Reference ABX hardware design and implementation

[voodooless]

Yes and no. Even an a-grade cosmos ADC gets best case -127dB in stereo, has a very low
input impedance, and needs a lot of care and feeding to realise it, with battery power supplies etc. I agree with the principle, but in practice it might just be too hard to standardise. The analog approach is much more plug and play and repeatable. And whatever you or I may think, a lot of people just wouldn’t accept a device for comparing DACs that starts by digitising the signal.

No ADC should be involved, though:

DAC audible differences are finally proven with ABX?

Still remains polarity. Devices have shown to be variable in this regard. Although this device seems to retain it. But DACs can vary. Good point. Polarity switching is easy for balanced inputs (just add more relays), but complicated for single ended, as it requires an inverting buffer.

www.audiosciencereview.com

 

[mcdn]

Quick update before I sign off for the night - full SPICE simulation and Monte Carlo testing is done, resulting in a few adjustments to the feedback/gain network structure. Primarily it no longer uses the stacked nets approach. Gain adjustment and matching still within 0.02dB across the 12dB range.

 

[mcdn]

No ADC should be involved, though:

DAC audible differences are finally proven with ABX?

Still remains polarity. Devices have shown to be variable in this regard. Although this device seems to retain it. But DACs can vary. Good point. Polarity switching is easy for balanced inputs (just add more relays), but complicated for single ended, as it requires an inverting buffer.

www.audiosciencereview.com

Oh yes, sorry, you were talking about upstream adjustment. Perfectly technically valid, but probably not satisfying for most target users.

 

[voodooless]

probably not satisfying for most target users.

Won’t that be the case for any solution you can come up with?

Then start with those two questions: Who are the target users? What would satisfy them?

 

[hardisj]

I don't know how feasible it is but I'll again suggest/request digital inputs. Namely, HDMI.

Speaking for myself, the ability to tell amplifiers apart is pretty much forgone. As I said in the other thread, in every case that I've heard a difference via an ABX test I was able to track it back down to frequency response measurements.

Since I use my ABX testing to provide in my reviews I would really like the ability to also test DACs/Preamps. Thus, my request for something more than what I currently have.

Also, 3 sets of speakers, I think you can cut that to two. If you've done real ABX testing of speakers you know you need the speaker switcher to place the speaker in the center point like Harman did. I don't think anyone here has such a contraption. Otherwise, you can easily tell which speaker is playing just based on where the sound comes from. Even when sat right next to each other (which messes with baffle step), you know it's coming from the right or left. Given that you're the one who set them up, you know. And if someone else set them up, you still know which side it's coming from.

So, *if* it's feasible, I'd love to have HDMI ABX capability.

 

[pma]

As I said in the other thread, in every case that I've heard a difference via an ABX test I was able to track it back down to frequency response measurements

That’s it. Same here.

Re power path, you rather need to switch between 2 amps into one pair of speakers.

 

[ElJaimito]

I'm understand it, but how to deal both camps? Otherwise it will turn out to be a device only for objectivists, who don't really need it.

There's at least four more ABX box designs on DIY audio, just to pick one site. I'm certain there's a lot more out there. AFAIAC such a device has to be above suspicion for the results to be acceptable generally, and that again means no active components in the signal path, whatever it is, high level, low level, and no clicks or bumps in the signal to cue switchover... Otherwise, what's the point? Just asking, obvs. everyone will do what they want! And subjectivists won't believe any of the results anyway. So there's that... perhaps someone somewhere will be convinced.

I take the point about single speaker testing as well. I am aware of various sources that insist (with sound (cough) logic) that single speaker testing is the only way to go for critical analysis, not least our very own Amir! Click-free cue-less switching with high level signals is an interesting trick to master... I don't see a way to do it without active devices and that IMHO defeats the point. With a stereo amp, having two single speakers to compare, with a mono signal bridged to both and low-level switching is achievable and gives a best approximation to one type of testing.

 

[pma]

I'm understand it, but how to deal both camps? Otherwise it will turn out to be a device only for objectivists, who don't really need it.

I do not understand this prejudice. Do you think that the fact that some of us are able to measure and have engineering background results in the fact that we do not listen and have mediocre hearing abilities? You might be quite surprised.

All in all, I do not believe in public projects discussed with all kind of audience. A closed group or blog might be an option. As now it goes to nowhere.

 

[KSTR]

I don't know how feasible it is but I'll again suggest/request digital inputs. Namely, HDMI.

And what does an analog ABX box do with these?

 

[pma]

And what does an analog ABX box do with these?

He is (IMO) submitting his wishes (with sample end price of $2k - $3k) .

 

[KSTR]

Can I suggest using an identical pair of external high quality pre-amps ?
Yes, they're not as good as a resistor network in theory, but they do have a few advantages:

• Available off-the-shelf
• High input impedance
• Low output impedance
• Known performance
• Better bandwidth, noise and distortion than any source.

And how do you want to (remote-)control A/B level balance and before that, each L/R to 0.02dB with these, or let's relax that to 0.05dB (0.5%), for calibration? No preamp I'm aware of would offer that range and precision, let alone remote controlled.
Besides that, it doesn't reduce variables, actually it introduces tons of unknown ones and totally violates the idea of a minimalist precision , signal path.

 

[KSTR]

He is (IMO) submitting his wishes (with sample end price of $2k - $3k) .

I simply don't understand the use case. What do you want to ABX with digital data that can't be handled with foobar's ABX directly (forgiving a few corner cases)?

 

[pma]

I am afraid that I am probably not the one to answer .

 

[KSTR]

associated flyback diodes

Side note on this, just FWIW, most of the time I see this done not ideally, with the typical reverse-biased diode across the coil. Back-EMF then can then create opposing current and slow down the opening, leading to longer arcing and quicker wear-out.
I've learned a different and better way. I protect the switching MOSFET with a Zener rated below the Vds breakdown. And I slightly reduce current slew rate (both for one and off states) with a gate series resistor and Miller cap. This extends relay life-time and also reducesv possible EMC issues because both dv/dt and di/dt are lower.

 

[KSTR]

@mcdn , what about my comment in post #12?

 

[hardisj]

And what does an analog ABX box do with these?

I'm asking for a digital feature as well for HDMI switching.

I'm submitting a wishlist.

Otherwise, frankly, I don't see the point. People seem to be chasing a "better" box just because of one man's test. A test that many - including myself - consider as being flawed based on the practically 100% success rate. Tell me the last time such a diatribe was collected over this ABX box (or any, for that matter).

It's like me saying "damn, these apples taste like oranges" and then everyone trying to make a better tongue just because there's likely something wrong with my tongue.

 

[KSTR]

As the main reason for having active electronics in the signal path is level matching, I would still stand by my initial proposal to do this in the digital domain. If DeltaWave is accepted for doing this, then other digital means should also be a valid option.

Sure, that can be done additionally and independently and does not interfere with this project, as the target is the other, the "source provider" project. As mentioned in the other thread, I'd suggest a small DSP which generates two SPDIF streams with delay adjustment, and A/B, L/R and polarity calibration (when using a UART control port then the ABX box uC could send commands to control the calibration there).

 

[KSTR]

I'm asking for a digital feature as well for HDMI switching.

I'm submitting a wishlist.

Erin, I'm asking the third time now what do you actually want to test, how would the setup, the signal path (in its two domains, digital and analog) and procedure look like. Notably, where is the conversion happening?

 

[KSTR]

Two 10k pots + relay switches - sorry. This is for absolute beginners and not a serious project as intended here.

(also @mcdn)
Well, it shows the basic idea one could build upon by removing its flaws:
- add AC coupling to the inputs
- add precision buffers to drive the level adjustment cells made from much lower impedance element to reduce noise
- restrict control range to 0...-10dB (with a degeneration resistor in the GND leg)
- use 10-gang low-value precision wirewound pots (I personally see no need for automated calibration, only for assisted calibration)
- add output buffer
- the reduced complexity of the attenuators could be re-invested in the buffer circuits
- ....

 

[pma]

... and you have completely new design.