Hello,
I have two projects that I am working on that may be of interest to some folks and would like to hear your perspectives on.
I am creating two ABx switchers- one for speaker level, the other for line-level. They switch (nearly) instantaneously, and contain zero-crossing (V or I) switching feature to minimize distruptions during switching. Both include double blinding testing capability, and the ability to store an "answer key" and "test results" which can be reviewed after the listening session.
The 1st features mercury displacement relays, silver or gold clad C10100 bus bars, and custom binding posts in silver or gold. It is specifically designed for high power equipment and systems (up to 680Volts / 60A).
The 2nd is specifically for low-power applications. It will provide the same sort of experience, but is a bit complex to explain since it includes both digital and analog low-level switching.
Here are a few questions:
If you were to build this, with "transparency at any cost" as the primary goal, what would you want to be sure it had/did, that would allow you to have faith in the test results?
How would you (if you had the desire) modify those which are already on the market?
Please note: I understand that some believe that ABx switching is inconclusive/untrustworth/faulty/emotionally distressing/etc and wouldn't own or use such a device. No worries if that's the case, I understand that perspective but those discussions seem to quickly spin off into the weeds and I would like to avoid that. If the thought of an ABx switcher causes consternation, please instead think of it as a convenience item, that is not trying to "prove anybody wrong" but rather something that is used to instantaneously switch between two sources- into the same load- purely for convenience. Thanks in advance!
I have two projects that I am working on that may be of interest to some folks and would like to hear your perspectives on.
I am creating two ABx switchers- one for speaker level, the other for line-level. They switch (nearly) instantaneously, and contain zero-crossing (V or I) switching feature to minimize distruptions during switching. Both include double blinding testing capability, and the ability to store an "answer key" and "test results" which can be reviewed after the listening session.
The 1st features mercury displacement relays, silver or gold clad C10100 bus bars, and custom binding posts in silver or gold. It is specifically designed for high power equipment and systems (up to 680Volts / 60A).
The 2nd is specifically for low-power applications. It will provide the same sort of experience, but is a bit complex to explain since it includes both digital and analog low-level switching.
Here are a few questions:
If you were to build this, with "transparency at any cost" as the primary goal, what would you want to be sure it had/did, that would allow you to have faith in the test results?
How would you (if you had the desire) modify those which are already on the market?
Please note: I understand that some believe that ABx switching is inconclusive/untrustworth/faulty/emotionally distressing/etc and wouldn't own or use such a device. No worries if that's the case, I understand that perspective but those discussions seem to quickly spin off into the weeds and I would like to avoid that. If the thought of an ABx switcher causes consternation, please instead think of it as a convenience item, that is not trying to "prove anybody wrong" but rather something that is used to instantaneously switch between two sources- into the same load- purely for convenience. Thanks in advance!
Those mercury contactors are really expensive. They also make quite a noise when they switch which could be a hidden signal as to which relay switched. Zero crossing on those contactors may be a challenge for audio frequencies.
I think the software would be the biggest challenge. It would need to be really flexible to accommodate various testing regimes. The hardware you describe would be pretty foolproof. However, some will complain about heavy connection terminals and others would demand them. Not to mention the battles of wire etc. but I'm sure you know all of that. Good luck with the project. I look forward to pictures.
I think the software would be the biggest challenge. It would need to be really flexible to accommodate various testing regimes. The hardware you describe would be pretty foolproof. However, some will complain about heavy connection terminals and others would demand them. Not to mention the battles of wire etc. but I'm sure you know all of that. Good luck with the project. I look forward to pictures.
AFAIK, mercury displacement relays are SPST, cost $80 and up, switch in 80mS and up, and can not be shipped to some US states.
80mS is a long time in audio. Hard to see how you can zero-cross few-mS audio with a hundred-mS switch.
Hello,
I'm also building the hardware for a ABx device. I'm helped by a friend for the software. One way to use our ABx switch is with a digital source. We are working of to get either a tune flowing during the ABx switches or to have the tune restarted.
I'm using common chinese relay without bigbang noises so far. A cheap 12 relay USB interface controlled by a PC/MAC.
regards,
PF
I'm also building the hardware for a ABx device. I'm helped by a friend for the software. One way to use our ABx switch is with a digital source. We are working of to get either a tune flowing during the ABx switches or to have the tune restarted.
I'm using common chinese relay without bigbang noises so far. A cheap 12 relay USB interface controlled by a PC/MAC.
regards,
PF
I built a simple stereo line-level ABX box with reed relays a long time ago. It has a push button, while the button is pushed you see what the last X was and it selects a new X. The sound is interrupted while the button is pushed, but I don't consider that a problem because it doesn't give you any clue what the next X will be.
The relays make a sound while switching. To make it as difficult as possible to hear what relays switch, I mounted them as close to each other as possible and put them in an
AB
BA
pattern.
Mind you, according to Linear Audio volume 13 reed relays have a measurable distortion, albeit a barely measurable distortion.
The relays make a sound while switching. To make it as difficult as possible to hear what relays switch, I mounted them as close to each other as possible and put them in an
AB
BA
pattern.
Mind you, according to Linear Audio volume 13 reed relays have a measurable distortion, albeit a barely measurable distortion.
Double pole switching, ie A and N switched for each line level input, IOW complete isolation when not engaged.
AC power and earthing point distribution from the ABX test box with integral power filtration (switchable) per individual AC outlet to prevent/reduce AC power 'crosstalk'.
That's a starter for ya.
If they don't have the above they are invalid IMHO.
Dan.
Last edited:
Thanks for the reply. I'm not sure if we talking about the same thing... These are more like relays- and quieter than electromechanicals by a good bit. The manufacturer calls them "contactors", but when I think contactor I think of the big knifeswitch-esque monsters that go "KACHUNG!!" when they're switched... not those. To me, they're just "relays"...
For these, I've used them before and the sound is quite pleasant. (Just a gentle "whooshing" sound.) This can be abated by application of rockwool around the relays, but in the last project with them it was unecessary as all relays were hidden away in a 3RU chassis.
I should clarify the "zero-crossing" claim. Agree completely that those big monsters will be difficult to drive/control precisely at zero. Plan is to simply have a window "close to zero" so that at least it's not quite as big of a step function. I have not yet tested them to see how fast they can really go. If I think about the mechanism of operation- it's dead slow. They literally drop a slug into a pool of molten mercury and wait for the level to go higher than the terminals.
I agree that the software will certainly be a challenge. In fact, to try and abate the slow switching times, I am considering using a look ahead buffer to allow a predictive switching event- but that's a pretty undeveloped thought. It may have zero chance of working, but we'll see. When I did this last time it was all done with analog (and had no smarts regarding storing info/reporting back/etc.). SW is always the pain in the neck...
And... YES, they are expensive. The displacement relays for the prototypes were several, several hundreds of dollars. I have some ideas about how to do it differently, but would require custom machined hardware and tons more development time and money. Maybe one day.
Thanks PRR!
Yes, very valid point. One thing working in our favor is that we're only going to be switching audio content, which is statistically slanted towards low-"ish" frequency content. If we were trying to switch a 16kHz sine at full scale we'd be dead in the water before starting. Still, even at 10ms for a 100Hz signal, I'm going to have to depend on a quiet transient somewhere in the content to just be "low-enough" to make the switch. In the end, I think it will be a trade-off in how much delay one's willing to accept after they press the button.
I actually think the act of turning one set of relays off and then the other on (i.e. preventing shorting the outputs of two amplifiers together) is going to blow the timing budget out of the water. That's the reason I was thinking of the look ahead buffer, but I'm not sure if that is going to buy us anything as slow as the big relays are.
Yes on cost- just stupid expensive. Probably not something a home gamer would use, but would be cheaper than most professional test equipment.
On shipping to US states, haven't heard of that. Definitely is "dead in the water" trying to go into Europe. I've been looking at alternative approaches, but nothing gets me something I can use in the short term faster than these big goofy relays. They are pretty linear (besides some field distortion that was abated by removing the mild-steel mounting bracket and replacing it with an aluminum one.) It is nearly undetectable measured on the AP, but I know some folks will swear they can hear them regardless of what we can measure/explain.
Sounds like a cool project PF! Could you please explain the sentence I marked in your quote above in red? I didn't quite follow.
On the relays, the concerns of course is switch bounce (which will create pretty gnarly harmonics as it bounces around) and the marring of the surface finish of the contacts when hot-switched. Over time, they will corrode and be audible (some would argue they'll be audible even from the jump). If you're ok changing them out as the contacts became faulty maybe it's no big deal? I guess it depends on the end user. I'm guessing that you plan to eliminate a portion of the contact "bounce" with a cap across the relay, right?
What is software is your friend writing his code in?
Seems reasonable. I'm showing my ignorance (I've thus far only listened to tube amps) but what would need to be done different to support this? I've not thought about this much, so forgive me if the answer is ridiculously obvious.
I agree on sound going away while switching not being a big deal. In the big ones previously, I was so concerned about shorting amplifier outputs against one another, I put a very conservative "dead time" in the switching event. It wasn't more than a second, but in audio it was an eternity.
I've also been thinking alot about the "mental stress of ABx listening" that I've read about on this forum. I recall hearing lots of arguments against the test, with listeners reporting that they "... had to live with the gear for a while. Here it on different days, different mind sets..." One way to abate that would be to have the ABx switcher just sit silently, and log information for several weeks. The listener could, at any moment, push a button that indicates "I like this" and behind the scenes the micro could keep up with which amp piece of equipment (A or B) had been playing at the time. Over a 1-2 month period, they wouldn't need to be stressed out- just indicate whether they liked or didn't like their experience for each listening session. Then get the results at the end of the test...
Seems complex, but worth doing to keep ourselves honest- even with ourselves.
Thanks Dan! I actually do the "complete isolation" now (if I understood correctly). Because many of the larger amps are full-bridge out, I have to switch both + and - legs to ensure amps are isolated from one another. Let me know if I misunderstood what you meant.
On the AC power and earthing point you mention. Do you mean, that the ABx switcher should provide "filtered and isolated" AC outlets that the two amps being tested to plug into? This is to prevent the two amps affecting one another via the mains voltage? That's an interesting idea... especially if it's make or break for some folks! It increases complexity quite a bit, but if it's necessary to have "buy-in" of enthusiasts, it should be considered for sure.
Most tube amp with output transformer need to be connected to a load/speaker when power up.
Dummy load for tube amp
Dummy load for tube amp
For a useful data point those relays are used in some surge generators switching 6KV at 3KA. In that service they are good for 500 to 1000 cycles. No audio amp short of a shaker amp for tanks and rockets could create that stress. I suspect they will last a while. The mechanical noise from them switching will be significant. And each may have a distinctive noise (What John Atkinson used to defeat an ABX type test).
If the system is software controlled it would be worth looking into getting a good statistically sound test model or 3 as part of the software. Testing human subjects may need some specific provisions (discarding the first few tests as the subject learns the process etc.). Given the claims in some audiophile circles, making a test cycle that runs for several days with changes in hourly cycles may be meaningful. It would be far less stressful.
Hey Demian,
Thanks for the response! To clarify, I was talking about:
above. I don't believe those are switching 6kV/3Va. I was assuming (incorrectly perhaps?) that common chinese relay with 12V coils wouldn't be used in that application. Audio application can be frustrating because, although the current will flow through the electromechanical relay and sound can be heard, the "sound quality" is reported to be affected long before complete failure of the device. For example, it's my understanding that unpassivated silver can be used in some switchgear because a 6kV/3VA switching would blow through the silver sulfide providing a self cleaning effect. Likewise, "self-cleaning" contacts can abate residue build up from arcing, but arguably not sufficiently to "not be heard". Neither of these two solutions to sulfide/oxide/etc. buildup could be trusted to put to rest the concerns of critical listeners that you want to believe in your product.
I think perhaps only 10-20% of addressing hot swapping issue in electro mechanical relays is maintaining a relatively low resistance connection, another 70-80% is addressing legitimate concerns of non-failure-related compromises that might deter critical listeners from using the product.
That portion is very interesting to me. Is there a writeup/article I can read about that? The design I put together previously always switched, but which relays were turned back on would vary so that the listener could not tell which amplifier was active. The other issue is the electrical signal of the electromechanical switch itself. This can be detected by critical listeners- rather the DUT's reaction to it can sometimes be detected. Addressing switch bounce can minimize (but not eliminate) the harmonics generatred during the "bounce". In the last version of the large switcher, the amplifiers' treatment of the harmonics generated allowed one listener to consistently identify one amplifier from another- and that was with the use of mercury displacement relays!! I had no issue referring to him as a true "critical listener" after that test run.
Agree completely! I didn't think about the discard of a few of the first tests- but think that can be very helpful. The "test that runs for weeks/months" is completely possible. I assume that some "audiophiles" will never be convinced, and that's ok. I just think we can address many more concerns than we are currently addressing- and my tolerance for jumping through hoops is slightly higher now than it was when I was younger.
Last edited:
Thank you and sorry for my bad explanation, my english is poor.
When experiencing ABX tests my panelist required some improvements based (or issued) from the mastering world. In mastering, when a sound engineer need to compare tunes he usually has two choices when switching between A and B. Either let the music flow or when he switch between A and B the music replay the exact pervious tune.
The way of doing is possible only with digital source, it's a common feature in ABX software, softwares are usually designed to check compression, artifacts, or even formats. When using an analog source (who still use that?) the "replay" fonction is lost.
Hope it's a bit more clear.
Concerning the relay, I replaced the chinese relays by high quality relays used in automatic test programmation automates. Try to google "ruthenium relay reed". Such high quality relays will switch MHz signals or nanovolt levels forever in a ABX tester, their performance will not degrade like chinese relay. Their noise is less than a swiss quartz clock. The relay configuration allows me to either "insert" (in and out) any audio component like preamps amps filters almost anything or use two sources like mains cables, players or two "loads" like a speaker or an active monitor. I build connectors adapters depending on the application
The load remain bridged with 12R resistor when not "on air" keeps tubes guru less nervous.
Regards,
PF
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.