The temperature coefficient does not change. This applies to a single resistor and to a pair of paralleled resistors and to a pair of series connected resistors.
The operating temperature of the resistive element depends on the temperature rise above ambient and this is determined by the power dissipated in the resistive element.
If any of the above situations (sole, parallel and series) has the same power dissipation then in all three situations the temperature rise will be the SAME.
The effect of the Tc will be the SAME.
If you reduce the dissipation in a single resistor by adding others to share the total dissipation, then the resistor will run cooler due to less power dissipation. That is where deltaT and reduced variation in resistance due to temperature changes will happen.
This sharing of the total power dissipation will come about in both the parallel pair and the series pair.
The reduced change in resistance, due to reduced change in temperature, is not restricted to ONLY paralleled combinations of resistors.
eg.
we require a 1W 150r, this is the same as two 500mW 300r connected in parallel and the same as two 500mW 75r connected in series. All three arrangements have the same total resistance. All three have the same power dissipation capability.
The operating temperature of the resistive element depends on the temperature rise above ambient and this is determined by the power dissipated in the resistive element.
If any of the above situations (sole, parallel and series) has the same power dissipation then in all three situations the temperature rise will be the SAME.
The effect of the Tc will be the SAME.
If you reduce the dissipation in a single resistor by adding others to share the total dissipation, then the resistor will run cooler due to less power dissipation. That is where deltaT and reduced variation in resistance due to temperature changes will happen.
This sharing of the total power dissipation will come about in both the parallel pair and the series pair.
The reduced change in resistance, due to reduced change in temperature, is not restricted to ONLY paralleled combinations of resistors.
eg.
we require a 1W 150r, this is the same as two 500mW 300r connected in parallel and the same as two 500mW 75r connected in series. All three arrangements have the same total resistance. All three have the same power dissipation capability.
This is wrong.
Go back to the connection between the dual polarity PSU and the dual polarity Power Amplifier. Compare the currents flowing in the TWO wires feeding the speaker load to the currents flowing in the THREE wires feeding the amplifier. You will find that the Flow and Return feeding the speaker exactly cancel out and you will find that the Flow and Returns (two Returns each working for ~50% duty cycle) of the Triplet exactly cancel out.
It is the canceling of the Flow and Return in the triplet that makes the triplet mandatory. Self got close to explaining it in 2nd ed (I don't have his later ed). Cordell explained it fully.
When you arrive at the PCB, the Triplet is still mandatory if you require cancellation of the Flow and Return Fields. Traces/tracks follow the same rules as wires.
I ran out of time for an edit.
The proposed back plane & power trace is close coupled and will reduce the radiation. But this is not as good as a canceling twisted pair of traces.
A pair of spread out traces each with their own back plane will achieve better attenuation of radiated fields than simple separated traces.
In my view the best is keeping the + & - traces as close together for as long as possible and adding a ground to cancel the remaining linear (summed + & - current flows) field.
The proposed back plane & power trace is close coupled and will reduce the radiation. But this is not as good as a canceling twisted pair of traces.
A pair of spread out traces each with their own back plane will achieve better attenuation of radiated fields than simple separated traces.
In my view the best is keeping the + & - traces as close together for as long as possible and adding a ground to cancel the remaining linear (summed + & - current flows) field.
You can shield electric fields. But magnetic fields, not so much.
And circuits that have gaps between the flow and return conductors (i.e. "enclosed loop area"), and have time-varying currents flowing in them, propagate time-varying magnetic fields. The converse is also true: Any circuit with a gap between the flow and return conductors (i.e. any conductive loop that encloses any geometric area) will have time-varying currents induced in it, by all time-varying magnetic fields that are around.
So shielding won't help much, since it's magnetic fields we're talking about, rather than electric fields. The answer is to try to remove ALL of the enclosed loop areas, in all of the conductive loops.
And what are ALL CIRCUITS? Well, "circuit" MEANS "loop". (Bummer.)
Of course, some loops are much more critical than others. The worst transmitting loops are usually the AC and rectified AC. So twist all of the transformer winding pairs, all the way to each end. And, for example, if the primary has one wire that's supposed to go to a switch and/or fuse, then the other primary wire MUST go along for the ride, and back, with the two wires tightly twisted all the way, even if it looks like it makes no sense for one of them to have to go there and back, "electrically". The other main conductor pairs that you don't want ANY gaps between are the low-level input signals. Otherwise they become antennas and the induced time-varying currents will in turn induce hum (and other) voltages across the input resistors of high gain power amplifiers... Yikes.
It's simple, really. It's just temporarily painful to realize that you have to think about something that you didn't think you had to think about. But it's also a nice club to join, because everyone else's amps at least occasionally have hum that they can't figure out. And it's also how and why "ground loops" work, at least in part. (I am starting to lose track of the number of times that people on diyaudio have had hum problems that none of the experts could solve, after dozens of posts, and I said just do this and this and they came back with something like "It's GONE!! Thank you!!".) The bottom line is that it's never just "the circuit", as in the schematic. It's exactly how you MAKE the circuit, too. It falls under "layout". But that's a much less meaningful skill unless you know about this loop-area thing (and also about the reason for star grounding, and parasitic inductance, and...).
And circuits that have gaps between the flow and return conductors (i.e. "enclosed loop area"), and have time-varying currents flowing in them, propagate time-varying magnetic fields. The converse is also true: Any circuit with a gap between the flow and return conductors (i.e. any conductive loop that encloses any geometric area) will have time-varying currents induced in it, by all time-varying magnetic fields that are around.
So shielding won't help much, since it's magnetic fields we're talking about, rather than electric fields. The answer is to try to remove ALL of the enclosed loop areas, in all of the conductive loops.
And what are ALL CIRCUITS? Well, "circuit" MEANS "loop". (Bummer.)
Of course, some loops are much more critical than others. The worst transmitting loops are usually the AC and rectified AC. So twist all of the transformer winding pairs, all the way to each end. And, for example, if the primary has one wire that's supposed to go to a switch and/or fuse, then the other primary wire MUST go along for the ride, and back, with the two wires tightly twisted all the way, even if it looks like it makes no sense for one of them to have to go there and back, "electrically". The other main conductor pairs that you don't want ANY gaps between are the low-level input signals. Otherwise they become antennas and the induced time-varying currents will in turn induce hum (and other) voltages across the input resistors of high gain power amplifiers... Yikes.
It's simple, really. It's just temporarily painful to realize that you have to think about something that you didn't think you had to think about. But it's also a nice club to join, because everyone else's amps at least occasionally have hum that they can't figure out. And it's also how and why "ground loops" work, at least in part. (I am starting to lose track of the number of times that people on diyaudio have had hum problems that none of the experts could solve, after dozens of posts, and I said just do this and this and they came back with something like "It's GONE!! Thank you!!".) The bottom line is that it's never just "the circuit", as in the schematic. It's exactly how you MAKE the circuit, too. It falls under "layout". But that's a much less meaningful skill unless you know about this loop-area thing (and also about the reason for star grounding, and parasitic inductance, and...).
Last edited:
Good stuff from Tom! In Bill Whitlock's paper on grounding (AN04?) I love that he starts it by saying that when there's hum most people fiddle around and then stop saying "That's good enough."
Starting with the right layout is SO critical. But I'd go further than Gootee and if I were producing a really top drawer amp I would shield the low voltage parts with mu metal shielding, just for those magnetic fields. Then I'd probably put something more around, maybe thin steel plate that's foldable, or the same in Aluminium. But these Aluminium cases that we are all using today are terribly overrated in the job they do. And there's quite a good argument for not using toroids and using a steel case.
Starting with the right layout is SO critical. But I'd go further than Gootee and if I were producing a really top drawer amp I would shield the low voltage parts with mu metal shielding, just for those magnetic fields. Then I'd probably put something more around, maybe thin steel plate that's foldable, or the same in Aluminium. But these Aluminium cases that we are all using today are terribly overrated in the job they do. And there's quite a good argument for not using toroids and using a steel case.
You're right, it's not. In both cases you get half the delta R. There's something slightly counter-intuitive about it working in series because you (or at least I) imagine that two may as well be one. But if they each have half the delta T, which they must do, then they must have half the delta R.
I had always imagined it as a property of parallelling, but it's actually just a matter of disippation. It's a jolly good argument against using weedy little resistors.
OK. I'm not sure how the economics works out on that, but I think I see where you are coming from. Start with a std 25ppm in SMRs and then bring it down to a tenth of that for 10p.
I was thinking more along the lines of having some thermal mass and keeping the temperature more constant. Ie. you want them to heat up to a reasonable mean but not have it bounce around too much. I had even thought that one could measure this effect by the radiation from them (like the in-ear thermometers) and then choose your power rating for the greatest stability. But maybe your idea trumps mine. Or it may just work out the same (except for the fact that it's difficult and pricey to get 25ppm in ordinary resistors). Yeah, that's worth a proper look.
The reasonable mean sentence is something which I think might be central to why amps sound better when warmed up. When cold, the resistance can only go one way, up. Later, it can go both ways. No doubt you'll disabuse me of that idea as well
Actually I quite like the idea of not doing a bit of copper track, but doing a run of resistors at 25ppm to cover the ground. That would be one for the Hi End lot.
Last edited:
No! I have not quitted yet, weelllllllllllllll.................................................
Ive been off for a long time but Im back with new design, its not doe but overall layout is clear...
fotos hochladen
Ive been off for a long time but Im back with new design, its not doe but overall layout is clear...
An externally hosted image should be here but it was not working when we last tested it.
fotos hochladen
Hi Mihkus, Have we pushed you too hard , to the point that your enthusiasm for doing this has died?
On what you've got above, which does look a whole lot better in principle, I would still put the Vbe multiplier transistor between the output transistors.
Now I know you want to be able to do this Class A as well (in which case you can hardly do without the Vbe multiplier, as that allows you to vary how much it is in Class A) but if you were happy with Class B, then it occurs to me that you could use the Quad current dumping method. That gets rid of the poor transition and has nearly zero quiescent current, so doesn't need diodes or a transistor in the middle.
It would take me a good few hours to work out how to do it properly, if not a day or two, but it is effectively a bridge, which happens to be nicely frequency independent. There's a bit of Maths involved, which I don't expect is hard, but it really is dead simple. It probably needs one of the meatier op amps as I think it ends up with quite small value resistors. But technically it is as good as any of the other solutions.
On what you've got above, which does look a whole lot better in principle, I would still put the Vbe multiplier transistor between the output transistors.
Now I know you want to be able to do this Class A as well (in which case you can hardly do without the Vbe multiplier, as that allows you to vary how much it is in Class A) but if you were happy with Class B, then it occurs to me that you could use the Quad current dumping method. That gets rid of the poor transition and has nearly zero quiescent current, so doesn't need diodes or a transistor in the middle.
It would take me a good few hours to work out how to do it properly, if not a day or two, but it is effectively a bridge, which happens to be nicely frequency independent. There's a bit of Maths involved, which I don't expect is hard, but it really is dead simple. It probably needs one of the meatier op amps as I think it ends up with quite small value resistors. But technically it is as good as any of the other solutions.
No, the idea is still in my mind...
I havent had much time, Ive been working hard since I started to feel better.
Damn you school!
I will built it soon, maybe even before chirstmas but im not sure tho, Time is the biggest issue right now.
I think the idea is fine, but I started to build and design something with one of those LME498... 10/11 thingy's
I think it would sount a lot better.. but the opamp idea is still in my mind..
I havent had much time, Ive been working hard since I started to feel better.
Damn you school!
I will built it soon, maybe even before chirstmas but im not sure tho, Time is the biggest issue right now.
I think the idea is fine, but I started to build and design something with one of those LME498... 10/11 thingy's
I think it would sount a lot better.. but the opamp idea is still in my mind..
An externally hosted image should be here but it was not working when we last tested it.
foto hochladen
Im just bored today... Nothing fancy...
I can make arcs now
Never knew that!Hi, Was just thinking about this!
That extra gap between them might prove useful in the end, and save you producing a hotspot.
I'll have a look at this shortly. Could we have a bit of annotation, please? And some sort of view of where the currents are going to flow. Also from i/p op amp to the drivers.
Anyway, I'm just getting a bit of energy and some clear space in front of me, so I may be able to give this a bit of time.
Best wishes and Happy Christmas.
That extra gap between them might prove useful in the end, and save you producing a hotspot.
I'll have a look at this shortly. Could we have a bit of annotation, please? And some sort of view of where the currents are going to flow. Also from i/p op amp to the drivers.
Anyway, I'm just getting a bit of energy and some clear space in front of me, so I may be able to give this a bit of time.
Best wishes and Happy Christmas.
I will try few of my ideas on breadboard soon...
I really have to get my hands on those parts, like that i can fully understand whats going on.
But i found this page:
AF Power Amps
The design is half done, not as loop-less as it can be but I'd like to give my best.
I really have to buy new keyboard, this one is just too damn stiff!
I really have to get my hands on those parts, like that i can fully understand whats going on.
But i found this page:
AF Power Amps
The design is half done, not as loop-less as it can be but I'd like to give my best.
I really have to buy new keyboard, this one is just too damn stiff!
Last edited:
I have to say that I'm not impressed with that website in the least and the 'products' look a frightful mess. We're going for a higher standard than that.
But I do think it's a good idea to get a breadboard and do the layout on that, and then it from there to put on stripboard. Both processes will make you more familiar with the circuit and how you want it to go to together. In some ways the layout will follow from that experience.
If it helps on the stripboard front, what I have done in the past is use 3 strips as a ground 'backbone', with two +ve and two -ve strips on either side. (You can go up to 5 Gnd strips but it gets more complicated). This is very convenient for using things like the LM317 (you'll work out why) and for having separate paths back to the main ground. You can also put some 1.5mm copper wire along the length of the strips when you need very low resistance. Stripboarding is an art, but you can make things very compact. I have managed to cram a whole voltage regulator, including power transistor, into 1 cubic inch.
But I do think it's a good idea to get a breadboard and do the layout on that, and then it from there to put on stripboard. Both processes will make you more familiar with the circuit and how you want it to go to together. In some ways the layout will follow from that experience.
If it helps on the stripboard front, what I have done in the past is use 3 strips as a ground 'backbone', with two +ve and two -ve strips on either side. (You can go up to 5 Gnd strips but it gets more complicated). This is very convenient for using things like the LM317 (you'll work out why) and for having separate paths back to the main ground. You can also put some 1.5mm copper wire along the length of the strips when you need very low resistance. Stripboarding is an art, but you can make things very compact. I have managed to cram a whole voltage regulator, including power transistor, into 1 cubic inch.
Christian, was this :
Circuit a bit unstable without all that filtering on the input of the opamp?
Working on the PCB bias and current amplifier part...
The resistor values will be similar, maybe a bit higher... not sure tho
Have to take time to think about grounding... I dont even know how to start.
The opamp part is a mess. What if opamp has its own supply wires from PSU?
An externally hosted image should be here but it was not working when we last tested it.
Circuit a bit unstable without all that filtering on the input of the opamp?
Working on the PCB bias and current amplifier part...
The resistor values will be similar, maybe a bit higher... not sure tho
Have to take time to think about grounding... I dont even know how to start.
The opamp part is a mess. What if opamp has its own supply wires from PSU?
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I will use OPA552... Datasheet:
http://www.ti.com/lit/ds/symlink/opa552.pdf
Mains variations? I bet Im going to use regulated power supply... If you mean that.
Got 2 LME49810 if this one fails permanently or fails my expectations probably one of the most expensive IC's ive ever played with...
37€ each.
http://www.ti.com/lit/ds/symlink/opa552.pdf
Mains variations? I bet Im going to use regulated power supply... If you mean that.
Got 2 LME49810 if this one fails permanently or fails my expectations probably one of the most expensive IC's ive ever played with...
37€ each.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.