I find it quite sad to see the Tripath bashers having a field day.
Speaking as a diy beginner, Tripath -- and operations like diyparadise, autocostruire/audiodigit and 41hz -- have been my way of getting into diyaudio with a fairly small investment. I would love to try UCD/NCD etc (and even the passdiy recipes), but the financial layout and technical demands would have been a bit high for me at the beginner stage. Perhaps these above all sound far better than Tripath, perhaps not. I don't know, and I'm keeping an open mind. But what I do know is that my modest 2020 amp sounds far far far better than a Marantz amp I bought 8 years ago for 15x the price and far far better than Class A/AB monoblocks I bought a couple of years ago for 6x the price. Tripath going under is a loss to the diyaudio community.
Speaking as a diy beginner, Tripath -- and operations like diyparadise, autocostruire/audiodigit and 41hz -- have been my way of getting into diyaudio with a fairly small investment. I would love to try UCD/NCD etc (and even the passdiy recipes), but the financial layout and technical demands would have been a bit high for me at the beginner stage. Perhaps these above all sound far better than Tripath, perhaps not. I don't know, and I'm keeping an open mind. But what I do know is that my modest 2020 amp sounds far far far better than a Marantz amp I bought 8 years ago for 15x the price and far far better than Class A/AB monoblocks I bought a couple of years ago for 6x the price. Tripath going under is a loss to the diyaudio community.
Other Class D Sigma-Delta
I came a bit late to the Class D party - as I remember years ago reading about the Sharp SM-SX100 - this rather small, *Revolutionary* amp...another review here...
http://www.stereophile.com/amplificationreviews/253/index3.html
From reading about the Sharp, I bought or built Panasonic (SA-XR25/45/50/57/70), LCAudio (ZAPPulse 2.1), Gainclones (LM3886/75, 4870), Tripath (TA2024/2022/2021, AMP5 & AMP2 - not started yet), Sharp (SD-EX101/111, SD-HX600), and UCD180.
They are mostly gone now, and what has remained in my house is the Panny XR45 (home theater only), the UCD180, and a couple Gainclones and Tripaths.
What do I like best? My UCDs are unbeatable on my maggies, but the most enjoyment I get is my Tripath TA2020 on a simple full range Fostex. Also have yet to start/finish the AMP5 and experiment with the "Guts" of a Moto DCP-501 (5 channels of Tripath TA3020)... Don't know why, but I just like it.
Also gone are my Luxman, Carver, Pioneer Elite, Onkyo (this was hard to part with), which were all sold after rotating through my setup, and playing around with the above amps.
I came a bit late to the Class D party - as I remember years ago reading about the Sharp SM-SX100 - this rather small, *Revolutionary* amp...another review here...
http://www.stereophile.com/amplificationreviews/253/index3.html
From reading about the Sharp, I bought or built Panasonic (SA-XR25/45/50/57/70), LCAudio (ZAPPulse 2.1), Gainclones (LM3886/75, 4870), Tripath (TA2024/2022/2021, AMP5 & AMP2 - not started yet), Sharp (SD-EX101/111, SD-HX600), and UCD180.
They are mostly gone now, and what has remained in my house is the Panny XR45 (home theater only), the UCD180, and a couple Gainclones and Tripaths.
What do I like best? My UCDs are unbeatable on my maggies, but the most enjoyment I get is my Tripath TA2020 on a simple full range Fostex. Also have yet to start/finish the AMP5 and experiment with the "Guts" of a Moto DCP-501 (5 channels of Tripath TA3020)... Don't know why, but I just like it.
Also gone are my Luxman, Carver, Pioneer Elite, Onkyo (this was hard to part with), which were all sold after rotating through my setup, and playing around with the above amps.
This is sad. I wasn't a huge fan of the units, but I thought they had potential. Which is why I bought BWRX's last set of boards, to do some tinkering and work/evaluation, to possibly produce a finished product. We had an app.-- and a desire to go there. For this app, we will have to go in some other direction, now. If we go ahead with it at all. The numbers might have been in the +100k range of released (sold) product/items, at the end of the given run, however many years it may have run.
There is a definite grunge that occurs under heavy load and a false sense of detail that appears under light signal loading. This as the apperance of spaciousness and micro-detail/microdynamics. Load the chip up with complex signals..and it really falls down.
I feel this is due more to the free running oscillator not being properly controlled and the components and layout, etc, all around the chip. If proper stabilization could be achieved, in a manner similar to what one goes through when attemtping to perfect a digital device like a CD player and all of it's issues, then the tripath chips, or class D in general, could really shine. If the system was internally better laid out inside the chip itself, or if it was broken up into two chips maybe, with proper thought given to the issues of making it 'high fidelity' design, then it would likely really show some stiff competition to the best out there. But then again, there is other product of similar design out there to run off and play with.
The problem is the long learning curve for most folks, with regards to this 'gaining perfection' in the given chip's use.
There is a definite grunge that occurs under heavy load and a false sense of detail that appears under light signal loading. This as the apperance of spaciousness and micro-detail/microdynamics. Load the chip up with complex signals..and it really falls down.
I feel this is due more to the free running oscillator not being properly controlled and the components and layout, etc, all around the chip. If proper stabilization could be achieved, in a manner similar to what one goes through when attemtping to perfect a digital device like a CD player and all of it's issues, then the tripath chips, or class D in general, could really shine. If the system was internally better laid out inside the chip itself, or if it was broken up into two chips maybe, with proper thought given to the issues of making it 'high fidelity' design, then it would likely really show some stiff competition to the best out there. But then again, there is other product of similar design out there to run off and play with.
The problem is the long learning curve for most folks, with regards to this 'gaining perfection' in the given chip's use.
thomaseliot said:
thomaseliot, to avoid any mismatching, let's open AD1994.pdf and TA2021.pdf simultaneously, because it's two 2x25W amps for the same cost. Well, dare to compare:
TA2021 <--| who is better |--> AD1994
"Digital revolution" <--
DS analogue modulator+logicTHD@1Khz/10W/14.2V=.035% 4x --> THD@1Khz/10W/15V=.008%
THD@6Khz/1W=.2% 20x--> THD@6Khz/1W=.01%
nois floor 1Khz=-110db 30x--> nois floor 1Khz=-140db
IMD plot ?x--> IMD plot
PSRR=65db = PSRR=65db
Output Offset Voltage 50mV 12x--> Output Offset Voltage 4mV
BTL NOT --> BTL enable
CROSSTALK -80db 10x--> CROSSTALK -100db
Vdd=8.5..14.6V --> Vdd=6.5..22.5V (add 5V required)
IVX said:
I saw the specs for that chip a little while ago and immediately thought to make a board for it until I saw that it only came in that leadless 64 pin package... I have access to tools that can solder that package (solder paste and reflow oven) but even then I couldn't guarantee every pin is soldered properly. It would either work or it wouldn't. Any kind of rework on those things really sucks.
One thing I thought was interesting about the AD1994 was that it was used complementary output devices. I don't know of any other integrated chip that does that.
IVX said:
Hi Ivan,
Tripath has doubled his prices last weeks
I agree with Brian, I'd like too to listen to this AD1994 and see how behaves with Brahms' Requiem (this is my personal most important spec).
What about TK2050?
THD + N: POUT = 30W/Channel, RL = 8Ω VCC = 30V 0.007 %
POUT = 70W Parallel, RL=4Ω VCC = 30V 0.005 %
IHF Intermodulation Distortion 0.05 %
SNR 103 dB
CS Channel Separation 0dB = 10W, RL = 8Ω, f=1kHz 95 dB
TDA1400 has similar specs but at 100W 8ohm.
You'll say that is not fair to compare more powerful amps with AD1994. The point is exactly there: if I like the sound of SDM to what have I to compare them?
P.S.
TK2050 is in the AMP4 kit from 41hz sold for 49 EURO including PS.
Yeah, in the <10W region AD1994 still looks better then TK2050 (THD@1w 1:10), maybe due to bigger mosfets (Qg etc) in the TK2050. I guess that almost all of the forum members understand, there is at least two basic ways to improving formal specs -more loop gain, and more precision of switching. Switching devices (typically mosfet) better and better day by day, so we can just await for 6-18 month, and semiconductors industry will offer at two times faster mosfets for the same Volts & Amperes. The loop gain is the more intelligent stuff, i.e. tricky loop design allows you more loop gain at the minimal swithching frequency, e.g. MUETA 350Khz with huge wideband loop gain, and after filter feedback, hence such amp could have two time less switching with four times bigger dead time, but show much better efficiency, output impedance, S/N and 1:100 THD vs TDA1400 (.002% vs .2% at 6Khz). BTW, MUETA pure analog sigma-delta (hysteresis modulator) also, like IR_D_AMP and many others (tripath included), and BTW2, if i remember correctly, MUETA is dead too, maybe frozen i don't know why. Actually, i completely lost any kind of the sacred trembling at THD .005%, because my class D exercises around tda8939 gave .0003% 1.5 years ago, without any digital revolutions -just a HUGE loop gain+2 month working, and don't forget - i am just a stupid diyer vs smart class D masters, that was founders of this Class D forum.
btw, BWRX, I would NEVER have used any aspects of the boards in production design without informing you of the situation, and taking things from there. The project never matured past the point of me obtaining the boards. I felt there was no use in mentioning anything, unless something came of my tinkering with the chip. Then I would have piped up. I don't like thieves, even if it was not to be seen as such, I do the stand up thing regardless. Ethics. In the end, all we have is our word. I don't play or mess with that.
Pardon me If I'm adressing the wrong guy. Did I get those custom tripath boards from you, or someone else on the forum?
Pardon me If I'm adressing the wrong guy. Did I get those custom tripath boards from you, or someone else on the forum?
IVX said:Actually, i completely lost any kind of the sacred trembling at THD .005%, because my class D exercises around tda8939 gave .0003% 1.5 years ago, without any digital revolutions -just a HUGE loop gain+2 month working, and don't forget - i am just a stupid diyer vs smart class D masters, that was founders of this Class D forum.
Hi Ivan,
As stated before, sigma delta modulation is more difficult to design: this explains worse specs with respect to PWM modulated amps. Maybe this explains too why there are not high power amps with sigma delta modulation except Tripath?
Sometimes I suspect that the number of zeros in THD spec are more important than high resolution in listening experience.
I tell you an anecdote. Some time ago I tried a molypermalloy toroid insted of ferrite, in a Tripath amp output filter: sonic improvement was really really big. A friend of mine reported the experience to a Tripath engineer. He answered that they knew that already, but MPP costs more and has no effects on THD, so they made no further research.
Ciao
Thomas
P.S
Quote from Mueta home page:
"MUETA� is changing the world�s perception of Class-D amplification in hifi products. It is a revolutionary step forward in audio design, not only in terms of efficiency and performance � but also in terms of ease of circuit design and manufacturing simplicity."
Marketing needs are unrelated to real product value. Mueta looks like interesting, have you more informations? I'm reading their pdf now.
Hi Thomas.
At the this forum was some info about MUETA in the 2003-2005 probably, try searching for "MUETA" string.
I can repeat, that if tripath = DS, then MUETA DS too, pretty hard (at least for me) to say, where isn't DS class D selfoscillating amp (clock, hysteresis or delay is no matter) and where is noise shaper, or just a feedback loop
At the this forum was some info about MUETA in the 2003-2005 probably, try searching for "MUETA" string.
I can repeat, that if tripath = DS, then MUETA DS too, pretty hard (at least for me) to say, where isn't DS class D selfoscillating amp (clock, hysteresis or delay is no matter) and where is noise shaper, or just a feedback loop
thomaseliot said:
Ahh. Never even made it that far into looking at the spec and design sheet. That possibly/maybe explains the issue of cloudy and 'greyish dirty sound' under heavy signal loading. Basically, it 'congests'. Methinks it's due to the Delta-sigma design. Note the sensitivity of the unit to transient current delivery issues. Extremely fast caps would be the saviour here, and the EXACT right mix of bypass vs main PS cap in terms of value mixes.
Change the bypass in value and type...listen to the same musical passage again. Remove. Clip next cap in. When the congestion is at the lowest on the diffcult passage, you then have your best combination of PS caps...for that chip and components/layout. Might be worth a test. You are dealing with potential resonant cavity issues here, in terms of how they might be viewed. You have to stabilize and negate the resonance and the harmonics at the same time you speed up transient current delivery. Ear tests work best here. Similar to standard audio rails but a different resonace is being dealt with. That freerunning oscillator cap becomes a serious issue as well. Square wave devices go through every single point of a given aspect of signal propogation in terms of RF and potential/mass, this, twice every single hz of the sample rate, whatever that may be.
This has a serious affect on, or, it is seriously affected by the limitations of the PS caps. Meaning, they 'see' every single aspect of the Cap's maximum limitations. This becomes the 'quality' of the signal. I'm not an engineer, but these things should be obvious.
This also goes for regular D/S chips in D/A conversion as well.
I completley redid (totally modded out) a Super T amp, the metal chassis one..and believe it or not... I threw a 2.5KVA balanced AC transformer on it....the believe it or not part..was the quality difference was NOT, I mean NOT subtle. You could easily hear the change in the quality of sound.. outside the room.
Try it. The AC power line noise..comes through, or it is neutral noise. Try it, it will blow your mind. Not subtle. Even guys who have never heard a bloody single piece of audio gear other than Sears and Wallmart crap, could pick it out 100% of the time in a blind test. Then figure out why.
I thought it would result in a big fat zero, ie 'no change' ...but boy...was I suprised!
I've experienced dramatic changes in sound with different supplies varying from small and large batteries (lead and nicad) to transformer based supplies in various sizes and SMPS in various voltage and sizes.
This was tested on various amps (TA2024, TA2021B, TA2020 and TAA4100A based...)
The SMPS was preferred most of the times, I believe they work with 'fast' caps most of the times, don't they?
This was tested on various amps (TA2024, TA2021B, TA2020 and TAA4100A based...)
The SMPS was preferred most of the times, I believe they work with 'fast' caps most of the times, don't they?
Their 'pulse' charge rate or slight rail 'noise' comes through, in my expereince with trying every single type of PS supply system under the sun. I've not spent much time on it in terms of figuring it out, but the balanced AC result was the most obvious change for the good...of them all. If you use a SMPS, the quality difference of the balanced AC is likely to still be there, for the most part, is my best guess.
As for the free running oscillator, it would come out as noise very much like clock jitter, under heavy loading, as the 'rate of change' per enacted bit is greater, thus highlighting the jitter aspects. Low levels of change stabilize the PS (or more correctly, press it less into it's various types of harmonic and current stability related distortions, which would equate to ..jitter, as it affects placement of the given 'bit') and the jitter become less of a factor. Both play together. Stabilize that oscillator!! Make sense?
basically..low load means less affecting of the two important components..the PS and the oscillator..and heavy loading affects both..thus a notable congestion in the sound. A seemingly non linear aspect to it, due to the doubling up and interconnectedness of the issues.
i looked at the schematic and design points, and immediately recognized that an inordinate amount of time spent on the oscillator would likely be worthwhile and rewarding, and at the same time realized that a secondary concern was the maximal quality aspect of the power supply..was paramount to getting a good sound out of the chips. Low voltage, high current, thus the caps are stressed to their very transient current draw limits in various and complex ways, and become a major factor in the sonics.
As for the free running oscillator, it would come out as noise very much like clock jitter, under heavy loading, as the 'rate of change' per enacted bit is greater, thus highlighting the jitter aspects. Low levels of change stabilize the PS (or more correctly, press it less into it's various types of harmonic and current stability related distortions, which would equate to ..jitter, as it affects placement of the given 'bit') and the jitter become less of a factor. Both play together. Stabilize that oscillator!! Make sense?
basically..low load means less affecting of the two important components..the PS and the oscillator..and heavy loading affects both..thus a notable congestion in the sound. A seemingly non linear aspect to it, due to the doubling up and interconnectedness of the issues.
i looked at the schematic and design points, and immediately recognized that an inordinate amount of time spent on the oscillator would likely be worthwhile and rewarding, and at the same time realized that a secondary concern was the maximal quality aspect of the power supply..was paramount to getting a good sound out of the chips. Low voltage, high current, thus the caps are stressed to their very transient current draw limits in various and complex ways, and become a major factor in the sonics.
It started out by brainstorming about sigma-delta modulation and Tripath chips being more sensitive to the quality of the power supply because of using this type of modulation. Therefore I don't think it's that offtopic to discuss this in order to perhaps find a matching alternative to Tripath chips.....
Thanks for pointing this out Thomas,KBK, and IVX, no offense to you offcourse Panomaniac....
Thanks for pointing this out Thomas,KBK, and IVX, no offense to you offcourse Panomaniac....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.