Douk A5 Budget Stereo Amplifier Review

[truth_soldier]

It isn't the same though.

People can feel free to disagree with each other's opinions, the issue is that people can paste large amounts of AI output with very little effort and it takes a disproportionate amount of time to wade through it all and refute all the errors.

OK I get you.

 

[RandomEar]

I realize all that and caveated that it's not necessarily completely correct - though maybe has some truth and good points in it.
the point is not to pit AI against you in an argument, it's to debate the points and possibly dig into the info for the purpose of learning.
and... I deleted the AI content because after I thought about it, and this site's frowning on posting AI content.

If you want to learn about the topic, read the right source material. These [1, 2, 3] are a good start for caps. If you are interested in SMPS, I can recommend [4, 5]. Using AI in a discussion just creates walls of text that other readers have to wade through to understand and agree or refute it. It's convenient for the AI poster and cumbersome for everybody else.

"the sum of training data" to me means that it's just a better search engine and as such it may be finding more information than any one person - including you - knows about. I think we should all be open to possibly learning something that we didn't know. ie, be skeptical but open to learn.

Again: If the sum of training data in a specific field is crap, the LLM output will be crap(pier) for that field. This is sadly the case for audio stuff.

let's just say I'm skeptical that your statement about what a power supply "likely" has in it means that adding this particular cap value must be redundant.
But I'm also not blindly believing what one person thinks they hear from that one cap addition. I think it's worth trying to see what happens.

It's fine to be sceptical. But in this case, I would rather be extremely sceptical of any audible improvement from such a mod until someone explains how this could possibly work in a plausible way.

To see if an X type cap is installed in the Douk PSU, someone would need to open it to check which would probably destroy it because the casing looks plastic welded or glued. But think about it logically: This is a noise suppression cap for the MHz range you want to install close to the DC input of the amp. But we are not listening to input voltages and we can't hear anything close to the MHz range anyway. The GaN PSU already delivers filtered 48 V on the input with ripple in the range of a couple dozen to maybe 200 mV (AIYIMA specifies less than 141 mV for their 48V/5A and less than 48 mV for the 48V/10A, see attachment). That's 0.42% of ripple, a very small amount.

If you are not worried about ripple, but rather voltage sag at the TI chip input, the location and type of this cap are incorrect as mentioned by @MCH . The cap needs to be extremely close - as in less than a handful of millimeters away at most - for this to work as specified. That fat 3300 µf cap is easily ten times as far away from the chip, maybe more depending on the trace routing on the PCB. In addition, such a "support cap" would need to be an MLCC and not a film cap, because the legs on a film cap have a signifcant enough amount of impedance that it makes it perform worse in this application.

My understanding of this is also limited. I'm not a PCB designer. But I still think this idea is nonesense and haven't seen any indication that it's not.

 

[truth_soldier]

Yes they are, they are C50 C52 in my version (v3)

Actually ... are you sure C50 and C52 are input PVDD caps, they look like pre-output filter caps (on the other side).

 

[truth_soldier]

That cap in the mod is way too far away

Btw, not so much in the A5 but in the A5 Pro (that originally brought up this bypass mod) the large Rubycon cap solder position is very close to the TPA3255 inputs, not far away. Just sayin'. The A5 Pro is not the subject of this thread.

 

[MCH]

Actually ... are you sure C50 and C52 are input PVDD caps, they look like pre-output filter caps (on the other side).

hey, yes, i can confirm continuity between the input 3300uF capacitor and C50 and C52.

The output caps you refer to in a previous post are most probably C1 C26 C28 C29 (version 3). I dont fancy dissasembling my unit to confirm this, though, but by their placement i am pretty sure those are the ones.
Note that C50 and C52, even being to the other side of the PCB. are still quite close to the power input pins as TI recommends. Normally in early TPA3255 amps you could see 4 ceramic caps on the same side of the PCB than the amp chip, but since a couple of years ago it became the trend to place the chip on the lower side in contact with the lower side of the enclosure for heat dissipation and the surrounding components on the top side.
If you see v1 and v2 of the A5, they still had the typical 4 ceramic caps like the TI eval board and most other tpa3255 amps, no idea why they replaced them for 2 electrolytic caps in v3.

 

[MCH]

The output caps you refer to in a previous post are most probably C1 C26 C28 C29 (version 3). I dont fancy dissasembling my unit to confirm this, though, but by their placement i am pretty sure those are the ones.

Actually, no need to dissasembly the unit, i tested continuity between C1 C26 C28 C29 from the window and the speaker terminals and indeed each one of them is connected to one of the terminals, so yes, confirmed, those are your output caps

 

[truth_soldier]

Actually, no need to dissasembly the unit, i tested continuity between C1 C26 C28 C29 from the window and the speaker terminals and indeed each one of them is connected to one of the terminals, so yes, confirmed, those are your output caps

OK so you're saying that the C50 and C52 have continuity to the Rubycon cap but not to the speaker terminals?

 

[MCH]

OK so you're saying that the C50 and C52 have continuity to the Rubycon cap but not to the speaker terminals?

correct

 

[DrLex]

I’m afraid that I can’t contribute to the technical discussion, but the positive review persuaded me to try one as a temporary replacement for a failing Arcam delta 290, and I am delighted with it, so much so, that I may not bother having the Arcam repaired. The HPF is ideal for use with my Rogers LS3/5A (11Ω ) and is encouraging me to add a subwoofer later in the year. Thank you, Amir (and your panther).

 

[truth_soldier]

I just noticed the V3 was available since at least May of 2025 (seen here).
so it came out way before @Douk Audio announced it here on ASR in September 2025

 

[RandomEar]

If you see v1 and v2 of the A5, they still had the typical 4 ceramic caps like the TI eval board and most other tpa3255 amps, no idea why they replaced them for 2 electrolytic caps in v3.

They wrote in a private message that they feared cracking of the ceramics, see here. I assume that could happen due to thermal cycling? On the other hand, there's still lots of smaller MLCCs close by for output filtering. Who knows.

 

[Douk Audio]

I just noticed the V3 was available since at least May of 2025 (seen here).
so it came out way before @Douk Audio announced it here on ASR in September 2025

Yeah, you are right.
The V3 version of the A5 first finished production on April 24, 2025. Therefore, if a customer received a V3 unit in May, it indicates that the product was shipped directly from China.

At that time, our overseas warehouses still mainly contained V1/V2 versions. If a customer purchased through channels like Amazon or eBay with local overseas delivery, they were unlikely to receive the V3 version. This is precisely why we scheduled the official release for September—by then, the inventory across most channels had been updated to the V3 version, ensuring consistency and avoiding confusion due to different purchase sources.

Hope this helps.

 

[DavidR]

I substituted one A5 into my 3-way dipole system on the woofers. I have a pair of 10" 8-ohm woofers in parallel on each side. Being dipole, it requires a big gain boost down to 20Hz. In fact, it's about 25db to be flat to 20Hz. Lowpass is 250Hz LR8 linear phase using the Ultimate Equalizer from Bodzio software. I use a Kenwood KM-X1 6-channel amp per side (100W @8 ohms rating). I think that they have excellent current capability, I've not found any strain at even fairly high power output. The A5 amps will be used on a wall/desktop system, but I wanted to see how one holds up on the woofers that are extremely demanding. I'm pleased to report that in admittedly limited testing they were the equal of the Kenwoods. I will probably try one on the midrange and maybe the tweeter, but I connect the tweeters directly to the Kenwood. Given the comments about turn-off noise I'm a bit concerned about that testing. I haven't yet measured for any DC offset, either.

 

[Kaameelis]

I have a pair of 10" 8-ohm woofers in parallel on each side. Being dipole, it requires a big gain boost down to 20Hz. In fact, it's about 25db to be flat to 20Hz.

It is totally off topic, but more important than high power amp is big Xmax and high sensitivity of woofers.

 

[truth_soldier]

but more important than high power amp is big Xmax and high sensitivity of woofers.

I agree than high sensitivity speakers are usually a good thing, but I don't think they need to have big Xmax.
High sensitivity large drivers with low xmax can still move a lot of air and sound great.
I have a pair of 15" 96db drivers in open baffles. They have only 6mm Xmax but sound wonderful.

here's another example Lii Audio 15" driver with 5mm xmax: https://liiaudiohifi.com/catalog/fast15
Also if you horn load a driver then you need even less xmax as the horn makes up for it.

 

[DavidR]

It is totally off topic, but more important than high power amp is big Xmax and high sensitivity of woofers.

I provided that info only to emphasize just how well the A5 functioned in that system. Off-topic, yes, the displacement is why I used two 10" drivers per side. What's important are the overall T/S parameters for the intended application. These were "Originally manufactured for a famous, premium brand of Finnish self-amplified studio monitors" according to Parts Express back in 2014 2010 when I bought them (Peerless). They fit the application well. That's all that need be said.

 

[Kaameelis]

I have actually same type, OB dipole speakers, only boost on my woofers is 20 dB and they are 12", 2 in one H frame and I can see increasing THD because cone amplitude is close to Xmax on moderate SPL levels. https://www.audiosciencereview.com/...-your-in-room-measurements.13540/post-1737514
My woofer amps (2 pc.) are 150W to 8 ohm and they are not very far from max power on moderate SPL levels.

 

[truth_soldier]

provided that info only to emphasize just how well the A5 functioned in that system.

+1, the stock/un-modified A5 sounds very good in my system w/high sensitivity speakers driven by the tiny and low cost SMSL D1 DAC.

 

[truth_soldier]

Does anyone have a good solution to keeping the lid propped open while being secure in place?

This was the first thing I found - plastic mesh cut from an "organizer" type basket, the type you can find at some office supply or home organize/decorate stores. Cost just a few bucks. It'll do until I can find something else that looks better. I cut it to press-fit so it doesn't fall out, has to be pulled.

 

[truth_soldier]

Tangentially related ... at the same store I found these $3 bamboo kitchen pots-n-pans "hot pads" (trivets)
I put some poly feet on them and use them under the Douk A5 and my tube preamp to aid airflow/cooling.
Plus I like the wood color makes a bit more visually interesting - esp combined with my SMSL DAC bamboo mod.