Ifi Micro iDSD

No. You can connect your source via an analogue in so use it as a normal amp (which is very good too) or you can connect digitally via usb, optical or coax.

I use it every way. Optical in for tv. Usb from computer and IPad (with a camera adapter) or a mini digital from digital out on my Fiio x3. Hi res is brilliant directly from the Fiio.

You can also connect it to active speakers with either analogue or digital in, so it is a pre-amp as well. It has two phono out sockets.

You can also adjust its power range quite considerably, so it is ok for even iem's as well as higher impedance headphones.

It's quite a versatile little amp and I find myself using it more and more with the th900.

It also has this 3d effect which is really nice with headphones. It kind of spreads the sound out and makes headphones sound more open.

I like it a lot.

Well, it sounds damned good whatever I do with it Simon. Mick recommended it to me and I know that he doesn't beat around the bush with these things. It's the best portable that I've had, although it is quite big.

It's pricey, I know, but I'm very glad I got one.

I've been using the Senn hd650 with the Micro and that gets pretty lively too. Very nice. The 3d effect brings the top end in nicely so it's not quite as closed down as it can be. Listening to some Jethro Tull on FLAC is just briliant. Really revealing. The Senn do such a good job.

I do like the Micro enormously. It really is a great amp. It has so many options for plugging in and high quality files are superb. The Nano isn't quite what I'd expected. It's better with higher impedance so that the noise is more under control. It's great with Senns, but I wouldn't walk down the street with them on my head!!

Not sure I'd pay that much for idsd pro. I have a psychological barrier once prices go beyond around £500. I used to have a headphone one of £300 but I've just blown that away though, so ignore me!!

Mick, if you're around ..... Do you use the bit perfect switch for DSD? I prefer the halfway setting on the th900 myself. Bit perfect seems to suck out the bottom. Don't fully get that switch or exactly what the polarity switch does to the sound.

According to the manual of the iFi Micro iDSD (TI BB DSD1793 based) has the following filters:

DSD: Extreme/Extended/
Standard Range (analogue) filters
PCM: Bit-Perfect/Minimum-Phase/Standard (digital)
filters
DXD: Bit-Perfect Processing (fixed) analogue filter


Tip: For PCM we recommend “Bit-Perfect” for listening
and “Standard” for measurements. For DSD, select
Extreme/Extended/Standard to find the one that sounds
best for listening and “Standard Range” for
measurements.

I'll start by saying DXD is 24/352.8 PCM so no idea why they put is a separate category.

Contrary to what their "tip" says I reckon Bit-Perfect should be used only when playing higher than 44.1/48KHz or there will be tons of artifacts as the analogue stage in the DAC cannot filter properly all the digital nasties. It will make sense when feeding either High Res of at least 2x (88.2/96KHz) but preferably 4x (176.4/192KHz) or optimally 8x (352.8/384KHz). With HQPlayer it is possible to upsample to 752.8/768KHz which the Micro accepts and will provide the best results in Bit-Perfect mode.

Minimum Phase and Standard I believe are the usual filters included in many DACs, the former reducing pre ringing and being theoretically more suited to studio recordings whereas the second would be plain vanilla linear interpolation and be more suited to acoustic music like classical. IMO either would be preferable over Bit-Perfect for CD or any other standard resolution signals.

As to DSD analogue filters (Extreme/Extended/Standard Range), they are explained in the DSD1793 data sheet (pages 40 & 41). I guess preference will very depending on input signal, the higher the DSD speed the less need to filter aggressively because the ultrasonic noise is displaced further up the freq range. With regular DSD noise starts to build up at around 30KHz but if upsampled to 2x DSD (DSD128) noise is pushed to 60KHz and filtering can be made less aggressive. The higher the upsampling, 4x (DSD256) or even 8x (DSD512)are supported by the micro, the better performance digital artifact and noise wise. Less aggressive filtering means better phase response.

AFAIK HQPLayer is the only software available that can upsample standard DSD to higher DSD rates without intermediate PCM conversion steps. It is also great for converting PCM of any sampling rate to high rate DSD with much higher accuracy than that of the DSD1793 - remember this chip is Sigma Delta and WILL ALLWAYS convert any PCM to its native SDM DSD like format - which can be sent with out further processing to D to A stage where it will be subject to the analogue filtering mentioned in the previous paragraph.

In later revisions of Foobar's SACD plugin it is possible to  upsample DSD real time but unfortunately it involves an intermediate PCM step: DSD->PCM->DSD which is the reason I didn't include in the Foobar guide. On the other hand it is possible to convert PCM to up to DSD512 with 8 different encoding options.

IMO the way to make the most of the Micro would be using HQPlayer set up to convert everything to DSD256.
As not everyone can afford it (it is cetainly not cheap), next best would be either SOX plugin in Foobar to upsample everything to 192KHz (and use Min. Phase or Standard on the DAC) or the SACD plugin to output either DSD256 or DSD512 and play with different micro modes to check which one sounds best to one's ear.

The "theoritical" advantage of native DXD over lower sampling rates is that there is no processing required, the moment you over/upsample that advantage would be lost as instead of "real" samples you'd be using "interpolated" ones.

Over/upsampling is done automatically by the DAC chips in most converters, usually in three 2x steps. If you feed 44.1 they will oversample 1st x2 to 88.2 then again x2 to 176.4 and again 2x to 352.8. If you provide, say, 88.2 the DAC will only do two steps to achieve the same result and in case of 176.4 it will only perform a single 2x step. This is due to the small processing capacity of the ICs which, unlike computers, can't do it in a single 8x or 4x step. Dither is also controlled by the IC and usually is TPDF.

As NOS is not a good idea from a technical point of view, most manufacturers skip the "Bit-Perfect" option. The micro allows to "bypass" this by using the external filter mode and then providing no external filter. Some people, mostly old R2R DAC owners, say NOS sounds better to them though from a technical POV it makes no sense, I haven't tried it myself as I have no NOS capable DAC.