Tripath EVM experience

[MurrayP]

I posted this on the end of an earlier thread but suspect it didn't get noticed. Should be of interest to anyone thinking of building a Tripath EVM board into an amp. Any comments welcome!!!

Like the others above I have been interested to follow recent developments in switching amps. Tripath and amplifiers using Tripath technology have had great press - and the evm modules are reasonably priced so I decided to give one a go.

Its summer here in NZ and despite a cold summer, 1kW of class A from a Plinius SA250 in my lounge is a bit much, so I though maybe I could try a Class D amp and maybe oust the Plinius and save a few bucks on the way past. My set up is not bad, SACD, TT, good valve pre, SA250 and N801's - and is going rather nicely (at last!). So I'm in a very good position to compare results from the Tripath amp and the Plinius (which I find increasingly hard to fault).

Just to add interest my profession involves me with design of very high power class D amplifiers (to about 750kW) - better known as inverters. The techniques of these are completely similar to those used in making class D work well - at least from an electrical engineering perspective.

I had little difficulty in getting the EVM module going (forget the number but it’s the largest one - about 300+W into 8 ohms). For the sake of peace of mind I fitted the speaker protection components which Tripath leave off the big module (I'm not running it near full power).

The result is a very "nice" sound. Very smooth, nothing offensive makes cds and records that are unlistenable through the Plinius quite pleasant by taking the harshness away. Unfortunately this is definitely a filtering process and the life and energy (air, feeling of energisation in the room) in the music go out the back door too. The bad cds I refer to are bad - it isn't the Plinius mucking them up - it just plays them as they are.

Most of the smoothing effect is in the highs. The mids seem OK but I think maybe recessed. The upper bass is not especially tight (but this may be my power supply not being big enough - I'll borrow 47kuF and add that tomorrow and see what difference it makes). My 801's need very well controlled bass or the 15inch woofers take over the sound - they sound relatively boomy with the Tripath amp.

I've incorporated all tweaks I could find on the net - except input transformers (but I have wound the input coax cable around large common mode cores appropriate for rf suppression at the frequencies used in the amp module).

I'm open to any suggestions as to how to get better sound out of the amp. Please let me know your thoughts!

My gut feeling is that the sound is quite dramatically different to that which I would expect from a normal amplifier, and I don't see that it is likely to be result of the fact that it is class D (Class T is really a marketing name - its a switching amp that shuffles the carrier frequency around and compensates for switching times - nothing very unusual in that - it has been done for years in inverters to stop the ear latching on to the carrier frequency (we use much lower frequencies 2 - 16kHz) and to produce reasonable fidelity output waveforms).

All the indications to me are that Tripath have an algorithm built in to their control chip that processes the sound to achieve the soft sound people like (in fact Tripath actually refer to this at their website). One can never be sure as someone stated above - once manufacturers start processing with DSPs they can do as they wish (just look at the cheap home theatre units with their various sound effects - "hall", "church", "jazz club" etc). This is a hobby horse for me as I recently heard a top end cd player which clearly (to me anyway) does the same thing - it is well know for sounding "live" - but in doing its thing it kills attack and makes electronica sound live too - which it shouldn't!!

Any tweak I have tried with the Tripath EVM has had little influence on the basic sound. Research on the net finds generally that others have found similar results with the Be Canto EVO 200.2 - which tends to point to a character of the Tripath processing rather than a character of the particular implementation (Bel Canto make their own pcbs which are quite different to the EVM modules).

Don't get me wrong - the sound is very good and outside of direct comparisons with hi-end gear or plenty of listening experience I think most people would find it quite beguiling and quite possibly a lot of people may prefer it to other amps whcih play the game more directly down the middle.

Anyway, this is just my experience in my set up. I have no barrow to push, except having invested the money in the kit I would rather have liked the new amp to be replacing my Plinius!

Look forward to comments/feedback/experience/suggestions from others!

MurrayP

 

[Brett]

Hi Murray,

I found your commens about the sound quality very interesting. Some people I know in the states and UK have had good results using the BelCanto and Veritas amps in systems similar to mine, so I'm going to give them a try.

I am looking at building some modules up, but going a slightly different path. I planned to try some TA2020's as I run horns with efficiencies around 110dB/W/m so I don't need a lot of power. The plan was also to quad amp them, 3 x 2020 on top with a 104 for the sub if I like the sound of them. And run the 2020's off a large deep cycle 12V battery I have. Hell of a lot cheaper than 4 tub or Pass amps, and much easier to get clean PSUs with the battery than my dirty mains (should see it on a spec-an!).

Interesting what you said about the algorithms being designed to make them sound 'softer'. One of the things that struck me when I first read the datatsheets was the shape of the noise floor, rolling off smoothly and shallow at both ends. Not unlike most of the tube amps I've ever measured. I feel from my experiences that this is one of the reasons why tube amps and reel to reel recorders might be preferred 'tonally' by some people, because depending on system architecture and gain, often the music information is not that far above the noisefloor. And poorly designed systems, especially some SE designs might actually modulate the PS and noisefloor, causing a harshness and muddiness I and others have heard before, which was rectified sonically by a PS redesign. Also might account for why the BelCanto reportedly sounds a lot better bridged, even driving HE speakers like Avantegardes which would only need a few watts at best (NOT 400), particularly as they have their own active sub below 200Hz or so.

Could your bass boom be a product of the different output Zs of the two amps? I am only familiar with the B&Ws by reputation and am wondering how much their tuning is affected by the amp and cable Z. I recall vaguely reading a range of reviews on them in the mags, and bass quality seems to have varied significantly iirc. Also did you try repositioning the speakers with the Tripath? I know that with my previous Tannoys, the difference in bottom end with a couple of SETs I tried, versus a number of SS designs could be changed a lot by some mnor repositioning. Of course, this may be a particular anomoly of my room / system interface.

I also think the input trannies' advantage lies in isolating the devices, removing any possible return earth currents and HF crap between the pre and power. Probably more of a problem in amps that use feedback.

But in the end I think your preferences might be down to parochialism. After all I've yet to meet a Kiwi where everything NZ isn't the best in the world
Sometimes it is too........hehehehe

Cheers!

 

[subwo1]

MurrayP, did you get the chance to increase the capacitors in the power supply? Do you think that the difference in bass reproduction as compared to your other amp may be due to a lower damping factor? It could be that the lower overall current rating of the outputs hinders their ability to control the inductive load as tightly. The damping factor may also be reduced a little by the very small dead-time needed between the switching of the outputs. Do these ideas seem possible?

 

[MurrayP]

bit more info...

Thanks for the ideas Brett and Subwo1.

In answer, well .... I might have take back my comment about DSP processing. I checked the amp at work today and found the output response to a square wave to show no significant delay 9maybe 1 - 3us but not enough to permit digital filter type algorithms to run). Guess the sound is just part of the territory for this amp. One thing I would warn about is that there is about 3Vrms of filtered pwm (ie a sine wave at whatever switching frequency is in use) on the output which makes measurement quite difficult - but Tripath have an App note about this.....

Absolutely I agree about the output Z's being responsible for the bass differences. More bass was apparent with the Tripath. Whether this good or bad for anyone will depend entirely upon their system - it just isn't accurate I feel, but then I use a valve pre to get the (inaccurate) euphony I like so who am I to judge! The 801's are very sensitive to this and I suspect this is one of the reasons they are often said to have boomy bass. Not so in my room on the Plinius (though i have room mode issues - another topic!). I think bass on the 801's is incredible but you must go by the rules - the cables (and so presumably the output impedance) should not exceed 0.2 ohms according to B&W.

Agreed on the trannies - just don't have any to try.

Parochial ?? A Kiwi?? Never!!! (You’re not an Aussie are you??? - must figure out how to get those damn smilies in there!)

Subwo1 - no haven't increased the caps yet, sorry, but do expect it will tighten up the bass. Will advise when I have tried this. A friend has designs on this amp for his subs so I guess we'll get to grips with how it goes fairly thoroughly in time.

We have an audio group meeting tonite and i hope to run it then so will report what others felt about it if we get a chance.

Rgds

Murray

 

[Brett]

Aussie?......strewth mate, I dunno

On Monday I watched a huge grey kangaroo bound along my street. He obviously didn't find what he was looking for, so he turned around and hopped back up it and on into the scrub. This really happenned, it's not a 'tourist tale', and it delighted my 2 1/2 yo nephew.

Earlier on the same day a large Maori fella delivered some furniture to my home, and told me that this was actually the West Island. He was too big to argue with.

Confusedly yours

PS: I agree with subwo1 about adding extra capacitance to the power supply.

 

[mrfeedback]

Hiya Brett ,
Hey brudda, I heard him white fella, call imself B&O ice power or sumthing or udda, make him switching mode power amps modules, yeah I heard em the udda day over my cousins place,eh.
I bin hear big plenty amps, but this one be bloody good eh, I reckon.
No noise, no colour, just dem koori music, like I heard em up the mission dat time.

A bit of Oz humour - more seriously, don't know the noise shaping characteristics and only relatively brief listen on good speakers and dodgy selection of cds, going through a DSP broadcast audio mixer, the result was nice, clean, polite, etc but uninvolving - music was presented but didn't give any goosebumps.
I'm keen to hear other appraisals of such digtal power amps.

Yeah, ok, Kiwis make ok beer.

MurrayP - 1kW of class A - is that audio output or room heating power - 20c + per hour running cost gets ugly in my experience.

"Just to add interest my profession involves me with design of very high power class D amplifiers (to about 750kW) - better known as inverters. The techniques of these are completely similar to those used in making class D work well - at least from an electrical engineering perspective. "

Hmm, they'd be fun at the next AC/DC show.

 

[MurrayP]

gudday

well, no roos here, just us kiwis. Great to hear from the West Island. Sorry to hear about your airline (bit contentious that one - don't relly think air nz stuffed them up - just stupidly bought into it!).

OK, so now I have 47000uf on each side so we'll see how the bass sounds now. Suspect actually it won't help as it should all be thoroughly decoupled by the amplifiers supply immunity but I'll try anything.... will advise.

Yes have long wanted to take our motor drive products (which have bandwidths to around a few hundred hertz) and run some good subs with them - lets see a three phase 900A per phase inverter should be good for about 200kW left and right channels, leaving another 200kW for a mid channel or maybe up back? Anyone want to volunteer some speakers???

Murray

 

[subwo1]

MurrayP, I look forward to learning the effect of those 4700 uF caps. An idea for your sub drivers would be to use some heavy duty motors coupled to some kind of coupler to drive something like a plywood circle, suspended by a motor cycle inner tube with a little air in it. That should give a few inches excursion each way if you could get the inner tube to stay attached.

 

[Brett]

subwo1 said,

use some heavy duty motors coupled to some kind of coupler to drive something like a plywood circle, suspended by a motor cycle inner tube with a little air in it. That should give a few inches excursion each way

Well lad, it's been done, but not quite the same way you envisaged it. Take a gander at <a href="http://www.servodrive.com/servo/bt7.shtml"> these babies</a>, designed by the great Tom Danley. They are the mother of all PA subs, and cost appropriately. The only person that tried them at home I know of, said they went beyond awesome, but the motor noise was too loud for domestic use. At 140 or so dB, I 'spose it's not an issue. Remember these are designed to be used in groups of four or six per side. Woof.

Murray em lad, you sez

as it should all be thoroughly decoupled by the amplifiers supply immunity

but I suspect thats only relvant for the Tripath module itself. The switching driver fets are still connected directly to the supply rails. So if they aren't stiff enough, during large current draws, the supply will dip like any other amp, compressing the dynamics and soundstage. I doubt the Tripath module compensates the drive to the FETs during these dips, but I bet it is fairly immune to them itself. I think my thinking is justified by the app notes that talk about PSU pumping, and inverting phase on one channel of a stereo amp to minimise this. 'Ave I thunk correct?

Because the amp has the outputs directly connected to the rail, and all the 'hard' work is already done in the chip itself, I think a very, very clean, low Z PSU is a must, that's why I thought about using batteries on the small modules. For a large amp, I would think BIG mono supplies, choke filtered, with good diodes is a must. I'd suggest at least some of the PSU caps were right at the module itself (shortest possible leads)., and very low esr.

My rule of thumb has always been 1 joule of storage per peak watt, per channel. I don't usually have many problems at this level of storage. Not sure how you have yours configured, but you must be getting there.
Oh, better get a good inrush limiter, or a lot of mains fuses. Then the amp can also double as an arc welder.

Cheers

 

[MurrayP]

47,000uF

sorry, didn't help (or make one iota of difference as far as I could hear). Bass from the Tripath evm is still much fuller and rounder than the Plinius, a sure sign of not having quite the same grip.

For interest my best bass test cd is "The Whitlams" No Ahprodisiac single (well 5 versions on the one cd actually). Ok, so the aussies have some good bands..... not including Split Enz or Crowded House who were Kiwis even if Oz did claim them of course! Actually a lot of good bands .. I'm just jealous!

Track 4 has easily the best bass of any cd I know of - rolls out the bottom at about 16Hz (assessed by trying to count the cycles of the floor wobbling - way below hearing it, but the waves in the floor are fun!)

Back to topic - the amp does go low it just isn't very tight - like many others. I suspect with the vast majority of speakers this would not even be noticable.

Incidentally a brief demo at our audio club in a different system showed the same character - nice and clean sound and very easy to listen to really - just falls short of the detail of a top amp.
For the $ of course it is a complete steal. We're talking 5 - 8 times the price for the Plinius. Even a complete DIY would be hard pressed to make so many watts for the $.


One more point - I am only running +/-65V - it was just a handy power supply I had around. maybe at +/- 90V things might get more detailed, tighten up - who knows.

Will post more if anything new comes to light,

murray

ps my wife just commented how our power bill seems to be high.... actually the amp runs around 1kW dissipation in class A but a micro detects lack of use after 20 minures or so and switches it to AB, but it probably still runs a 100W or so and it never gets turned off.

 

[subwo1]

Murray, your thoughts make a lot of sense. One thing I have been thinking about switching amps is that those using overall feedback should be able to provide more control over the load when the power supply has a voltage reserve. Then maybe it can supply extra power to damp the inductive kick of the speaker through the output resistance of the mosfets.

 

[MurrayP]

feedback

yes I agree completely. Actually since the EVM is not destined to stay here a mate who wants an obnoxiously large subwoofer amp has designs on it - but is hoping to close the loop on this amp. He should be able to do it as long as it is rolled off low enough not to come anywhere close to the natural roll of of the amp. Then he will have an incredibly solid dual sub amp I suspect.

It'll be interesting to see how Class D amps evolve - I think Tripath are on to it vs. the alternatives such as TACT which rely on power supplies that are rock solid and regulated to control volume (as far as I know) which must add cost. The virtues of class D include low losses but also low cost done the Tripath way. It has taken class A and AB a long time to get where it is and amps still don't sound the same since none seem perfect. In the latest Stereophile there is a $17k BAT monobloc review - pretty impressive amp but it has an output impedance of around an ohm (all from memory) - so very very obviously these are designed to achieve something, but perfect amplification in engineering terms is not the goal!

Murray

 

[subwo1]

That seems like a good deal. You got to test out the EVM and can add it to your list of experiences.

I have been playing around with class D circuits. I started experimenting with them about 15 years ago. Slow switching speed of the outputs has been a problem which new, more advanced components seems to be helping. But more speed is still needed. The key is high current output transistors, like mosfets, which can switch within 10ns without failing. A new type of transistor may be needed