Re: Web links? already? ...
Hey I like that, thanks 🙂 I'd prefer you just link to the homepage though rather than a specific page.
I will email you a better logo to use as well.
FastEddy said:
Hey I like that, thanks 🙂 I'd prefer you just link to the homepage though rather than a specific page.
I will email you a better logo to use as well.
Hi,
I preferred the star grounded version.
It better separates the charging pulses from the regulating section. Although distance is in your favour.
You have removed the two snubbers. Why?
The negative diode discharge protection looks a bit untidy. Can it be cleaned up? Positive looks fine.
I preferred the star grounded version.
It better separates the charging pulses from the regulating section. Although distance is in your favour.
You have removed the two snubbers. Why?
The negative diode discharge protection looks a bit untidy. Can it be cleaned up? Positive looks fine.
I removed the fill plane. It looks like there's not an overwhelming reason to use it at this stage, and I also like the looks of the starred version. Furthermore, I think it is easier to understand and maintain, at least, using ExpressPCB software. So, let's table the filled plane idea (for now, anyway).
As for the snubber, I have to give credit again to richie. He's done this already, so I trust his design choice. Also, adding the more complex snubber was going to be problematic in terms of size. I'm trying to leave some room around the input/output connectors for terminal blocks.
One thing that really became apparent to me last night as I was looking up materials is that the large polyprop or polyester caps are prohibitively expensive and large when you go to 10 uF. The only way around this is to use radial electrolytics (which will be too big for this board), tantalums, or smaller capacitance (1uF) poly caps. I have changed the regulator cap body outlines to reflect this. The distance between the outer holes is now 10 mm, while the inner distance is 5 mm. I've also added extra holes on C3 and C8 to make pitch sizes of 2.5 mm (fairly common spec) possible with additional jumpers.
Andrew, I cleaned up the protection diodes. Do they seem o.k., now?
As for the snubber, I have to give credit again to richie. He's done this already, so I trust his design choice. Also, adding the more complex snubber was going to be problematic in terms of size. I'm trying to leave some room around the input/output connectors for terminal blocks.
One thing that really became apparent to me last night as I was looking up materials is that the large polyprop or polyester caps are prohibitively expensive and large when you go to 10 uF. The only way around this is to use radial electrolytics (which will be too big for this board), tantalums, or smaller capacitance (1uF) poly caps. I have changed the regulator cap body outlines to reflect this. The distance between the outer holes is now 10 mm, while the inner distance is 5 mm. I've also added extra holes on C3 and C8 to make pitch sizes of 2.5 mm (fairly common spec) possible with additional jumpers.
Andrew, I cleaned up the protection diodes. Do they seem o.k., now?
Re: No need for high voltage caps?
I think it will be more complicated. Transformer saturation will happen. Most transformers already use 70% .. 80% of max flux density at normal operation. If we consider furthermore lower mains frequency in Europe, 50Hz instead of 60Hz, then already most of the margin is eaten up. Transformer will saturate already at voltages short above 120V. You will have not double output voltage. Depending on the plug in moment (means at which mains line angle you hit, when plugging in) you will get a more or less massive output voltage overshoot and then moderate overvoltage and transformer saturation. Trafo will draw excessive currents from line, causing the fuse to trip or smoking the primary of the transformer.
FastEddy said:richie00boy: " ... There would be no point using 100V caps as you would never put anything like that much voltage on the board. For 100% reliable operation the max voltage each regulator input should see is about 40 volts. ... "
Yes, absolutely ccorrect. Using a 30 VAC wall wart or transformer, the peak to peak voltage is about 52 volts ... but if someone accedentally plugs that 120 VAC primary / 30 VAC secondary into an outlet in Europe ... oops, grab that fire extstingisher ... secondary voltage goes to 104 VAC or more
Anyway, there are always opportunities for a screwup like me to do the wrong thing and cook the electronic goose.
I think it will be more complicated. Transformer saturation will happen. Most transformers already use 70% .. 80% of max flux density at normal operation. If we consider furthermore lower mains frequency in Europe, 50Hz instead of 60Hz, then already most of the margin is eaten up. Transformer will saturate already at voltages short above 120V. You will have not double output voltage. Depending on the plug in moment (means at which mains line angle you hit, when plugging in) you will get a more or less massive output voltage overshoot and then moderate overvoltage and transformer saturation. Trafo will draw excessive currents from line, causing the fuse to trip or smoking the primary of the transformer.
Beat me to it! I was just going to point out there will be more to worry about than blown caps if you plug it in double mains voltage socket.
The star concept vs. the plane concept
Personally, I think the star is overrated, because noise isn't just electrons but electromagnetic waves, which means it will have little trouble jumping from one trace to the next, but if you're not doing both power and ground planes, the star might be a better bet than a ground plane alone; and if you're doing the star, then ritchie00boy's design concept, which you're following, is the right way to do it IMHO. If you were doing planes, I'd do three planes, ground sandwiched by input positive and negative under the filter caps, and sandwiched by output positive and negative under the regulators and associated circuitry; but this is expensive with a capital E and probably not worth the effort in this application.
OTOH, I'm not sure I agree with separate traces going to the common pins of (R5, C8, and D5,) and similarly to (R3, C5, D3, and D4,) and (R4 and C4,) and (R6 and C7.) I've grouped these with parentheses in defiance of typographical standards to make it perfectly clear which sets I mean. I would use islands of copper to connect these, rather than traces as you have done; I don't know that moving the components is necessary, and I think it might be undesirable. This concept is a compromise between the plane concept and the star concept, and I think offers some benefits of each with mitigation of the drawbacks of each. The two sides of the ground star between the regulator devices look fine to me, all things considered, and they would be disturbed by moving things around, so that's why I say don't move the components.
I will say that I have a power supply board, cannibalized from commercial equipment, single-sided copper, on which there are no conventional traces, but instead everything is a plane with small lines between the different signals to separate them electrically. I use this board with a fairly conventional 7809 regulator circuit to provide power to my guitar effects chain prior to input to the preamp, which is one of the most noise-sensitive applications readily imaginable, right up there with phono preamps in that regard, and it is remarkably quiet; I get comments from other players on its quietness, in fact, and I have regularly seen it outperform battery-powered effects chains (who knows how good their cables were, but still and all, fairly impressive) from a noise standpoint. So my opinion (and it is an opinion, though somewhat substantiated) of ground planes, at least in power circuits, is better than my opinion of stars; but I must acknowledge that stars do seem to work on ground, and one of course has little choice in a hardwired circuit if one is not to avoid resistance between ground points (steel is a remarkably bad conductor compared with copper, and people tend to forget this, thinking that anyplace on the chassis is just as good as any other). I just don't know if I'd make them quite so... well, ubiquitous is probably the right word.
Personally, I think the star is overrated, because noise isn't just electrons but electromagnetic waves, which means it will have little trouble jumping from one trace to the next, but if you're not doing both power and ground planes, the star might be a better bet than a ground plane alone; and if you're doing the star, then ritchie00boy's design concept, which you're following, is the right way to do it IMHO. If you were doing planes, I'd do three planes, ground sandwiched by input positive and negative under the filter caps, and sandwiched by output positive and negative under the regulators and associated circuitry; but this is expensive with a capital E and probably not worth the effort in this application.
OTOH, I'm not sure I agree with separate traces going to the common pins of (R5, C8, and D5,) and similarly to (R3, C5, D3, and D4,) and (R4 and C4,) and (R6 and C7.) I've grouped these with parentheses in defiance of typographical standards to make it perfectly clear which sets I mean. I would use islands of copper to connect these, rather than traces as you have done; I don't know that moving the components is necessary, and I think it might be undesirable. This concept is a compromise between the plane concept and the star concept, and I think offers some benefits of each with mitigation of the drawbacks of each. The two sides of the ground star between the regulator devices look fine to me, all things considered, and they would be disturbed by moving things around, so that's why I say don't move the components.
I will say that I have a power supply board, cannibalized from commercial equipment, single-sided copper, on which there are no conventional traces, but instead everything is a plane with small lines between the different signals to separate them electrically. I use this board with a fairly conventional 7809 regulator circuit to provide power to my guitar effects chain prior to input to the preamp, which is one of the most noise-sensitive applications readily imaginable, right up there with phono preamps in that regard, and it is remarkably quiet; I get comments from other players on its quietness, in fact, and I have regularly seen it outperform battery-powered effects chains (who knows how good their cables were, but still and all, fairly impressive) from a noise standpoint. So my opinion (and it is an opinion, though somewhat substantiated) of ground planes, at least in power circuits, is better than my opinion of stars; but I must acknowledge that stars do seem to work on ground, and one of course has little choice in a hardwired circuit if one is not to avoid resistance between ground points (steel is a remarkably bad conductor compared with copper, and people tend to forget this, thinking that anyplace on the chassis is just as good as any other). I just don't know if I'd make them quite so... well, ubiquitous is probably the right word.
Yes just check some of your traces - try and thing about current flows and what you want to flow where and in what order. You don't need to star everything, in fact doing that may be worse in some instances. This is particularly true of the reverse protection diode, the idea of which is to shield the regulator from reverse voltage, so connecting it to the load or load capacitor would relise this in the best way.
O.k. That was fun. It now looks like pangea. I'm not sure if this is the way to go. Any votes for this look?
LOL, well, you kinda went whole hog with it, but yeah, that's kind of what I had in mind. I dunno if I wouldn't have left the star in between the regulators... Andrew and ritchie might tell you to put it back in, and I wouldn't argue. I was mostly concerned about the ends of the component sets I named out by the regulators. Nice idea getting the island out to the inside ends of those resistors, though- I'd leave that in. I'd just put the ground star back like I said.
I think you're real close here.
I think you're real close here.
I still think you need to heed a bit more what I said in post 107, but what you have there will work OK. Personally, I don't like the look of that kind of board as it reminds me of something laid out by hand in the 1970's 😀 I like to see either 45 or 90 degree angles only, but that does impose some restrictions on laying out traces then.
Thanks, guys. I modified the design, trying to keep more "sharp" lines. I also moved around the diodes. Those are tricky, especially D6. I couldn't see how to make the layout totally symmetric on both +/- sides, but perhaps, that doesn't matter. So, how does it look now?
That's a bit more reasonable.
I didn't envision the islands going so much beyond the component pads, just being big enough to cover all the pads that were on that signal, but its not necessarily a problem. Let's see what others think.
I didn't envision the islands going so much beyond the component pads, just being big enough to cover all the pads that were on that signal, but its not necessarily a problem. Let's see what others think.
One advantage I was thinking (let me know if this is right) is that if some of the optional components are left out (such as the regulator caps and diodes), then you just have a few holes in those filled planes, but no traces left "hanging". Another good thing is that it makes it a little easier to lay out the components without having to worry about making the traces look "pretty", if you know what I mean.
More diodes? .... how about
So, you have added diodes to the regulator side ... which is good polarity protection and might lend to some slight improvement if Schottky types.
How about a couple more in the input side? Won't make any differences when the circuit is used as filter / regulator for AC or DC wall wart ... and turns the circuit into a full wave rectifier (instead of half wave) ... 1N4003 x 2 ... No?
So, you have added diodes to the regulator side ... which is good polarity protection and might lend to some slight improvement if Schottky types.
How about a couple more in the input side? Won't make any differences when the circuit is used as filter / regulator for AC or DC wall wart ... and turns the circuit into a full wave rectifier (instead of half wave) ... 1N4003 x 2 ... No?
richie00boy said:
Yeah. (Except I'm not using electrolytics on the regs, right?) By the way, what is the size of your board? I'm guessing it's quite a bit bigger than mine, because of those larger caps.
At this point, I think I should really consider stopping further pcb development (except if there is a major issue) until I have the board physically in my hands. I will order it this week, and some parts from digikey, and maybe by next weekend, I can show a working result.
FastEddy, I tried adding the bridge option, but placement of the thire input connection is problematic.
I added the extra diodes, but only by adding a trace on the top layer. I couldn't figure out any other way given the current board size and my requirement to have enough space around the input connectors for adding terminal blocks. I suppose the snubber could be removed, but that's not a good compromise to me. Anyway, does the trace on the top layer bother anyone especially? Also, more importantly, do I have the diodes connected properly?
O.k. I think the board is ready to go. If there are any last comments/suggestions/requests, let me know ASAP. I will order the board by the end of today. Here's the latest links (PNG):
Top + Bottom + Silkscreen:
http://www.cellandtissue.com/ezdac/dual_supply_board.png
Top:
http://www.cellandtissue.com/ezdac/dual_supply_top.png
Bottom:
http://www.cellandtissue.com/ezdac/dual_supply_bottom.png
Schematic:
http://www.cellandtissue.com/ezdac/dual_supply_schematic.png
Thanks, everyone for your help! When I get the board I will post some pics, and then update the thread with some scope measurements after I get it wired up.
Top + Bottom + Silkscreen:
http://www.cellandtissue.com/ezdac/dual_supply_board.png
Top:
http://www.cellandtissue.com/ezdac/dual_supply_top.png
Bottom:
http://www.cellandtissue.com/ezdac/dual_supply_bottom.png
Schematic:
http://www.cellandtissue.com/ezdac/dual_supply_schematic.png
Thanks, everyone for your help! When I get the board I will post some pics, and then update the thread with some scope measurements after I get it wired up.
so into the vat
I believe that the boards you are going to produce will have decent appeal for several different construction types, half wave or full wave rectification, large capacitance or modest, DC or AC input, fixed or adjustable regulation .... all by adding the proper mix of parts.
... Good show, now get to work ...
I believe that the boards you are going to produce will have decent appeal for several different construction types, half wave or full wave rectification, large capacitance or modest, DC or AC input, fixed or adjustable regulation .... all by adding the proper mix of parts.
... Good show, now get to work ...
Can I get one?
Can I get one ?
The true test of any production design is ... if someone else wants one.
"Anything worth doing is worth doing for money ..." - Alfred E. Newman (aka William Gains - Mad Magazine)
Can I get one ?
The true test of any production design is ... if someone else wants one.
"Anything worth doing is worth doing for money ..." - Alfred E. Newman (aka William Gains - Mad Magazine)
- Status
- Not open for further replies.