I just asked them, since Bruno Putzeys mentions a power supply as one of the many things in the pipeline in one of the interviews. They did not commit to any release date, but Kim Nordtorp Madsen told me that it is going to have regulation on the main rails and be roughly the same size as the Hypex SMPS1200 series. Since Kim Nordtorp Madsen advised me to not get too close to 70V when I was choosing a power supply, and pointed at the 54V Connex SMPS800RE as a good choice, I infer that the regulation on their future SMPS is going to be quite sophisticated.
This occurred...
This occurred during very moderate load testing, but it does not happen at the lowest levels. For my loudspeakers i need all (or nearly) all the power available from the Purifi modules, so I need the full rail voltage. Also, even if the one lowers the voltage, if it is jumping around all over the place, thta kind of defeats the purpose of the "regulated" supply?
With your efficient speakers of course this may not be a problem with your set up, have you monitored the Connex rail voltage during high level music playback, I am just curious?
I am a little surprised that Audiophonics had this problem with the Hypex supply though? As i have seen full power tests of the Purifi modules, powered by the Hypex supply where there was no actuation of the protection circuits whatsoever? The ASR website did full power testing with the Purifi demo amp, and never once had any issues with the protection circuits kicking in...so I have to conclude that the Hypex supply did not have the same over voltage problem?
This occurred during very moderate load testing, but it does not happen at the lowest levels. For my loudspeakers i need all (or nearly) all the power available from the Purifi modules, so I need the full rail voltage. Also, even if the one lowers the voltage, if it is jumping around all over the place, thta kind of defeats the purpose of the "regulated" supply?
With your efficient speakers of course this may not be a problem with your set up, have you monitored the Connex rail voltage during high level music playback, I am just curious?
I am a little surprised that Audiophonics had this problem with the Hypex supply though? As i have seen full power tests of the Purifi modules, powered by the Hypex supply where there was no actuation of the protection circuits whatsoever? The ASR website did full power testing with the Purifi demo amp, and never once had any issues with the protection circuits kicking in...so I have to conclude that the Hypex supply did not have the same over voltage problem?
I absolutely agree. However, tweakers will tweak. I'll need to take a look at the layout, but I don't think adding the remote sensing option to the PCB will cost me anything up front (aside from two more resistors). The support cost could be significant, however.
If you choose to use that feature, you really have to be careful with the wiring. It would be an option for advanced builders only. The only documentation of this feature would be "see the Hypex/Purifi data sheet for details". Basically, if you have to ask how to use the feature, then please don't use the feature. For that reason alone, I might also choose to omit that feature - or to include it but not mention or support it. I need to think about this.
I prefer to set my customers up for success. Providing them with a path to destroy their speakers is not exactly congruent with this philosophy.
Well, yes and no. It is to be expected in a power supply where the VCC and VEE are coupled by a transformer, but only one of them is regulated. You could build a regulated supply where both VCC and VEE were regulated. Such a supply would not suffer from supply pumping.
Tom
Of course. This is what an on-board 1 ohm link is meant to eliminate. You could have a thin trace bypassing it for the default on-board feedback, which an 'advanced' builder must cut when desired. Plus a through hole to facilitate wiring the external feedback. But the 1 ohm R remains as insurance against any ill effects from potential poor external loop connection. This is already better than what is provided on the Purifi FE02B board. On my EVAL1 builds with panel mounted connector feedback (again, like demonstrated by Purifi), I am adding 1R at R8&9_A1&2 for just such insurance. Those tiny SMDs are a pain for me. Again, just my thoughts.
Only in the sense where both rails are regulated in a manner such that they can both, both, source and sink current either to each other or back to the input supply. Otherwise the pump is still on.
Driving channels on the same supply out of phase is common practice and goes some way to ameliorate supply pumping. The assumption is that low frequency music content, the biggest contributor to pumping, is in phase on both channels. Pumping involves the transfer of power from one rail to the other. Inverting the phase on one channel causes it to pump in the other direction assuming both channels are reproducing the same signal. Having inverted the phase of that channel you have to connect its loudspeaker in reverse to restore the phase. Bridging two half bridges completely prevents pumping since the amplifier is behaving as a full bridge where the issue does not arise. Beyond that you might consider taking an active route the simplest of which would be to run a dummy Class D amplifier to regulate the rails against relative variations using current mode control and driving into a zero ohm load. Conceptual piccy...
...
...
Attachments
Connexion SMPS... I see a pattern now actually. I have had incidents. I blew not less than 3 units of "digital amp" using a SMPS300 with the output close to max voltage for the amps. I have used one 800R for a another amp, also quite close to the max. All these 4 amplifiers has gone south within just a few days/weeks.
These are just observations... but maybe these PSs has some trouble keeping thier voltage in check.
It might be that they don't sag under load which is really attractive but maybe they "peak" in a unfortunate way during certain circumstances!?
??
//
These are just observations... but maybe these PSs has some trouble keeping thier voltage in check.
It might be that they don't sag under load which is really attractive but maybe they "peak" in a unfortunate way during certain circumstances!?
??
//
It is not uncommon to see single ended conversion where the control loop is happy to regulate against output transients within a specified range, 20% load to 80% load but go off on random walks when you start hitting limits, full load down to zero. The output voltage can suffer from serious overshoots. For a basic off-line convertor there is no reverse path. Ideally you would generate a bulk DC on the secondary side and then post regulate with a synchronous buck convertor which will regenerate, boost back, to the bulk supply. It's all added complexity though.
Connex SMPS 800 re
Here is exactly what happened here with the SMPS 800 re:
I was playing a brown noise track which i use for break in and testing, it has a lot of energy in the lower midrange and upper bass (excersizes woofer cones very well). The nominal voltage of the supply is +/- 65 volts, and the specification for regulation is < +/- 3 V. At very low levels, everything was fine, as I advanced the volume to no more than medium levels (below normal listening levels) the amp modules started cutting out, with the indicator LEDs going out with the cut outs. I monitored the negative rail and it was bouncing all over the place very quickly. I saw peak values well below -70 V, and this was tripping the protection of the amp.
Connex has suggested adding an additional 4700 µF to the rails to stabilize it. I will give that a try, very carefully, but I have also ordered a Hypex supply to be safe.
I did not encounter the problem playing music signals, but I was only testing, playing at very low volumes as the Purifi modules were not mounted up to any case/heat sink for this testing. And I did not play any music during testing with considerable lower midrange upper bass energy.
The Connex supply has some pluses, just by heat output alone it is clearly more efficient than the Hypex ones (I am very familiar with both SMPS 600 and 1200 from Hypex), and the idea of voltage regulation is attractive in theory (if it really can be made stable) as this should result in better dynamic response. As to noise, they make claims, but provide no specs, and neither does Hypex.
Here is exactly what happened here with the SMPS 800 re:
I was playing a brown noise track which i use for break in and testing, it has a lot of energy in the lower midrange and upper bass (excersizes woofer cones very well). The nominal voltage of the supply is +/- 65 volts, and the specification for regulation is < +/- 3 V. At very low levels, everything was fine, as I advanced the volume to no more than medium levels (below normal listening levels) the amp modules started cutting out, with the indicator LEDs going out with the cut outs. I monitored the negative rail and it was bouncing all over the place very quickly. I saw peak values well below -70 V, and this was tripping the protection of the amp.
Connex has suggested adding an additional 4700 µF to the rails to stabilize it. I will give that a try, very carefully, but I have also ordered a Hypex supply to be safe.
I did not encounter the problem playing music signals, but I was only testing, playing at very low volumes as the Purifi modules were not mounted up to any case/heat sink for this testing. And I did not play any music during testing with considerable lower midrange upper bass energy.
The Connex supply has some pluses, just by heat output alone it is clearly more efficient than the Hypex ones (I am very familiar with both SMPS 600 and 1200 from Hypex), and the idea of voltage regulation is attractive in theory (if it really can be made stable) as this should result in better dynamic response. As to noise, they make claims, but provide no specs, and neither does Hypex.
Last edited:
Oh that’s cool. I can easily do that. Invert a channel and then the corresponding speaker output? This reminds me of something perhaps.
Review and Measurements of Nord One NC500 Amp | Audio Science Review (ASR) Forum
So why is one of the two outputs wired opposite here? My theory is that the Nord folks heard they had to do that for single supply amps, and then they just carried over. However, Amir M. did not check whether also the inputs were inverted. This is actually fun :-D
Just FYI, Purifi confirms using a 1 ohm resistor to connect the sense feedback "is quite OK". Also, as would be expected, there is no need to fight with tiny SMD parts. One can use a leaded resistor from the PCB feedback hole to one of the three output holes.
Amazing Tom! You have a winner! I have a feeling these are going to sell like hotcakes.
I hope so. The performance will be good and I think I have settled on a good feature set.
The assembly quote is in. It'll be a bit more expensive than I thought, mostly because of all the connectors. They're inserted by hand, and at Western wages that adds up.
My current thoughts on the pricing is $149/each (QTY 1) with discounts at QTY = 2, 3+, 5+, and 10+. I should be able to hit $99/each for QTY = 10+.
That said, I think the performance and the opportunities to tweak (or not, if that isn't your thing) will be well worth it. The raw PCB is manufactured in Canada - just outside of Toronto. It's fully electrically tested before it's shipped to me. I buy the parts at Mouser and Digikey and deliver them personally with the boards to my assembly house in Calgary. I use good materials, good subcontractors, and genuine parts. Made in Canada!
Tom
Yes... It does look rather suspicious. Perhaps Amir missed the trick. Bear in mind that the technique is only applicable to a stereo pair of half bridge amplifiers operating from the same bipolar, +ve, GND, -ve, supply and you have to invert both inputs and outputs on just one channel. Not sure how easy that is with single ended inputs but for balanced it should be a simple thing to do.
Single supply bridged amplifiers do not suffer from rail pumping. One half of the bridge pumps in one direction whilst the other half of the bridge pumps in the opposite direction. Since they are driving the same bridged load with the same signal, one side inverted by definition, from the same supply the pumping effect cancels naturally.
Last edited:
Regarding the evaluation amps Purifi circulated using the EVAL1, I don't think so. The black output sockets of both channels go to OUT- on their respective 1ET400A boards, and are in common with ground. So, the inputs could not be inverting on one of the two channels, unless they were distributing systems with out-of-phase channels. Also, if the inputs were configured that way on the FE02B board, they should have mentioned it on the datasheet, as it would be an uncommon feature that they would want to document (and claim).
Kim, from Purifi has just shared this with us:
"First the 1ET400A02B have supply monitors that will shut down in case the supply voltage gets too high – just as the case you link to – so no worries the amp is protected.
The Hypex SMPS1200A400 is made specially for single ended amps like the NC500 or the 1ET400A02B – and there will be no pumping – as it does not use rectifiers in the supply but instead have mosfets that can sink current.
In case with the Connex or a transformer/rectifier/capacitor supply you need to connect the channels in reverse phase, and/or have plenty of capacitance…
So with e.g. a 4700uF it should not be an issue when playing music."