Buckeye Purifi Eigentakt 1ET9040BA monoblock power amplifier Stereophile Measurements

[Hiten]

Ok, I studied briefly working of SMPS power supply. It can provide enough current when required by adjusting the voltage. So it seems power supply is not at fault in this recent test.

 

[mocenigo]

What’s your evidence for that claim?

The data sheets of the boards and of various buffers. Most single ended opamps have noise and distortion levels which are higher than the most recent Purifi and Hypex modules, and if you take into account the gains of the latter, they all have too much noise and distortion. If a board has (warning, random number coming) say, 104Db SINAD, an opamp 110, and the gain of the module is 13Db, then you end up with 97Db SINAD. You need at leat 117 (actually 118) from the opamp.

A THD+N <0.00006% @ 100W, 4Ω, 1kHz (as in the 1ET9040BA) means a SINAD of 124DB (but you can do the same game with bust SNR or THD, or IMD). This board has a gain of 14.4Db. Hence, you get improvements with better buffers until a bit after 138.8Db in the buffer. Of course these are only measurable and not audible, but they are there. You can get values like that, especially with distortion (noise is a different matter) only with ad-hoc designs — usually composite amplifiers.

This is one of the reasons I got the Neurochrome Universal Buffer with my 1ET400A build (bypassing the Purifi adapter board's onboard buffer).

Of course this is overengineering, even with my 97.5Db/2.8V/m speakers I have no audible artifacts, and I had no audible artifacts also with my previous 108Db horns from the listening position. With the latter, I could hear some hiss by putting my head INSIDE the horn (LOL) using with the Purifi adapter board (single ended OPA1612), but no such hiss with the Neurochrome Universal Buffer.

But if the goal is to extract the maximum from the available tech, well, then overengineering it is, even if from a listening standpoint there is no difference. For me, the only reason to move to the 6225 or 9040 would be bragging rights, and in that case I do not want the buffer to be the bottleneck.

 

[mocenigo]

That's fair. I have looked at their reports, gear they use, licensing they have to operate under major EU/US companies to run their testing and I see no reason to be worried at all. As I mentioned, they are good enough to satisfy US/EU regulator requirements so can't ask for more.

Sorry I lost track. Purifi, Microaudio, or both?

 

[goryu]

Let me try again...

Nothing has changed with the discussion on my end...others alluded that the 9040 should have a non-limiting SMPS due to its high-end nature and I merely reminded that we offer such an SMPS as an upgrade.

@Buckeye Amps
Can you elaborate or point me to any comments you have made about thermal testing the 9040 modules with the Micro smps assembled in the case? I would assume you have done thermal testing to verify temps are within a safe zone when driven hard. Thanks.

 

[Buckeye Amps]

Let me try again...


@Buckeye Amps
Can you elaborate or point me to any comments you have made about thermal testing the 9040 modules with the Micro smps assembled in the case? I would assume you have done thermal testing to verify temps are within a safe zone when driven hard. Thanks.

Apologies for missing this.
If I had no talked about this preciously in the official owners thread that is my oversight.

Yes thermal testing has been done. Both by me and even more especially by Micro Audio. Sami stress tests his SMPS' to a level of "insane" torture.

 

[mocenigo]

Nothing has changed with the discussion on my end...others alluded that the 9040 should have a non-limiting SMPS due to its high-end nature and I merely reminded that we offer such an SMPS as an upgrade.

Some people do not realise that the difference between, say, 1400 and 1200 W is 1.3Db. And that would be only on peaks that, with most speakers and in most environments, would rattle the windows, if not shatter them. Not to speak that the SMPS1200A180 can probably delives up to 1600W for a very brief time anyway. So, this criticism is twofold wrong. The Microaudio SMPS is for those (like me!) that like overengineering.

I did not care, however, that in my 1ET400A build, using a Connex 600RE (at 54V) instead of the SMPS1200A400, I may have lost 2.1Db on the peaks. And have to thank Kim Nordtorp Madsen of Purifi that assisted me in making the choice. This amp is now 4 years and 8 months up and running...

 

[Sokel]

The only thing I'll add as a sort of rebuttal to a comment regarding Eastern standards/practices a few posts above:

Money plays just as much, if not more, of a role in industry as the political/societal structure of a producing country where lack of accountability may be called into question. Which is why Western standards/production/safety is no less prone to failures or lack of oversight as money can buy a lot of leeway when it comes to oversight/accountability.

So even as an American, I know better than to start throwing rocks...

Just don't try it in Denmark.
In a similar occasion about power claims the guy got busted when tried to bride the expert that was about to testify.

About the above,is not so much about countries,is more about a company's complex owning status and financing.Imagine for example the same company that owns a amps company also owns a cert lab.Ir been financed by the same entity who wants combined results.
That can't do in EU.Thus the trust*.

*Edit:MY trust,I don't talk for others.

 

[amirm]

Indeed. That is why I prefer European products rather than Chinese, Russian or US products.

Labs don't produce a product. They are agents approved by western governments to conduct safety and emissions compliance local to where products are produced as to increase rate of compliance. No one in US ships their product to China to get it tested. They pay the higher expense locally.

Further, the work requires technical skills and is not one performed by underpaid farm worker in an assembly line.

Net, net, I don't know what nationalistic has to do with lab testing. Your priority should be that it is lab tested. Not where the lab is. Insisting that an overseas product always be sent to western countries for testing makes no sense.

 

[al2002]

Anyone know whether the 7040 modules also have thie IMD problem?

 

[Buckeye Amps]

Anyone know whether the 7040 modules also have thie IMD problem?

They do not

 

[fatoldgit]

There is a comment in the review comment section about the Purifi modules being out of spec and they will be updated so eliminate the issue.

Peter

 

[pma]

Following my post from yesterday, I have measured maximum power of the 1ET400A, with SMPS400A180 power supply (2x46V), in the 20Hz - 20kHz range, for THD and THD+N <= 1% and at least 10s duration at each measured frequency. This is the result:

The datasheet plot for the maximum power vs. supply voltage is this:

In my opinion, the module meets this specs with 8ohm load. Please note much higher maximum power above 10kHz, and it is measured with wide 90kHz bandwidth, without additional LPF filter, so not restricted to usual 20kHz.

4ohm load was not measured, as I do not intend to exceed SMPS maximum output current (10A), which would pollute the maximum power measurement.

I assume that the 1ET9040BA module would meet the power specs as well, if properly supplied from the PSU. However, please let us not forget a big difference - 1ET9040BA is a bridged topology, though 1ET400A is single ended. 1ET9040BA BTL outputs thus see half of the load impedance (2x higher current), as is the rule for the BTL topology. So, the independent check would be welcome as well.
------------

One more measurement of maximum power of the 1ET400A, with SMPS400A180 power supply (2x46V), in the 20Hz - 20kHz range, for THD and THD+N <= 1% and at least 10s duration at each measured frequency. Now with 4.7 ohm lad (I do not have a 4 ohm load configuration at the moment). I suspect that this measurement into 4.7 ohm load is influenced by SMPS400A180 output current/voltage capability, but please note again a big rise of power at high frequencies.

 

[Lars Risbo]

the HF rise is probably due to the bandwidth limitation from both the LC filter and analyzer: harmonics are suppressed and the amp can be driven more into clipping before hitting 1%

 

[pma]

the HF rise is probably due to the bandwidth limitation from both the LC filter and analyzer: harmonics are suppressed and the amp can be driven more into clipping before hitting 1%

That might be so, in case you speak about LC filter in the Purifi module. I did not use a LPF filter before the analyser. And yes, measuring BW=90kHz allows only for 4th harmonic at 20kHz.

 

[pma]

the HF rise is probably due to the bandwidth limitation from both the LC filter and analyzer: harmonics are suppressed and the amp can be driven more into clipping before hitting 1%

That might be so, in case you speak about LC filter in the Purifi module. I did not use a LPF filter before the analyser. And yes, measuring BW=90kHz allows only for 4th harmonic at 20kHz.

..... - or it is probably more complicated than the measuring BW.

 

[RickS]

Buckeye offered brass connectors as the fix. They were the only change between the initial Buckeye Purifi and the follow up corrected unit measured on ASR. The results were also verified by Rick Sykora's testing. Obviously, SpeakOn offers even better connections for sure and the original Buckeye Purifi with SpeakOn parts never had the issue.

Sorry, just saw this as was down with the flu...

Correct, your statement about previous root cause testing is accurate. Was working with Buckeye at the time and we comprehensively tested other potential solutions (tightening, soldering, etc.) before the designer found the ferrous content of the binding post tab. A summary of that investigation was posted here. The "faulty" tab is part of the binding post and was determined to have ferrous content. So fortunately, ended up a simple fix.

Newer Buckeye amplifiers (like the Stereophile test unit) use an improved binding post and it has fewer parts. None of the new binding post parts have ferrous content.

 

[amper42]

Sorry, just saw this as was down with the flu...

Correct, your statement about previous root cause testing is accurate. Was working with Buckeye at the time and we tested other potential solutions (tightening, soldering, etc.) before the designer found the ferrous content of the binding post tab. A summary of that investigation was posted here. The "faulty" tab is part of the binding post and was determined to have ferrous content. So fortunately, ended up a simple fix.

Newer Buckeye amplifiers (like the Stereophile test unit) use an improved binding post and it has fewer parts. None of the new binding post parts have ferrous content.

Rick,
I am glad you are feeling better. You are a wonderful asset to all of us on ASR!
Stay well my friend.

 

[Lars Risbo]

..... - or it is probably more complicated than the measuring BW.

yes probably

Here is another hypothesis: The LC filter when loaded with 8 ohms may have a peaking frequency response (Q above 0.71) and this means the output voltage can exceed the rail voltage at high frequencies. Could explain the rising power vs frequency graph you showed.

 

[SMPSGuy]

I have measured maximum power of the 1ET400A, with SMPS400A180 power supply (2x46V), in the 20Hz - 20kHz range, for THD and THD+N <= 1% and at least 10s duration at each measured frequency. This is the result:

The high feedback used in these designs not only reduces distortion in general but also reduces the effect of power supply fluctuations.

As we all know Class D gain depends on the supply voltage. Some have suggested it might be used as a volume control for open loop, zero feedback, designs because that's nice.

Whilst measurement bandwidth limitations have been mentioned, and countered, I wonder if the rise in output power with frequency might be due to loop performance of the SMPS or more specifically filtering on the board itself.

In the first instance bandwidth of an SMPS is limited so it may not be good at controlling high frequency demand. AHA! you say Capacitors. Indeed that might be the reason. At higher demand frequencies the capacitors take over, the supply ripple goes down so the distortion goes down letting you draw more power before hitting your 1% limit.

But.... With apologies for linking to another Purifi Product.

NAD M28 7 Channel Purifi Amplifier Teardown

I recently reviewed the NAD M28 multichannel amplifier based on Purifi modules. As promised, this is a tear down of the unit with kind permission of the owner. The amp was a delight to open with beautiful screws holding the attractive top down. Once removed, we are greeted to a nice and...

www.audiosciencereview.com

Spot the two thick wire rod inductors to the right of the main output inductor. On the right of them are a bunch of local filtering electrolytics. This appears to be a standard feature of Purifi Modules.

There is a thing with Switch Mode Power Supplies, a Class D amplifier is an SMPS...

Ridley Engineering | - Middlebrook1976

ridleyengineering.com

https://ridleyengineering.com/images/pdf/Middlebrook1976BritishLibrary.pdf

They interact with their inputs. Middlebrook deals with input filters but I would not be surprised if the same applies to an input SMPS supply that has its own frequency dependencies.

As a result my own wet finger would suggest that Purifi include these inductors on the board, along with the on board electrolytics, in order to isolate the amplifier from the connected supply and keep the amplifier happy whilst allowing the module to be used with a range of disparate supplies.

Perhaps these were the wrong inductors that were fitted?

In terms of the power going up with frequency in order to hit a specific THD again I am going to suggest it is down to ripple on the capacitors but with this input filter the SMPS is not important. Yes it may give you long term grunt but that filter limits the overall supply HF performance.

Should you wish to check my blither then it might be worth having a look at the HF ripple on the supply post that local filter.

Either way I think that supply ripple as reduced by capacitance may, outside of regulation, be the reason why you can get higher power out with THD as the limit.

Don't forget the important side of the coin is that that filter on the board itself will, by and large, determine the limit. Whacking more capacitance on your SMPS is likely to be a waste of time.

 

[pma]

yes probably

Here is another hypothesis: The LC filter when loaded with 8 ohms may have a peaking frequency response (Q above 0.71) and this means the output voltage can exceed the rail voltage at high frequencies. Could explain the rising power vs frequency graph you showed.

I have been investigating that phenomenon for some (not very short) time. First, I suspected the power supply voltage drop. But, as shown in this post, it was not the case, the supply voltage remains 46Vdc approx., with some but not important rise of ripple. Yes the linked example is for 8ohm, but with 4.7 ohm I measured almost same behaviour, so I am not sure it might be a LC(R) resonance due to Q. At frequencies <5kHz you may see sudden rise of amplifier spectrum noise floor just after the moment when THD vs. level plot hits the distortion knee (and at the same moment you may see higher frequency ripple at the peaks of oscilloscope sine wave, probably as approaching the clipping level and also resulting in big shifts of switching frequency). This behaviour stops at signal frequency near 10kHz and higher. Reflected in THD vs. frequency at constant power 100W/8ohm (2x46Vdc PSU) it looks like this:



Anyway, good to have one of the module circuit designers here. Would it be possible to disclose the reason of high CCIF IMD odd harmonics (skirts) in the Stereophile test of the Buckeye 1ET9040BA (probably sub-optimal) sample? It also reflected in quickly accelerated THD above 10kHz, much higher than that measured with standard 1ET400A. I am sure you definitely read the review. Maybe this should be rather dedicated to an email?