rha61: PI filtering that is just my kind of solution! Works every time, instead of just parallelling... 😎
Steven344: As an engineer i am more interested in inventing new technology, than protecting the older solutions.
So i would be fully willing to publish my designs. But unfortunately the Class D amplifiers i have on the market at this point are sold to a third party, (L C Audio) and so of course i don't have the rights to publish them.
But i will ask them....! I have them right here on my computer! 😀
They will probably not let me publish the current 3rd generation core, found in ZAPpulse 2.3SE and 700XE. But maybe older ones....
Anyway i have been working a lot looking for the best OPAMP to use in a Class D modulator for audio. And after testing dozens and found the best ones (= low THD) had propagation delays of less than 100nS, i really dont think SG3524 or TL494 will give a good result! 😉 Those ones with higher delays gave a really bad result! Sorry!
Chris: Since the rectifier caps are meant to remove the noise from diode recovery, you don't need any caps when using the FRED's. Caps will do more harm than good....!
Steven344: As an engineer i am more interested in inventing new technology, than protecting the older solutions.
So i would be fully willing to publish my designs. But unfortunately the Class D amplifiers i have on the market at this point are sold to a third party, (L C Audio) and so of course i don't have the rights to publish them.
But i will ask them....! I have them right here on my computer! 😀
They will probably not let me publish the current 3rd generation core, found in ZAPpulse 2.3SE and 700XE. But maybe older ones....
Anyway i have been working a lot looking for the best OPAMP to use in a Class D modulator for audio. And after testing dozens and found the best ones (= low THD) had propagation delays of less than 100nS, i really dont think SG3524 or TL494 will give a good result! 😉 Those ones with higher delays gave a really bad result! Sorry!
Chris: Since the rectifier caps are meant to remove the noise from diode recovery, you don't need any caps when using the FRED's. Caps will do more harm than good....!
Lars Clausen said:
Hi Lars,
Would those be of the LT solutions? They seem lightning fast. You're not going to keep it a secret are you? 🙂
Mr. Pass say's otherwise but I don't think his writings referred to T-networks, so thanks for the tip I think I'll leave the rectifiers alone, am very happy with them as I am with the T-networks, one song I tested with yesterday was just insanely fast like nothing I'd heard yet. Can't even remember which one it was.
Right now I"m back to rooting out EMU 1820M line noise, it sends surges to the input of my amp even with it's output muted, I'd almost rather junk it than use a line conditioner.
Sounds like you've got something in the works for a new amp these days, I think that's great. Look forward to hearing more about it.
Regards,
Chris
Lars Clausen said:
Hi Lars:
Do you recommend the PI filtering for the zap2.3SE modules? If so ...
1. Is a 4mH inductor the size that you would recommend?
2. Do you recommend an aircore, or ferrite core, or either inductor?
Lars, is a FRED bridge better than using a conventional bridge with rectifier caps?
Hi Chris: Thanks for the info. It looks like I had better buy a FRED bridge? The one you are using sounds excellent, so ...
rha61 said:
Hi Alain
Hope you don't mind, just a couple of questions ...
1. Just wondering, how much better was the pi filter when compared to the snubber?
2. How did you decide that 4mH was the right value to use?
3. Why did you choose to use an aircore rather than a ferrite core inductor?
Thanks
LARS, take a look at old class d amp
http://www.angelfire.com/indie/jcnoernberg/digitalcaramp1.pdf
Lars can you take a look at this schematic and tell me what you think,this is an amp from a kit 20 yrs ago ,I wanted to know if you think it can be improved,still works and the sound is good,would like your eng input
http://www.angelfire.com/indie/jcnoernberg/digitalcaramp1.pdf
Lars can you take a look at this schematic and tell me what you think,this is an amp from a kit 20 yrs ago ,I wanted to know if you think it can be improved,still works and the sound is good,would like your eng input
KLe said:
Are you sure a pi filter will work well for a class D amp?
In a class D amp the rail current varies much more than in tube amps or class A amps (tube amps sometimes use large coils to filter out power supply ripple). The CLC filter may form a resonant circuit that gets excited by the varying rail current, this may cause resonances or ringing of the supply voltages. Using an additional R in series with the L may dampen this but still there will be voltage drops when suddenly a large rail current is required. When suddenly a lot of current is required, only the last cap of the CLC can supply that current, it will take some time before the enough current can be supplied via the L.
What I would use is CRC type of networks to improve the filtering, actually I"m using that after I tried simulations with CLC type of networks and gave up because of the resonant behavior. Actually I'm using a number (4) of RCR networks in parallel with R=0.1 Ohm and C=8200uF and after that a single 10.000uF cap. This works great and can suppress switching noise (although less well than a CLC of course) , has a lower series impedance and reacts faster than a CLC network.
Best regards
Gertjan
ghemink said:
i've no problem with the CLC filter , even at a very loud volume (at the limit of the UCD180 on a 4ohms load)
in my case Rs=1 to damp the LC resonance
and the effect of the pi filter is surprising ( beautiful midrange , deep soundstage ...)
i will try your solution to compare
Alain
ghemink said:
Hi Gertjan
Yes, you are right, but, Alain tried the CLC filter (I believe) and said that it was better than the snubber that he tried, and, Lars appears to have given it his approval, also? 😕
So, I thought, perhaps the size of the inductor is important?
Hopefully, Alain and Lars will reply to my posts/questions?
"Actually I'm using a number (4) of RCR networks in parallel with R=0.1 Ohm and C=8200uF and after that a single 10.000uF cap. This works great and can suppress switching noise (although less well than a CLC of course) , has a lower series impedance and reacts faster than a CLC network"
Gertjan, that's is an interesting way of doing the PS filtering ... thankyou for that information ... very much appreciated🙂
What type of bridge rectifier are you using? What are your thoughts on using rectifier caps? I am using rectifier caps at the moment, and, I found that it reduced the zener noise dramatically.
Chris, is using a FRED bridge, and, it would appear that these produce very little zener noise. What are your thoughts on FRED bridges?
KLe said:
They rock 🙂 Mine in particular do seem to be well regarded.
I don't make the claim it's the best because it's not, but it's more than good enough.
Bypass caps as your'e saying are generally a requirement, you didn't make a mistake there.
What kind of supply are you playing with anyway?
Regards,
Chris
classd4sure said:
Hey Chris
PS is 1KVa toriod(40V sec) c/w 2x0.1uF sec filters, 2x35 bridges c/w 0.22uF bridge caps, and 2x10,000uF BHC T-Network filter caps.
I will definitely get a pair of 68 amp 600V FRED rectifiers from IXYS. They sound very impressive and it appears that they do not need rectifier caps?
KLe said:
Nice. "c/w" = come with? I had to ask.
Is that for a monoblock?
I'm a huge fan of "massive overkill" on the rectifiers 🙂 Some say they switch so cleanly that the bypass caps aren't required, others say it pays to further smooth their turn off even more (make the recovery even softer) by bypassing them anyway.
I guess Lars was saying with T-networks you don't need to/probably shouldn't. That would provide better isolation to the secondaries I'm assuming... what we want.
I'm not currently using any and it's got a great sound!
I haven't even yet built my supply to the level you have, just a plain transformer (with mu metal shielding and static shield) dual bridges and a set of T networks which I did tie to ground on their input side. I may try floating it, but the last time I did (non t-networks) it created a small ~2Volt offset. Floating it would be ideal since it would offer full isolation, but I'm not sure if a static shield on it's own would provide enough isolation between primary and secondaries.
Anyone know?
After I find that out and do whatever experimenting I have to do I'll be wiring up the Jensen 4 poles.
I think they were likely invented for SMPS, better efficiency and lower EMI, however I don't think you'd notice as big an audible difference since they aren't directly inline with the filter caps, then again maybe you would?
Regards,
Chris
classd4sure said:
Thanks Chris
Its the PS for 2xZap2.3SE modules. Yep, the PS is floating. Forgot to say that I use a DC blocking cap on the neutral lead.
The sound is now very nice, but, I think that it could still be better. I get very little noise from my speakers, which are 91/92db.
I haven't used a snubber yet, but, I was thinking of trying what Carlos suggested ... 100nF (MKT) and 0.1R resistor + 3.3nF (MKT) cap across each 10KuF PS cap?
I am also interested in what Gertjan has suggested ... using a number (4) of RCR networks in parallel, where R=0.1 Ohm and C=8200uF, prior to the 10000uF cap. Gertjan says that this works great and can suppress switching noise (although less well than a CLC) , has a lower series impedance and reacts faster than a CLC network.
Actually, I found that the T-Network were revealing the bridge zener noise so much, that I had to use rectifier caps (4 per bridge).
Paralleling the Jensen 4 poles should be very interesting ... keep us informed. It would be interesting to parallel the BHC T-network caps, but I only have 2 of them to play with, and I am afraid of them smoking or at worst, exploding.
Kle ,
i used 4mH because that was what i had at home and when testing , it was so effiscient that i let it there
you can see a lot of articles about pi filtering ( CLC or CRC ) on the net
http://www.passlabs.com/downloads/articles/powersupply.pdf
a famous french MosFet amps designer :
http://perso.wanadoo.fr/jm.plantefeve/alim.html
Alain
i used 4mH because that was what i had at home and when testing , it was so effiscient that i let it there
you can see a lot of articles about pi filtering ( CLC or CRC ) on the net
http://www.passlabs.com/downloads/articles/powersupply.pdf
a famous french MosFet amps designer :
http://perso.wanadoo.fr/jm.plantefeve/alim.html
Alain
rha61 said:
Hi Alain
Thanks for the info. I know that CLC's are used for ClassA solid state and tube amps, but, you have used it for classD and it has worked extremely well, so ... its time to have a read.😀
I believe that you are trying Gertjan's idea, using a number (4) of RCR networks in parallel, where R=0.1 Ohm and C=8200uF, prior to the 10000uF cap, to compare with your CLC. Please, keep us informed of the results.
thanks
rha61 said:
Hi Alain
Sorry, can't read French
... do you have anymore info for me to read. Appreciated KLe said:
Hi Alain
Thanks for the info. I know that CLC's are used for ClassA solid state and tube amps, but, you have used it for classD and it has worked extremely well, so ... its time to have a read.😀
I believe that you are trying Gertjan's idea, using a number (4) of RCR networks in parallel, where R=0.1 Ohm and C=8200uF, prior to the 10000uF cap, to compare with your CLC. Please, keep us informed of the results.
thanks
hello all,
Using RCR networks is not my idea. I think I saw it first in the LCaudio supply (the virtual 4 pole supply). However, it is probably a quite common technique as I saw it in more supplies. I use this RCR filtering after my SMPS. It would be interesting to see how the power supply waveform looks with a music load in case of the LC filtering in comparison with the RC filtering. I expect more variations with the LC filtering, although there will be better high frequency noise suppression.
Best regards
Gertjan
- Status
- Not open for further replies.