Play time: NAP140 with SMPS

[Mike Hanson]

Long ago I upgraded my NAP140 to use Avondale NCC200 boards and their corresponding power supply. That left me with a pair of NAP140 boards doing nothing. I've also got a spare Pesante case from Modushop. After watching @laverda play with the Mean Well switchmode power supplies, I thought I would jump into the water. This is my first time trying to use an SMPS in a build, so I'm really not sure what to expect.

First of all, I wasn't sure that the NAP140 boards would still work, as they've been bouncing around my office, standing in for NCC200, VBE, HCR200, etc. in various layout plans. Therefore, I didn't want to buy big supplies, if it turned out they would be unnecessary. Consequently, I got two inexpensive 36V 50W units (a.k.a. LRS-50-36), so I could do a proof of concept. 100W is likely insufficient for two NAP110 boards used in anger. However, if this test worked, then I could add two more and go dual-mono, so each channel would have 100W.

I dialed them up to 40V, and bridged them to create 40V-0-40V. (I believe the original NAP110 supply is closer to 42V-0-42V, but I didn't want to push the little units too far.)

Here's the current state of affairs:

[URL='https://avzine.com/images/nap140-smps/a.jpg'][/URL]

All the voltages measure fine, and the speaker output DC offset voltage is <= 20mv for each channel. I'm waiting for some banana connectors, so I can't actually try to play music through them yet.

I did notice something interesting, though. When the SMPS units are not connected to the boards, then they are dead silent. However, when I hook them up to the two boards, they both emit a piercing whine almost like a dentist's waterpick. I've got countless switchmode supplies in my house, and none of them make noise like this.

I'm wondering if this means they are struggling to power the boards. There is no input or output connected at this time, so this should be a quiescent state which I assume requires very little power.

Is my eventual plan to power them with four 50W SMPS modules misguided? I have enough room to put two larger MW units in there, but not four.

Your thoughts?

 

[Mike Hanson]

I just did a quick test, in case one or the other of the NAP boards was causing the trouble. I disconnected each of the boards in turn, and the whine/squeal was approximately half as loud with only one. Then I disconnected both, and there is actually a small buzz (not a squeal like when the boards are connected).

I also remembered that I have a pair of LRS-200-48 units, destined for another project. I might be able to dial these back to 42V for the NAP140 boards. I would have to rejig the loom significantly to test it, though, so I don't want to take that step until I'm nominally confident that it will help.

 

[MJS]

+ and - 38V is more typical of unregulated Naim amps.
The squealing could be the boards oscillating. Check the earths carefully. SMPS under load also squeal depending on the switching frequency/quality of design etc.
You really need a scope on the amp output to see what it's doing.

 

[Mike Hanson]

It's definitely the power supplies, not the NAP boards.

 

[martin clark]

+1 to that; the smps without a load might be OK; but usually each such thing these days include a 'y' cap for EMC reasons from output +vdc to ..one of the input mains lines.
As a result - it is not a given they will be happy with outputs connected in series.

LF audible squealing to me suggest there's an issue with each attempting to start-up; actually knowing more than that might well require a 'scope, and a willingness to work within live-to-Mains equipment.

Bit of a minefield...!

First things first - given target amp board draw is around 38mA at idle, give or take, from a 40v supply, give or take:, lets approximate that with a purely -passive impedance of 1Kohm: - do your SMPS still howl if you use two 1K/3w resistors in series across the -40/0/+40v rails of your supplies, in place of the amp boards in your proposed set-up..?


ETA: is there a minimum recommended load current for your smps? They might also be squealing because they are not loaded sufficiently at the amp boards idle current/from not enough current drawn, and have dropped into a discontinuous mode. Perfectly normal for much efficent smps design these days (e.g its why the chord mojo internally hisses a bit when nearly-fully-charged)

 

[Mike Hanson]

It's not really "LF", at least not in an audio sense. Here's a little video demonstrating the irritating whine:

 

[Mike Hanson]

lets approximate that with a purely -passive impedance of 1Kohm: - do your SMPS still howl if you use two 1K/3w resistors in series across the -40/0/+40v rails of your supplies, in place of the amp boards in your proposed set-up..?

I haven't got any suitable resistors on-hand to do that test, so I've just ordered some. They should arrive in the next few days.

 

[Mynamemynaim]

What about connecting two supplies to one board ....to double the current supply...and see if they still squeal ?

 

[Mike Hanson]

What about connecting two supplies to one board ....to double the current supply...and see if they still squeal ?

I tried that already:

I just did a quick test, in case one or the other of the NAP boards was causing the trouble. I disconnected each of the boards in turn, and the whine/squeal was approximately half as loud with only one. Then I disconnected both, and there is actually a small buzz (not a squeal like when the boards are connected).

 

[laverda]

Something’s ‘a miss’ there Mike I am still running my NCC300 monos off two Meanwell SMPS (totally silent but they run warm) and listening to them as I type this.
I’ve also used the Hypex SMPS600n400 which is a dual rail +-65v on the NCC300 mono, A better unit altogether IMO.

double check grounding/earthing. I’m sure you have...... many times...

Don’t you just love DIY

 

[Mike Hanson]

First things first - given target amp board draw is around 38mA at idle, give or take, from a 40v supply, give or take:, lets approximate that with a purely -passive impedance of 1Kohm: - do your SMPS still howl if you use two 1K/3w resistors in series across the -40/0/+40v rails of your supplies, in place of the amp boards in your proposed set-up..?

The resistors arrived, and I just did this experiment. I hooked a 1K resistor between ground and +40V, and another between -40V and ground. There was no squeal, so the SMPS units appear to be fine.

With that eliminated, what's my next step? The ground is typically the culprit. I'm using a lever nut terminal block as the 0V between the two 40V supplies. It's intended to host 8 connections to ground:

• From 0V on one SMPS to 40V on the other (resulting in 40V-0-40V).
• Two grounds each to the NAP140 boards
• Two speaker negative outputs

Here's @laverda's video (start @ 3:19):

You'll see that he has the power supplies on top of each other, with a solid core wire running directly between the two units. All the other grounds are soldered to that. Is my different configuration likely to result in this noise?

I do eventually plan to add another two SMPS units, so that each amp has its own pair. Therefore, I could rig them up more like laverda's. Note that when I hook up a single board it still squeals, so the the only real difference I would be testing is the shorter direct link.

 

[martinab2]

I'm wondering if the problem is how the SMPS outputs are referenced to mains earth internally. They may not like working in series as Martin Clark pointed out earlier. The way to find out would be to disconnect the mains earth and completely isolate the casing of the lower SMPS unit (the one you are using to make the -40v supply) so that it is floating electrically. Don't touch the unearthed SMPS casing while you are doing this and make sure it can't come into contact with any other components or casework.

 

[martin clark]

The resistors arrived, and I just did this experiment. I hooked a 1K resistor between ground and +40V, and another between -40V and ground. There was no squeal, so the SMPS units appear to be fine.

WIth the resistors still connected on your +/0/- supply, connect to the amp - so that the smps has to suply the idle current load and the added 40ma draw on each rail from the resistor load. Does that stop or diminish or changes the noise ..?

 

[Mike Hanson]

I'm wondering if the problem is how the SMPS outputs are referenced to mains earth internally. They may not like working in series as Martin Clark pointed out earlier. The way to find out would be to disconnect the mains earth and completely isolate the casing of the lower SMPS unit (the one you are using to make the -40v supply) so that it is floating electrically. Don't touch the unearthed SMPS casing while you are doing this and make sure it can't come into contact with any other components or casework.

Given laverda uses the same brand of supplies (albeit a different volts and power) in this configuration, I'm assuming it's possible to do it this way.

I've also confirmed that there isn't continuity between the earth input and either output. Finally, if it were connected when running, then I assume it would have gone BZZT! when I hooked it all up.

 

[martin clark]

Mike - sorry perhaps I havent looked closely enough - presumably the smps perforated metal cases are earthed/bonded to ground at the mains wiring connections.
Is it possible you are getting some ground loop pickup - do you get the noise if the smps cases are isolated from the metal chassis in your video..?

Just an afterthought

 

[martinab2]

Finally, if it were connected when running, then I assume it would have gone BZZT! when I hooked it all up.

Not necessarily. The problem may have something to do with how the current/voltage sensing on the SMPS is configured so may be forcing the supplies into some sort of fault protection mode when connected which could account for the strange noises. Maybe it's nothing to do with the mains earth but it's an easy one to eliminate from the list of potential suspects.

 

[Mike Hanson]

WIth the resistors still connected on your +/0/- supply, connect to the amp - so that the smps has to suply the idle current load and the added 40ma draw on each rail from the resistor load. Does that stop or diminish or changes the noise ..?

The noise is the same.

 

[Mike Hanson]

Mike - sorry perhaps I havent looked closely enough - presumably the smps perforated metal cases are earthed/bonded to ground at the mains wiring connections.
Is it possible you are getting some ground loop pickup - do you get the noise if the smps cases are isolated from the metal chassis in your video..?

Just an afterthought

The case of the SMPS is definitely connected to the unit's ground input, so mains ground, chassis ground, and SMPS input ground are all the same (a.k.a. "earth"). That's completely separate from the "ground" that the load sees, which is a connection between +V of one SMPS and -V of the other.

 

[laverda]

Shite photo but its the only one I have of my first experiment with SMPS's powering NCC220's. I used 200/48v units and run them at 50v or so. It worked very well for the low cost at the time (£18 each) and I just cobbled them together just to see if the smps's would complain. They didn't so on I went with the NCC300 build using them. I recall I mailed Meanwell at the time to see if they advised a bit of capacitance between the supply and the amp. They replied none should be used as the smps might keep trying to shut down while charging the caps. Eventuality they would boot up but not not advised.

They run complete silent but do generate some heat so need to be in ventilated case IMO. The resultant sound was OK albeit if a bit light in weight in the lower registers as I recall, the 300's fixed that.

 

[Mike Hanson]

Shite photo but its the only one I have of my first experiment with SMPS's powering NCC220's. I used 200/48v units and run them at 50v or so. It worked very well for the low cost at the time (£18 each) and I just cobbled them together just to see if the smps's would complain. They didn't so on I went with the NCC300 build using them. I recall I mailed Meanwell at the time to see if they advised a bit of capacitance between the supply and the amp. They replied none should be used as the smps might keep trying to shut down while charging the caps. Eventuality they would boot up but not not advised.

They run complete silent but do generate some heat so need to be in ventilated case IMO. The resultant sound was OK albeit if a bit light in weight in the lower registers as I recall, the 300's fixed that.

Thanks for the confirmation that I'm on the right track. I notice that with this prototype you're not using the chassis ground input, but you do in your video, so I'm disregarding that discrepancy.

I suppose it's possible there's a difference with these little units that I'm using compared to the bigger ones. Mine are just 50W, as I figured two together would generate a maximum of 100WPC, which should be enough to drive the NAP140 boards. Perhaps I should be using 75W or 100W instead. Of course, at this point I'm dealing only with quiescent current, so 100W is definitely enough.

Perhaps I'll try connecting a NCC200 board instead, just to see if the behavior is different. I suspect it will be the same, thought, given both NAP140 boards are doing the same thing, and I can't imagine that they're both defective.