They just can't say it, can they?

[Don Richard]

On a side note - if you really want the true holy grail, then the amplifiers should be strapped to the drivers inside the speaker and use the voice coil as the output inductor (the only reason they don't do it in standard amplifiers is because of the radiated emissions requirements).

This sort of thing is what's been done with studio monitors for a while. Straight out of the DAW with AES/EBU digital to the powered monitor. The setup I saw used Genelec monitors.

[Mallette]

Ah, Mike, reading your post is MUSIC! Let me know when you IPO.[H]

Dave

[DesmoDave]

You are of course correct. Re-reading my post I am clearly taking myself too seriously. I will up the medication and endeavour to lighten up.[8-)]

[Chris A]

Is it easier to design a high quality DAC that drives a 10kohm load with <5V, or one that drives a 2 ohm load with >20V?

The highest end converters are operating with noise floors that are limited by quantized energy states of the switches in the converter (24-Bits from 5V is really that crazy). Over-sampling also has a huge list of benefits, and that requires the physical switches to be small (lower capacitance). Adding a requirement of sourcing a lot of power requires the physical parts to be bigger, and therefore your over-sampling rate needs to be reduced. Increasing the rail voltage also requires the parts to get better and that just further complicates the output lattice and ultimately increases the noise; which nets reduced resolution and less over-sampling.

Also, the Digital to Analog conversion process is limited by how clean the power supply is - and unfortunately, this isn't something that can just be iron-fisted to be better. I tried to hint at it earlier, but the lack of power supply rejection on these digital chip amps is a very complicated limitation.

Mike, you lost me here. Are you saying that you are worse off doing all your signal processing in the digital domain (starting with digital source material), then converting to power output stage in one stage? I'm not following-Maybe there is something here that I'm missing.

Additionally, you throw in oversampling in the same conversation. Did you mean to do that - or are you assuming something that I missed again? Are you saying that you have a problem with SNR doing digital-analog in one stage? How about two stages- after all the other signal processing (including preamp/EQ)?

Chris

[Danzo]

It's funny i came across this thread now as i am about to order a Spectron Musician 3 (hardly possible to find a 2nd hand one in Europe);

I was pretty impressed by their design especially their feedback use and power supply isolation,

including 2 seperate low level power supplies; anyone interested can read a bit of Tech Talk on their website, but

you lot probably were already aware of this manufacturer anyways?

Of course, i mean to use this amp with B&W 802M and VSA VR-5HSE, but i doubt i will resist trying

it with Forte 2s also..

By the way, i quite like the sound of Forte 2s driven by Tripath that i picked up for cheap (Panasonic SA-XR700)

after some members here had recommended similar Pannies, it was a good shout! Quad and Croft valves are having a vacation for now.

[DrWho]

Is it easier to design a high quality DAC that drives a 10kohm load with <5V, or one that drives a 2 ohm load with >20V?

The highest end converters are operating with noise floors that are limited by quantized energy states of the switches in the converter (24-Bits from 5V is really that crazy). Over-sampling also has a huge list of benefits, and that requires the physical switches to be small (lower capacitance). Adding a requirement of sourcing a lot of power requires the physical parts to be bigger, and therefore your over-sampling rate needs to be reduced. Increasing the rail voltage also requires the parts to get better and that just further complicates the output lattice and ultimately increases the noise; which nets reduced resolution and less over-sampling.

Also, the Digital to Analog conversion process is limited by how clean the power supply is - and unfortunately, this isn't something that can just be iron-fisted to be better. I tried to hint at it earlier, but the lack of power supply rejection on these digital chip amps is a very complicated limitation.

Mike, you lost me here. Are you saying that you are worse off doing all your signal processing in the digital domain (starting with digital source material), then converting to power output stage in one stage? I'm not following-Maybe there is something here that I'm missing.

The nomenclature here is a bit non-existant, so let me see if the following helps. Basically, I'm trying to differentiate between two types of class D implementations:

1. Digital input to the "amplifier" which then directly creates a PWM signal that is used to drive the gates of the output stage FETs.
2. Analog input driving a comparator stage that creates the PWM signal

My comments were focused on the "digital" approach, which is being touted as the method for keeping the signal digital for as long as possible. Once you use a digitally generated PWM signal, you're basically in the realm of a classic DAC - or in this case, a DAC with enough oompf to drive a speaker load. I am proposing that even in the theoretical world, doing a classical DAC before the (analog) class-D amplifier should yield better results (approach 2) since the requirements for driving a speaker load dictate compromises elsewhere.

Additionally, you throw in oversampling in the same conversation. Did you mean to do that - or are you assuming something that I missed again? Are you saying that you have a problem with SNR doing digital-analog in one stage? How about two stages- after all the other signal processing (including preamp/EQ)?

Ya, the SNR will suffer trying to do the digital-analog in one stage. Basically, a high performance low signal DAC will have better SNR than the digital/analog/speaker drive solution (approach 1). If you split it up into two stages: small signal DAC feeding an analog input class D amp (approach 2), then you should be ablet to achieve a total system performance that is on par with the DAC. In other words, the DAC is the limiting component. I think the assumption that was missing is that the lattice output of a normal DAC is far better than a pwm output equivalent (otherwise converters would be going the pwm route).

[Chris A]

Mike,

Thanks for that explanation. I can't disagree with you on this one. Note that I believe there will probably be two analog amplifier stages needed, at least.

I don't know much about these class "D" integrated amps (or receivers) in terms of their amplifier structure. Also note that the NAD "all digital" device was difficult to understand in terms of the topology of the amplifier section. I'd like to hear one, but note that it would probably take two of these unit on the Jubs plus some sort of digital input active crossover. It's not conducive to supporting bi-amping as for as I could ascertain.

While I believe that, in general, it is better to have fewer stages, it's also important not to lose oneself in the metaphor. If you count the typical number of analog amplification stages going from digital source to speaker terminals, any reduction is probably going to do good things to the apparent fidelity of the reproduction, but trying to do amplification in one DAC-analog amplification stage is probably not the way to try it.

Chris

[LousyTourist]

Erik, it's nice to still see you around here. I am in need of another pair of 2A3 SET amps (something close to the sounding of the Moondogs). Lets me know if you or somebody have a pair for sale. Thanks.

PM sent...

[cjgeraci]

I'll go ahead and say it...my chip amp was sent to the floor, where it is still mopping and moping. A little too sterile.

Of course, given your cache of tube gear, you would say that, wouldn't you? []

Ahem, unlike my good friend Sheltie Dave, there are still some fans of the little chip amps left here in the midwest.

I agree with Mike that the power supplies on the class D amps have a large effect on their sonic signatures. For a number of years, I have been using two little "chip" amps (Super T and Trends) on my horns in my main system, with upgraded power supplies (the KingRex PSU). The little amps provide great sound for my horns, but I am not asking them to do too much. I run active front and back, and the little guys are only being asked to run single drivers (Radian 850s up front and Altec 902s in the rear).

One of these days I will have to try one of the Panasonic amps that started this thread.

Carl.

[Danzo]

My mistake, Panasonics sa xr don't seem to be Tripath.

[Quiet_Hollow]

My mistake, Panasonics sa xr don't seem to be Tripath.

On an interesting note, some models of Panasonic home audio devices (not XR's) did happen to utilize Tripath technology. Most notably, the SA-BX500.

[djk]

They use the equibit technology from TI.

http://www.ti.com/sc/docs/news/2001/01073.htm

Equibit is a digitally controlled Class D electronic amplifier or switching amplifier technology,
developed by Toccata
Industries and first implemented in the TacT Audio
(later Lyngdorf)
amplifiers. Toccata and their Equibit technology was acquired by Texas Instruments in 2000.

[DrWho]

Here's a fun article for you guys that want a switch signal coming straight off the digital audio feed:
http://www.hypex.nl/docs/papers/PaperAESDenmark.pdf

Some more interesting articles:
http://www.irf.com/technical-info/appnotes/an-1071.pd
http://www.hypex.nl/docs/papers/AES118BP.pdf

[DesmoDave]

Mike B - whew! this is what you read for fun!?

Seriously, I have heard a lot of hype around the Hypex amplifier design but I have no idea what it is that makes them different. Are you able to shed some light on what makes the Hypex design special? Has anyone on the forum built one of these amps around an nCore module?

Thanks in advance.

David

[DrWho]

What, you're not finding it a good read? [] Way back when I started posted on this forum, I made it a goal to read a whitepaper every day...which then turned into once a week and now it's about every other week since it's getting harder and harder to find new material. I certainly haven't comprehended it all, haha.

The Hypex design is the only one I know of that takes the feedback after the output filter, so you don't get the speaker specific frequency response shifts.

In plain english, the Hypex is less speaker dependant and will never sound harsh / overly bright the way some of the other options might. Likewise, it will never sound dull or lacking in high frequency extension the way some other other-options might.

The only thing I'd do differently on the Hypex is to use an old school bridge rectified AC power supply rather than the little digital switcher thing he offers. Big and oldschool are a bit against the design elegence of the small size that class D offers, but the way the front end is architected, you're better off with a little more oompf and a quieter supply (it'll help remove some of the hash that shows up in the low level detail during complex passages). It's also helpful to isolate the various channels from each other.

Ironically, I see that he writes about similar ideas in his nCore marketing blurb:
http://www.hypex.nl/docs/papers/ncore%20wp.pdf

Btw, the measurements in that document are crazy insane for an amplifier - especially the IMD. I've only heard the UcD series - I think I might try out one of these nCore amps when they come out. All of the changes he talks about I 100% agree with and my own research has indicated exactly the same things.

If anyone needs help putting together one of his kits, I would be more than willing to help out if you don't mind me listening to it at home for a few hours []

[djk]

I can send you a kit one of these to play with:

IRAUDAMP7S
25W-500W Scalable Output Power
Class D Audio Power Amplifier Reference Design
Using the IRS2092S Protected Digital Audio Driver

You may keep it as long as you need to.