New P482 Stereo Power Amplifier

Alan March

Alan March

Founder of March Audio
Staff member
Hi everyone

Exciting times at March Towers. The new Purifi 1ET6525 modules have finally arrived. A special thanks to Fedex for losing them for 4 weeks!!!

Anyway, we have been busy testing them in our new P482 amplifier. This replaces the P422 in a few weeks. The new amp has a revised power supply and features our "ultra buffer" that was developed for our flagship P801 monoblocks. The new buffer is required to take advantage of the improved performance these latest generation Purifi modules.

We are confident our engineering skills have made this is the lowest noise, lowest distortion and highest power output implementation of the Purifi 1Et6525 modules.

20241113_131134mod.png



Headline specs (actual measured performance *NOT* Purifi module datasheet numbers):

Measurements taken with 21.0dB gain setting and BOTH channels driven.

Power:
246 Watts 8 ohm 1% THD
490 Watts 4 ohm 1% THD
540 Watts 2 ohm 1% THD

Output Noise:
7.6uV RMS (20Hz to 20kHz, input shorted)
DC Offset = 0.3074mV

THD:
-133dB (20Hz to 20kHz)

THD+N (SINAD)
-115dB @ 4.475V RMS, 4 ohm load, 5 watts

Gain Options (selectable on rear panel switch)
21.0dB (full power output with 4.0V RMS standard domestic XLR input/source)
27.0dB (full power output with 2.0V RMS standard RCA input/source)

Be wary of other manufacturers "gaming"" the SINAD figures with unusably low gain settings.


As you possibly know, we are very close to the measurement limits where it becomes difficult to discern between the noise and distortion of the measurement system and that of the device under test. Below is some of the data.

As discussed previously regarding our P801 amplifier, due to their very low noise it is very difficult to accurately measure the SINAD. The gain of the input buffer and power amp module amplifies the inherent noise of the signal generator/DAC. Most amplifiers noise masks and swamps the noise of the signal source, but with these amplifiers it needs to be taken into account.

Below I have shown the amplifiers inherent noise, measured with the input shorted. I have also shown the amp output noise with the signal source/generator connected:

Input shorted = -115.45dB
With signal generator connected = -112.45dB

Due to this the measured THD+N is lower than it should be due to the signal generator noise. Hence to calculate the actual THD+N / SINAD we need to add the measured THD and the amp noise measured with the input shorted:

(-134.3dB) + (115.34dB) = -115.3dB

As you can see, the final SINAD is dictated by the noise and not the THD.


Graphs

Output Noise with input shorted

7.6uV

1731464110754.png




Noise, Input shorted dBFS.

FS = 4.6565V rms = FS = -115.8dB

REF 4.475V RMS (5 watts 4 ohms) = -115.45dB

1731464110776.png




Noise with signal generator connected

REF 4.475V RMS = -112.45dB

1731464110788.png




THD / THD+N

REF 4.475V RMS (5 watts 4 ohms) = -111.9dB

CORRECTED = -115.3dB


1731464110805.png



Power curves to follow.

Needless to say, it sounds incredible :) !
 
Hey Alan,

I thought it was time to join the community, it's Bobby here.

Looking good, anxiously waiting for this release :)

I had noticed the Gain stage was initially noted as 23dB & 29dB but now corrected to 21dB & 27dB, was that inadvertently copied from the P801's Ultra Buffer specifications?
 
Oniiz86 said:
Hey Alan,

I thought it was time to join the community, it's Bobby here.

Looking good, anxiously waiting for this release :)

I had noticed the Gain stage was initially noted as 23dB & 29dB but now corrected to 21dB & 27dB, was that inadvertently copied from the P801's Ultra Buffer specifications?
Click to expand...
Welcome!

Yep, my bad. Brain fart, I was thinking of the P801 amps.
 
Hello Alan
I have a question if that's OK? I have been discussing your test results on Audio Science Review. One of your competitors has posted a plot showing their equivalent to the P482 having more noise, distortion and lower power. He says the difference in performance isn't statistically significant and seems to be saying the Purifi modules vary significantly in power 210 to 230 watts. Have you found this?
Thanks
 
@james dyson I believe the difference regarding power output is because most competitors are using Hypex's SMPS1200A400 which is only 63V so cannot achieve full rated output, I believe BoXem did touch upon this but Alan's custom SMPS solution is closer to the 1ET6525SA module's rated voltage of 65V, somewhere between 64V-65V & it is the only one touted to have Active Synchronous Rectification. ;)
 
Last edited:
james dyson said:
Hello Alan
I have a question if that's OK? I have been discussing your test results on Audio Science Review. One of your competitors has posted a plot showing their equivalent to the P482 having more noise, distortion and lower power. He says the difference in performance isn't statistically significant and seems to be saying the Purifi modules vary significantly in power 210 to 230 watts. Have you found this?
Thanks
Click to expand...
Hi James

I tend not to pay any attention to ASR, and I certainly don't want to get into an argument with another manufacturer. However, I will say that this doesn't correlate with our experience.

We have built thousands of Purifi based amps and each one is tested at production. This means we have a large database of performance results. Whilst it's true to say the modules are not identical in performance, the differences are small, very small.

Ok, we havent tested a large number of 6525 modules yet, but I see no reason why the 6525 module would be any different to the other modules. To assert that the module is varying 10% in power output is pretty out there and wrong in my experience.

@Oniiz86 hit the nail on the head. If this manufacturer is using a Hypex power supply then it is too low a voltage. 65volts is required, the Hypex is only 63volts at the nominal mains supply voltage of 230v (110v). It has very poor voltage regulation and will wander all over the place depending on the incoming mains voltage. The Hypex spec sheet says 49 volts to 72 volts! 49volts will significantly limit the power output. 72 volts actually goes above the modules recommended supply range. Bus pumping with reactive speaker loads could easily make the voltage rise to 75v which would trip its protection mechanism shutting it down. Alternatively, when heavily loaded this psu will drop in output voltage reducing power output.

So, 210 watts is about what I would expect at the nominal 63 volts. It's the psu causing the power output variations, not the module.

Our power supplies have very tight voltage regulation WRT the incoming mains voltage and output load, hence we dont see these problems. The amp performance remains consistent regardless.

Hope that helps.

Alan
 
Last edited:
@james dyson @Oniiz86 The question about power supply voltage prompted me to perform a bit more testing. I have set-up the PSU for 63volts and ran power curves on the 1ET6525 modules.

8ohm 1% THD+N = 203 watts
8 ohm 63v psu 1%.png

8ohm 0.1% THD+N = 165watts
8 ohm 63v psu 0.1%.png


4ohm 1% THD+N=405watts
4 ohm 63v psu 1%.png

4ohm 0.1%THD+N = 311 watts
4 ohm 63v psu 0.1%.png


So my estimate of 210 watts at 1% THD+N into 8 ohms was about right. So thats all you will get out of a Hypex PSU running at nominal mains input voltage (230v/110v). This is just one reason why we dont use them and have our own custom PSU.
 
Last edited:
@Alan March A curious observation I noticed of when the Award-Winning P422 was reviewed by Australian Hifi, is that it also achieved 246 watts at 8 Ohms with the Hypex SMPS1200A400, perhaps the difference of 230V/240V plays a part in the final output power too, I believe this review took place before the transition to the custom PSU in late 2021.
P422 Review.png
P422 Review Newport Test Labs.png
 
Oniiz86 said:
@Alan March A curious observation I noticed of when the Award-Winning P422 was reviewed by Australian Hifi, is that it also achieved 246 watts at 8 Ohms with the Hypex SMPS1200A400, perhaps the difference of 230V/240V plays a part in the final output power too, I believe this review took place before the transition to the custom PSU in late 2021.
View attachment 187
View attachment 190
Click to expand...

Yes, the P452 did utilise a Hypex PSU. Note the mains voltage during the test - up to 253 volts! As mentioned, the Hypex supply varies its output voltage proportionally with the incoming mains voltage because its unregulated. So its a bit of a lottery depending on the mains voltage at your house as to whether the amp will reach its power output spec or not.

Taking into account voltage drop due to load, you would need over 240v (120v) for the Hypex to output enough voltage to make the 1et6525 reach full power output.
 
Alan March said:
Yes, the P452 did utilise a Hypex PSU. Note the mains voltage during the test - up to 253 volts! As mentioned, the Hypex supply varies its output voltage proportionally with the incoming mains voltage because its unregulated. So its a bit of a lottery depending on the mains voltage at your house as to whether the amp will reach its power output spec or not.

Taking into account voltage drop due to load, you would need over 240v (120v) for the Hypex to output enough voltage to make the 1et6525 reach full power output.
Click to expand...
Thanks for that Alan, silly me for not picking up on that, I also didn't realise that Australia is officially a 230V country from way back in 2000, I thought we were 240V, I believe WA is still 240V & Queensland only recently transitioned to 230V back in 2020, I need to educate myself with Australia's current electrical power standards. ;)
 
237v here in WA as I write. Remember there is a tolerance its allowed to operate within. For example, whilst the UK is formally 230v and was supposed to drop to there from 240v to harmonise with Europe, it never did because 240v is still within the allowed tolerance.
 
@Alan March It appears that Purifi are almost ready to list the 1ET6525SA officially on their website, it appears the 8/4 Ohm power rating specs have been slightly revised to 250W 8 Ohm/450W 4 Ohm from 230W 8 Ohm/480W 4 Ohm, perhaps you can verify with them to see it may in fact be a misprint as to avoid another potential model number snafu ;)
Purifi 1ET6525SA OEM List.png
 
Oniiz86 said:
@Alan March It appears that Purifi are almost ready to list the 1ET6525SA officially on their website, it appears the 8/4 Ohm power rating specs have been slightly revised to 250W 8 Ohm/450W 4 Ohm from 230W 8 Ohm/480W 4 Ohm, perhaps you can verify with them to see it may in fact be a misprint as to avoid another potential model number snafu ;)
View attachment 192
Click to expand...

Hi

We are happy with the data from our tests. Remember, as discussed above, power output does depend upon the behaviour of the power supply. Plus of course the limiting behaviour of the module protection system.

A production batch of modules have been shipped form Purifi, so sales can go live with the P482 (and new P282) next week.
 
Last edited:
Hey Alan,

It appears Purifi now explicitly mention to utilise a "Thermal Interface Material" (Thermal Grease or Pad) in the 1ET6525SA datasheet to optimise heat dissipation, this isn't something previously documented for their other modules like the predecessor 1ET400A, I thought the aluminium enclosure was wholly sufficient for Purifi modules & the new 1ET6525SA has Idle Losses of only 1.6W which makes it the coolest & most efficient on the market :)
1ET6525SA Thermal Requirements.png
 
Our enclosure provides perfectly adequate cooling. We can run full power sine waves which you should remember are not representative of real world use. Music has a high crest factor which leads to much lower RMS power levels than sine waves and therefore much lower heat dissipation.

However, some manufacturers have smaller less heat dissipative cases which may require careful heat management and optimisation.
 
Hey Alan
On ASR there was a question regarding the validity of your corrected SINAD figure. The poster said it wasn't valid because the noise floor increases with increasing power output. Therefore just taking the shorted input noise figure doesn't account for the increased noise at higher power output. Whats your response to that?
 
You must log in or register to reply here.
Back
Top