Options for Cheap ABX Boxes

[BrEpBrEpBrEpBrEp]

As we've seen in other threads, it's easy to set up a simple ABX/AB for DACs/amps (have a relative or friend switch the inputs). Such a test is perhaps suitable for someone looking to tell differences for themselves.

However, there are many possible ways to accidentally break the "blindness" of such a test, and an ABX/AB done as said above does not stand up to scrutiny.

Therefore, I thought it would be good to open a thread in order to collect cheap "ABX switcher" devices and methods, if they are available. That way members would be able to relatively easily set up their own reliable blind testing of physical devices, for fun and profit.

Hopefully we can find some devices, people who own them, and maybe get some measurements (crosstalk at least - should be pretty easy to check) to make sure they're suitable.

As @bboris77 pointed out in another thread, some apparently have audible tells on only one or two of the inputs, which would also disqualify them, so we should check for such issues by ear.

 

[BrEpBrEpBrEpBrEp]

Here is one box to start, that has already been reviewed:

Nobsound MC103 Pro (3in-1out XLR)
https://www.audiosciencereview.com/...und-3-in-1-out-xlr-audio-switch-review.11062/
https://www.audiosciencereview.com/...eardown-of-douk-audio-mc103-pro-switch.11020/

Using balanced connections, crossfeed seems to be at most -120 dB, identical to RME ADI-2 Pro. Seems SNR is degraded from ~130 to ~120 dB when playing white noise in an unused input, so crosstalk between inputs should be inaudible too (correct me if I'm wrong).

 

[BrEpBrEpBrEpBrEp]

It also seems to me that with some relays and a microcontroller you could construct a box that allows a single person to conduct a double-blind ABX on their own. I don't know if such a device exists - if anyone knows of one, please let me know.

 

[Bob-23]

A-B Switch Box For Amplifier Comparisons

© August 2000, Phil Allison, By Rod Elliott​

"Introduction
This is probably going to be a very controversial device. Its purpose is to prove people wrong and that is very confronting. If you don't wish to have your cherished beliefs about amplifiers and audio generally challenged then do not build or use this unit.

Many of you will know about the ABX system for doing audio comparisons. No doubt it is a very fine piece of design but out of reach for the average person. Some years ago I felt that a much simpler device would at least allow me to do comparisons on power amplifiers while the music played in a similar way to ABX. This device was the outcome. After using it for a few seconds my attitude to audio listening tests changed forever. If you are game for a challenge then try it yourself. It will cost you under $50 to build. It may be the best or worst fifty bucks you ever spent depending on your attitude.

The idea is that the listener, you, sits in the 'hot seat' and concentrates on familiar music on your favourite speakers in your own lounge room with the ability swap power amps over without moving more than one finger. If there is even a small change in timbre or definition it should be instantly audible. The stereo image and any other factor can be checked precisely since you can sit totally still during the switchover. Well at least that was what was expected to happen."

Source: Elliott Sound Products (ESP) - Projects List - Last Project
https://sound-au.com/

 

[JSmith]

Not cheap... but looks the real deal;

https://www.dagogo.com/audio-by-van-alstine-abx-comparator-review-part-1-audio-store-wiring/

An similar thread from 2020;

https://www.audiosciencereview.com/...o-you-know-of-any-serious-abx-switcher.14788/



JSmith

 

[SIY]

Why restrict things to ABX? There's a lot of other formats for controlled tests that may, depending on the thing being tested, be easier to implement.

 

[BrEpBrEpBrEpBrEp]

Why restrict things to ABX? There's a lot of other formats for controlled tests that may, depending on the thing being tested, be easier to implement.

Well, I started this thread looking for a cheap off the shelf solution and I figured that ABX was a common test and likelier to have more options. Plus, anything that can ABX can probably also AB/ABC so that's a few options covered.

 

[BrEpBrEpBrEpBrEp]

A-B Switch Box For Amplifier Comparisons

© August 2000, Phil Allison, By Rod Elliott​

"Introduction
This is probably going to be a very controversial device. Its purpose is to prove people wrong and that is very confronting. If you don't wish to have your cherished beliefs about amplifiers and audio generally challenged then do not build or use this unit.

Many of you will know about the ABX system for doing audio comparisons. No doubt it is a very fine piece of design but out of reach for the average person. Some years ago I felt that a much simpler device would at least allow me to do comparisons on power amplifiers while the music played in a similar way to ABX. This device was the outcome. After using it for a few seconds my attitude to audio listening tests changed forever. If you are game for a challenge then try it yourself. It will cost you under $50 to build. It may be the best or worst fifty bucks you ever spent depending on your attitude.

The idea is that the listener, you, sits in the 'hot seat' and concentrates on familiar music on your favourite speakers in your own lounge room with the ability swap power amps over without moving more than one finger. If there is even a small change in timbre or definition it should be instantly audible. The stereo image and any other factor can be checked precisely since you can sit totally still during the switchover. Well at least that was what was expected to happen."

Source: Elliott Sound Products (ESP) - Projects List - Last Project
https://sound-au.com/

https://sound-au.com/absw.htm

Looks like that might be an option for DIY, but only AB sadly, no ABX.

 

[amirm]

I bought the parts to build this but never found the time: https://github.com/oxtoacart/pyabx

I think that is our best shot at building something like this: a program on the computer that controls external relays. The builder from what I recall put some countermeasures against relay click sound by making them toggle like craze or something for a bit which may be an issue.

 

[amirm]

Well, I started this thread looking for a cheap off the shelf solution and I figured that ABX was a common test and likelier to have more options.

ABX boxes were a "thing" years and years ago. These days they are next to impossible to find. Someone should build a cheap one.

 

[amirm]

Not cheap... but looks the real deal;

That's the problem. I was tempted a number of times to get it but $1,000 is just a lot of money for the occasional use.

 

[BrEpBrEpBrEpBrEp]

I bought the parts to build this but never found the time: https://github.com/oxtoacart/pyabx

I think that is our best shot at building something like this: a program on the computer that controls external relays. The builder from what I recall put some countermeasures against relay click sound by making them toggle like craze or something for a bit which may be an issue.

Could you use SSRs to avoid the click, or do those have performance drawbacks that would be apparent in audio? I see a few USB SSR boards for sale online.

Other than that, it seems like it would be ideal if members could build a list of "known good" boards that work for this. That way someone can buy one 'blind' and be relatively confident they're performing a useful ABX test.

 

[Geert]

I bought the parts to build this but never found the time: https://github.com/oxtoacart/pyabx

I think that is our best shot at building something like this: a program on the computer that controls external relays. The builder from what I recall put some countermeasures against relay click sound by making them toggle like craze or something for a bit which may be an issue.

Had a quick look at the code. Seems like it introduces 50 random relay switches between each test. It's easy to adapt the program to make it behave as expected.

 

[Blumlein 88]

Why restrict things to ABX? There's a lot of other formats for controlled tests that may, depending on the thing being tested, be easier to implement.

Could you elaborate? Are you thinking of up-down reversal testing and such?

I think more people would take and be convinced with testing that starts with something clearly audible and works toward the inaudible.

 

[SIY]

Could you elaborate? Are you thinking of up-down reversal testing and such?

I think more people would take and be convinced with testing that starts with something clearly audible and works toward the inaudible.

Sorting immediately came to mind.

 

[BrEpBrEpBrEpBrEp]

Sorting immediately came to mind.

Seems like other tests could trivially be implemented in software with the solution @amirm posted.

 

[Wes]

I suggest thinking of the components required to build an ABX box (or one connected to a computer - e.g. a pseudo-random number generator) and then designing a kit with components that can be easily assembled.

 

[BrEpBrEpBrEpBrEp]

I suggest thinking of the components required to build an ABX box (or one connected to a computer - e.g. a pseudo-random number generator) and then designing a kit with components that can be easily assembled.

I bought the board for PyABX that Amir posted above. It's available on AliExpress for cheap and SainSmart for quite a bit more.

However, PyABX only works with that model of board that has an FTDI USB-Serial chip. Software would need to be rewritten for use with different relay boards.

 

[Wes]

I'm talking about something the average subjectivist could put together...

 

[BrEpBrEpBrEpBrEp]

I'm talking about something the average subjectivist could put together...

Yep, that's why I was originally looking for an off-the-shelf solution. PyABX only requires some stripping of wires and connecting to screw terminals, so I figured I'd buy a board and do a "how to" basic tutorial with photos so that anyone can set it up.

Unfortunately, this still restricts it to people who are willing to strip wires... but when it's in kit form, I think that's about as low as the bar will get.