Hypex UcD180HG HxR or 400HG HxR ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2008
I've performed some listeing tests of the Hypex and ICE 125ASX2, level- and frequencyresponse-matched within 0.2 dB of course, anything else would be unserious.

My general impression is that the difference is not all that big, but the ICE is generally a bit more laid-back and "warm" compared to the Hypex, while the Hypex is a bit more transparent.

What you like better depends on your priorities, but I would chose the Hypex.

The next thing up now is to order an other pair of UcD's and an SMPS, and build two stereo amps to drive my mids and tweeters.
 
Given I know so little about such things, I assume to power the six drivers in my new active speakers I would need six modules? There are a couple for sale in the Swap Meet, which I am looking at buying.

Why are you moving away from the Icepower? If you have any to sell that you wont be using please let me know:)

David
 
Last edited:
Disabled Account
Joined 2008
If you have a stereo 3-way active system, you will need six amp modules, and enough power supply capacity of course. I would think that a Hypex SMPS400A180 supply should be good enough for three amp modules at least.

I am moving away from ICE because I like the sound of Hypex better, its more to my tastes in sound. The downside is that Hypex is far more expensive, but sometimes you just have to pay for improvements.... ;)

No plans for any ICEpower sale yet. Most likely I will sell the finish amps I have, not just the modules (and B&O ICEpower dont allow that I sell the modules either).
 
This thing is interesting because it may shed some light on one reason why amps sound different.
I find it surprising such comparative measurements aren't more common. Haven't had occasion to do a lot of them myself, but 2dB is typical of the differences I've found even between pretty good amps. I've also found audible differences between amps even when SPLs are matched and nonlinear distortions, noise, and changes in driver loading are negligible. How do multitone measurements of the two amps compare?

A tone structure of f_{n} = 1.88 f_{n-1} produces the greatest correlation between objective and subjective measures of performance in the DS and Rnonlin metrics. But personally I've found it more useful to measure the impulse response directly---it's a good approximation of transients in music, has thorough spectral coverage just like music, it's easy to make time domain comparisons, and I've found direct measurement of the IR shows differences that correlate well with subjective audio quality even when the IRs reconstructed from swept sine measurements are identical. Square waves are also interesting. I've gotten reasonable correlation between square wave differences and subjective quality there too. It is a bit more involved to look at a bandlimited square wave on an oscope or RTA and reason about its quality than an IR, though.

Hypex doesn't publish phase data but the ICEpower modules are characterized and exhibit significant phase rotation by 20kHz. I don't know the high frequency extent of your hearing, but if the 180HGs are flatter phase it's possible some of the difference you're picking up on is harmonic misalignment on the ICEpowers.
 
Disabled Account
Joined 2008
New pair of Hypex amps finished at last! (lots of work...) Think I'll need a break now, at least until next year..... ;)

There are two UcD400 and two UcD180 in each box, each pair with its corresponding 180 and 400 W SMPS. I have two of these boxes placed just behind the speakers. The 180's power the tweeter and upper mids, the two 400's power two pairs of low-mids/woofer in my 4-way dipole system.

Sound is really really good - transparent, smooth and detailed.

I'm a bit surprised by how much heat the modules produce, even on idle. Good thing my boxes have heatsinks on the sides.

An externally hosted image should be here but it was not working when we last tested it.
 
New pair of Hypex amps finished at last! (lots of work...) Think I'll need a break now, at least until next year..... ;)

There are two UcD400 and two UcD180 in each box, each pair with its corresponding 180 and 400 W SMPS. I have two of these boxes placed just behind the speakers. The 180's power the tweeter and upper mids, the two 400's power two pairs of low-mids/woofer in my 4-way dipole system.

Sound is really really good - transparent, smooth and detailed.

I'm a bit surprised by how much heat the modules produce, even on idle. Good thing my boxes have heatsinks on the sides.

An externally hosted image should be here but it was not working when we last tested it.

Well done looks good.

You would think at idle they would hardly use anything.

Looking forward to your next adventure.

Happy New Year

I went for the Coldamp version in the end, as I couldn't find an affordable supply for the Hypex. I am now impatiently waiting for them to arrive.

David
 
Good reading here, thanks for all the infos and sharing your latest project, Stig Erik.

Anybody knowing if any of the Hypex SMPS can be used with the two outputs floating - meaning, if the common mass of a power supply module can be split ?

With linear PSU's there's considerable sonic improvement if you take the effort to keep mass to PSU separate for each rail and even if you keep mass to transformer and caps separate - though "technical explanation" for that IMO is a mixed bag at best.

Stig Erik, as far as I remember your ICE was driven from linear PSU ? - I'd be very much interested in your findings with them attached to the Hypex SMPS - if thats any possible -

- and you already did have your "break" of course :D

As an aside, I've been pretty happy with my ADAMS Monitors some time back which also used Hypex modules and (a tiny) AMT tweeter - which now - also - seems to become available for the DIYer branded as Expolinear OHT-25 (EUR 76.-/ each, according to a German DIY magazine)

Michael
Happy New Year to everybody
 
Last edited:
Disabled Account
Joined 2008
How do you split "mass" (ground) for each PSU rail? They have to be connected to each other at some point... :confused::confused:
If you have a schematic of this split it would be nice!

The ICE amps I used were all self-powered, they had their own on-board switched PSU. For the record - I'm allergic to big power transformers and capacitors..... :cool:
 
My wording was not the best - what I meant is how to wire PSU mass points with respect to star mass point.

the upper part in the pix is the "common wiring topology" (daisy chaining) for linear PSU's
the lower topology in the pix is what gives me consistently better results

psu_topology.png



Of course I do not use one simple diode, but you get the picture I guess...


Michael
 
Last edited:
Disabled Account
Joined 2008
Ah, I see ... star grounding. You could do that with the Hypex SMPS if you use the Ground pin at the SMPS as "star", but I'm not sure if it matters since there is only one ground connection at the UcD module, and the speaker ground shall NOT be connected at power supply ground, but rather at a designated speaker negative terminal.
 
Disabled Account
Joined 2008
A little extra tip for those of you who consider building such an amplifier: dont forget ventilation and cooling. You MUST mount the UcD modules on heat-sinks. Hypex says this clearly in the UcD datasheets, but dont forget the power supply... the SMPS dont need an extra heat-sink, but be sure to have enough air-flow around it. My boxes have some ventilation holes, but they are not providing enough airflow obviously, since the 180W SMPS started to shut down. I hope this problem is solved after drilling some holes in the bottom plate and off-setting the top cover a little bit so there's a gap between the cover and the front plate for improved air-flow. I also fitted some 2 cm tall feet to lift them off the floor, and improve flow through the bottom vents.
 
Ah, I see ... star grounding. You could do that with the Hypex SMPS if you use the Ground pin at the SMPS as "star", but I'm not sure if it matters since there is only one ground connection at the UcD module, and the speaker ground shall NOT be connected at power supply ground, but rather at a designated speaker negative terminal.

Yeah, to use the SMPS mass as star ground would certainly be contra indicative.
:)

So its only left to use two separate modules - too expensive for me, considering it would only be an experiment in combination with an A/B amp that already has a sophisticated PSU topology...

I wonder why audio circuit designers do not pay more attention to such subtle details - no time for any listening ?? - no sensible audio environment ?? - no feedback from users ?? - too expensive ?? - hey, its only a few extra traces and connection pins to provide !

I'd even be happy *if* the layout would allow for a DIY mod.


Michael
 
With linear PSU's there's considerable sonic improvement if you take the effort to keep mass to PSU separate for each rail and even if you keep mass to transformer and caps separate - though "technical explanation" for that IMO is a mixed bag at best.
Which explanation(s)? After all, there's a bunch of different factors in play. Generally what I've found within the amp proper is amps lack sufficient PSRR+bypassing to avoid running into trouble with the ripple from their quiescent current draw at typical output powers. Layout exacerbates that if the references aren't well placed or there are crosstalk issues between nets. But it's my experience the average amp has bigger problems with lack of CMRR due to unbalanced interconnects and poor SnR due to higher than necessary gain.
 
You could ask Hypex support about this, they usually respond very fast to emails.

support@hypex.nl

Thanks - thats an option.
;)

On the other hand - I had the feeling Jan-Peter might even be *exactly* the right person to discuss such topics with :D ?


Which explanation(s)? After all, there's a bunch of different factors in play. .

:D

With linear PSU's there's considerable sonic improvement if you take the effort to keep mass to PSU separate for each rail and even if you keep mass to transformer and caps separate - though "technical explanation" for that IMO is a mixed bag at best.

Generally what I've found within the amp proper is amps lack sufficient PSRR+bypassing to avoid running into trouble with the ripple from their quiescent current draw at typical output powers. Layout exacerbates that if the references aren't well placed or there are crosstalk issues between nets. But it's my experience the average amp has bigger problems with lack of CMRR due to unbalanced interconnects and poor SnR due to higher than necessary gain.

- The interconnect issue certainly is a whole topic of its own - I'd leave that aside here (assuming its taken care of already).
- PSRR is another issue as well - I certainly would like to see load and line regulation plots of the Hypex SMPS - but again thats also an issue at its own I'd leave aside here (assuming its taken care of already).
- Layout in terms of inductive and capacitative crosstalk is another issue as well - but again thats also an issue at its own I'd leave aside here (assuming its taken care of already).

Whats left is simple resistive coupling due to charging current peaks.
Though its easy to simulate and to make up your mind - it does not correlate to sonics IMO.
One other aspect is output impedance of PSU - split or daisy chained - but same here - given the additional wire lengths needed, it also does not correlate to sonics IMO.

So might be I'm either totally off - or my perspective on electric aspects is tooo limited...
:)
Possibly you might have to do the mod on one of your amps (with linear PSU) and give it a listen. Many transformers provide the two wires if the insulation is cut back slightly - easy to split the two secondaries this way.

Michael
 
Last edited:
Whats left is simple resistive coupling due to charging current peaks.
So we're focused narrowly on connection of the secondaries here. When I've looked at this optimum placement of star center isn't entirely clear but it seems preferable to put it on the amp board to maximize the extent to which transformer and wiring non-idealities result in "float" on the secondaries rather than ground bounce within the amp. What I've done in my layouts is to pour a ground plane under the power inlet and bypass caps and bring out individual channel grounds from a common service point on the plane. That way any deviation from optimal star centering is low impedance and should have minimal consequences.

The op amps I use for control have 135+dB CMRR and 140+dB PSRR at 100-120Hz. They're also differential input with the output reference being the connector pin to the speaker. So I'd be surprised if I could measure a difference modding my amps between the two wiring topologies. If I end up with a sacrificial trafo I'll fish around with the oscope and see if I can find differences in power plane behavior.

It's been a few months since I did any searches in this direction but at the time I didn't find good characterizations of transformer behavior during charging peaks. And I haven't done measurements myself. My suspicions tend to fall more on parasitic inductances and inter-winding capacitances than resistive coupling, though the increase in a winding's effective source impedance as the core moves towards saturation might be significant.
 
So we're focused narrowly on connection of the secondaries here.

As from my side : yes !

What I haven't mentioned yet but actually is my "favorite" in this bunch of interactions is cross talk between positive and negative rails when current is drawn at load or at charging.

In any case - if you have not split the PSU mass connect to the star point, there is always a cross talk to the other rail, which is not there if the mass is split - meaning - with typical discrete power amps and their rather lowish PSRR this is what seems at least to be *one* valid explanation.


When I've looked at this optimum placement of star center isn't entirely clear but it seems preferable to put it on the amp board to maximize the extent to which transformer and wiring non-idealities result in "float" on the secondaries rather than ground bounce within the amp.

basically the placement of the star point does not matter.
In practice - one would choose a location where all that many mass points can be put together - with a discrete design, if done extensively you end up with several dozens of wired you have to bring together at a certain *point*.



What I've done in my layouts is to pour a ground plane under the power inlet and bypass caps and bring out individual channel grounds from a common service point on the plane. That way any deviation from optimal star centering is low impedance and should have minimal consequences.

So what you do is actually kind a RF layout with caps additionally (?) shorting to a ground plane ?


The op amps I use for control have 135+dB CMRR and 140+dB PSRR at 100-120Hz. They're also differential input with the output reference being the connector pin to the speaker. So I'd be surprised if I could measure a difference modding my amps between the two wiring topologies. If I end up with a sacrificial trafo I'll fish around with the oscope and see if I can find differences in power plane behavior.

You rather will have no luck with measuring IMO - and due to this high PSRR the sonic effect may be less - wold be interesting for me if you still can detect anything...


It's been a few months since I did any searches in this direction but at the time I didn't find good characterizations of transformer behavior during charging peaks. And I haven't done measurements myself. My suspicions tend to fall more on parasitic inductances and inter-winding capacitances than resistive coupling, though the increase in a winding's effective source impedance as the core moves towards saturation might be significant.

There is a nice tool for transformer simulation at the Duncan page, the PSU Designer II :

PSUD2

And of course there is LT-Spice, PSpice, ... where you can jump into transient analysis and model as much as you like (and as much data you have from your transformer).

Quite a while back I did this with PLITRON output transformers, when I was more interested in tube amps:

Ausgangsuebertrager Innenwiderstand, Frequenzgang und Phasengang bei verschiedenen Beschaltungen verdeutlicht anhand eines PAT4070-CFB
(no transient analysis shown and in German too)


Michael
I guess I'll quit with my off topic rants here :)
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.