Thanks for the help, now I see, why things go wrong, I feel very stupid today, I am used to put snubbers over the speaker, not to ground.
I keep in mind for future.
regards
Hi, I had a strange result with that prior circuit. When I added a bit of dc offset to the grounded pins of the comparator to add a bit of overlap, it increased the crossover distortion, regardless of the direction of the offset.
I observed how the Vswitch drive was high when low was needed for off-state, for one of the Vswitches. That was causing the very bad cross-conduction. That was on the both the prior and most recent circuits.
My thought about the switching of the output going from rail to rail rather than just from rail to output is that the output choke is resetting itself through flyback. I made some changes on the latest circuit to remove the 8kva shoot-through, and the amount of flyback was so great that it was cancelling out almost all of the audio output with a 10khz audio input.. Apparently, the 8kva shoot-through energy had been absorbing the flyback energy prior to that.
That definitely showed some distortion.
Still I have problems, the promised book respons is not there I did try this like crown amp does, but also here strange things happen, the bias adjustment however do work, changing the triangle offset against th audio input signal do give bias, I can even adjust for 70 amp bias
But I do not see the right respons of output stage as a 3 level has to do.
Oops that was a wrong plot.
regards
Attachments
Last edited:
The last pic of crown, here I did again see errors with sim, the nodes where mess up, like comparator output was on mosfet gate, en vica versa was all troubled, the output however was as it has to be.
The IR2110 mosfet output I use was from @Tekko I hope he don't mind.
Distortion of open loop triangle version is quite low. I have use parts of the crown schematic, like the resistors on input comparators and #tekko output.
regards
The IR2110 mosfet output I use was from @Tekko I hope he don't mind.
Distortion of open loop triangle version is quite low. I have use parts of the crown schematic, like the resistors on input comparators and #tekko output.
regards
Attachments
Last edited:
I am not interested in triangle generated D class at the moment .., only in self oscillating, those self oscillating are much better. (I presume so)
I am refering to three loops in relation to self oscilating version, no triangle generattion ..
I do try both, the version we have do work, needs some polish and more, 3 level and self oscillation is quite tricky.
a triangle version is for shure not more bad then self oscillation, with a liniair triangle I get quite nice results, and it do not drop, it stays where it is, and double the frequentcy also, so using 400 Khz it doubles to 800 Khz, with a 3 level, self oscillation do not.
I am on this time do learning about more kinds of technologies, so I need time. and also understand, mine goal is no feedback if possible, using air coils and oil cap lowpass, so we need 3 level for much as possible suppression of carrier, special because of aircoil. keep follow me.
your help is welcome, thanks already for the simulation files.
regards
a triangle version is for shure not more bad then self oscillation, with a liniair triangle I get quite nice results, and it do not drop, it stays where it is, and double the frequentcy also, so using 400 Khz it doubles to 800 Khz, with a 3 level, self oscillation do not.
I am on this time do learning about more kinds of technologies, so I need time. and also understand, mine goal is no feedback if possible, using air coils and oil cap lowpass, so we need 3 level for much as possible suppression of carrier, special because of aircoil. keep follow me.
your help is welcome, thanks already for the simulation files.
regards
Last edited:
- It seems, that STA326/328 DDX are closest to ideal???
-No signal NFB, at least 2 level's modulation, signal synchronous main frequency ....
Or not???
For my ears - STA326 sound very well ... but it's depend from other construction's details ....
Is Abletec/Anaview stuff even around anymore? Only outside the US? I can't seem to much of anything from them or about them.
And is there any real audible difference between the various Hypex UcD offerings: UcD2k, 700, 400, and 180?? Perhaps HxR would make a difference but otherwise shouldn't they all sound about the same? Just more more power or less? What is UcD+?
Anyways, here's my WAG start for an updated list based on various forum comments. Let's keep this limited to DIY options only that do >100wpc into both 4R and 8R loads. Let's also add whatever commentary we can comparing to the best OEM offerings from any Class (A, AB, tube, etc.).
This could be fun.
And is there any real audible difference between the various Hypex UcD offerings: UcD2k, 700, 400, and 180?? Perhaps HxR would make a difference but otherwise shouldn't they all sound about the same? Just more more power or less? What is UcD+?
Anyways, here's my WAG start for an updated list based on various forum comments. Let's keep this limited to DIY options only that do >100wpc into both 4R and 8R loads. Let's also add whatever commentary we can comparing to the best OEM offerings from any Class (A, AB, tube, etc.).
- Hypex NC400 (Have listened to the nCore NAD M27 and it was sublime, but it was most likely the NC500)
The rest are in no particular order: - Hypex UcD2k/700HG/400HG/180HG w/HxR
- TI TPA3255/3251
- Connex CxD500
- Sure IRS2092
- Connex IRS 2092
This could be fun.
For all people, very low distortion do not mean it sounds well, most because of feedback, like this own pic here I did sim, high openloop and you get there, however, it acts like a break on the music, slowing down all kind of things.
Using hoever poles in the feedback path and using low overall feedback and bigger puls output feedback (oscillation loop) do wel.
The muata amplifier is also such a low distortion design and ha capacitor feedback (current) and for frequencies until 1 Khz it has overal feebback, this idea uses much opamps, also something I do not like.
The best transistor amplifier I had, and sounds complete 3D was the alexander based amplifier with feedback, even tubes did not beat it. I did use ringt emittor transistors and super emittor follower, the TIM was very very low, the second HD was also, TIM is nasty, in a class d it looks like we can beat it easely, TIM excist because of that nasty feedback.
I plans to go build such amp, I go made two parts, a full bridge power part with mosfet driver included and a modulator pcb, that way I can test all the versions, I did receive some parts already.
This tread is for diy builds, to see what is best, I did also had the idea to made a open loop, the triangle generator needs to be very high quality, maybe digital generated because then I can adjust linearity.
I need two triangles in fase, one above 0 volts and one above 0 volts, this way I get a very good signal for the 3 level modulation. do someone now how to do this the best way? maybe output of generator into two other opamps decoupled with caps, thanks.
Using hoever poles in the feedback path and using low overall feedback and bigger puls output feedback (oscillation loop) do wel.
The muata amplifier is also such a low distortion design and ha capacitor feedback (current) and for frequencies until 1 Khz it has overal feebback, this idea uses much opamps, also something I do not like.
The best transistor amplifier I had, and sounds complete 3D was the alexander based amplifier with feedback, even tubes did not beat it. I did use ringt emittor transistors and super emittor follower, the TIM was very very low, the second HD was also, TIM is nasty, in a class d it looks like we can beat it easely, TIM excist because of that nasty feedback.
I plans to go build such amp, I go made two parts, a full bridge power part with mosfet driver included and a modulator pcb, that way I can test all the versions, I did receive some parts already.
This tread is for diy builds, to see what is best, I did also had the idea to made a open loop, the triangle generator needs to be very high quality, maybe digital generated because then I can adjust linearity.
I need two triangles in fase, one above 0 volts and one above 0 volts, this way I get a very good signal for the 3 level modulation. do someone now how to do this the best way? maybe output of generator into two other opamps decoupled with caps, thanks.
Attachments
Last edited:
The trianlge wave is not needed to invert, one above and one beneath the zero, I did invert one but is not needed, this because of double sided modulation what occurs with a triangle, a saw tooth is an other cookie who need inverted then.
Using one triangle who is above zero or from zero and up and invert the audio do work also, I use then the half sinus of signals.
Book are a little confusing about 3 level modulation, but it is quite interesting because it doubles the carrier frequency and and make ripple suppression a lot more easy.
If I do see well, the triangle can be used the way as on pic, inversion or not inversed change nothing, because of the double sided modulation, I am right?
Attachments
Last edited:
I am busy with one who has three feedback paths, and a way to let it oscillate very tightly, it can do three level but still do not want wel, I presume that possible the comparators are the cause in ltspice, I use one who is for one supply only, I do not now if the signal for three level is also present on the comparator, strange behavior in spice with it.
Signal is as in pic, oke, has to work as papers say, and the other is when I change the output stage, now suddenly I lost the signal shift in pic two. But I get closer, so I con on learning about class d.
regards
Attachments
This is quite a long thread allready and very few answers the original question like you do: Which Class D amp sounds best? Hats of to you, because of that.
But you could have classified the other amps too.
I have only limited experience with Class D myself. I did a couple of LC ZAP amps, which sounded ok, but very ´Rocky´ (kind of what JBL mostly stands for ).
One of them still sits in a very High End system here, because the new owner didnt find anything he liked better. That was one with dual power supplies.
Then I did LC´s new high end module (Now NewClassD) with discrete opamps in the input, in a dual mono setup. That sounded very good, but were very expensive.
And now I have just finished a Dual Hypex UcD180 with just a single SMPS400/180.
I thougt 400 meant 400 watts, but it doesn´t, and it is a pain in the neck, with all those custom connectors, and no good advice how to connect the things.
(But I found that on DIY Audio).
For me the NewClassD high end modules sounded absolute the best, then UCD180, but not nearly as good, and then the old ZAP modules. I would very much
like to hear the new NC400 modules.
I am running the UcD180 with balanced inputs, they are silent as the grave, but they dont do saxophones that well. That is a good test for listening fatigue.
Best regards
Arthur.
Last edited:
Hi Arthur.
I have never test and hear the modules you did mention, the goal was to get the diy made class d and some ideas to made them etc, maybe the name is wrong, and give confusion, I am afcourse convinced that these modules are wel made and that the people behind it di put all there effort in it, I amd quite new to class D, but after some reading I think it is interesting, special with the new carbide mosfets.
But the lineair hybrid amp I did made in the past with only plate followers (yes I wdid rule out almost every feedback) did sound the best ever, using a 2 mu high current tube and loaded with mu followers and dc coupled to mosfets with a servo chip and the circlotron allfet who did sound not so well as the all plate amp.
But time is a issue because do care for mother .
regards
I have never test and hear the modules you did mention, the goal was to get the diy made class d and some ideas to made them etc, maybe the name is wrong, and give confusion, I am afcourse convinced that these modules are wel made and that the people behind it di put all there effort in it, I amd quite new to class D, but after some reading I think it is interesting, special with the new carbide mosfets.
But the lineair hybrid amp I did made in the past with only plate followers (yes I wdid rule out almost every feedback) did sound the best ever, using a 2 mu high current tube and loaded with mu followers and dc coupled to mosfets with a servo chip and the circlotron allfet who did sound not so well as the all plate amp.
But time is a issue because do care for mother .
regards