Dual Hypex UcD180 amplifier build

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The psu, as told me by Hypex, doesn’t need to be grounded to the chassis because their psu (SMPS1200A700) is “class 2 safety” product (they told me that the electrical contact between the chassis and the metal plate underneath the psu is enough) and so also the amp modules. [...]

So I would advice you to try to go this way first, then if you don’t obtain a satisfying result, try otherwise. But I think you will raech you target, cause I think Hypex products are very reliable.

Thanks Gaetano, this is reassuring and I will no doubt be pleased in the end: they are a very reputable manufacturer after all. The SMPS400Axx0 datasheet (rev 14) – my PSU – states that:
When mounting the module in an enclosure, a minimum safety distance of 6mm from the SMPS to all possible conducting parts must be ensured to keep compliance with Safety Class 2. This includes parts on the top, the bottom and all sides of the board. [...]
If the enclosure is limited in height
[...] one shall need to use smaller spacers and provide a layer of insulation both above and below the SMPS with a minimum thickness of 0.4 mm to comply with the Class 2 Safety directive.
So if I understand correctly, this “class 2 safety products don't need to be grounded” thing is only re safety & shunt to earth ground (green wire). On the other hand, the chassis-PSU contact made by the metalic spacer specified on the UcD180+SMPS400A180 schema likely implements the PSU ground to chassis connection recommended in AES48.

Stephen Macatee's excellent Rane Note 151 explains that:
Many audio manufacturers, consciously or unconsciously, connect balanced shields to audio signal ground [...]. Any currents induced into the shield modulate the ground where the shield is terminated. This also modulates the signal referenced to that ground. [...] Amazingly enough, acceptable performance in some systems is achievable, further providing confidence for the manufacturer to continue this improper practice. [...]
Signal-grounded shields on balanced equipment create ground loops in the audio path and modulate the audio ground, wrecking havoc in most systems.
[...] Again, some systems with signal-grounded shields work acceptably, causing further and future bewilderment.

My concern (here) is not so much safety but rather hum/buzz prevention, and as part of a whole sound system where most of the equipment is indeed balanced.

Gaetano, I could be wrong but so far I believe the wiring scheme that you implemented introduces such a loop in the audio path. (More on this below.) It might not cause problems in your particular setup but may in mine. Again, Hypex themselves advocate the implementation of AES48:

[Hypex's Legacy pin 1 problem application note] strongly advocates the implementation of AES48... This is in contradiction with their recommended wiring scheme (as of schema 1 & schema 2) and confused me further.

But actually I'm not so sure that there is a plain contradiction here anymore, rather possibly an oversight. Rane Note 151 further explains that:
It is easy to confuse chassis ground and signal ground since they are usually connected together, [...] let's discuss how to connect them. [...]
The shield noise currents flow through the chassis and shunt to earth ground on units with 3-conductor line cords. The key issue is that these noise currents do not flow through a path shared by any audio currents. [...] The hard part is implementing the proper layout scheme.
Connecting signal ground to the chassis can only be done in
one place in each unit. If done twice, one leaves the possibility open that the noise currents will flow through a path shared by audio.
There are two schools of thought on where to connect the signal ground to the chassis. They are both versions of the star ground scheme [...]. The first connects a trace (or wire) directly from the audio power supply ground terminal and connects to the chassis ground point (see Figure 4). It is important, in both "schools", that no other signal currents be allowed to flow through this trace. Do not allow this trace to share any other return currents from other signal-grounded circuit points, such as the input or output circuit's ground. This keeps chassis noise currents from flowing through the same trace which is a return path for an audio signal. [...] Another school of thought on where to connect signal ground to the chassis simply moves the center of the star ground to the input jack's ground.
Both wiring schematics (links above) look like Hypex advise to connect the star ground center at the audio input connector (second school of though). However as I said above, I believe that the chassis-PSU electrical contact made by the metallic spacer specified on the UcD180+SMPS400 wiring schema implements a PSU ground to chassis connection and thus also connects the audio ground to the chassis, via the PSU board.
But as Macatee explains, connecting signal ground to the chassis can only be done in one place in each unit[; if] done twice, [it] leaves the possibility open that the noise currents will flow through a path shared by audio and cause equipment compatibility issues within a whole audio system.

Is anybody able to confirm if I'm right or wrong, and why?
 
It has been 1½month since I emailed Hypex but they have not got back to me.

I don't actually believe that the Amplifier signal input wiring Application Note contradicts AES48 per-se anymore. Assuming that the PSU ground is only connected to the chassis via the UcD board, the system should indeed be ground loop free. However that scheme does not mention the metallic spacer specified on the Wiring UcD180HG SMPS400A180 schematics. Moreover, the Wiring UcD180HG SMPS400A180 schematics only shows a 2-conductor mains.


I just pinged them again, specifically asking to (1) clarify the purpose of the metallic spacer which makes a contact between the PSU board and the chassis, and (2) confirm where to tie the earth shunt of a 3-conductor line cord to the chassis in order to respect AES-48.
 
It's taken another 1½ month but I think I have made sense of wiring. First - Hypex were quite quick to get back to me after the second email mentioned in my previous message:
  • The metal spacer is needed to improve EMI performance. (I could not get any more info on that)
  • The chassis is earthed near the mains inlet.
I have also decided how I want to wire my integrated dual UcD180 system together. Without getting down to the nitty-gritty, it looks like Hypex's scheme can be extended to two UcD180 PCBs within a single chassis without problems indeed. Details were discussed here, but in short Bruno Putzeys' G Word paper on grounding practices in audio electronics clarified the following things for me:

  1. He states again that At some point [pin 1 problems] became so prevalent that the AES had to enshrine the obvious into a standard. Called AES48, it patiently explains that the shield should be connected to the chassis via the shortest possible route and that connections between the PCB ground and the chassis should be made elsewhere.
    Although this "elsewhere" is actually clearly labeled and wired as a star point on the schematics of AES48-2005, Putzeys' paper highlights the flaws of an internal star ground scheme.
  2. He presents an on-PCB differential signal wiring technique whereby audio circuitry has no audio ground as such and is insensitive to pin 1 problems. The UcD180 was designed by the paper's aurthor so I think it is sensible to assume that he implemented this technique.
  3. He also voices the opinion that the biggest overlooked opportunity for differential signalling is inside the box where small signals and large currents slug it out in cramped quarters. So I presume that the manufacturer's advice to use a shielded twisted pair between the audio input connector and the UcD180 is to provide the balanced & shielded connection inside the box that the paper advocates.

The above would explain why fairly liberal internal wiring schemes still provide good results with Hypex amplifiers. I reckon it also explains why there is no need to galvanicaly isolate the UcD180's heatshrink from the chassis. I have extended the official SMPS + UcD wiring schematic to reflect how I intend to wire my amplifier.
 

Attachments

  • wiring.png
    wiring.png
    529 KB · Views: 1,719
I just read the following info in the Appendix II: Connection guidelines to the NC400 datasheet rev 4 (page 16). It further clarifies what I wrote in my previous message re star ground schemes in class D amps as well as differential signal wiring:

9.2 Input configuration
[...] Star grounding only works at low frequencies. The longer the wires to the “star”, the lower the frequency at which supply decoupling becomes ineffective. The maximum permissible trace length of decoupling capacitors in class D amplifiers is measured in millimetres and the value of the decoupling capacitances is large. To make a long story short: you cannot connect the audio signal using a single wire and hope to solve the ensuing drama using star grounding. It won’t work.

The solution adopted by the NC400 is a floating input, consisting of a “hot” and a “cold” connection, which can be connected to the source like the primary of an isolation transformer. This breaks any current flow that might otherwise flow in the audio cable. You would never directly connect the cold pin of the input connector to the ground pin in the same way as you would never locally ground input of an isolation transformer because it would no longer provide isolation.
Now that I am set on a wiring scheme, I'm looking at implementing it physically. The above document also gives generally applicable wiring guidelines (some of which concurs with advice & pictures previously posted in this thread, e.g. Fox510's, zman01's and ghitus') such as:

9.1 Cable dressing
The NC400 module has exceedingly low distortion. This makes it very easy for extraneous causes to add much more distortion and colouration than the amplifier itself. The first major cause of such distortion is direct magnetic crosstalk from the supply cable into the audio input or the loudspeaker output. This is minimized in several ways:
  • Run the audio and power supply cables away from each other.
  • Tie-wrap the supply cable to form a tight bunch.
  • Tightly twist all loudspeaker cabling inside the chassis.
At this stage I reckon that layout option C seems like the simplest and most sensible layout.
 
Last edited:
I've had them out a couple of times in a setup not entirely dissimilar to yours :)
attachment.php



It sounded like they should sound great, although there was a very noticeable distortion. For now, I want to think this was due to routing high power speaker signal through the breadboard. I reckon it'll all be good once it's wired properly.
 

Attachments

  • Screenshot_2018-08-04_10-05-04.png
    Screenshot_2018-08-04_10-05-04.png
    109.7 KB · Views: 1,764
I guess I just got lucky, I have the 'V5' SMPS without the EMI (100nF to GND) spacer/connection.

Nothing on the Hypex is connected to mains earth, my pre-amp is though..

Pre-amp end is RCA with cold & screen connected to RCA housing, cable terminated directly at the HG180 plug hot/screen/cold.

...or I should say 'was terminated', I've just changed from Hypex to Anaview modules (in proper boxes this time...)

Plan is to build boxes for the Hypex's and keep them as spares, I'll probably face your dilemma once that happens.:D
 
there was a very noticeable distortion
I have the 'V5' SMPS without the EMI (100nF to GND) spacer/connection.
Indeed this distortion could be a consequence of not filtering some switching noise aliasing down into the audible spectrum? Don't know. Like I said, I would be surprised if it was still there once the modules are wired & grounded properly.


Plan is to build boxes for the Hypex's and keep them as spares, I'll probably face your dilemma once that happens.:D
My questions are due to integrating two modules in a single box but if you have two power supplies, you shouldn't need to :)
 
To be complete and in case someone else wants to know how well this worked -- I have enjoyed an 2-channel, single PSU Hypex UcD180 wired as discussed above for a couple of months now (see attached image1). No problem, sounds very quiet (as some posters predicted). What's more, it sounds very nice.

1 yes the speaker outputs were not wired yet
 

Attachments

  • IMG_0413 - Copy.jpg
    IMG_0413 - Copy.jpg
    891.2 KB · Views: 449
Thanks zman01.


Source/preamp: so far laptops with realtek I/O peripherals -- nothing HiFi
loudspeakers: EJJordan Eikona (2nd generation) housed in their Reflex 7 bookshelf cabinet design.


(They are my first full-range drivers and I am a total convert. Eliminating crossovers resulted in the best dynamic response I have ever heard by far. Consequently they are a little dim in the treble, but as far as I am concerned the dip above 10K only makes them more comfortable to listen to. Superb amp/drivers combo IMO.)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.