cdl: Thanks for this question. The NCD has no DC trap built in, nor is it in my opinion the right place to avoid such problems.
When you design a DAC or preamp, you should always ensure, through means of a DC servo, or DC blocking capacitor, that there is no DC out. A good way of checking this, is when you have a relay in the output for delayed connection, there should be no pops when the relay switches on. If there is more than some 0.1mV DC offset, you will hear a pop. This is bad practice.
It would be easy to build in a DC trap, to help all those who have a little DC out of their preamp, and many Class D modules have this. But then since it will always eat a little bit of the sound quality, you would be punishing those who have their preamp free of DC. That's why you see so many users of the other modules, having to modify them to make them play well.
When you design a DAC or preamp, you should always ensure, through means of a DC servo, or DC blocking capacitor, that there is no DC out. A good way of checking this, is when you have a relay in the output for delayed connection, there should be no pops when the relay switches on. If there is more than some 0.1mV DC offset, you will hear a pop. This is bad practice.
It would be easy to build in a DC trap, to help all those who have a little DC out of their preamp, and many Class D modules have this. But then since it will always eat a little bit of the sound quality, you would be punishing those who have their preamp free of DC. That's why you see so many users of the other modules, having to modify them to make them play well.
Lars, thanks a lot for your reply 🙂
Oh I agree, no part of the audio chain should output DC. Not always so in the real (DIY) world though...
Suggestion for NCD 2 though: how about including a DC trap that could be enabled or disabled w. a couple of solder bubbles? Should make everyone happy 😉
/cdl
Lars Clausen said:
Oh I agree, no part of the audio chain should output DC. Not always so in the real (DIY) world though...
Suggestion for NCD 2 though: how about including a DC trap that could be enabled or disabled w. a couple of solder bubbles? Should make everyone happy 😉
/cdl
If DC traps alsways involve a soundquality penalty, why not do it only at the last piece of equip before the power amps? That is, if the dc output of the other components are within reasonable values...Lars Clausen [/i]When you design a DAC or preamp said:
Cheers,
Hans.
We don't want to start wars here over whether DC servo or capacitor is better, do we now... 😀Hans L said:
Hans L said:
Works fine as long as the other components can handle the DC. But you always have to make sure that the DC is taken care of somewhere - can't just plug your equipment into any system .
/cdl
Hans L said:
If only the last equipment has a DC trap, you will get a loud bang each time you plug something into it, as what you are plugging in has 1 - 100 mV DC on it. This DC makes the loud bang, no matter if your last piece of equipment has a DC trap or not. Not very nice to live with....
There are circuits that perform better than just a simple capacitor. To understand what this is about you have to have the understanding of a capacitor, as a complex component with many properties, a side from the capacitance. Especially the electrolytic fluid, which the signal current (electron flow) actually has to go through, has a lot to say in sound quality.
Black Gate capacitors use graphite power in the paper, along with the normal electrolytic fluid, to reduce the effect of the fluid, to some extent. And that is what makes these caps special, and sounding better than the cheapest chinese capacitor, with the same specifications. However there are of course even side effects to that, as has been discussed also here on other threads. (Short lifetime of the capacitor).
If you take the equivalent circuit of the capacitor you have in mind, then you can start playing with a combination of capacitors, resistors and maybe also other components, to make a 'capacitor circuit', that is closer to a perfect DC blocker.
Lars,
Just got your caps today, ty.
I just got other supplies for it yesterday as well, so yep, we're going fully differential with "quality" wires.
Also picked up some solder braid, solder vac, and a new to me Metcal iron.
So, looks like I'm in for a long night. I feel confident I'll be able to provide everyone with a much more complete listening impression just as soon as I've got that all fixed up.
I'll keep it in pre filter feedback mode for the time being, and re-evaluate it later in post feedback mode.
I've also procurred some lab equipment, so I may throw up a few waveforms in a month or two's time. Nothing fancy like an AP would do, I might dabble with rightmark later though, but it's not a priority of mine at all.
BTW I really hope you all appreciate what I had to go through at "The Suck, Circuit Shitty" (the now canuck version of rat scrap).
"This is the Mic cable I would like 50' of"
"Can't sell you that unless you want the whole roll"
"That's not even a whole roll buddy, I bought 50' off it a month ago"
"No no can't do it this is what you want here"
..hands me some junk half the diameter, points to the sticker ... see, 50', "mic cable", exactly what you asked for"
"That's nice, no, this is what I'm guetting"
"Then you're getting the whole roll"
"Look I can go talk to your manager over there since he's the one who sold me 50' off that roll a few months ago.... if you like, or you can go talk to him first, but I'm getting 50' off this and I'm not buying what's left of the roll, and I want the bashed up section that wasn't wound cut off too".
Sometimes I wonder, what force of nature dictates that all the girls with beards work the late shift at Tim Horton's drive through, and all the goofs without a clue migrate to radio shack. Rest assured, there was enough wire left on the roll the hang themselves with.
Just got your caps today, ty.
I just got other supplies for it yesterday as well, so yep, we're going fully differential with "quality" wires.
Also picked up some solder braid, solder vac, and a new to me Metcal iron.
So, looks like I'm in for a long night. I feel confident I'll be able to provide everyone with a much more complete listening impression just as soon as I've got that all fixed up.
I'll keep it in pre filter feedback mode for the time being, and re-evaluate it later in post feedback mode.
I've also procurred some lab equipment, so I may throw up a few waveforms in a month or two's time. Nothing fancy like an AP would do, I might dabble with rightmark later though, but it's not a priority of mine at all.
BTW I really hope you all appreciate what I had to go through at "The Suck, Circuit Shitty" (the now canuck version of rat scrap).
"This is the Mic cable I would like 50' of"
"Can't sell you that unless you want the whole roll"
"That's not even a whole roll buddy, I bought 50' off it a month ago"
"No no can't do it this is what you want here"
..hands me some junk half the diameter, points to the sticker ... see, 50', "mic cable", exactly what you asked for"
"That's nice, no, this is what I'm guetting"
"Then you're getting the whole roll"
"Look I can go talk to your manager over there since he's the one who sold me 50' off that roll a few months ago.... if you like, or you can go talk to him first, but I'm getting 50' off this and I'm not buying what's left of the roll, and I want the bashed up section that wasn't wound cut off too".
Sometimes I wonder, what force of nature dictates that all the girls with beards work the late shift at Tim Horton's drive through, and all the goofs without a clue migrate to radio shack. Rest assured, there was enough wire left on the roll the hang themselves with.
Hello,
If it was Lars' intention to provide the absolute highest quality amp on the market, it's my opinion the supply for it is of a lesser standard. If you can, you stand to benefit from building your own. Adding the T-networks as I did was alot of work. The rectifiers used are great, the way in which they're used is far from optimal, as my preference would be having them isolated from one another. There is no snubbing on them at all. While they are FRED rectifiers which relaxes these requirements, they do benefit from aditional snubbing. You better know what you're doing if you plan on adding some. I'm not providing a how to.
Having done so, HF hash has cleaned up a good amount. To that end I've also twisted all the wires from the mains right down to the speaker terminals.
It's still very much suffering from a Nerf transformer, at this point I've no intention of replacing it... my money went elsewhere (so sorry). However the additional caps for my 4 ohm loads seems to have helped it a good deal. The deep lows are no longer straining to keep pace, they're not at all as they ought to be, but at least they're there when they should be, if that makes sense.
The combination of BHC T-netwoks, added bridge rectifier snubbing, and change of supply wiring has seemingly rendered it much more transparent, still very musical. So I think I did a decent job of it.
I also threw in some bleeders on the supply, to help tame any ringing in the ground plane of the supply that I can't do anything about, and really no other reason. They're high enough in value that they hardly represent any load at all.... they're just 1/4W bleeders.
I have more work to do before you get the full review, working on getting it differentially driven now, and still have to replace the caps for the input stage. Lars gave me some panasonic FC 25V 100uF caps. Guess I'll use em. It seems I've also been giving the oppertunity to test some black gates, possibly nx series.. will have to check. 4.7uF 50V caps though, so I'll stick the FC's in one module, BG's in another, see which I prefer and then make them all the same.
I'm not going to humor the BG's with a 10000 hours burn in time, so by tomorrow or the day after at the latest I"ll give a much more detailed listening impression before taking it to post filter feedback mode and doing it all over again.
One thing can certainly be said for it, the dynamics are truly explosive.
l8tr
If it was Lars' intention to provide the absolute highest quality amp on the market, it's my opinion the supply for it is of a lesser standard. If you can, you stand to benefit from building your own. Adding the T-networks as I did was alot of work. The rectifiers used are great, the way in which they're used is far from optimal, as my preference would be having them isolated from one another. There is no snubbing on them at all. While they are FRED rectifiers which relaxes these requirements, they do benefit from aditional snubbing. You better know what you're doing if you plan on adding some. I'm not providing a how to.
Having done so, HF hash has cleaned up a good amount. To that end I've also twisted all the wires from the mains right down to the speaker terminals.
It's still very much suffering from a Nerf transformer, at this point I've no intention of replacing it... my money went elsewhere (so sorry). However the additional caps for my 4 ohm loads seems to have helped it a good deal. The deep lows are no longer straining to keep pace, they're not at all as they ought to be, but at least they're there when they should be, if that makes sense.
The combination of BHC T-netwoks, added bridge rectifier snubbing, and change of supply wiring has seemingly rendered it much more transparent, still very musical. So I think I did a decent job of it.
I also threw in some bleeders on the supply, to help tame any ringing in the ground plane of the supply that I can't do anything about, and really no other reason. They're high enough in value that they hardly represent any load at all.... they're just 1/4W bleeders.
I have more work to do before you get the full review, working on getting it differentially driven now, and still have to replace the caps for the input stage. Lars gave me some panasonic FC 25V 100uF caps. Guess I'll use em. It seems I've also been giving the oppertunity to test some black gates, possibly nx series.. will have to check. 4.7uF 50V caps though, so I'll stick the FC's in one module, BG's in another, see which I prefer and then make them all the same.
I'm not going to humor the BG's with a 10000 hours burn in time, so by tomorrow or the day after at the latest I"ll give a much more detailed listening impression before taking it to post filter feedback mode and doing it all over again.
One thing can certainly be said for it, the dynamics are truly explosive.
l8tr
Good job!
Chris,
More late nights for sure. Some real and significant progress here.
How do you like the Metcal? Makes life a little easier I bet. Which model did you end up with?
I do have a question, are you sure the BG’s were intended for decoupling and not as input replacements? There is a real disparity in ESR between them and the FC’s.
Sorry to hear there is no plan for a better transformer. I also think it will be a real limitation to performance.
Roger
Chris,
More late nights for sure. Some real and significant progress here.
How do you like the Metcal? Makes life a little easier I bet. Which model did you end up with?
I do have a question, are you sure the BG’s were intended for decoupling and not as input replacements? There is a real disparity in ESR between them and the FC’s.
Sorry to hear there is no plan for a better transformer. I also think it will be a real limitation to performance.
Roger
Has anyone seen a list of ESR's for BG's? My own measurements show, that the BG's actually are not that good in ESR, comparable to tantalums. But FC's are actually very low ESR. (Not a quality in itself, in audio relations).
Hey Chrisclassd4sure said:
Snubbers are great, they just calm everything down and allow the music to sound more real, without loosing any dynamics.
If you don't mind me asking, what snubber did you use
The BHC T-network caps is where some of the dynamics are coming from and the rest will be coming from the NCD's. They are really (and 1 more really, so, really, really) fast
I wonder if Lars have some spare to try paralleling them
Lars, what do you think
Twisting all the PS wires is also very important. What effect did this have on its quietness, ie. the background hiss
thanks
The T-Networks are incredibly expensive, but i am wondering if it would be worthwhile making a variant of the power supply board with T-Net's.
It would end up costing like double up on the existing board, and the capacitance still just 10000uF/63V.
It would end up costing like double up on the existing board, and the capacitance still just 10000uF/63V.
Hey LarsLars Clausen said:The T-Networks are incredibly expensive, but i am wondering if it would be worthwhile making a variant of the power supply board with T-Net's.
It would end up costing like double up on the existing board, and the capacitance still just 10000uF/63V.
The T-Networks I have are 10KuF (per rail) and, yes, they are incredibly expensive, but, putting them in parallel would give 20KuF per rail 😛 Is that correct
Unfortunate, but they are really, really fast. I wonder if they are better then the Jensen 4poles, which, can be paralleled 😎
Chris has done this already
Actually Lars, the NCD1 must be pretty darn good considering considering the tranny is only 200VA. 😎 The fact that it performed so well when pitted against your friends shop's 4/5K commercial amp is an excellent result for your NCD1
Here's to Chris's next mod and your next improvement
Can it get better ... well it is right before our eyes 😱
excellent and here's to the future
thanks
Lars Clausen said:
FM's should be even lower.. would you like me to try those instead? 220uF 25V panasonic FM?
Snubbers used are almost the same as on my UCD amp, they're not properly designed, just thrown on based on what I had kicking around. So that's about 22 ohms half what (two // 1/4 w resistors) in series with a 470nF MKP X2 "EMI supression" cap. Soldered to the pins of the rectifiers on the top side of the board such that they're across secondaries at the rectifiers, same as on my UCD.
All the changes I mentioned were done in one shot, because it's obviously not the type of thing you need to test one at a time, just basic implementation if you ask me.
What's it done for the noise floor? Not a damn thing, nor did I expect it to.
It seems to have done something towards heterodyne noise ... the HF hash I mentioned. It's still there to some degree but less apparent.
I removed the original plugs on the PSU, removed the additional Fast On tabs, the 20k 1/4w bleeders go from the plug to the fast on tab solder areas. The inputs to the T networks are soldered directly to the original cap terminals on the underside of the board with with 16 or 14 awg (can't recall) cardas litz wire doubled up /twisted all the way, they snake out between the rectifiers to the Networks which are sitting horizontally on top of the PSU, one on either side of the original caps. The brackets I made up for them are bolted to the original mounting bolts of the rectifiers and allow the caps to nestle in nice and close. The output wires from rail to rail go in between the original caps, all wires are the same gauge litz doubled up, the output grounds are also doubled and create an output common at the module itself. That's to say the ground wires from the positive rail extend to the module, as does the ground wires for the negative rail. I had to make use of an "extra" ground location on the plug of the module to get them all in there... Lars wasn't good enough to tell me if it was built to handle high current so I just gambled and stuck it there anyway... no smoke so far.
Each ground wire is twisted with it's respective rail. It really makes for a messy looking spagheti incident given there's so many wires ... because each channel has 4 ground wires etc. Works really good though.
If you decide to make a supply with T-net caps please by all means... do them justice! There's also the possibility of a better supply with standard caps.
BTW on the underside of the PCB on the pinch points of the wires I also used a bit of heatshrink.... in effect it is triple insulated at pinch points because it's litz wire.
Either BHC or Jensen's can be paralleled in all the exact same ways.
I'll get a picture of all that when I can, sadly you wont' see the underside (nice work too I must say) because it's all assembled now. Anyway this isn't a how to, but the info is given so you can see it's nothing secrete, no BS envolved, just plain and simple needed to be done upgrades.
It's also by far not the ideal way to use the T-networks, in terms of the first 2 pole caps having a common node in the PCb ground plane, ideally they'd not. Also there's a way to parallel them which eliminates what Lars calls the woppling effect or whatever.. I didn't bother with that because of wiring constraints.
So it could be better still, but it's much better than it was.
I spent 150$ on cheap radio scrap mic cable and Neutrik connectors in order take it fully differential and further reduce the hiss seen with the RCA's. So that's the next move and then I'll swap all the bypass caps on the input stage so they match and I can get a feel for the imaging for the first time.
After that I'm going to a serious look at my audiodock 😉
All the changes I mentioned were done in one shot, because it's obviously not the type of thing you need to test one at a time, just basic implementation if you ask me.
What's it done for the noise floor? Not a damn thing, nor did I expect it to.
It seems to have done something towards heterodyne noise ... the HF hash I mentioned. It's still there to some degree but less apparent.
I removed the original plugs on the PSU, removed the additional Fast On tabs, the 20k 1/4w bleeders go from the plug to the fast on tab solder areas. The inputs to the T networks are soldered directly to the original cap terminals on the underside of the board with with 16 or 14 awg (can't recall) cardas litz wire doubled up /twisted all the way, they snake out between the rectifiers to the Networks which are sitting horizontally on top of the PSU, one on either side of the original caps. The brackets I made up for them are bolted to the original mounting bolts of the rectifiers and allow the caps to nestle in nice and close. The output wires from rail to rail go in between the original caps, all wires are the same gauge litz doubled up, the output grounds are also doubled and create an output common at the module itself. That's to say the ground wires from the positive rail extend to the module, as does the ground wires for the negative rail. I had to make use of an "extra" ground location on the plug of the module to get them all in there... Lars wasn't good enough to tell me if it was built to handle high current so I just gambled and stuck it there anyway... no smoke so far.
Each ground wire is twisted with it's respective rail. It really makes for a messy looking spagheti incident given there's so many wires ... because each channel has 4 ground wires etc. Works really good though.
If you decide to make a supply with T-net caps please by all means... do them justice! There's also the possibility of a better supply with standard caps.
BTW on the underside of the PCB on the pinch points of the wires I also used a bit of heatshrink.... in effect it is triple insulated at pinch points because it's litz wire.
Either BHC or Jensen's can be paralleled in all the exact same ways.
I'll get a picture of all that when I can, sadly you wont' see the underside (nice work too I must say) because it's all assembled now. Anyway this isn't a how to, but the info is given so you can see it's nothing secrete, no BS envolved, just plain and simple needed to be done upgrades.
It's also by far not the ideal way to use the T-networks, in terms of the first 2 pole caps having a common node in the PCb ground plane, ideally they'd not. Also there's a way to parallel them which eliminates what Lars calls the woppling effect or whatever.. I didn't bother with that because of wiring constraints.
So it could be better still, but it's much better than it was.
I spent 150$ on cheap radio scrap mic cable and Neutrik connectors in order take it fully differential and further reduce the hiss seen with the RCA's. So that's the next move and then I'll swap all the bypass caps on the input stage so they match and I can get a feel for the imaging for the first time.
After that I'm going to a serious look at my audiodock 😉
Re: Good job!
Yeah I know, as much as I'd love to Plitron is just too much of a freakn rip off so to hell with em. I've got some serious lab equipment on the way and I"ll start working on SMPS.
I haven't even used the metcal yet 🙂 It's the SP 200 I think. Very nice indeed.
As per the bypass caps I don't know if there's additional RC snubbing to keep ringing down, perhaps lowest possible ESR is not in fact the best idea, which is also why I"ll try the BG's. 4.7uF ought to be enough to let the bass breath don't you think?
Anyway, I never heard BG's before, and I have a Metcal I want to get dirty 🙂
sx881663 said:
Yeah I know, as much as I'd love to Plitron is just too much of a freakn rip off so to hell with em. I've got some serious lab equipment on the way and I"ll start working on SMPS.
I haven't even used the metcal yet 🙂 It's the SP 200 I think. Very nice indeed.
As per the bypass caps I don't know if there's additional RC snubbing to keep ringing down, perhaps lowest possible ESR is not in fact the best idea, which is also why I"ll try the BG's. 4.7uF ought to be enough to let the bass breath don't you think?
Anyway, I never heard BG's before, and I have a Metcal I want to get dirty 🙂
Lars Clausen said:The T-Networks are incredibly expensive, but i am wondering if it would be worthwhile making a variant of the power supply board with T-Net's.
It would end up costing like double up on the existing board, and the capacitance still just 10000uF/63V.
If you could, I think the slickest idea would be a more universal PCB that can accept Tnetworks. Then you can use the same one for a cheaper version, or include the T networks at a higher price, or users will at least have the option of upgrading at a later time when they can better afford to.
Those so inclined will always build their own supply too. Don't shoot yourself in the foot by building a supply no one can afford.
That is of course possible, but one thing that should be considered is that upgrading is not possible. When you solder something into a solid power plane, it's as good as impossible to desolder it without killing the THP.
But still it's good for building different versions. 🙂
But still it's good for building different versions. 🙂
- Status
- Not open for further replies.