. How much difference this will make, to the final result is hard to quantify, our main concern is EMC and signal integrity, so the test the designs are subjected to are way over what we would be able to do at home, and the kit used way beyond most budgets.
What does happen though when a ground loop apears is the wire comes out and some strategicly placed wires help ease the symptoms, if not find a cure.
The big problem is when hooking up kit that we havn't made, such as headphones or similar, where w you have no controll over the connectivity.
Like you I have to do things on a budget, and try and get things when I have the funds.
As I said there will be hundreds that dont agree with me on star grounding as a universal panacea, but it does prompt some interesting debate.
OK, so this is where I have got to...
o Low frequency noise/hum will be dealt with by careful topology and screening.
o Final stage of PSU (regulators) will be on pre-amp boards - one board per channel.
o Individual regulators for each board fed from common PSU (capacitor smoothing and first order LPF filtered plus resistor damping for diode switching).
o I have decided to make all reasonable efforts to minimise effects of RFI for reasons stated above:
- capacitor filtering at enclosure boundaries
- unbroken copper ground plane with distributed grounding for each board
- electrically sealed enclosure as much as possible.
In terms of the original question (PCB vs matrix vs...) I have come across something called "dead bug construction". It seems to be referred to by a number of different names but I will be using the variant which builds off a continuous copper-clad board. I can pick up a ground point wherever I happen to be on the ground plane; I can keep path lengths really short; very compact construction; robust; gives me a decent ground plane without having to go through painful (and let's face it quite boring) PCB layout and prototyping; I can test parts of the circuit as I progress through the build...
What's not to like? Anyone got an argument against using "dead bug" construction for a phono pre-amp?
o Low frequency noise/hum will be dealt with by careful topology and screening.
o Final stage of PSU (regulators) will be on pre-amp boards - one board per channel.
o Individual regulators for each board fed from common PSU (capacitor smoothing and first order LPF filtered plus resistor damping for diode switching).
o I have decided to make all reasonable efforts to minimise effects of RFI for reasons stated above:
- capacitor filtering at enclosure boundaries
- unbroken copper ground plane with distributed grounding for each board
- electrically sealed enclosure as much as possible.
In terms of the original question (PCB vs matrix vs...) I have come across something called "dead bug construction". It seems to be referred to by a number of different names but I will be using the variant which builds off a continuous copper-clad board. I can pick up a ground point wherever I happen to be on the ground plane; I can keep path lengths really short; very compact construction; robust; gives me a decent ground plane without having to go through painful (and let's face it quite boring) PCB layout and prototyping; I can test parts of the circuit as I progress through the build...
What's not to like? Anyone got an argument against using "dead bug" construction for a phono pre-amp?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.