Part I: 3e Audio - A5 Stereo and A7 Mono Class D + PFFB, TI TPA3251/3255 Amplifiers - Features and initial measurements.

Every once awhile, a manufacturer will connect to see if I might be interested in testing out one of their upcoming products. Recently, 3e Audio (their mission being to Enhanced, Exceed, Enjoy) based in Shenzhen reached out for an honest, independent, evaluation of their new amplifiers. As usual, most of my hardware discussions are based on objective results which I believe significantly reduces personal subjective biases and advertising hype - no YouTube videos where reviewers make silly facial expressions to draw attention to the thumbnail image here folks. 🙄

You probably haven't heard of this brand, but the engineering-focused company has been around for a number of years selling boards to the DIY market with DSP, power supply, and Class D amplifiers. In August, they announced the release of their line of five fully-assembled "A5x A7x Series Next-Gen PFFB Amplifiers". In the picture above, we see the pre-release A7 Mono (260W into 4Ω 1% THD+N, single TPA3255, PBTL) and A5 (150W x 2 into 4Ω 1% THD+N, dual-TPA3251, PBTL) models sent here for evaluation.

As you might know, in this blog over the years I've been covering a number of these very reasonably priced TI TPA chip amps from the early generation TI TPA3116, to the early TPA3255 models (like the AOSHIDA A7), to the very high performance Topping PA5 MkII+ (TPA3251), to the recent inexpensive and higher powered Fosi Audio V3 Mono (TPA3255) with Post-Filter Feedback (PFFB). The inclusion of PFFB to 3e Audio's amplifiers is not new for this company as this feature has been available in their DIY boards since at least 2022. 3e Audio is aiming to exceed the performance of others that have come before; of course, we'll see about that on the testbench!

I'm not sure exactly what date 3e Audio will make these amps available or the prices. At the time of these tests, I was told that they will be taking orders on their AliExpress store in early December 2024. Final prices might change, but I was told that the amps will range from US$199 for the A5se (140W x 2 into 4Ω, 1% THD+N, single-TPA3251, BTL), to US$300 A7 models with power supply.

Clearly we can see the family resemblance here. The boxes looks identical except for the model name and obviously this amplifier can accept both RCA single-ended and XLR (combo XLR/TRS) balanced inputs as per the front indicators. It's in a matte black, aluminum case, measuring 17cm wide x 16cm deep with speaker terminals and volume knob included, 5cm tall. The A7 Mono weighs 835g, and A5 is 900g. This pre-production version looks understated without any fancy company logos. The inclusion of clipping indicators is a nice touch and this will let you know if you are in need of more power - indication for an amp upgrade if this turns on too often!

There's a large, smooth turning, central volume knob. Press down on the knob for 2+ seconds to turn on/off, a short push will select balanced or unbalanced inputs.

There's no marker on the small "BYPASS" switch, slide left (towards the speaker terminals) to bypass the volume knob.

The family resemblance extends to the rear.

Obviously the A7 Mono only has a single pair while the A5 (Stereo) has two pairs of speaker binding posts; they feel solid, of good quality. My banana plugs fit snugly with reassuringly tight contact. To the left of the speaker binding posts are both balanced combo XLR/TRS and single-ended RCA analogue inputs. Both amps are true balanced designs so there should be benefit to the signal-to-noise with the use of XLR/TRS input (we'll test this later). Input impedance is 10kΩ RCA and 20kΩ XLR/TRS.

Notice the RCA gain switch which allows the user to set either +23/+26dB for the A5 models, or +26/+29dB for A7 models. The usual rule of thumb is to keep it at the lower setting if this is adequate since higher gain might add a little more noise.

On the right of those speaker connectors are a "Bypass Vol" switch (turn it to the left to bypass) which removes the volume control from the signal path, allowing potentially lower distortion. I see the manual recommends that you only switch this with the power off to avoid "pop" through the speakers. To the far right is the power input. For the more powerful A7 models, they can accept up to 52VDC and lower-power A5 models up to 38VDC.

Notice that on this pre-production model, there's a 12V trigger input connector. The device actually has automatic signal detection already (power-saving sleep mode after 5 minutes without input) and I was told that this trigger input will be removed in the final production models as it's likely redundant.

Let's look at the power supplies I was sent:

They're both GaN (gallium nitride) models, relatively compact, efficient Level VI switching power supplies. The larger 52V/9A (468W) one weighs about 1kg, measuring 21cm long x 9cm wide x 4.5cm tall; this is meant to power the A7 models. The smaller one provides 38V/5A (190W) meant for the A5, weighing 450g.

"Compact" size is relative of course. Here's a comparison of the 38V GaN
power supply compared to the 38V from the Topping PA5 Mk II+ amp.

For both amplifiers, when turned on and not playing music, idle power draw is <10W (manual specifies 8W for A5, 6W A7 Mono). It goes into a <0.5W power-saving mode when there has been no signal for 5 minutes (signal threshold minimum of 2mV RCA, 4mV balanced).

Okay, I think that's a good enough look at the externals of the product.

I have not opened the boxes up to have a look inside yet. Based on their webpage, it looks like there are socketed op-amps that can be replaced should you feel that this is beneficial. The manual speaks of the usual options like TI OPA's, LME's, JRC MUSES, Analog Devices, etc. Since what I have here are close-to-production units, it would not be surprising if the final products may have cosmetic / minor internal changes.

As usual, I spent time listening to the amplifiers for a few nights to create a subjective feel for the sound quality which I'll talk about next time since the subjective stuff isn't something I would hang my hat on. Much more useful IMO is to get a more objective sense of the products' capabilities and what it might bring to the family of inexpensive Class D amplifiers.

Let's start with some initial objective tests which will expand in Part II next time.

Never mind the mess of wires, dummy loads, and Linear Audio Autoranger back there.🙂 I'll be using the Sabaj A20d 2022 (ES9038Pro) as hi-res DAC for signal generation. E1DA Cosmos Scaler and Cosmos ADC for signal capture.

I. Volume Gain, and Oscilloscope View

Here are my measured results for volume gain from the two amps:

3e A5:

     XLR: +16.5dB (official +17dB)

     RCA Low: +22.6dB (official +23dB)

     RCA High: +25.7dB (official +26dB)

3e A7 Mono:

    XLR: +19.6dB (official +20dB)

    RCA Low: +25.4dB (official +26dB)

    RCA High: +28.9dB (official +29dB)

Notice the difference between the A5 and A7 Mono with the A7 about +3dB higher gain. This is consistent with the official figures in the manual. Likewise, the RCA Low/High gain results are consistent with the product's specs.

Let's hook the amp up to a digital oscilloscope to have a look at some waveforms. Here's the 3e A5 reproducing a 1kHz sine, both channels at 1Vrms, I purposely did not bypass the potentiometer:

A couple things to notice. First, the channel balance is excellent with only a 0.01V difference (<0.1dB). Secondly, notice that even though this is a Class D amplifier, and I'm not using any low-pass filtering, the switching noise in the signal is visibly low, significantly lower than the Fosi V3 Mono and more similar to the Topping PA5 Mk II+.

In the same way, the 3e A7 Mono has very low switching noise - let's have a look at a 1kHz square wave:

The switching noise is low enough that we can easily make out that the DAC is using a linear phase filter (running at 24/96) with pre- and post- ringing, again without me adding any low-pass filtering to the signal.

I didn't get a chance to catch the frequency of the switching noise but confirmed with the company that this is at 600kHz, as seen in other TI TPA325X-based amplifiers.

II. Output Impedance and Damping Factor

With Post-Filter Feedback (PFFB), these amplifiers have been designed to be load-invariant and we should see very low output impedance.

Indeed, wow! We're looking at output impedance on average just over 10mΩ. We can convert that graph to a damping factor into 4Ω as I have done in the past:

With such low output impedance, we're looking at a damping factor in the hundreds now across the whole audible spectrum befitting of high-quality solid-state amplifiers. It should be obvious then that these amps will be able to perform predictably across a broad range of speakers.

III. Frequency Response

That very high damping factor should essentially result in a "flat", at least audibly insignificant, frequency response with different loads.

Indeed, <0.5dB variation across the loads from 20Hz to 20kHz and with good ultrasonic extension beyond 30kHz  (if one cares 😉). Regardless of the dummy or speaker loads, there's barely any effect on the frequency response except those tiny differences above 10kHz which are altogether irrelevant (the Y-axis on the graphs above are just +/-3dB range). Notice that I have included a 2Ω graph here also. These A5 and A7 Mono amps are Parallel Bridge-Tied Load (PBTL) designs and they can provide more current for difficult loads, the company specifies stability down to 2Ω.

For reference, here again is the impedance, phase, and EPDR curves for that Sony SS-H1600 bookshelf speaker used in the measurements above:

Here's the 3e A5 amplifier connected to the Sony SS-H1600 bookshelf speaker showing both the frequency response (10Hz to over 40kHz) and phase:

0° phase means there's no polarity inversion with these amps.

The result of the PFFB implementation on these TI TPA325X Class D amplifiers by 3e Audio looks excellent.

IV. Crosstalk (A5 Stereo amplifier)

Obviously there's no measurement for crosstalk with the A7 Mono. For the A5 amplifier with 2 channels, let's have a quick peek at the crosstalk using a strong 0dBFS 4kHz tone in the left channel, and 0dBFS 300Hz signal in the right driven simultaneously; let's see how much seepage there is going into the other channel (the Sabaj A20d 2022 DAC itself shows a limit of -95dB RCA and -120dB XLR so that would be maximum). A more taxing test than typically looking at crosstalk in the other channel with only one channel driven.

Notice the very clean noise floor with XLR in, no 60Hz + harmonics power line hum.

As I am a "first watt is most important" guy (we're not talking about PA amps so low-level nuances most important), I typically default to 2Vrms into 4Ω with many of my tests. Clearly, we can see signs of the superiority of using balanced input with 22dB lower crosstalk. Notice that the higher frequency 4kHz signal tends to seep into the other channel more than the 300Hz tone.

The measurements above are with the volume bypass enabled (ie. volume control disabled). If we switch off the bypass, allowing the volume control back into the audio chain, let's see if that affects anything:

Good to see that crosstalk has not worsened with volume control enabled. The RCA result was exactly the same. It's interesting that the crosstalk value actually decreased (improved) compared to the bypassed mode with the XLR input.

These results are certainly encouraging to see, suggesting that the volume control has been implemented well and might not significantly impact the sound quality - we'll look in more detail next time.

This is what the clip warning indicators looks like when presented with a strong signal to RCA input, volume bypassed. As usual, in testing, I often push amps to their limits until shutdown. Yeah, in the past I've blown an amp here and there.😢 So far, these 3e amps have withstood the punishment well.

V. Summary, Part I

As typical, I'll split this detailed review into two parts. No need to rush!

Thus far, the 3e Audio A5 (Stereo) and A7 Mono appear to be well-thought-out Class D amplifiers enclosed in high-quality compact metal enclosures, have balanced XLR/TRS and unbalanced RCA inputs with bypassable smooth-turning volume knob. I believe 3e Audio might also sell the amplifiers without power supplies since some of us might already have one or two on hand from previous purchases; I guess we'll see once their online store opens. A good way to upgrade amps for less money.

The A5 is a lower-power dual-TI TPA3251 design (150W x 2 into 4Ω, 1% spec) compared to the more powerful A7 Mono based on a single TI TPA3255 (260W x 2 into 4Ω, 1% spec). Both are PBTL designs and the company indicates that they can handle down to 2Ω loads. We'll see next time the measured power output and distortion amounts.

Compared to earlier generations of TI TPA325X amps, these have implemented Post-Filter Feedback (PFFB) and we're seeing excellent performance in both these models resulting in load-invariant behaviors within the audible frequencies. Output impedance is low, averaging at just over10mΩ across 20Hz to 20kHz. Indeed, frequency response of various resistive loads (2/4/8Ω) as well as the reactive load (actual Sony bookshelf speaker) remains basically flat with no expectation of audible difference.

I can confirm that these amps maintained "absolute polarity".

Another characteristic of these 3e Audio amps I was impressed by is the well-attenuated noise level of the 600kHz switching frequency compared to the majority of other Class D amplifiers I've examined over the last few years. The only comparable TI TPA device with such lower noise was the Topping PA5 MkII+ I looked at a little over a year back.

Finally, the crosstalk on the A5 Stereo is good with an average of -85dB using the RCA input and better than -100dB with balanced XLR using my measurement method.

[Crosstalk is one of those measurements we can easily do but as an objective reviewer, need to remind everyone that for excellent stereo separation, even something seemingly very low like -25dB would likely be enough through speakers. There is intrinsically high crosstalk with speaker playback which is why we might want to apply crosstalk cancellation to improve soundstage presentation and conversely add crossfeed with headphones to make it sound more like listening to speakers in a room. Comparatively, on good vinyl systems, one might only get -30dB crosstalk.]

The only caveat I must remind everyone is that I am evaluating pre-release units so minor esthetic and design updates would not be surprising. The company however has assured me that the technical performance will be equivalent to the final product. So far, so good, as I continue listening and testing these amplifiers!

Off to do some more listening now. Next time in Part II, we'll examine the power capabilities of these amplifiers, have a look at the distortion characteristics, plus discuss some subjective impressions.

I hope all the American friends had a nice Thanksgiving. Wow, December here already, year end cometh!