Jan, what do you say about the "best" method in distributing power, tied together or some isolation?
We are nice.....janneman said:
This is a perfectly good method if you are interested in the lower frequencies. Inductance of a short wire is nothing (OK, not much) at 100 Hz, 1 kHz or so but you should allways be carefull in analyzing results. What are you measuring really. Never hurts to think about it.
If you are testing a high speed regulator, the test setup is a little bit more advanced, right Fred?
Re: I wish I was that smart...........
Not only close, it is correct. Where on earth did you dig that
up? Don't tell me you are studying swedish in secret. 😉
Fred Dieckmann said:
Not only close, it is correct. Where on earth did you dig that
up? Don't tell me you are studying swedish in secret. 😉
Comedy this good you can't get just any old place.....
Unless you look in the "newer op-amps" thread.
Between watching Phred and Per pulling each other's leg, John Curl stating what a lot of us have known for years (which seems to baffle others), Jan trying to keep his sanity while looking at bogus measurements, and the mystery of Babel to compound it all.............
Yep, I love this place.
Back to our regular entertainment.
Jocko
Unless you look in the "newer op-amps" thread.
Between watching Phred and Per pulling each other's leg, John Curl stating what a lot of us have known for years (which seems to baffle others), Jan trying to keep his sanity while looking at bogus measurements, and the mystery of Babel to compound it all.............
Yep, I love this place.
Back to our regular entertainment.
Jocko
not quite.......
"Not only close, it is correct. Where on earth did you dig that up? Don't tell me you are studying swedish in secret."
No..... actually I am studying Swedes in public.
"Not only close, it is correct. Where on earth did you dig that up? Don't tell me you are studying swedish in secret."
No..... actually I am studying Swedes in public.
Dear Jan,
“You thought you were measuring the regulator, but in effect you have been checking the dynamic impedance of a cap with parallel inductance with DC offset.”
No, I am measuring the regulator in practical setup. Would you measure audio amp with resistive load or nevertheless with real speaker?
“I have pleasure in trying to wring the last drop of performance from a circuit” Please be aware there are a lot of bottles in the cellar still…
“You thought you were measuring the regulator, but in effect you have been checking the dynamic impedance of a cap with parallel inductance with DC offset.”
No, I am measuring the regulator in practical setup. Would you measure audio amp with resistive load or nevertheless with real speaker?
“I have pleasure in trying to wring the last drop of performance from a circuit” Please be aware there are a lot of bottles in the cellar still…
Dimitri,
I don't understand your answer. You are measuring the supply after 50cm of wiring. As I said, you reaslly measure the cap and the inductance of the wire, which appears in parallel with the cap. Whatever you measure doesn't say a lot about the quality of the supply. So, you'r not really measuring the supply. Don't take it personally.
This is apart from the fact that connecting a supply with 50cm of wire means you lose any high quality regulation that you could get from a Jung-type regulator. A standard 7815 is cheaper, easier and gives the same results.
Jan Didden
I don't understand your answer. You are measuring the supply after 50cm of wiring. As I said, you reaslly measure the cap and the inductance of the wire, which appears in parallel with the cap. Whatever you measure doesn't say a lot about the quality of the supply. So, you'r not really measuring the supply. Don't take it personally.
This is apart from the fact that connecting a supply with 50cm of wire means you lose any high quality regulation that you could get from a Jung-type regulator. A standard 7815 is cheaper, easier and gives the same results.
Jan Didden
Jan is right. Dimitri, you must establish where the regulator "ends" and where your test gear starts. I don't consider 50 cm wire as part of a regulator.
But once again it depends what you are really interested in to measure.
But once again it depends what you are really interested in to measure.
Let us consider typical preamp (mixer, RIAA, mic). We try to place active circuit board as near as possible near the input jacks, we even make gain control spindle longer. We try to place xformer with rectifier, caps and regulator as far as possible from the input jacks. What will be the length for dc supply wires, including pcb conductors? Ok, don’t like 50cm, ok, 20 cm. Somebody in this thread speak about square mm of pcb conductor and uV. How can we compensate wires – by 4 wire circuit, but what will be when we try to feed
multiple ICs? We don’t need ultimate parameters for two specific nodes on the regulator output, we need good parameters for the whole supply rail.
I personally use 4 wire circuit for the output op-amp, because I have now idea what it load will be. Other op-amps feed through RC filters, but I try to switch their output stage in class A (or to use op-amp with appropriate quiescent current). When I don’t have cost restriction I use precision regulator with normal bipolar output stage (not the single follower) operating in class A with quiescent current value higher, than will flow in the circuitry.
multiple ICs? We don’t need ultimate parameters for two specific nodes on the regulator output, we need good parameters for the whole supply rail.
I personally use 4 wire circuit for the output op-amp, because I have now idea what it load will be. Other op-amps feed through RC filters, but I try to switch their output stage in class A (or to use op-amp with appropriate quiescent current). When I don’t have cost restriction I use precision regulator with normal bipolar output stage (not the single follower) operating in class A with quiescent current value higher, than will flow in the circuitry.
GOsh, that seems sort of odd. I try to put the regulator adjacent to the regulated circuit, not 20 or 50cm away.
The thread just died when Dimitri and I asked this.
Not even Fred has a practical suggestion
If the pcb is 200 mm or more AND there are several power consumers and we don't want (or can afford, space money cost) one regulator to each opamp or whatever.
Can Jan or anyone else suggest and practical solution?
My personal choice would be R+C.
Not even Fred has a practical suggestion
If the pcb is 200 mm or more AND there are several power consumers and we don't want (or can afford, space money cost) one regulator to each opamp or whatever.
Can Jan or anyone else suggest and practical solution?
My personal choice would be R+C.
dimitri said:
Dimitri,
That's all fine and dandy, but the fact remains that you claimed to measure regulator performance where you actually didn't.
Now, to your post: Indeed, take the typical RIAA etc. One stage, two stages? How close are the supply points of those two stages? 1/2 inch apart? How close can you place the regulators? Anothe 1/2 inch away? Total 25.4 mm give or take a few angstrom. A far cry from 500 or 200 mm.
And come on, give me a break. We didn't talk about how far the xformer was away. As far as I am concerned, that can be in the next room. We were talking about the regulator distance.
But, apart from being a story that isn't holding water, it also has no bearing on the original issue. You'll have to do better than that.
Anyway, I'm sure you know the score. I'm not going to spend any more time on this thread explaining the obvious. You have the last word.
Or maybe Per-Anders
Jan Didden
Start with ohms law.........
"we need good parameters for the whole supply rail."
Power distribution is the same principle as single point grounding. The first thing is to determine which circuits are the most sensitive to power supply noise and put them close to the central hub of single point ground and power supply plus and minus voltage outputs with short traces. Segregate large power supply signal currents from sharing common traces with sensitive low current signals. Using separate power supply traces to each device when possible is a very good idea. It keeps signal currents from sharing a common trace impedance, which creates an error voltage that is common to both devices, and is a function of both devices power supply currents. This from their separate signal currents across a common trace impedance. Designing for low crosstalk from currents is the key. Think of the voltages at each device's supply terminals in terms of signal currents and the trace resistances through which the currents travel from the point of low impedance output of the supply. RC filters are effective at filtering noise from the central supply but are real compromise since signal currents through series resistors will cause variations in voltage at the device drawing the signal currents. The biggest mistake is not understanding the supply currents to each device's supply terminals, and failing to plan the ground and supply path before the rest of the audio circuit's placement. The circuit should be built around the power supply distribution early in the layout process. You cannot go add the power distribution afterwards. This sounds a little abstract and you need apply it to the particular circuit you are designing.
A couple of good references are:
http://www.analog.com/UploadedFiles/Application_Notes/135208865AN-202.pdf
POWER UP:
An Overview of Power Supply Considerations
R.N. Marsh The Audio Amateur 3/83
"we need good parameters for the whole supply rail."
Power distribution is the same principle as single point grounding. The first thing is to determine which circuits are the most sensitive to power supply noise and put them close to the central hub of single point ground and power supply plus and minus voltage outputs with short traces. Segregate large power supply signal currents from sharing common traces with sensitive low current signals. Using separate power supply traces to each device when possible is a very good idea. It keeps signal currents from sharing a common trace impedance, which creates an error voltage that is common to both devices, and is a function of both devices power supply currents. This from their separate signal currents across a common trace impedance. Designing for low crosstalk from currents is the key. Think of the voltages at each device's supply terminals in terms of signal currents and the trace resistances through which the currents travel from the point of low impedance output of the supply. RC filters are effective at filtering noise from the central supply but are real compromise since signal currents through series resistors will cause variations in voltage at the device drawing the signal currents. The biggest mistake is not understanding the supply currents to each device's supply terminals, and failing to plan the ground and supply path before the rest of the audio circuit's placement. The circuit should be built around the power supply distribution early in the layout process. You cannot go add the power distribution afterwards. This sounds a little abstract and you need apply it to the particular circuit you are designing.
A couple of good references are:
http://www.analog.com/UploadedFiles/Application_Notes/135208865AN-202.pdf
POWER UP:
An Overview of Power Supply Considerations
R.N. Marsh The Audio Amateur 3/83
Thanks, Fred, that makes perfect sense to me. Don't know why
anybody would find it abstract though?
anybody would find it abstract though?
Because grounding is hard. Sometimes you won't even realize the problem until you've built the prototype and measured it. How best to wire power rails and grounds? When is a ground plane good and when is it a disaster? etc.
That, and the members of this board are generally unable to apply Kirchoff's Law.
That, and the members of this board are generally unable to apply Kirchoff's Law.
Nice post, Fred.
Very similar approach to how I design pcbs and harnesses. As much as possible (when it makes sense), I feed each block of amplifier circuitry from its own regulator. When this isn't feasible, I get rid of every shard of common impedance that I can ferret out. And when separate regulation is feasible, I still do the same.
regards, jonathan carr
PS. I normally keep each regulator no more than about 8mm away from its load, and I also use local bypasses (lots of 'em).
Very similar approach to how I design pcbs and harnesses. As much as possible (when it makes sense), I feed each block of amplifier circuitry from its own regulator. When this isn't feasible, I get rid of every shard of common impedance that I can ferret out. And when separate regulation is feasible, I still do the same.
regards, jonathan carr
PS. I normally keep each regulator no more than about 8mm away from its load, and I also use local bypasses (lots of 'em).
Because grounding is hard.
It can be just that. I did a two layer line card. (low cost! 4 layers with ground plane would have got me fired) The grounding required digital, analog, transient (10s of amps), RFI and ESD grounding. The only thing missing was coffee grounds.
The grounding strategy could not quite be single point ground and had to share some common traces. I spent a long time thinking very hard about return current path, in order to avoid my boss having grounds to fire me. I never have pulled it off if I wasn't doing it with the best PCB designer I will ever get to work with. He could tell what I wanted, even when I couldn't quite explain it while sitting next to next to him in front of the monitor. You will never see me that grateful again for someone that talented to work with.
It can be just that. I did a two layer line card. (low cost! 4 layers with ground plane would have got me fired) The grounding required digital, analog, transient (10s of amps), RFI and ESD grounding. The only thing missing was coffee grounds.
The grounding strategy could not quite be single point ground and had to share some common traces. I spent a long time thinking very hard about return current path, in order to avoid my boss having grounds to fire me. I never have pulled it off if I wasn't doing it with the best PCB designer I will ever get to work with. He could tell what I wanted, even when I couldn't quite explain it while sitting next to next to him in front of the monitor. You will never see me that grateful again for someone that talented to work with.
It's allways a pleasure to let pros do the work for you. My "pro" is a toolmaker for plastic molding. He knows molding tools which I don't.
Fred, may I ask what the pcb did? The application?
Fred, may I ask what the pcb did? The application?