I like the sound signature of my ICE 1200AS2 also, which sounds surprisingly good on my 107dB sensitive Radian horns, a combo I tried while my 300B SET amp was being updated. Ultimately, there is no need to use such a powerful amp on such a sensitive horn, and the tubes do give that extra richness and harmonics (coloration for some).
Not really.
In connection with the devellopment of a class D amp of own breed, I bought both nCore 500 OEM with buffer/converterboard and ICE 1200 AS2.
Sound comparison was done for a period of time, but it was pretty clear from the beginning, that the ICE module had more substance, than the nCore, which in comparison sounded more anemic but also very clean.
And that is actually what I hear from others to.
ICE and Pascal never really got a reputation in the DIY communities, mostly because of availability. Their policy is, that they only deliver modules and technology that is to be sold as a complete finished product.
They don´t want to see their modules retailed as is.
As a further consideration on these modules, I don´t like the layout of the nCore. It doesn´t really look as a high finish layout, sorry to say.
ICE has clearly considered a more complete and optimised layout from the very beginning, the same goes for Pascal.
So you are partially affected by aesthetics? Curious. I doubt Hypex would have missed a trick in that area, but could be wrong.
Hello all,
Very interesting thread I see the ICE 1200 mentioned 600 Watts! in the same sentence 107dB @ 1 Watt speakers. Hmmmm! at an average 85dB they would consume about 0.007 Watts and even this verges onto the edge of hearing damage.
Be very careful folks hearing can easily be irreparably damaged. Both IOS and Android have free dB apps the Android one is slightly better and works down to 0.4 sec, both will give you a good average.
Find a dB hearing damage guide Google will find them and use that in conjunction with your dB app.
A 20 Watt amplifier is good enough for any speaker of 84dB or above. If you are unsure about what is actually coming out of your amplifier use a multimeter and check the Amperage output from your power supply; divide by 2 and then take off the efficiency wastage for class D use 10% if you don't know. You will be very surprised!
Cheers
Very interesting thread I see the ICE 1200 mentioned 600 Watts! in the same sentence 107dB @ 1 Watt speakers. Hmmmm! at an average 85dB they would consume about 0.007 Watts and even this verges onto the edge of hearing damage.
Be very careful folks hearing can easily be irreparably damaged. Both IOS and Android have free dB apps the Android one is slightly better and works down to 0.4 sec, both will give you a good average.
Find a dB hearing damage guide Google will find them and use that in conjunction with your dB app.
A 20 Watt amplifier is good enough for any speaker of 84dB or above. If you are unsure about what is actually coming out of your amplifier use a multimeter and check the Amperage output from your power supply; divide by 2 and then take off the efficiency wastage for class D use 10% if you don't know. You will be very surprised!
Cheers
It is not really the aestetics that matter.
It is the logical reason behind the layout.
Placing the speaker output terminals on the converterboard, to which the amplifier module is connected through very thin pin connectors, does not seem to be the best choice to me.
Also the buffer is placed on the converterboard, where the supplies for the amp board also connects.
Basically the beauty of layout is not really a matter of looks, it is a matter of very thorough consideration, which takes into account every detail of performance, production and ease and uniformity of implementation.
And of course to elliminate possibilities for SPOF´s
I hope you understand.
Best regards
Soeren
It is the logical reason behind the layout.
Placing the speaker output terminals on the converterboard, to which the amplifier module is connected through very thin pin connectors, does not seem to be the best choice to me.
Also the buffer is placed on the converterboard, where the supplies for the amp board also connects.
Basically the beauty of layout is not really a matter of looks, it is a matter of very thorough consideration, which takes into account every detail of performance, production and ease and uniformity of implementation.
And of course to elliminate possibilities for SPOF´s
I hope you understand.
Best regards
Soeren
Last edited:
You can see measured performance for the ncore400 and the 1et400a (same modular design as the nc500) over on audiosciencereview. Clearly the layout is not compromising the amplifiers performance.
So yes, you're judging on aesthetics.
Maybe you would prefer it to be done a different way but their way is not making the module sound worse than it should.
So yes, you're judging on aesthetics.
Maybe you would prefer it to be done a different way but their way is not making the module sound worse than it should.
I know they measure very good, but one of the reasons is the low gain.
And I think they will stop measuring that impressive if gain was set to normal values. And then I think the layout might become a problem then.
So IMO you are right, but not really anyway.
The design with that low gain will also need a buffer matching the performance of the module. That will take one of the super op-amps on the market i.e. OPA1612 or like. That done the performance is no longer supreme to all others.
Still very good and among the best, but not superior in the same way.
Designing a discrete buffer providing the missing gain, will rarely match the modules performance.
Talking about eastetics is somewhat difficult, but you are in some sense right about this. I find a thought through design where everything has been taken into consideration beautifull, where others might consider it a peace of fiberboard with small gizmo things on it.
Anyways I find the layout unnecessary complicated with seperate boards, several kontact points and so on. This could IMO have been placed elegantly on a single board.
Just my five cents
And I think they will stop measuring that impressive if gain was set to normal values. And then I think the layout might become a problem then.
So IMO you are right, but not really anyway.
The design with that low gain will also need a buffer matching the performance of the module. That will take one of the super op-amps on the market i.e. OPA1612 or like. That done the performance is no longer supreme to all others.
Still very good and among the best, but not superior in the same way.
Designing a discrete buffer providing the missing gain, will rarely match the modules performance.
Talking about eastetics is somewhat difficult, but you are in some sense right about this. I find a thought through design where everything has been taken into consideration beautifull, where others might consider it a peace of fiberboard with small gizmo things on it.
Anyways I find the layout unnecessary complicated with seperate boards, several kontact points and so on. This could IMO have been placed elegantly on a single board.
Just my five cents
I'm surprised to read through 3 pages of replies with no mention of I2S input amplifiers, in the context of "state of the art" regarding class D amplification.
It seems everyone's wrapped around the axle of "preamp -> analog cable -> power amp" component architecture. Admittedly, with I2S you lose the choice of DAC + any analog signal massaging via tube stage, cable choice or whatever, however I have to believe that among I2S input amplifiers, there's got to be a good one.
It seems all digital signal path is more state of the art - if that means "modern" in a sense - than 110 yr old analog.
It seems everyone's wrapped around the axle of "preamp -> analog cable -> power amp" component architecture. Admittedly, with I2S you lose the choice of DAC + any analog signal massaging via tube stage, cable choice or whatever, however I have to believe that among I2S input amplifiers, there's got to be a good one.
It seems all digital signal path is more state of the art - if that means "modern" in a sense - than 110 yr old analog.
To use the I2S protocol, you´ll need a digital amplifier.
Class D amps are purely analog.
Digital amplifiers sounds like a great idea in most peoples ears, because as you mentioned, you won´t need a DAC.
But there is some trouble in doing digital amplifiers, that might worry most engineers.
First of all, it is not possible to use global feedback, so you need to do it as an open loop amp.
An open loop class D amp will have 0 dB PSRR, which calls for very stable and low noise powersupplies. And as a volume control, you´ll have to use the supply rails. That is how many digital amplifiers are done today.
Modern class D amps are all selfoscillating closed loop types, where the PSRR is very high, in theory almost with no end.
Also noise is distributed over a large frequency range without high amplitude peaks, which is just the other way arround in digital amps.
TI makes a bunch of digital chip amps though, where they use some kind of closed loop. How that works I don´t know, but i.e. Steinway Lyngdorf is designed as I described.
I think that´s the reason for not using I2S inputs directly into a digital amp.
The limitations will amongst other things mean limited power, and high amplitude supersonic noise peaks, which has to be filtered out with an output filter outside fedd back loop, because there is no feedback loop.
That makes the transferfunction of the filter dependant of the speakerload.
For these reasons and even some more, I don´t really fancy digital amps.
And I believe that the overall performance of a good DAC and power amp will outperform digital amps both in powerrange and THD+N, Zout and so on.
Class D amps are purely analog.
Digital amplifiers sounds like a great idea in most peoples ears, because as you mentioned, you won´t need a DAC.
But there is some trouble in doing digital amplifiers, that might worry most engineers.
First of all, it is not possible to use global feedback, so you need to do it as an open loop amp.
An open loop class D amp will have 0 dB PSRR, which calls for very stable and low noise powersupplies. And as a volume control, you´ll have to use the supply rails. That is how many digital amplifiers are done today.
Modern class D amps are all selfoscillating closed loop types, where the PSRR is very high, in theory almost with no end.
Also noise is distributed over a large frequency range without high amplitude peaks, which is just the other way arround in digital amps.
TI makes a bunch of digital chip amps though, where they use some kind of closed loop. How that works I don´t know, but i.e. Steinway Lyngdorf is designed as I described.
I think that´s the reason for not using I2S inputs directly into a digital amp.
The limitations will amongst other things mean limited power, and high amplitude supersonic noise peaks, which has to be filtered out with an output filter outside fedd back loop, because there is no feedback loop.
That makes the transferfunction of the filter dependant of the speakerload.
For these reasons and even some more, I don´t really fancy digital amps.
And I believe that the overall performance of a good DAC and power amp will outperform digital amps both in powerrange and THD+N, Zout and so on.
Didn't Bruno Putzeys write an article that concluded the classic low level DAC + power amplifier was simply an easier way to achieve performance than a "power DAC" such as you describe?
I am sometimes frustrated by the way that so many people just copy whatever everyone else does.
I know it saves effort but people also just copy dumb stuff.
So the lack of "power DACs" may just reflect sheep-like follow-the-leader.
But I suspect it may indicate that it has no real benefit.
On the other hand I am a little surprised there is no evident improvements in Class-D amps.
I would expect that some of the latest SMPS ideas would work well as Class-D.
Stuff like coupled inductors to reduce ripple or resonant zero volt switchers to reduce loss.
I suppose the best Class-D doesn't really need it, distortion and noise already below any practical requirement and acceptable efficiency.
But I also like to see clever new ideas and it feels like there are improvements still to be found.
Best wishes
David
Last edited:
The low gain probably helps but it would be easy to add gain to take it to a typical value with no detectable impact on the distortion.
So I suspect your subjective opinion of "sound quality" may be influenced by your view of the boards.
Hard to believe the measured performance of the Purifi and NCore 400 would create any audible difference.
Of course, I am always ready to be convinced otherwise by a properly conducted (unbiased!) test.
Best wishes
David
To be fair, the 107 dB @ 1 watt @ 1 metre is typically a mid horn and you wouldn't use 600W on it.
But it would probably be matched with a woofer, say 90 dB @ 1 @ 1 so for 105 dB peaks we need ~ 30 W.
And unlikely to be at 1 metre say ~ 3 metres, already at 300 W.
Nice to have the chance to add some equalisation to extent the bass without an enormous cabinet.
And theatre soundtracks have additional LFX boost.
So it adds up.
But yes, watch out for levels.
That power is there for headroom not ear injury.
Best wishes
David
The gainstage done by Hypex triples the noise, and that is one of the Linear Technology super op-amps.
Anyways it is still very quiet, but not revolutionary anymore.
But as I said already, I´ve been comparing both ICE and Hypex alongside.
I clearly prefer the ICE, even though the Hypex has better specs. And I honestly think, that the specs would have greater impact on personal bias than looks would have.
I´m not that visual minded.
My criticism towards the Hypex modules goes more on holistic thoughtfullness.
And now speaking of better specs, I think one should be aware, that for the very good class D designs, noise, distortion and also Zout isn´t really a matter of further development. These specs are so good, that I don´t see any reason to fight for further improvement.
But where else should we then look for possibilities?
Group delay is mostly an unconsidered specification, for some strange reason.
A completely uniform group delay throughout the audioband is actually very important, but I´ve never heard of anyone notice this.
Just my 5 cents
Best from me
Soeren
Anyways nice to discuss these matters with you
Last edited:
Hi Soeren, thanks for the comments.
Triple the noise is unimpressive, what op amp is used?
Of course the tests must be done "blind", that is unbiased, so we are not influenced by looks or prejudice about which one has better specs.
Yes, I commented much the same in the post #30, just before my reply to you.
Better efficiency, less EMI, use less materials - but what I would most like is less complexity, a clever simplification is somehow beautiful.
Surely the amp will be very close to minimum phase up until a reasonable fraction of the switch frequency?
So if we have a flat response then there is no appreciable delay.
An amp that roles off fast just past audio frequency will have predictable minimum phase delay in the audio band, of course.
Best wishes
David
Triple the noise is unimpressive, what op amp is used?
Of course the tests must be done "blind", that is unbiased, so we are not influenced by looks or prejudice about which one has better specs.
Yes, I commented much the same in the post #30, just before my reply to you.
Better efficiency, less EMI, use less materials - but what I would most like is less complexity, a clever simplification is somehow beautiful.
Surely the amp will be very close to minimum phase up until a reasonable fraction of the switch frequency?
So if we have a flat response then there is no appreciable delay.
An amp that roles off fast just past audio frequency will have predictable minimum phase delay in the audio band, of course.
Best wishes
David
Last edited:
Over in the Lounge, "Jim_The_Oldbie" says "I understand the misconceptions that many seem to have about digital volume, and I'm not hearing any degradation in the actual music - it's as lovely as ever."
I tend to agree from my recent experiences using I2S input class D "power DAC" style amplification. Both from TI's TAS line.
Admittedly, I know nothing about just how soft volume control is done, though I have heard of the variable power supply method. Both my I2S amps are on linear supplies with current capability far exceeding what I'll ever pull under normal listening SPLs.
I didnt know about the power supply rejection ratio Achilles heal of these amplifiers.
Thanks Kurt and all for the discussion!
I tend to agree from my recent experiences using I2S input class D "power DAC" style amplification. Both from TI's TAS line.
Admittedly, I know nothing about just how soft volume control is done, though I have heard of the variable power supply method. Both my I2S amps are on linear supplies with current capability far exceeding what I'll ever pull under normal listening SPLs.
I didnt know about the power supply rejection ratio Achilles heal of these amplifiers.
Thanks Kurt and all for the discussion!
Leo of Orchard Audio gave our London Audiocircle a Zoom presentation on his products a few days ago. One of which was a class d amp using Gallium Nitride o/p devices, Not a lot of detail about the modulator tech but it looks conventional ie self oscillating with two feedback loops. Switching freq is 800khz.
Performance looked impressive on paper said to be due to the use of GN but has low gain (16db) and low i/p impedance 5k so places demands on the pre amp, The devices used look like a Texas Instrument part with integrated drivers.
He has offered our group a loan amp so we should get to find out if its as good as the specs.
Ive used Tripath amps in the past and now use a Ucd400 but only as a bass amp. My mid/treble is a SE 300B valve amp. I went for this set up as I found the Ucd a bit wearing in the treble but in combination with the valve works very well for me.
It will be most interesting to see if the Orchard Audio amp solves the problem.
Performance looked impressive on paper said to be due to the use of GN but has low gain (16db) and low i/p impedance 5k so places demands on the pre amp, The devices used look like a Texas Instrument part with integrated drivers.
He has offered our group a loan amp so we should get to find out if its as good as the specs.
Ive used Tripath amps in the past and now use a Ucd400 but only as a bass amp. My mid/treble is a SE 300B valve amp. I went for this set up as I found the Ucd a bit wearing in the treble but in combination with the valve works very well for me.
It will be most interesting to see if the Orchard Audio amp solves the problem.
@Dave Zan
The op-amp used is LM4562.
It is one of the stellar Linear Technology op-amps, which shook the world, and were then taken over by TI.
I agree, blind testing is just the right way.
Simplification ca be beautifull, but sometimes you need a little bit more than that.
Looking just at todays gatedrivers, you´ll probably realise, that nothning related to class D will be simple in our time.
It is a very anti intuitive subject, which takes a lot of math to do right.
To me it is still a technology to admire. The basic idea was crazy, but the refinement has been genius.
The group delay of almost every class D amp, and many other amps as well, is not linear.
Have a look at the datasheets for even the best.
The phase shift should actually be shown as a straight line, with an angle corresponding to wavelength.
Most important thing, it should not bend in the audioband.
Best regards
Soeren
The op-amp used is LM4562.
It is one of the stellar Linear Technology op-amps, which shook the world, and were then taken over by TI.
I agree, blind testing is just the right way.
Simplification ca be beautifull, but sometimes you need a little bit more than that.
Looking just at todays gatedrivers, you´ll probably realise, that nothning related to class D will be simple in our time.
It is a very anti intuitive subject, which takes a lot of math to do right.
To me it is still a technology to admire. The basic idea was crazy, but the refinement has been genius.
The group delay of almost every class D amp, and many other amps as well, is not linear.
Have a look at the datasheets for even the best.
The phase shift should actually be shown as a straight line, with an angle corresponding to wavelength.
Most important thing, it should not bend in the audioband.
Best regards
Soeren
Hi jjasniew
Digital volume controls are many things.
In a preamplifier it is an extra component, creating both noise and distortion.
If the VC is placed directly in the signalpath of a DAC, things are different.
Placed in the DAC signalpath, the DVC can absolutely compete with an analog one. The best DVC´s can sure be good, if the signal is up-sampled to i.e. 32 bit or so.
Crystal Semi has even made a DVC for DSD, which is completely non decimating.
So DVC´s are not all bad.
The open loop class D really has no rejection.
The self oscillating class D, which is always closed loop, has a fantastic PSRR. That goes with the topology.
So closed loop is not always a guarantee for high PSRR, but self oscillating is more or less.
The best of them can run excellent on almost rediculous size PSU´s.
I.e just covering the output power with a crestfactor of 10, and taking the loss in the amp in consideration. That will do in most cases.
And it will not really bring anything good to oversize anyways.
It´s a new world
Eee, do you differentiate between a digital volume control and a digitally controlled volume analog volume control. When you say "In a preamplifier it is an extra component" it seems like you mean e.g one of these: DS1802 Stereo Digital Volume Control ICs? These are all analog signal path but the menas with wich ypu alter the attenuation is by digital control. So no upsampling or anything here.
Then we have the digital volume control in the PCM domain - here the PCM stream is recalculated to represent a lover level - hence, volume. In a DAC this is preformed in the DAC IC or in an FPGA - no special "DVC" IC is used to perform this here.
So, which one is it that you talk about?
//
Then we have the digital volume control in the PCM domain - here the PCM stream is recalculated to represent a lover level - hence, volume. In a DAC this is preformed in the DAC IC or in an FPGA - no special "DVC" IC is used to perform this here.
So, which one is it that you talk about?
//
@TNT
Sorry I wasn´t very precise.
It is the analog chips, that I don´t like.
It all came up because we discussed I2S input poweramplifiers, where volume control is a problem, if you don´t want to loose resolution.
Some of these digital amps use the supply rails as volume control.
Anyways if DVC are of interest, Dustin Foreman of ESS made a nice explanation in this presentation.
http://www.esstech.com/files/3014/4095/4308/digital-vs-analog-volume-control.pdf
Sorry I wasn´t very precise.
It is the analog chips, that I don´t like.
It all came up because we discussed I2S input poweramplifiers, where volume control is a problem, if you don´t want to loose resolution.
Some of these digital amps use the supply rails as volume control.
Anyways if DVC are of interest, Dustin Foreman of ESS made a nice explanation in this presentation.
http://www.esstech.com/files/3014/4095/4308/digital-vs-analog-volume-control.pdf
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.