Class-D 8-channel amplifier build

[gainphile]

I'm starting a new thread for this amp which will amplify my MiniDSP XO outputs.

So I think I've figured out my rough design. I'll elaborate more on my build thread but it basically goes according to the picture I lovingly drew. I'm thinking of also using those tripath amps but need to find out how gainphile powered them. All-in-all it could turn out to be a very cheap build (objectively)

I will be using switched Power supply unit from sure electronics.

"24V DC 14.6A 350W Regulated Switching Power Supply"

It just arrived today so I will have the whole amps assembled this weekend.

The whole thing is experimental to me. Both the Class D amps and the PSU especially the fan (it's supposedly temperature-controlled, not on all the time). It's because I don't have the energy to build and solder gainclones which I know is sonically transparent (although I already have all the chips and audiosector boards).

The amps are two of "4*100W TK2050 TP2050 Tripath D-class Amplifier Board" from Sure Electronics (eBay).

Total cost for the 2 amp boards (8 channels total) + 1 PSU is $160. I'll add $80 for casing and connectors.

The driven speakers are 4-way dipoles. Today they still lack the last bottom octave as a 3-way. But ok 90% of the time except synthesized music stuff. The subwoofers would be dipole W-frame subs for 20-40z as used in SL's Phoenix.

[ATTACH=CONFIG]24875[/ATTACH]

Maybe at that time I'll add the 2nd PSU.

[LogicprObe]

But...........it's class D..........

[gainphile]

Yes. One my objective/experiment is to validate that generally electronics are very accurate/transparent compared to electro-mechanical transducers such as Loudspeakers (+ room acoustics).

[LogicprObe]

At least you can use it to power 8 subwoofers later!

[Paul Spencer]

A very interesting project there Andi!

One thing I wonder about is using Class D for tweeters - is that the plan? I'd be a bit wary of that. You might notice that Adam speakers, although they like to promote class D amps in general, they use AB for the tweeters.

I tend to agree that the electronics are hardly the worst offenders.

[gainphile]

Yes that's the plan. All I can say now is "we'll see ..."

[Guest Drizzt]

Excellent stuff gainphile. I look forward to reading your results.

Definitely good value for money if the performance is up to your standards.

[Westy87]

Oh sweet. I think I'm gonna just buy an amp for my main speakers, but this could be fun for the bedroom or workshop

[gainphile]

It's done ... and I've listened to them for few hours

[Wembley]

So tell us how they sound!

I'm actually about to embark on a similar experiment myself very soon. Will be modifying a set of speakers with the miniDSP and a similar class D chipamp. So very interested to hear how easy it all is/was, and how you like the listening experience at the end.

Seems my decision not to buy an AVR with pre-outs might be one I live to regret though. I'll post in a seperate thread once I start on my own experiment, as see how the results compare to your own.

[gainphile]

I've introduced few variables since using my gainclones so I need to carefully asses them.

1. Previously the gainclones have pots so I didn't need to use the digital attenuation for tweeter level matching. Now I do and this supposedly is bad.

2. Previously the bass section uses different amp. Now all 3 channels uses the ClassD

So I've done all the adjustments I could but I'll listen carefully tonight when it's quiet and will post report tomorrow .

I can tell you few things now though...

- It's very pleasant to build system based on kits. Like "modules".

- No hums, turn on/off thumps etc. which normally takes me weeks to troublehoot

- hiss is at very slightly higher level like gainclones. I can hear it 10cm from the tweeters but after that none. I classify this as quiet.

- Hum is nonexistent, period. Better than my gainclone implementation.

- No DC voltages at the output

The response of my speakers in room to make sure there are no funny things going on... No room EQ, just speakers'. They look very similar. The Class D lower bass region is better, understandably due to similar amp modules being used.

With Class D

[ATTACH=CONFIG]24926[/ATTACH]

With Gainclones

[ATTACH=CONFIG]24927[/ATTACH]

[Paul Spencer]

Do they have any turn on delay?

[LogicprObe]

Do they have any turn on delay?

Dunno about the amps...................but my missus does!

[Antripodean]

Nice work Gainphile! Interested to know how often the fan comes on and the apparent volume it produces. How close are you to your amp when listening?

[Wembley]

Very nice work there Gainphile. Nice to know that the job isn't too tricky.

I'm also very impressed with the measured in-room response. I'd be interested to know if that was measured with a close microphone or from the listening position. My room doesn't have any treatments yet, so never get anything that nice. Treatments will be next on the cards after I play with my speakers.

Now to see if the listening impressions match the improvement in the FR charts. Also be curious if a mix of Class D and gainclone (driving woofer and tweeter respectively) is different again, assuming its a test you care to perform.

[Paul Spencer]

What kind of measurement is that Andi? Gated @ 1m? Outdoors? Looks ultra flat but then the y scale is huge.

[gainphile]

Well it's 3am in the morning and I have my conclusions

When I initially listened to the new amps with the above "variations" yesterday it did not sound as clean as when I used the gainclones. Digital attenuation for level matching was a suspect so I removed them and soldered few pots to control them analogue-ly instead.

After that the amps reproduced the music flawlessly. They are clean and transparent.

I thought the sound is a bit dry and more resolving than gainclones, but normally I care less about these kinds of observations. I paid attention too on the bottom punch as Roger advised me the day before at the GTG. I do find them rolled off below 20hz quicker than Gainclones. As evidence those LP rumbles are no longer threatening even without 18hz highpass filters. These are not the amps if you want to reproduce 14hz (the specs do say it's -3db at 20hz).

Operationally the amps are great. Aside from the intial impressions I posted in terms of hum, hiss, etc. they do not operate hot (the case just slightly warm) and I have never heard the fan of the SMPS switching on yet. They do not have turn-on delays, but thumping is nonexistent both on power on/off.

So in conclusion I'd say they are as good as my gainclones, and are better proposition in terms of value for money and ease of implementation.

Now some pics ...

The modules fully wired up. There's space left in case I want to add the second SMPS. DIY can't get any easier than this....

I thought long and hard whether to implement RCA (my pet hate) or not and glad I used two 4-pin DIN connectors. Tidy and easy! But it makes it hard to do A/B comparation with other amps.

How the amps look like along with the crossover. I use neon switches as LED is another pet hate of mine. I still need to tidy up the power cables.

About the measurements, they are taken using pink noise, 1m at the design axis which in my case is between the tweeter and midrange. No gating. I found this reliable for level matching, but not for dipole eq investigations, for that I only trust outdoor, non-gated measurements, 2m from the ground (yes a pain). This is the measurement using potentiometers to adjust tweeter and midrange levels, this time on a 10db scale.

[Antripodean]

Great write up Gainphile! It answers all of my questions and more Seems like it is something that an electronics newbie could put together quite successfully

[gainphile]

I have grown fond to these amps. Did a few research and apparently the Tripaths made it to IEEE "25 chips that shook the world" list. That is quite an impressive achievement.

http://spectrum.ieee.org/semiconductors/processors/25-microchips-that-shook-the-world/5

It still beggars belief that something can be had for that kind of price and ease.

There is also an old review from Stereophile on one of these Tripath chip based amplifiers. It gives quite a good description on their operation, qualities, and shortcomings. For example the waveforms rings and distorted (but so is 4th order crossovers)

http://www.stereophile.com/solidpoweramps/442/index.html

I generally do not trawl the forums for Amplifier stuff but a quick look shows there are 2 observations when it comes to these amps. First is that they are generally more resolving, but also that they are clean (as in "sterile") sounding. I will leave these observations to the experts but that is too what I'm thinking about them. As always there are people who likes the sound of Class AB and there are who swears by Class D. For me they are minute, if any different at all and I can easily live with either.

[andreasmaaan]

Many thanks for the interesting observations gainphile. partly on the basis of your thread, i've just bought the same unit and power supply, and have spent about 30 min testing so far.

Initial observations are that the amp has a lot of grip and detail, and very precise imaging. On the other hand, things sound a little shrill and unnatural to my ears.

I suppose I could be generally summing up tripath chip-based amps here, but these features (both good and bad) seem even more pronounced with this unit than with my dayton t-amp (sonic impact clone based on ta2024).

Anyway, would you agree with my observations? And did the sound change (ideally, mellow out a little) over time?

Cheers,

Andreas

[gainphile]

Hello,

No I do not observe any 'shrillness'. I had dome tweeters on my speakers and lately have been using waveguide+compression driver. The CD is Selenium D220Ti which has been reported to be "shrill" due to titanium diaphragam. Yet I do not hear that even with the Class-D. I'm running active though.

In regards to imaging, I think this is the domain of speakers especially its polar response and room interaction rather than electronics.

[jimwhite]

Any news? I am waiting with bated breath !

[DQ828]

I'm with Jim. I need to power my new speakers eventually and am searching for amp options. People seem to be pretty divided over the whole D class sound quality issue. I also believe it takes a while for anybody to fully assess a me toy, unless there really bad.

David

[zog]

I've got a few Tripath modules, from a few different suppliers. A lot of this stuff would be familiar to readers of DIYaudio forums, but you have to wade through thousands of posts to get the info!

Current one I'm using is a HiFiMediy T2.. same Tripath chipset as the Sure Electronic TK2050 amps, but better quality components (caps, etc) "out of the box", doesn't sound as thin (as a Sure unit I tried with the same speakers) to me. I'm using it with a 160VA toroidal transformer, and have added some extra PSU capacitors filter caps. It's a standby while I sort out my ta3020 v3c, but it sounds good enough that I'm in no hurry.

http://hifimediy.com/

Also using a Lepai LP-2020 amp with my computer.. this one was tweaked up with most of the DIYaudio forum recomended tweaks by my friend Col for me. It's sold as a cheap 20W amp but there are a whole raft of tweaks to build it up into a nice low powered amp.

I've also got 2 ta3020 v3c amp (150W) modules from ConnexElectronics. These sound quite stunning. But I broke one of them trying to "make it better" (fried the tripath chip while attemping to upgrade the input caps). I ordered another one and it was D.O.A. They appear to be quite fragile units.. but the sound is amazing. A tad noiser than the HifiMediy unit (which is dead quiet!), but by far the best sounding amp I have heard, in detail. Got this in a steel chassis with a soft start module, and a 500VA toroidal transformer.

Note: I dont have a huge frame of reference, my previous amps have been older Rotel units, and a ART SLA-1, which is a PA/Studio amp that aso plays fine with Audiophile gear. Except the higher noise floor the ta3020 v3c better, fuller bass, more control, more detail... if only I had a working one.

[gainphile]

Remember I'm driving my speakers actively with these amps. This is the most ideal condition that any amplifier can expect (unlike passive).

The output *from each individual drivers* matches what I wanted. It drives my woofers, for example at 20hz with Q = 0.5. That's -3db at 20hz driven to full xmax without clipping (speakers are OBs which also demand much less than boxed speakers).