Hypex Ncore

Status
Not open for further replies.
uh, unless it ends up costing as much as a small car, then I'm out :eek: :D
OK how about a motorcycle then? J/K :p

Have you considered an all in one power amp/power supply combination similar to the ice power modules for driving subwoofers ??

Given the ease with which an SMPS and amp can be wired together I've never understood the fascination with combos, but the demand is there and UcD + SMPS projects are being started. With all the aux supplies and whatnot.
 
What about a well tuned discrete input stage? Or use a top notch opamp like the LM4562 biased into class A.

Use of high quality components (eg. low ESR caps for power supply decoupling).

Its also a bit painful to get the earth's right on the UCD's (at least for the original models I use) when using non balanced connections, so better grounding arrangements for unbalanced inputs would help.

As I have an amp using the first generation UCD's (standard not HG) with OPA627's biased for class A, I'm considering swapping out the modules for these latest ones. What changes to the sound should I expect comparing the old UCD400 with these new modules (note I didn't say better - I'm pretty happy with the old UCDs but still interested in what has changed)?
 
Last edited:
Its also a bit painful to get the earth's right on the UCD's (at least for the original models I use) when using non balanced connections, so better grounding arrangements for unbalanced inputs would help.
It can't get any easier than it is. If you have two ground paths, for instance your supply is grounded to chassis and so is your RCA input, just don't connect your RCA ground to the ground pin of the amplifier module as well because then you get current flowing through the RCA cable. Use the "cold" input as ground instead and for Pete's sake don't connect the cold pin to the amplifier ground directly. Once you get your head around how differential inputs work you'll realise that it already does exactly what you want, that is solve ground loop problems for you. There used to be an app note on the site for this, I'll have a look round where it went and put it back.
 
RCA to Balanced - Hypex AppNote

Here is the "old" wiring appnote from hypex site. I think for quite a few of us it will still take trial-error approaches.
My "gut-feel" is that right approach is the one on right hand side (power supply/amp/grounded to the amp-chasis). Balanced input connector has ground connected to the amp input (and from there it is grounded to amp-chasis)). Unbalanced input, you build a cable as per the picture (unbalanced ground/negative goes into the "cold" in balanced input, and shield also connects to the unbalanced source ground/negative at the source, and at the balanced input on the amp gets effectively connected to chasis).
Hope i got it right ;) ?

tx
 

Attachments

  • wiring.pdf
    95.5 KB · Views: 611
Its the bit about not connecting power ground to the chassis, for DIY implementation I'm not comfortable with a metal chassis (which also provides good EMI shielding) not connected to power ground. I ended up with the arrangement you suggested, where the chassis was connected to power ground and the module grounds including RCA ground were isolated from the chassis ground. I did find, as suggested, that the earth loop was eliminated by connecting the RCA ground to the amp cold input, not amp earth.
 
Hi,
These connections are governed by the rules.(depend upon the class of the psu and the class of the product).
chassis must be connected to the earthing.
(this requires that the SMPS (if using an EMI filter balanced), must have the central pole in the chassis (earth) and not on output gnd.
Therefore, in the case of RCA, must be isolated from the chassis.

Regards

Roberto P.
 
Last edited:
IMNSHO the word "ground" should be forcefully removed from the school curriculum. The conflation of "chassis potential", "earth potential", "power supply return" and "reference potential" into that one little sneaky cowardly word "ground" is the source of an absolutely incredible array of engineering problems the world over. Conversely, the simple act of not using the same galvanically connected network for both reference potential and any combination of the other three solves all problems. Chassis, earth and PSU null may be interconnected in any way you like provided that actual signals are accompanied by another wire carrying the reference potential of the source. The receiver, of course, should be capable of subtracting the voltages between the two wires, not attempt to take a shortcut by forcing one of the two to its own "ground" (or to try and force its own "ground" to follow the potential of the reference wire).

Class I equipment needs direct earthing of the chassis for safety, class II equipment does not. However, in the latter case you may not use earthed mains inlets. You may have noticed that mass market consumer audio gear has class II wiring and a 2-prong mains cord. This is because unbalanced connections simply do not mix easily with earthing.

If you cannot construct per class II, floating the secondary ground is a commonly used and very bad practice. Very bad because it requires star connections which are a source of HF problems. It is much better to use a differential input to float the reference terminal ie the RCA shell and sense between the shell and center pin without making a direct connection to chassis or any other of the irrelevant "grounds". One would place a cap and perhaps a resistor to chassis to limit the voltage between the two and stop the connection from becoming an antenna.

That way the differential input acts like an input transformer: only the voltage between two input pins matters, not the voltage with respect to whatever one may think is "ground". There is always an easy solution that does not require fancy floating schemes if you have a differential input at your disposal. There is neither a simpler nor a better solution than the one already in place in the UcD modules. The only thing we can do is write a very long and detailed document explaining the operation of differential inputs. Trust me, once the penny drops you'll bang your head and say "if only I knew it was that obvious". One thing you're guaranteed to do, for instance, is immediately adopt the scheme on the right hand side of the wiring application note. This is much better than any floating secondary or differential RCA connection you can dream of.

So let's move back to the topic of what the Ncore DIY module is going to look like.
 
Last edited:
From a practical point of view, maybe a heatsink which can be acessed (screwed) from both sides? On the UcD400 you have to drill across the chassis; or add a second, larger, plate before being able to fix it from the inside onto the chassis or thick external heatsink.
 
For home hifi i builded and played with Hypex UCD180 with HXR modules. I remember with these modules that the control was really impressing. Though, i couldn't get used to the overall sound of class D. There was a lack of liveness, warmth, involving character of the sound comparing to modern class A/B amps. Not to say that the modules where bad, i suppose that the effect has to do with the absence of (harmonic) distortion? I don't know.

What can be said about the sound of the new Ncore amp in this area?

thx
 
@Shaman, Yes I think indeed we're go for the discrete buffers. And probably for the loop amplifier too while we're at it (Ncore isn't fully passive like UcD).
I'm still collecting votes on the power issue though. There's some who think 400W is already intimidating and others who can't get enough power... I was wondering though, suppose I make it 400 at 4 ohms but like the 1200, low impedance capable (so you get 600~700W into 2 ohms with a realistic PSU). Would that be a compromise? Power hungry folks just add woofers :)

For the output connectors I'm considering having custom gold plated screw tags made, a П-shaped bridge with a screw through the top that pulls a square nut upwards, so the wire goes between the nut and the top. The nice thing is I can connect the power stage on one leg and take the feedback off the other so you can screw in multiple sets of wire for bi-wiring and really have only 1.5 milliohms of common impedance between them.

The input and power connectors are another kettle o' fish because now there's also a +/-15V supply and I'm not that happy with the current Molex KK for the input. So there I still haven't found a decent alternative to using a high quality industrial type and including preconfectioned cables with the module.

@brubeck
I find it hard to describe the sound of something that essentially has no sound of its own. The music's just there. I suppose it's what people call "effortless", but in a manner that doesn't even try to impress. It just refuses to draw attention to itself.
 
I vote for the 400W + version as long as it is easy to implement (initimidation for me is rather the complexity of possible connections for beginners than the power itself). As for low impedance, probably it depends if there are a more speakers with low impedances (which would benefit from this approach) than low sensitivity (which would ask for even higher power) in the diy community ?
 
I'm still collecting votes on the power issue though. There's some who think 400W is already intimidating and others who can't get enough power... I was wondering though, suppose I make it 400 at 4 ohms but like the 1200, low impedance capable (so you get 600~700W into 2 ohms with a realistic PSU). Would that be a compromise? Power hungry folks just add woofers :)

If you have to choose only one power class, that seems to be a nice choice!
 
I thought we had solved that. More power than the UcD400 (ideally at NC1200 levels), premium components (where it matters) and the best discrete input you can design (aka world class). :p :D

TheShaman talks to my heart ;)
Being the owner of a dual mono UcD400HG - HxR for four years,
there are not much more to ask for than these!

I can't wait to be shocked again by the performance...

A quick question:
Is there in your plans a matching linear or regulated SMPS power supply?
 
I'm still collecting votes on the power issue though. There's some who think 400W is already intimidating and others who can't get enough power... I was wondering though, suppose I make it 400 at 4 ohms but like the 1200, low impedance capable (so you get 600~700W into 2 ohms with a realistic PSU). Would that be a compromise? Power hungry folks just add woofers :)

A better compromise would, perhaps, be a 300@8/600@4 module.
I'm a bit "greedy" on this subject because of my past experience.
This is going to be a statement amplifier so it's very likely it'll be used on statement loudspeakers. I've heard UcD amps in various setups and whenever we had such a speaker, like a big B&W (and I'm not talking 800 but even 802N), a big Focal or a big ATC or even a good old Infinity, the UcD400s would start to "sweat" as soon as we turned the volume higher than normal (see: over 100dBs). 700s had no such issues, of course. Most people and esp. those who have big speakers want to hear LOUD every now and then - and even if they don't, they want to know they can and not that their amp is not up to the task. ;)
And it's not like ClassD is really attractive to the high-sensitivity (horn etc.) loudspeaker crowd, to begin with.

I assume the existing UcD line of products will still be available for purchase so someone with lower-end speakers could opt for the soon-to-be-updated (half-way between the existing UcD and Ncore, if I get it right) lower powered UcDs. [Not that I believe there is such a thing as "too much power", but I can understand the "too expensive" term]

Just my 2c of course.
 
Last edited:
As mentioned earlier as current design considerations I would also opt for easy access of standard industrial connectors that are easily obtainable across the globe.

Also mentioned I second that the mounting on heatsink could be much better than for the current UcD range. The UcD OEM range looks much better in that aspect IMHO.

Perhaps the On/Off control could be handled via a jumper on the module itself in case one does not want to use it for thump free operation in order to avoid using any cables for that?

Furthermore I can perhaps represent a small group of people in Sweden and request that the modules should be both low impedance capable but even more important they should be low frequency capable. On another forum we have serious problems finding amplifiers with a low enough bottom extension that are stable at low frequencies. For instance the ICE-Power 1000ASP fails and goes into weird clipping/cut-out behaviour when driven hard at low frequencies.

Regards
UrSv, who still misses the two single digit serial UcD180STs gone MIA on return to Hypex
 
Last edited:
@TheShaman
Actually I've always felt that the effect of UcD400's "sweating" had more to do with current than with voltage, which is why I set so much stall on 2 ohm performance. My hunch there would be that a 600W/2 ohm module would be preferable to 600W/4. Something to ponder. The UcD product range remains in place.

@UrSv: Yes let's open the mounting can. For EMC reasons I prefer plate mounting like the NC1200 and the current UcD OEM range. The only thing this can't answer is the question what to do if you want to fasten screws from the inside out. So:
@JMBulg: If we were to decide on the plate construction, I see two options. One is that the user makes a second plate that's bigger than the module (e.g. 1cm around) onto which the module mounts normally (screws through the bottom) and then that whole construction is bolted into the chassis from the module side. The other option would be a kind of mounting brackets to clamp the module down. Do you think you could live with either? Other suggestions welcome too btw.

@SP502, SMPS, but not regulated. The amplifier itself is a pretty capable regulator in its own right...
 
@UrSv: Yes let's open the mounting can. For EMC reasons I prefer plate mounting like the NC1200 and the current UcD OEM range. The only thing this can't answer is the question what to do if you want to fasten screws from the inside out.

A larger plate than module is fine with me. It really doesn't have to be much larger as 10 mm would probably be fine on two edges.

Would through PCB mounting be bad? I mean four holes that you reach through the PCB itself?
 
@JMBulg: If we were to decide on the plate construction, I see two options. One is that the user makes a second plate that's bigger than the module (e.g. 1cm around) onto which the module mounts normally (screws through the bottom) and then that whole construction is bolted into the chassis from the module side.
This is actually what I did and it is feasible. It demands however more mechanical work, not necessarily within reach for occasional DIY, and makes a possible source of problems (two contact surfaces instead of one for the hear transfer).

Suggestion: make your heatsink 10mm larger on two sides only (probably longer since the module is probably shorter than the SMPS. Making it higher is more susceptible to add further constraints on case heights) and make the adequate holes there.
 
Status
Not open for further replies.