Apologies - that was a short hand. I meant the interaction of the switching frequency with the crossover impedance.
I'll bow out now, recognising that I currently have a limited understanding of the specifics of what you are investigating.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
I'll bow out now, recognising that I currently have a limited understanding of the specifics of what you are investigating.
No, we don’t know for sure what exactly causes this. We do know that multiple speakers have this effect on various amps using the same chip. All of them have a speaker impedance approaching a short at high frequencies.
Solving the impedance solves the issue.
I call that a good enough indication of what is going on here. We can call for more investigation, but unless someone has the equipment to dig deeper, that just remains wishful thinking. So let’s just be pragmatic, and solve the issue as best as possible with the information we have.
Solving the impedance solves the issue.
I call that a good enough indication of what is going on here. We can call for more investigation, but unless someone has the equipment to dig deeper, that just remains wishful thinking. So let’s just be pragmatic, and solve the issue as best as possible with the information we have.
As Voodooless said most probably we already know the issue. I have to put resistor of minimum 1.1 Ohm in series to get rid of that noise sound (and protection engage on higer volume) but it obviously affect audible high frequency. Maybe they stop the production of this amplifier because of this kind of problems and actually this is shame because I still think that this amplifier has superior sound quality (but not in all applications). I have compared it with ADCOM GFA-5802, Advance Acoustic X-i95, Vincent SV-737. Vincent was only one to compare (with 1200$ Marantz CD+Streamer) in term of sound quality but for 10x price. I think that speakers manufactures should have in their minds that Class-D/Full Digital Amplifiers are more common and they should design crossovers more carefully considering implications of speakers behavior. I don't blame Sabaj (it's actually more or less Axign reference design of the circuit) because they couldn't test it with "all of the word" speakers design. I have bought mine from hifi-express for 400$. Now you cannot buy amplifier that could compete for less then 1000$ (this is of course my subjective opinion). I wonder if SMSL VMV A2 (more expensive brother with similar design and 3x price) have any advantage in this area.
P.S. Next amplifier that I would like to try with my Magnepans would be full digital amplifier from Lyngdorf.
Voodooless....... Sound quality with good 1.2 Ohm resistor in series is actually really nice. It's bit less High Frequency but the quality of sound is much better then earlier with those cheap 0.07 euro resistors.
Conclusion.... I should try first 0.01mH coil for minimal audible frequency change ("sound stage"
) and if it wont help i should try some bigger value like 0.04-005mH? Do we all agree?Thank you, @Monterr, Yes... an incremental process/method, isn't it :=)Conclusion.... I should try first 0.01mH coil for minimal audible frequency change ("sound stage"
L (in Series, after Capacitor, before Tweeter/between)=R (at Freq)/(2 x Pi x Freq).... from Freq, Impedance rises (a little bit before) and 6dB/Octave Music Frequency (gentle) roll off
- try 0.01mH/10uH (about Freq/26Khz)
- not enough, try 0.015mH/15uH (about Freq/20Khz)
- not enough, try 0.017mH/17uH (about Freq/18Khz)
- not enough, try 0.018mH/18uH (about Freq/17Khz)
- too much, try 0.011mH/11uH (about Freq/22.5Khz)
- not enough, try 0.017mH/17uH (about Freq/18Khz)
- etc
- not enough, try 0.015mH/15uH (about Freq/20Khz)
- etc
Last edited:
But consider this, the speaker having low/zero impedance at 100kHz frequencies would not be a problem if there was no 100kHz content in the signal and the amplifier clearly should not be generating this stuff but instead just amplifying the audio signal.
So, if this is what is going on, which does seem likely, then the primary fault is the amplifier is emitting non-audio content outside the audio band which we suspect to be an artifact from its switching implementation. This fault becomes exposed when we have a speaker that has essentiually non-existent impedance at that beyond audio frequency so that excrssive current is drawn from the amp but will still be present with other speakers as well, even if that doesn't trip the amplifier.
So, if this is indeed what is going on, the bandaid we need to do is a low pass filter of some sort in front of the tweeter(s) to cut out that 100KHz signal and that is what the inductor, or inductor plus resistor, suggestions above are attempting to do.
I did a test with self made inductor from speaker wire (I didn't measure it yet) and it works !!!
Sound is better than with resistor.
Should I buy some better quality crossover coil or leave it like that?
Attachments
Last edited:
Thank you, @Monterr, Fabulous, Yes (better quality crossover coil), but we need to know the value of the self made inductor. Let us know what the value is....I did a test with self made inductor from speaker wire (I didn't measure it yet) and it works !!!
Sound is better than with resistor.
Should I buy some better quality crossover coil or leave it like that?
If it is 0.01mH/10uH then, as an incremental process/method, you should now try 0.009mH/9uH (about Freq/27Khz) or 0.0085mH/8.5uH (about Freq/28Khz), why?.... if the sound further improves, without affecting the A30a, then this can be considered as further success, can't it? The consideration is to lift the Tweeters minimum Impedance to be above 2ohms, preferablely above 2.5ohms (but only where necessary, as necessary), reasonable? Note that an octave below 26Khz is 13Khz and and an Octave below 28Khz is 14Khz where the R/Impedance of the Inductor will slowly begin to rise and after 26/28Khz quickly rise, reasonable.
Ok... Probably tomorrow I will borrow impedance meter from my brother. I will let you know but I think I'm on the "edge" of the "enough" impedance. I have tried shorter wire coil and loud noise was still there
P.S. Sound is now fabulous... I am shocked how those magnetostatic speakers perform now with Sabaj. And it's "dead silence" even on extreme high volume level.
Thank you, @Monterr, knowing the self made Inductor value will allow this to be a more Scientific Process/Method, rather than Trial & Error, won't it?
Fabulous that you are now enjoying the Magnepan 1.7i speakers with the Sabaj A30a. This suggests that the Sabaj A30a was not the issue, doesn't it? Glad that we have been able to assist you, with the Magnepan 1.7i speaker issue. Note that the Sabaj A30a has very poor (no Lid) Ventilation, so best, at least for the moment, to keep an eye on the Lid's Temperature, isn't it?
As I wrote few posts earlier: " I think that speakers manufactures should have in their minds that Class-D/Full Digital Amplifiers are more common and they should design crossovers more carefully considering implications of speakers behavior. I don't blame Sabaj (it's actually more or less Axign reference design of the circuit) because they couldn't test it with "all of the word" speakers design. I have bought mine from hifi-express for 400$. Now you cannot buy amplifier that could compete for less then 1000$ (this is of course my subjective opinion)"
As of temperature it actually never get hot. It's warmer when you listen quietly (because fan isn't working) then loud. It's always comfortable touch temperature (like maybe 40-45C degree). I don't know if it's possible but it looks like it's now generating less heat with this inductor when it's powered without any music.
Thank you, @Monterr, Yes, the Tweeter Circuits minimum Impedance is now Higher (with the Inductor, R/Impedance, and Riseing after the (set/calculated) Freq), reasonable? The A30a will be Happyer, won't it, and less heat (thermally generated) is an Indicator of Efficiency (Circuit/Implementation, Amp/Speaker), reasonable?
Fortunately @GaNMaster achieved the following with his (early) implementation, which includes the AX5689C, which suggests that the AX5689 (itself) is not an issue....
Last edited:
On the contrary, you should blame the Sabaj because this whole problem is seemingly spurious, audio unrelated, out of band noise from the amplifiers switching circuit being generated entirely within and then being output from the Sabaj.
Arguably it would be nice if the Maggies were built more defensively with a crossover that protected themselves from this sort of abuse but they were designed back when this sort of amplifier design deficiency wasn't so common and now the poor owner is having to work around the amplifier's misdeeds. Ideally this fix of filtering out the unwanted high frequency out of band rubbish would seem better done where it belongs, in the amplifier, not in the speaker.
Thank you, @gwing, would you say that this graph indicates what you are suggesting or other graph....
Last edited:
Interesting, I'm assuming here that this graph relates to the Sabaj we are discussing rather than some other amplifier.
The suggestion that the 140Khz signal originates from the 140Khz switching frequency of the amp doesn't come from me but was offered earlier in the thread, it is however seeming to be a highly plausible theory.
The graph *might* illustrate it, unfortunately the graph is for SINAD not just noise so it is not definitive. It does show a precipitous rise of THD+noise at about 140kHz and *if* that is from noise then yes I would criticise the amp for generating that noise and clearly it should not be paired with a speaker having low impedance at the 140kk+ frequence. On the other hand if the rise is caused by amplification and distortion of the 140 kHz input signal I would be more generous and only say the amp should ideally not be doing it, but if it is only designed to amplify the audio band and only publishes specs fo 20 - 20k or so then we haven't really got any right to damn for it struggling to cleanly amplify frequencies well outside its design range (although t would be more better if the amp filtered out the incoming unwantd high signal rather than amplifying it). In both cases I primarily blame the component generating the 140Khz signal in the first place (which is most likely the amp although this graph doesn't prove it).
** Actually, looking harder at the graph, that says the SINAD measurement is for a 45kHz bandwidth. If that means the input signal only reached 45Khz and there was no 140KHz input then the spike at 140k must be generated in the amp and it would in that case seem clearly proven to be at fault.
There is no spike at 140 kHz, because the graph shows Watts, not frequency.
My bad. Its amazing how the brain can see and read what it expects rather than what is there isn't it!
Thank you, @gwing, there are a few No/s mentioned in your Post....
- 140Khz switching frequency.... No, Switching frequency is about 4 times 140Khz unless you can provide Data/Information that suggests otherwise, reasonable?
- 140 kHz input signal.... No, the Graph indicates Output not Input, reasonable?
- it would be more better if the amp filtered out the incoming unwanted high signal rather than amplifying it.... No, you will need to provide Data/Information that suggests that the Sabaj is not filtering out the incoming unwanted high frequency signal, reasonable?
- I primarily blame the component generating the 140Khz signal.... No, you will need to provide Data/Information that suggests that the Sabaj is Outputing a 140Khz Signal, reasonable?
Update: I've gone back to my 2019 Sabaj A20a amp, and I now prefer it to the A30a.
Similar threads
