Peachtree GAN400 Amplifier Review

[uwotm8]

Excuse me but seems like a garbage for that price. Performance (except power output) is on a typical chi-fi TPA3255 level.
Load dependance, distortion rise vs frequency, on-off pop - such a bad D-class bingo LMAO
I'd say again that for me it's rather 60 dB SINAD due to HF THD mess

For 10x less price well maybe...

 

[pma]

This is a review and detailed measurements of the Peachtree GAN400 stereo balanced (input) amplifier. It is on kind loan from a member and costs US $2,499.
View attachment 271528
I am not generally a fan of peachtree audio product looks but this one in coffee stain looks pretty good. Other than a power switch, not much else is there. Ditto on the back side:
View attachment 271529
The claim to fame of this amplifier is use of GaN transistors which in theory simplify the design of a class D amplifier (a bit) and improve its efficiency (a bit). And in theory, it can also produce lower distortion (in open loop anyway). The drawback is much higher cost of the transistors compared to MOSFETs. Let's see if the performance is there.

Peachtree GAN400 Measurements
I performed all of my testing using XLR input. Let's start with our usual input of 1 kHz with output into 4 ohm to the tune of 5 watts:
View attachment 271530

There is fair bit of inconsistency between the channels with respect to SINAD. I switched my loads and issue persisted so it is not my setup. Averaging the two we get 84 dB which is slightly average for all amps tested but not where performance should be:
View attachment 271533

Frequency response is disappointing as it shows load dependency:
View attachment 271534

This means that the high frequency response will be speaker dependent which is not good. I expect this kind of performance in ultra budget amplifiers (under $100), not at this price.

Noise performance is good but again, not where it should be:
View attachment 271535

Same for multitone distortion:
View attachment 271536

Crosstalk however, is superb indicating two independent amplifiers:
View attachment 271537

There is lots of power both at 4 and 8 ohms:
View attachment 271538
View attachment 271539

As noted though, the curve starts to go up prior to clipping which means distortion is increasing.

The amplifier more than meets its 400 watt spec at 1% THD:
View attachment 271540

The shape of THD+N graph is very smooth which perhaps could be attributed to GaN transistor design:
View attachment 271541

There is no pop on power up but there is when you shut it down:
View attachment 271542

The amp is stable on power up so no need to leave it on:

View attachment 271543

Peachtree GAN400 Power Stress Test
With the owner's permission , I subjected the amp to varying load angle (phase) and down to 2 ohm impedance. It did very well:
View attachment 271544
Variations of ±60 degrees did not bother it at all at either 4 or 8 ohm. A bit strangely, at 2 ohm it would not run with 0 load angle but did produce half the output at other angles (hard to see in the 3-D graph). Since minimum impedance is 2.5 oh, we can't fault it for this. FYI I tested a couple of amps at my disposal and the both got quite unhappy with varying phase angles, shutting down and such. Whereas the GAN400 marched through this test without once going into protection.

Conclusions
I just bought a 65 watt GaN based USB-C charger and am amazed at how small it is. The buzz from there has travelled to audio and has resulted in GaN based amps like this Peachtree. Alas, there is no performance advantage that can be seen in this implementation. Noise+Distortion is only slightly better than average and far cry from top performing amplifiers we have tested in the past. Combine this with quite high price of $2,500 and this is a difficult offer to swallow. Good news is that there is plenty of power at 400+ watts independent of the load (4 or 8 ohm).

Personally I don't see a reason to pay so much for this level of performance so I can't recommend the Peachtree GAN400 amplifier.

-----------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Amir, I agree with your conclusions.

- noise is higher than it should be for SOTA amplifier (5W/4R/1k spectrum),
- distortion in blue channel is higher than it should be (5W/4R/1k spectrum),
- FR dependency on resistive load 4-8R is high and will be even worse with speaker complex impedance, most likely audible in a DBT test,
- THD+N vs. power at 1kHz (BW22kHz) is not bad but not very good,
- THD+N vs. power and frequency (BW45kHz) is bad,
- no overload recovery test is performed,
- no output DC offset protection test is performed.

"Fast" GaN transistors are not the way how to design a SOTA class D amplifier. Sophisticated feedback network like in case of NCore or Purifi is to be used.

I am voting "not terrible".

 

[amirm]

The 3D graph doesn't seem intuitive though. Is there an option for 2D plots, with the different impedance loads being different coloured traces, phase on X axis, and voltage on Y axis?

Yeh, it is pretty bad. Alas, it was written by AP and it has no other choices whatsoever. I can export the data though and plot in Excel. Will do in the future.

 

[Soniclife]

Stand Alone – AudioGraph

www.audiograph.se

It is an eval unit. Have to decide if I am purchasing it. Will probably do a stand-alone review for it.

Will we get to see how some hypex and purfi perform as part of this eval?

 

[amirm]

Will we get to see how some hypex and purfi perform as part of this eval?

There is complication there. Those amps go into protection and stop. Unfortunately the script for this load doesn't detect that and keeps running generating bogus results. May have to write a custom program to solve this.

For now the Hypex was nowhere as resilient as this amp.

 

[HarmonicTHD]

Another proof that proper engineering is required and not reliance on some hyped technology. New technology can lead to improvements but only when engineered correctly. Same applies to other audiophile hypes eg Op Amps, certain tweeters etc.

 

[DSJR]

Is there any support for asking if a simulated speaker load could be re-introduced here a la Stereophile, you know, one with a rollercoaster impedance curve in th ebass and lower midrange, peaking at 20 ohms or more in the upper hundred to lower kHz region then settling at 4 - 6 ohms or so at extreme hf? Judging by the basic responses (and levels?) shown here into a fixed impedance, this amp would fare very badly indeed, but obviously I'm just guessing here. I just think the graphics as Stereophile present would help more general viewers to see what actually happens when amps of inappropriate design turn into graphic equalisers when driving a real loudspeaker...

 

[EdW]

Thanks for the review @amirm . I didn't know that you were now able to run power stress tests. Congrats for the new step in amplifier testing!

Perhaps we shouldn’t be too surprised that 2 ohm zero phase causes a class D amp to collapse. The maximum power drain from the power supply occurs into a resistive load for class D. A purely reactive load should place a light load on the PSU. Quite different from a conventional class AB non switching amplifier.

 

[mcdn]

Is there any support for asking if a simulated speaker load could be re-introduced here a la Stereophile,

in principle that could be derived from the output impedance of the amp and applied to any speaker you like. Not sure if Amir can measure output impedance easily, but worth asking.

 

[DanielT]

It was fun with an amp test, thanks for that Amir.
_____
Slightly higher distortion, or SINAD, I can easily accept because it is still the speaker's distortion that is decisive, as long as it is not a question of a horribly bad amplifier (at worst, poor construction with very little power).

BUT, and the other has been on, a dubious FR, which can be audible (depending on the type of speaker) is bad. Nop, US $2,499 with that risk I won't take anyway.

Peachtree GAN400 has decent power, which is an important aspect so that there is enough power so that in normal case It is not driven into clipping. If this happens, with a low powered amp, however, a real nasty and audible distortion can occur. Anyone who has, for example, had a crappy car stereo and cranked the volume knows how bad it sounds.The same nasty type of clipping can occur with a clock radio, mobile or tablet.

 

[Sokel]

It was fun with an amp test, thanks for that Amir.
_____
Slightly higher distortion, or SINAD, I can easily accept because it is still the speaker's distortion that is decisive, as long as it is not a question of a horribly bad amplifier (at worst, poor construction with very little power).

BUT, and the other has been on, a dubious FR, which can be audible (depending on the type of speaker) is bad. Nop, US $2,499 with that risk I won't take anyway.

Peachtree GAN400 has decent power, which is an important aspect so that there is enough power so that in normal case It is not driven into clipping. If this happens, with a low powered amp, however, a real nasty and audible distortion can occur. Anyone who has, for example, had a crappy car stereo and cranked the volume knows how bad it sounds.The same nasty type of clipping can occur with a clock radio, mobile or tablet.

Funny thing is that the same company offers a better or similar specd and also powerful one at 1000$ less!
What would be the discriminator for someone to choose this over the cheaper one?

 

[pma]

in principle that could be derived from the output impedance of the amp and applied to any speaker you like. Not sure if Amir can measure output impedance easily, but worth asking.

Output impedance is easy to measure, provided that you are willing to do it - necessary condition.

Output source impedance
Characteristic to be specified
The internal impedance measured between the output terminals under specified conditions. The manufacturer shall state the rated value in the specification.
Method of measurement
The amplifier is brought under standard measuring conditions. Then the source e.m.f. is
reduced to zero and the rated load impedance is disconnected.
a) A sinusoidal current source of internal impedance at least 10 times the expected value of
the output source impedance is connected, in series with an ammeter, to the output
terminals of the amplifier. A voltmeter is also connected to the output terminals. The
current is then adjusted to the value which would flow through the output terminals under
standard measuring conditions.
b) The value of this current may be calculated as the current which would produce a voltage
level of −10 dB referred to the rated (distortion-limited) output voltage across the rated
load impedance.
c) The voltage across the output terminals U2 is then measured.
d) The measurement may be repeated at other signal frequencies.
e) The output source impedance is then calculated according to the following formula: |Z| = U2/I2

With any software that generates frequency sweep it is even easier. Send 1A to the output, measure voltage at output terminals and you have modulus of output impedance, 1V = 1ohm. Like this:

 

[Multicore]

I am not generally a fan of peachtree audio product looks but this one in coffee stain looks pretty good.

Hmm. I think it looks like a modem from around 1995.

 

[Multicore]

Yeh, it is pretty bad. Alas, it was written by AP and it has no other choices whatsoever. I can export the data though and plot in Excel. Will do in the future.

If you keep the fancy load box and do this testing in future then I would be grateful for some tutorial text on how to understand the test and its results.

 

[xaxxon]

There is complication there. Those amps go into protection and stop. Unfortunately the script for this load doesn't detect that and keeps running generating bogus results. May have to write a custom program to solve this.

For now the Hypex was nowhere as resilient as this amp.

How about a benchmark ahb2?

 

[Multicore]

Output source impedance
Characteristic to be specified
The internal impedance measured between the output terminals under specified conditions. The manufacturer shall state the rated value in the specification.
Method of measurement
The amplifier is brought under standard measuring conditions. Then the source e.m.f. is
reduced to zero and the rated load impedance is disconnected.
a) A sinusoidal current source of internal impedance at least 10 times the expected value of
the output source impedance is connected, in series with an ammeter, to the output
terminals of the amplifier. A voltmeter is also connected to the output terminals. The
current is then adjusted to the value which would flow through the output terminals under
standard measuring conditions.
b) The value of this current may be calculated as the current which would produce a voltage
level of −10 dB referred to the rated (distortion-limited) output voltage across the rated
load impedance.
c) The voltage across the output terminals U2 is then measured.
d) The measurement may be repeated at other signal frequencies.
e) The output source impedance is then calculated according to the following formula: |Z| = U2/I2

Let's see if I understand this test right... With the amp turned on but producing no output, attach test gear to the amp's output that pushes current into the output and see what voltage develops across it for a given current?

With any software that generates frequency sweep it is even easier. Send 1A to the output, measure voltage at output terminals

This test sounds different. Test signal is applied to the amp's input and the amp itself produces the current at its output on a test load?

The latter is the arrangement we are familiar with from Amir's and much other modern testing.

 

[Matias]

This shows how GaN alone is not the last word in class D amplification and solves all problems, as per Q&A 8.

 

[pma]

Let's see if I understand this test right... With the amp turned on but producing no output, attach test gear to the amp's output that pushes current into the output and see what voltage develops across it for a given current?


This test sounds different. Test signal is applied to the amp's input and the amp itself produces the current at its output on a test load?

The latter is the arrangement we are familiar with from Amir's and much other modern testing.

You understand, but maybe partially. One power amplifier (or second channel of the DUT) sends current to the DUT output. I think it should be easy to understand. Sine sweep is sent from driving channel and response measured at the output of the DUT (forced) channel. At DUT, 1V = 1ohm, because 1A x 1ohm = 1V.

 

[AudioSceptic]

Is there any support for asking if a simulated speaker load could be re-introduced here a la Stereophile, you know, one with a rollercoaster impedance curve in th ebass and lower midrange, peaking at 20 ohms or more in the upper hundred to lower kHz region then settling at 4 - 6 ohms or so at extreme hf? Judging by the basic responses (and levels?) shown here into a fixed impedance, this amp would fare very badly indeed, but obviously I'm just guessing here. I just think the graphics as Stereophile present would help more general viewers to see what actually happens when amps of inappropriate design turn into graphic equalisers when driving a real loudspeaker...

I like that idea in principle, but how do we decide what that simulated load should be? A worst case of available speakers, an average, or what?

 

[TNT]

Peachtree GAN400 Power Stress Test
With the owner's permission , I subjected the amp to varying load angle (phase) and down to 2 ohm impedance. It did very well:
View attachment 271544
Variations of ±60 degrees did not bother it at all at either 4 or 8 ohm. A bit strangely, at 2 ohm it would not run with 0 load angle but did produce half the output at other angles (hard to see in the 3-D graph). Since minimum impedance is 2.5 oh, we can't fault it for this. FYI I tested a couple of amps at my disposal and the both got quite unhappy with varying phase angles, shutting down and such. Whereas the GAN400 marched through this test without once going into protection.

Yuiipiiie!! Finally - real world amplifier load.

Now you just have to re-test the top 20 on the SINAD list ;-D

More seriously... Stress test is interesting but I which you would also do 5 watt check also - thats where we play most of time...

You dint even have to show these curves - just report that nothing to see at 5W... but if there was - please show us...

//