TSS Two Chassis Super DAC

That is a photo of the analog board of the DAC. Remember, it’s a two box DAC. There will be no USB input on the analog box.

There was limited horizontal space without using a daughter board (or a mezzanine board…) Still I wanted to have at least the inputs that a DS had, but there are also more and more companies using two AES3 inputs in parallel to handle double rate DSD, so I wanted the hardware to be able to handle that without weird adapters…

PS Audio is getting away from using XMOS. This will allow them to much more easily add things like quadrate (no promises on the timeframe, yet). The display processor will have plenty of power to do bridge like functions and USB so we can keep those off of the main board. The network, USB, etc. connectors will be above the power inlet.

The limit isn’t necessarily the DAC, it’s the weakest link of the source, the cable and the receiver. We’ll be using a better quality receiver that should support 192k if the source and cable do.

The random noise on the outputs of analog devices can be averaged out to some extent by paralleling them. Each doubling of the number of devices lowers the (random) noise floor by 3dB. The DS Sr is measurably 6dB quieter than the DS Jr because it has four digital switches compared to one on the Jr. Doubling that again for the TSS should get at least another 3dB of S/N over the DS (there are other things that will lower the TSS noise floor some more.)

None, and (except for mute) that’s the largest number too. :slight_smile:

There are three analog attenuation levels: -10dB, -20dB and -30dB. That should let more systems avoid a preamp if they wish. (The output being the standard 4VRMS for balanced and 2VRMS for single ended should help as well.) The attenuators are implemented as shunts to ground so (arguably) the output signal doesn’t go thru the shunt resistors or the relays that control them.

The other relays are grounding options and hence don’t directly “touch” the signal.

For single ended the barrel can be hooked to the audio output ground for normal unbalanced use or the negative output side of the transformer for balanced use. That seemed like a fair compromise.

Within about 0.1 uS of a failure in the round trip connection between the digital and analog boxes the analog box will mute. It does this in two stages - the software mute starts on the next sample and the hardware mute happens as fast as the relay can change state. (4ms?) There’s also a slower watch dog timer in the analog box that resets the analog box if the digital box is talking non-sense. The analog box is muted while it’s in reset and until the digital box starts making sense again.

4 Likes

Ted gave us pics of both the analog and the digital boards. My questions can be assumed to relate to one or the other based on context – there’s no USB connector visible on the digital board. But there are 2x AES/EBU, space for one coaxial SPDIF, a Toslink and 2x I2S over HDMI.

Dual AES for DSD is interesting, I haven’t come across that yet. Do players like that even matter for the TSS given how “bits-are-bits” it’ll be? But still – it’s among the cooler answers you could have given :slight_smile:

Is that the display processor on the digital board? Fingers crossed the USB Audio device class implementation will be broadly compatible. No offence intended, but PS Audio’s recent track record with non-FPGA software hasn’t exactly been seamless.

Re Toslink I was specifically asking about the receiver component. The one in the DS Sr is only qualified to 96kHz but certain individual components would run successfully at higher rates, like the one in my DAC. I’m hoping that the TSS will use a component that’s qualified all the way to 192.

The output switches, technically analog, but aren’t they just gates that swap between +/- 15V? How much variation do you see on their individual outputs?

Relays – wow, that’s cool. It’s like you’ve built audiophile equipment before… :wink:

And finally, I am mostly reassured about the safety. I guess the right thing for me to do is much like I already do: mostly I turn on the DAC and double check its volume setting before I switch on the power amp.

Thanks yet again.

Ted, I was just wondering if you have considered building an analog domain volume control in the analog box? Could you drop in the tube preamp circuitry for the volume control? This would be an economical proposition for folks like myself that have just one digital input. I know for a fact a lot of audiophiles will NOT use a digital domain volume control as a matter of principle, so having a very good analog domain volume control could be a great marketing move. I believe MSB also has an analog domain volume control on their ultra expenditure DACs and it is part of their value proposition. For me, it would be an ideal architecture. If it is as good as an in line is audio tube preamp of course. Have you thought about this?

The idea is that there isn’t enough bandwidth for double rate DSD over a single AES3 connector, by using two in parallel there is. The funny thing is that some use one for left and the other for right and others interleave them. That’s all software so I don’t have to figure it out yet.

The digital board has a PIC, but it’s not for the display. The display is it’s own device - I give it 5V (from it’s own supply) and it talks to the FPGA and digital board PIC with I2S, SPI, etc.

There isn’t really any such thing. TOSLink doesn’t have options like faster - (it was designed for 44.1 and 48, it was already a stretch to get to 88.2 and 96k.) On the other hand I’ve not seen any of the brand we’re using not perform at 192k.

The variation doesn’t matter - any independent random noise on them can get attenuated by averaging multiple copies. They each are fully differential and I use them in opposite polarities, etc. so the common mode noise on the power supplies should already be attenuated. The differential mode noise should be random and hence amenable to averaging. (Never minding theory the DS Jr and DS Sr show that to be true.)

It’s always good to be paranoid about the output levels before turning on the power amps. But you won’t get a pop worse than starting a track at somewhere other than the beginning if the optical connection fails or resurrects. There are no buffers of that can be played over and over, etc. in the analog board - it handles everything just one sample at a time. And four missing samples in a row causes a mute.

2 Likes

Yes I considered it, but getting a, say, $10K preamp in a $10K DAC for a total of $10K just isn’t going to happen. I suspect the DS’s digital volume control may convert some of the non-believers. Any analog volume control will add more distortion. The biggest problem with the DS’s digital volume control is that there’s a fixed analog noise floor and preamps in general move the noise floor down with the volume. The fixed noise floor in the TSS will be lower than the one in the DS and the TSS will have more options to match the noise floor to the user’s system so the ones that are willing to experiment may well find that the digital control is better for the typical excursions you might do when listening.

Also FWIW the TSS will allow you to losslessly raise the volume of for quiet material as well, that will further ease matching levels with the rest of the system.

2 Likes

Ok. We will just have to rely on the judgement of the maestro to come up with the best architecture.

Another option would be a passive analog volume control. I believe this is what msb uses.

That’s in essence what the attenuators already are and I don’t know how to make them better than a quality shunt resistor. The TSS will have more resistors of better quality than the DS for the attenuators but still running a passive volume control typically down around -20dB or -30dB will never be ideal. And the digital volume control is absolutely better for small changes.

1 Like

Wow, that’s a nice feature to have. There are a ton of albums that I’d love to be able to digitally (and losslessly) raise the volume. Do you know if this is a feature found in any of the popular audio applications (Roon, JRiver Foobar200)? If so, can it done be losslessly there as well?

VLC does it. There’s no reason in principle that any digital path can’t provide gain. The down side is what to do with overflows and other range problems. Some provide compression to keep the loud from getting too loud. Some just crap out. If the source is 16 bits and was dithered to get there the dither noise might become a problem with too much gain.

In the TSS we’ll have a VU meter which (among other things) will show green when the signal after the volume control is below 0dBFS, yellow between there and (probably) 3 or 4dBFS and red above that. The sigma delta modulator implicitly does a soft limit, but if the signal stays too high it will mute rather than clip. Anyway that should help people that might have forgotten that they have the volume higher than 100…

1 Like

Funny thing is a $20 plastic toslink cable would most likely sound better then all that - though you might need a $50 usb - toslink converter also. Bandwidth would be limited though. It’s amazing how inferior the USB input is to the toslink on the DS. Guess that’s why it’s Ted’s input of choice. I wasted a bunch of money on an expensive USB cable - regen etc and an old radio shack toslink cable made it all sound silly - go figure…

2 Likes

Many seem to feel that way - I’m not one of them, even for the 24/96 it’ll handle. And, yes I have couple of good glass cables.

That’s strange - I tried to go back to USB a few weeks ago thinking maybe it was me. Soundstage collapsed and sound hardened and was grainy. Sounded more like my old brooklyn dac. Who knows maybe it’s my source. My Aurender only has 1 USB output. I ordered the Matrix converter so I can try the i2s output. Hoping that will equal or surpass toslink so I can listen to my DSD rips again. Didn’t miss them - sonically at least. Redbook via toslink smoked DSD via usb - for me anyway.

1 Like

For sure, YMMV. And I’m talking IMS (in my system) and to my taste. I may just have never heard a good implementation of it. One issue is the Toslink I/Os are typically very cheap. Hence folks’ interest in the full on fiber method (forget what its called). Descriptors such as “hard/tilted/glassy/hi-fi-ish” always come to mind and I can’t listen to it for long. Plus not doing higher res is a non-starter for me.

When I got into galvanic isolation of devices a few years back, it was brought to my attention that it was inherent in optical, so I figured I could save a lot of expense and hassle…bought good glass cable, etc. but it never sounded right to me. Could be due to my application being from my dedicated Mini server, which may have crap optical. This is despite the fact that I think it was the modder of my Mini that recommended it. Anyways - I got the USB chain sounding good, and that’s been working for me for streaming/server duties. Always prefer DMP when possible.

I know Ted has discussed this a bit before and in essence his opinion is that all of the inputs can sound very good, however designing them to sound good is easier or harder to do depending on which input we’re talking about. He mentioned that I2S is one of the easiest ones to get to sound good. Here is a post where he discusses this. Personally I’d be using toslink if it had the same sample rate/bit depth (and DSD) input options as I2S and USB had. It doesn’t, so I use I2S as my choice of input. I think it sounds better than all of the other input options, though that may reflect on my source component.

Been told by 2 dealers it will be more than $15k based upon what they have been told. $20k is a much more realistic estimate. I would definitely have interest in hearing it but there are some awesome DACs in that price range. At 3X the DSD it would have to be jaw-droppingly good.

JMO.

I think the main thing is at this price it has to be upgradable. The expected sound and the upgradability that’s offered here and nowhere else in this way would be at least a reason for me to never buy another DAC in such a price range.

As the DS has this upgradability, too, it’s a bargain for me.

But I guess even the TSS can’t be made upgradable in terms of other aspects (future sampling rates etc.)

I moved to DS after doing a shoot out between a $2K DAC, a $6K DAC and a $20K DAC years ago. Despite all the hype, I found the differences to be relatively small, and concluded digital technology is progressing so fast SQ is converging and it would never make any sense to spend $20K on a DAC.

However, I am strictly an audio pragmatists (as opposed to dogmatic). If Ted’s wizardry proves me wrong and he can deliver step change improvement with the TSS over the DS, I’m all in. The proof is always in the pudding (i.e. the listening).

1 Like