Allright then...
First, there have been reported QC issue with the specific PSU in this very thread (which I read from first page).
Second, I do not trust any SMPTS not specifically made for audio to perform flawlessly wrt pollution spread on the mains from which I power the rest of my audio equipment (which has a dedicated power line). And even those...
Noise introduced into the mains. Levels as low as permitted by law are not good enough for audio.
Telstar, FWIW can we assume the SMPS86 sold by Tom for use with his lower powered models (i.e. 24V rails) is “designed” specifically for audio? Note that it uses MeanWell modules, as well as soft-start and output inductors, so is admittedly not just a stock LED lighting supply.
I’ve had enough public and private conversations with him on the subject to get the sense that he’s a good enough engineer to not recommend something that would be a major compromise.
As with his recent decision to move to professional assembly of at least the surface mount components on his semi-populated boards, I think he’s looking to preserve the reputation of his products’ performance.
I’ve had enough public and private conversations with him on the subject to get the sense that he’s a good enough engineer to not recommend something that would be a major compromise.
As with his recent decision to move to professional assembly of at least the surface mount components on his semi-populated boards, I think he’s looking to preserve the reputation of his products’ performance.
I have dealt with noise from switching power and find that noise level is not the effecting factor. As a matter of fact, linear supply transformers have much more influence in sound than switching power. You would be lucky if you can select the best transformer for a specific design. I look at some Mod builds that use an R core transformer and wonder how it sounds, because when I tried one out on an NCD amp built, it totally compressed the sound whereas on O core from the same manufacturer and of the same rating was quite nice.
So the question would really be, what are the spec differences that really matter in audio. I know from experience that noise is not the dominant factor.
Really? I don't recall us discussing any quality control (QC) issue in this thread. I'm not aware of any QC issue on any of the Mean Well supplies. Would you provide a link to the post or any credible source for this, please? I'm sure Mean Well would be very interested to know as well.
Unless I blanked and missed that post, it sounds like FUD (Fear, Uncertainty, and Doubt).
Ah. So it's a trust issue. I am curious what you base this claim on.
Noise (both emitted and conducted) is an issue in many applications. That is in part why we now have medical grade switchers. The medical grade switchers, such as the Mean Well RPS-400 series offer better isolation between primary and secondary (not killing the patient is important) and is also held to a much stricter standard regarding emissions and conduction of noise (not disturbing the measurements is important). You can see the Mean Well data sheet for the specific standards.
In the medical world, signals of very low amplitude are measured routinely. EEG signals (brain waves) are in the low uV range (gamma waves), for example, and we don't have any issues measuring those (as long as we can get the electrodes to make good contact with the scalp of the patient anyway). EEG signals are around or below the noise floor of common audio gear - including good gear - and we're still able to resolve these signals with precision ... using gear powered by SMPSes.
There is nothing that makes the various natural laws work differently for audio. Ohm's law doesn't care if the resistance is in an audio circuit or in an RF circuit. I've designed RF circuits for use in applications requiring -160 dBc or better noise floors. The switching supplies worked fine there too... What makes audio so special? I suspect this is where the psychology comes in, but I'll leave that as a topic for another thread.
What do you base this on? Facts, please.
I think the bottom line is that you've decided that you don't like SMPSes (for whatever reason). That's fine. I have no plans of forcing anyone to use a power supply they don't like. What would I possibly gain from that?!
Don't like SMPSes? Don't use one! Don't like what I offer? Great! There are other options and you seem clearly capable of designing your own power supply. More power to you! (No pun intended).
I will continue to make recommendations, which are grounded in science, measurements, and solid engineering. That's how I operate. This also means that my future designs will be rooted in science, measurements, and solid engineering. How you feel about that is really up to you.
Tom
I dunno... Using common commercial practice of 1VA per W and 10.000uF per rail per 100W, we get to a 800VA 2x25V transformer and 160.000uF worth of cap for a stereo amp (we could use less than that for 8r loads).
An Antek 8425 (which is actually good for 960VA on 60hz) is 85$. A 22.000uF/50V 85° cap from a decent brand is about 6$ at mouser, so 48$ (it gets up to about 64$ using 105° caps). Add a pair of strong bridges, connectors and some copper and it looks pretty much like a tie against a pair of se-600-36. A softstart for that big transformer might tip the financial balance in favor of the smps by a few dollars.
The smps still looks better for most loads and conditions but the linear one will handle more gracefully overload. The se-600 shuts down and restarts at 105-125% overload. Which is not impossible with a stereo build on 4r speakers.
Last edited:
I guess Parasound doesn't follow the "commercial practice" then. Their A23 has 2x10000 uF per channel and 1 kVA shared between channels. The transformer is plenty, but the 2x10000 uF per channel is considerably below your recommendation as the amp is rated for 125 W into 8 Ω and 225 W into 4 Ω.
As I measured the other day and reported here, the difference between 2x22000 uF and 4x22000 uF is a few watt (0.04 dB at 300 W) with 4 Ω load for the MOD686. There's no measurable difference in performance aside from that 0.04 dB higher power at clipping due to the lower ripple voltage with the larger caps.
Agree 100 %.
That depends on the amount of cap on the MOD686 and the bandwidth of the current limiter in the SMPS.
Tom
Last edited:
The cynical side of me would suggest that parasound picked a 1kVA transformer for the added weight and bragging rights in a strange marketplace. Their 1000VA transformer is what they need to continuously play sinewaves onto 4r at their rated continuous power. Which is rather pointless but well..
More seriously, I actually picked very generous values to quickly see what a "suitable" ps would cost. 1VA per W is certainly more than I've found in most amps I've opened and 10.000uF/100W a whole lot more (I shouldn't have called that last bit commercial practice that's true). Since the mod686 behaves that well on increased ripple, it just reinforces my point that a linear supply isn't actually more expensive than smps.
More seriously, I actually picked very generous values to quickly see what a "suitable" ps would cost. 1VA per W is certainly more than I've found in most amps I've opened and 10.000uF/100W a whole lot more (I shouldn't have called that last bit commercial practice that's true). Since the mod686 behaves that well on increased ripple, it just reinforces my point that a linear supply isn't actually more expensive than smps.
This is precisely why I didn't want to talk about SPMTS here.
And I wont.
BTW, someone experienced issues with the MainWell psu, I'm not dreaming about it, but I have no time nor desire to dig for it. the mere presence of a FAN would bar me from using it in the same room where resides my audio setup.
Yes, toroidal, R-core, EI-core... they all sound different and there is no scientific explanation that I know of, besides some measured magnetic fields.
I know more about digital and there noise is king.
I think everyone is free to try whatever they want once they get their own stuff, but I think the recommended switching power does not have a fan. I specifically checked. I think switching power will give more consistent experience sonically.
With single end circuits, there are some additional consideration with return current. The linear supplies take care of it to some degree, and switching does need something similar as well, just plugging the unit in is not the best approach, which is probably you have problems with it.
With single end circuits, there are some additional consideration with return current. The linear supplies take care of it to some degree, and switching does need something similar as well, just plugging the unit in is not the best approach, which is probably you have problems with it.
Absolutely! Cooking your own power supply and whatnot is all part of DIY. I make recommendations based on the best of my knowledge at the time of the recommendation, but it's not like I travel the world to inspect that every amp has been completed according to my recommendations. That would be kinda cool to do though. Especially if I could get someone else to provide the travel budget. I'd love to meet my builders in person...
SMPS definitely provides the most consistent performance and is more plug-n-play than a typical unregulated supply. That said, with the Modulus-series of amps, the unregulated supply performs just as well up to clipping. Both are valid solutions.
Yeah... That's why I normally recommend using the differential inputs.
Regarding supply capacitance: Duly noted. I'll definitely consider adding more capacitance. Maybe 4x22000 uF. Even if science says it isn't necessary for the Modulus amps, market perceptions do factor into the equation as well. I "just" need to make sure the soft start circuit can handle the additional energy storage of the caps.
Tom
Last edited:
Tom, regarding the RPS-400 series. It looks like they are single voltage. Don't the Modulus amps require dual voltage? Or am I interpreting this wrong?
Also, what is your opinion on Connex's SMPS offerings (if you have one)?
You're interpreting correctly. You'll need two RPS-400-36 in series to form ±36 V.
I only have experience with the SMPS300RE. It's okay. I generally prefer to buy products which have an actual data sheet or at least detailed list of specs, which the Connex supplies don't.
Also note that the Connex supplies are specified in "music power". I.e. their "300 W" supply is suitable for a 300 W Class D amp playing music with a 10 dB CF, so it's really only a 100 W supply. I found this out the hard way when I load tested the SMPS300RE I purchased.
Turns out if you draw 250 W from the "300 W" SMPS300RE, it'll overheat and fry the two secondary diodes. Easy fix, but annoying with the "music power" spec.That said, the SMPS300RE works great as long as you stay within its limits. I'm not aware of a Connex product that'd be a good fit for the MOD686. Maybe the SMPS800RE with a custom ±36 V output voltage. Even that's pretty marginal (800/3 = 266 W).
Hypex specs their supplies in "music power" as well. Apparently, that's a Class D thing. None of the Hypex supplies is suitable for the MOD686 as they exceed the ABS MAX ±42 V of the LM3886 at high mains.
Tom
Follow-up question. How do you set one of the supplies as the "-36 V" side? Does the -V terminal of the two supplies get joined together? I looked at the datasheet for the RPS-400-36 but didn't see anything.