Hey, guys. I recently found the "the wire" here on this forum and wanted to build one for myself. Given the lack of boards and the fact that I have loads of through-hole parts laying around, I decided to try my hand at designing a through-hole version of the PSU and SE-SE boards. I tossed it together in eagle. Everything's included in the zip. If anyone feels like doing some sanity checking or giving some feedback on any improvements that could be made, I would appreciate it. Thanks.
I've got one like that too...A LME49990 driving two parallel BUF634.
I did a board for mine. I have heat sinks on my BUF634s so it's not
very photogenic, all you see are the heat sinks...
So, I had a look at the amp board.
You did the LME49990 and LME49600 in SMD
and everything else in PTH.
I'm not sure of the reasoning behind those choices...
The power traces are too small.
The left and right negative power rails are connected
by the thermal area. You put a thermal relief in your
thermal area...kind of defeats the purpose.
I didn't look at the power supply yet.
You did the LME49990 and LME49600 in SMD
and everything else in PTH.
I'm not sure of the reasoning behind those choices...
The power traces are too small.
The left and right negative power rails are connected
by the thermal area. You put a thermal relief in your
thermal area...kind of defeats the purpose.
I didn't look at the power supply yet.
Last edited:
Thanks for the input. I made the board that way due to having some LME49600's laying around and the LME49990's being SMD. I can do through-hole plating myself as well. I see your point, however. Kinda defeats the point.
I realized that I also happened to have a few DIP8 BUF634's in the parts bin as well, and that I could always just use the LME49990's on adapters, so I decided to make a single sided board. Needs a few jumpers, but otherwise all single sided. The ground does meander a bit, and I'm not sure if that's an issue or not. I could always add a ground plane to the top layer and some vias.
I made some changes to my original board for kicks, as well as another version halfway between the two just because all I had to do was click save after I changed the opamps.
Feel free to look at what I've attached and leave comments if you like. If not I'll probably just go with what I've got and stop posting my random cruft. ^_^
Thanks again for any help.
I realized that I also happened to have a few DIP8 BUF634's in the parts bin as well, and that I could always just use the LME49990's on adapters, so I decided to make a single sided board. Needs a few jumpers, but otherwise all single sided. The ground does meander a bit, and I'm not sure if that's an issue or not. I could always add a ground plane to the top layer and some vias.
I made some changes to my original board for kicks, as well as another version halfway between the two just because all I had to do was click save after I changed the opamps.
Feel free to look at what I've attached and leave comments if you like. If not I'll probably just go with what I've got and stop posting my random cruft. ^_^
Thanks again for any help.
I dont see how this is 'the wire' at all, its bad enough publicly publishing a design that youve blatantly lifted the idea for and probably not asked permission or even informed Owen, but to just use the same name for a different design that has gone under no testing? I really dont know why you would do that? the layout is not worthy of the name IMO, traces crossing, no power or ground planes, no useful heatsink, vias all over the place, decoupling caps that are too far away, too physically large, on the end of a long trace with a via before the pin that will have very limited usefulness (OTT high quality decoupling and tight layout is what makes this amp shine), long feedback paths, traces that are too thin. huge long, thin traces from the main reservoir caps that are at the end of the board that has the lowest power demands and highest PSRR?
will all lme49990+lme49600 designs be called 'the wire' from now on? Owen designs very clean deliberately SMD for high performance and measures them to within an inch of their lives. I dont see that throwing a couple of the same parts onto a DIY PCB with no testing has anything to do with 'the wire'. and you are not the first to do this.
will all lme49990+lme49600 designs be called 'the wire' from now on? Owen designs very clean deliberately SMD for high performance and measures them to within an inch of their lives. I dont see that throwing a couple of the same parts onto a DIY PCB with no testing has anything to do with 'the wire'. and you are not the first to do this.
and you know it has 'easily the same performance' how? sorry i'm not buying a perfboard amp having the same performance as a multilayer SMD amp when it comes to DC coupled wide bandwidth amps the parasitics are crucial. 'the wire' SE-SE is still the highest measured performance opc has ever measured on his AP APx585 and various other test gear, it tested his ability to get proper measurements.
its cool you like your amp, but making baseless throwaway statements like the above in reference to a design that was designed very objectively, with nothing to back them up... tweaks my ******** probe.
btw 0.02Ω is not even close to the same, let alone easily the same. the SE-SE is in the fractions of a miliohm. will that alone make a huge difference? no probably not; but its not the same by any means. also the SE-SE has a noise floor of -160 and -164dBV and here are the rest of the results starting here and filling the page
you have frequency response like this? thats 2 channels overlayed, so the channels match impeccably, not going to happen on perf. the bump at 80khz is the AP, not the amp ...
Last edited:
you can change the title yourself, the first post remains editable indefinitely to the OP, not sure about the subject but the content can.
its all good if you are asking for advice, no worries, but certainly speaking for myself you would get a more productive and less uptight response if you just present it as an amp you are playing with, mention being inspired by the wire' in the first post; whatever, but there is no need to call it 'the wire TH'.
Bare in mind Owen is making a go at making fully commercial amps too
its all good if you are asking for advice, no worries, but certainly speaking for myself you would get a more productive and less uptight response if you just present it as an amp you are playing with, mention being inspired by the wire' in the first post; whatever, but there is no need to call it 'the wire TH'.
Bare in mind Owen is making a go at making fully commercial amps too
@qusp
You know, it kind of a complement that people want to associate something they built
to "The Wire". I'm not saying it right or even acceptable, but how often have we asked
for an Asprin when were really know we are getting so other brand of pain killer.
How often has an amp been referred to as a CMoy evey though it was not even close
in design to what Chu Moy built? It might be inevitable that any op amp/IC buffer amp
gets referred to as "The Wire".
@Cyko500
Just find your own name for it and your fine. Just don't try and associate it with
"The Wire" or the "Ventus EZ" or the "Dragonfly V2" and probably others I don't know about...
You know, it kind of a complement that people want to associate something they built
to "The Wire". I'm not saying it right or even acceptable, but how often have we asked
for an Asprin when were really know we are getting so other brand of pain killer.
How often has an amp been referred to as a CMoy evey though it was not even close
in design to what Chu Moy built? It might be inevitable that any op amp/IC buffer amp
gets referred to as "The Wire".
@Cyko500
Just find your own name for it and your fine. Just don't try and associate it with
"The Wire" or the "Ventus EZ" or the "Dragonfly V2" and probably others I don't know about...
Sure, sometimes with other designs I have seen the same, but with people somehow assuming/claiming that their design will have the same high performance, just because the same parts and/or schematic are used. that isnt really what we see here, just a case of mistaken identity and seems pretty harmless.
I guess it just hurts my head that opamp+buffer = the wire, when really it had very little to do with the parts. as you say and as ive said in these cases before, the parts themselves, which is usually all the designs have in common (sometimes only vaguely), are pretty standard high performance ICs arranged in a fairly standard format. what makes it the wire is the performance, execution and OTT objective (ie. not frugal) design philosophy, all are missing here.
perhaps Owen wont even use the name for his commercial designs, it would be difficult now and thats a pity to not be able to leverage the well respected 'brand'. flattery doesnt pay the bills
of course it might help his cause if he was able to get them out this year too before another 5-10 copies are out there.
I guess it just hurts my head that opamp+buffer = the wire, when really it had very little to do with the parts. as you say and as ive said in these cases before, the parts themselves, which is usually all the designs have in common (sometimes only vaguely), are pretty standard high performance ICs arranged in a fairly standard format. what makes it the wire is the performance, execution and OTT objective (ie. not frugal) design philosophy, all are missing here.
perhaps Owen wont even use the name for his commercial designs, it would be difficult now and thats a pity to not be able to leverage the well respected 'brand'. flattery doesnt pay the bills
of course it might help his cause if he was able to get them out this year too
before another 5-10 copies are out there.
Last edited:
so anyway honestly Cyko500: I think you have a lot of work to do here on this design, i'll have a look tomorrow.
one thing I will say, just using a small value, low impedance capacitor type doesnt make for good HF decoupling, the enemy of proper HF decoupling is inductance and distance. inductance comes in the form of vias, thin, long traces and the package inductance, you have all of these on your decoupling caps
one thing I will say, just using a small value, low impedance capacitor type doesnt make for good HF decoupling, the enemy of proper HF decoupling is inductance and distance. inductance comes in the form of vias, thin, long traces and the package inductance, you have all of these on your decoupling caps
Last edited:
Here's mine,take a look...
Specs:
An externally hosted image should be here but it was not working when we last tested it.
Specs:
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
right, so first of all congrats on a cute little amp and backing yourself with the measures, but its not even in the same ballpark sorry mate. Everything is significantly worse; noise is 60dB higher, crosstalk under load is pretty ordinary and ~24db worse, THD ~25-30dB worse, the channels are not matched, FR is something like 90-100x or more worse and the measurement only goes up to 15kHz so it probably gets more sketchy, while the wire is ruler flat, has a 0.002dB wiggle.
I wont go on, every single measure is like that, guess you are sharing a ground on the channels on the board to get crosstalk issues like that under load, or are using a mini/phono jack with bad crosstalk
noise may be better than the above, if the ADC you are using to test it with is limiting the measurement, but that seems unlikely, as most soundcards will have better spec than that.
I wont go on, every single measure is like that, guess you are sharing a ground on the channels on the board to get crosstalk issues like that under load, or are using a mini/phono jack with bad crosstalk
noise may be better than the above, if the ADC you are using to test it with is limiting the measurement, but that seems unlikely, as most soundcards will have better spec than that.
Last edited:
"No Load" represents the exact specs of my soundcard in loopback.Sorry,I don't 10000$ AP.
Besides,I'm curious about the device that has -160 dbA noise floor?I measured the latest AP on Electronica 2012 exhibition,and it did -126 dbA.
Edit : I can confirm absolutely flat FR up to 10 Mhz.
I found the spectrum measurements:
As you can see,it just repeats the soundcard output at any load.The card is Audiophile 2496.I don't think it's that bad,and I'm sure it will do better with better ADC.
Besides,I'm curious about the device that has -160 dbA noise floor?I measured the latest AP on Electronica 2012 exhibition,and it did -126 dbA.
Edit : I can confirm absolutely flat FR up to 10 Mhz.
I found the spectrum measurements:
An externally hosted image should be here but it was not working when we last tested it.
As you can see,it just repeats the soundcard output at any load.The card is Audiophile 2496.I don't think it's that bad,and I'm sure it will do better with better ADC.
Last edited:
it isnt close to flat up to 15khz ... how do you explain the huge difference in crosstalk with somewhat higher power? that 8Ω loading wont make any difference to the soundcard. if everything else there is your soundcard, that makes the measurements pointless except to show gross error and you cannot truthfully make the claim you have.
re 160dBa, the wire does I presume your DbA measure is at 1khz? should be the same as standard dB in that case. not sure what Owen used for that measurement, but given i've seen measures from Ap2, the AP above, an incredible HP analyser and knowing where he works, I tend to trust it.
edit: re -160dB, its probably done with a long FFT
you can go on thinking that perfboard with no ground plane and pretty red film decoupling caps is just as effective if you like, I disagree and without some meaningful evidence, we can only take your pretty crappy measurements above as evidence to the contrary
btw again, no worries cool little amp, but why the compulsion to make claims of objective performance that are just biased assumptions?
re 160dBa, the wire does
I presume your DbA measure is at 1khz? should be the same as standard dB in that case. not sure what Owen used for that measurement, but given i've seen measures from Ap2, the AP above, an incredible HP analyser and knowing where he works, I tend to trust it. edit: re -160dB, its probably done with a long FFT
you can go on thinking that perfboard with no ground plane and pretty red film decoupling caps is just as effective if you like, I disagree and without some meaningful evidence, we can only take your pretty crappy measurements above as evidence to the contrary
btw again, no worries cool little amp, but why the compulsion to make claims of objective performance that are just biased assumptions?
Last edited:
@qusp,I think we have a misunderstanding.I'm not pushing anything,in fact,I had even read your posts on the previous page before posting photos and specs of the amp.That post was generally appointed to the thread,not to you.I just wanted to represent some good practices with this kind of designing and building,as I acquired them with experience.
When I was building this amp 2 years ago(I started before "The Wire" appeared),I had about -60dB crosstalk without load.I was told by many "specialists",that this is the limit a perfboard can go,so I should leave it that way.Later I've build a second version with larger decoupling caps,and went to -85 dB,then I was told that the problem is with stereo jacks.My third version(above) has better grounding and does -94 dB with headphones on.Of course,at 8 Ohm it's much worse,but that's over the roof anyway.At since it repeats the performance of a professional soundcard,I can honestly say that perfborads are not that bad for a single project(which,I presume,is what the author aims at).
I'd ask how you measured 160db with AP's "24-bit Analog Devices ADC",but I don't think that's necessary.I'm sorry for loading the thread with useless posts.Have a nice day!
When I was building this amp 2 years ago(I started before "The Wire" appeared),I had about -60dB crosstalk without load.I was told by many "specialists",that this is the limit a perfboard can go,so I should leave it that way.Later I've build a second version with larger decoupling caps,and went to -85 dB,then I was told that the problem is with stereo jacks.My third version(above) has better grounding and does -94 dB with headphones on.Of course,at 8 Ohm it's much worse,but that's over the roof anyway.At since it repeats the performance of a professional soundcard,I can honestly say that perfborads are not that bad for a single project(which,I presume,is what the author aims at).
I'd ask how you measured 160db with AP's "24-bit Analog Devices ADC",but I don't think that's necessary.I'm sorry for loading the thread with useless posts.Have a nice day!
Last edited:
Definitely a case of FFT gain, yes. Remember it's pretty healthy, 30+ dB even for N=16384.
In The Wire SE-SE, noise is resistor-limited, with a bit of extra current noise contribution (the LME49990 has in=2.8 pA/sqrt(Hz)!). When fed from a noiseless 0-ohm source, effective input voltage noise works out to about 6 nV/sqrt(Hz). Good, but not particularly close to what the LME49990 could do (en<1 nV/sqrt(Hz)). Noise would be the same with a LM4562 or NJM2068. At unity gain, dynamic range does remain very high though (about 127 dB for a 2 Vrms signal @ 20 kHz BW, or around 140 dB at the maximum of about 9 Vrms, certainly distortion limited at this point). Nobody's gonna hear sub-1µV noise, that's for sure.
I'm just wondering what people are supposed to drive the thing with while not compromising noise or distortion or both. Maybe that's the point of calling it "The Wire". But you're still missing a gain stage and volume control.
@GTrenchev, what sort of opamp are you using in there? Even if the LME49990 were available in DIP (it isn't), I'd assume a 100+ MHz GBW opamp at near unity gain wouldn't take kindly to sockets and a perfboard layout. Almost 80 dB of chsep driving 8 ohms is quite impressive though. That has to be power amp territory, in spite of a shared ground.
Last edited:
thats not the point, the point is to have it so that it doesnt leave any finger prints on what you feed it. I said nothing about the performance being necessary, I simply objected to the baseless claim of equal performance with perf
only if you actually NEED gain and a volume control.
unity gain is more than enough for most headphones with many modern sources, which also often have digital volume control, which is why the design is how it is.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.