|
Post by puffin on Nov 21, 2017 21:32:24 GMT
I have made several Class D amps over the years. Some will only take a max of 12vdc and so I never bothered much with the PS, using a regulated wall wart or SMPS. These were the early Tripath chips TA2024 (only good for about 5W but ridiculously good....up to a point) TA2020 and latterly TPA3116. I had a TK2050 based board knocking around that hadn't been used for serious applications. This amp can take 12-36vdc, but had been used occasionally with a 19vdc old laptop SMPS.
I thought it might be worth trying it with a decent linear supply. I had a dual supply that I made for some bridged gainclones, but it gave out 38vdc and unregulated. Class D amps are not very happy with overvoltage, but the datasheet said that these had overvoltage protection, so I thought I would give it a go. The PS is rectified with a board from a group buy on PFM and is a CRCRC with 60,000uf of capacitance.
Testing with some old bookshelf speakers and an mp3 player showed that it did not catch fire or give up the ghost, in fact it drove the speakers to silly levels without effort and i had never heard them sound so good.
So it is now doing duties in the main system and sounds very acceptable, a lot gutsier and more MOSFET than Class D usually sounds IMO.
Anyone else running ClassD with a linear PS?
|
|
|
Post by MartinT on Nov 21, 2017 21:58:07 GMT
Yes, on my PC system. I have a rather cute SMSL SA-S3 class T amp being powered by a Paul Hynes SR3 module regulating at 14V. It sounds very nice indeed with my single driver Fostex speakers.
|
|
|
Post by Clive on Nov 21, 2017 22:49:48 GMT
Yes I run Temple amps with LPS and am about to try supercap versions.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 22, 2017 8:52:10 GMT
Without question a decent linear PS is the way to go with PWM amplifier, especially chip amps. My personal amplifier runs 94V rails via a custom 3.5Kva tx the main amplifier module is happy between 45 and 110V. I can change the power output by reprogramming the module (along with the switching frequency, which is much higher than the usual 1/2 a gigahertz) and turning down the rail voltage What most PWM amplifiers using a SMPS do is to make the switching frequency of the SMPS pretty close to that of the amplifier module itself, in theory this cancels out the vast majority of the switching frequency noise (commercial decision there I feel!) Pay care attention to the bridge's make sure they are ultra low switching noise and if your a PFM guy then snubber caps are useful, keep the circuit pathway as short as possible. The trick here is having a very fast switching time (below 32 n/s if possible), yet being able to keep the internal noise down to a absolute minimum, also having a especially clean clean rail supply voltage to the control module is essential, plus keeping the build fully screened as much as possible. Paying very care attention to the zobel network, understanding the exact components required to keep the balance of bandwidth and sound quality not an easy task I can say, From my experience most commercial PWN switching amps have a limited bandwidth of 50Khz +/- 3 dB (there are exceptions) Main smoothing, storage caps I would suggest using 105C as these tend to give a better ripple current rejection and a low tan plus the ratings at max temp are much better, which means at more realistic temperatures (IE around 40-50C) these will have a big improvement with RCR compared to 85C rated components . I have just finished building the latest update of my long term amplifier, and changed to these Kemet's (they bought out BHC a while back) Quality capsHaving changed to the 200V 33000Uf caps, the only issue was to build a wiring harness that incorporated ring eyelets to accept the screw terminals. Interesting project that was, took around a full day to make one up, as you have extra section to produce your 0 volt rail. You guys may not need such large caps, however there is a good selection there to choose from. I would also suggest using a top quality metal film or Teflon capacitor as bypass (depending on your budget)
|
|
|
Post by puffin on Nov 22, 2017 9:46:50 GMT
I had always been a bit wary of running chips at near to max voltage, but I built 2 LM3875 Gainclones P2P and only had a couple of 30vac trannys avaialble which after rectification would give around 42vdc. I built some discrete voltage regs using info from Nick Wetstone's Decibel Dungeon and I have had no problems and they run nice and cool. Am I imagining it or is there a benefit to running at near to max voltage?
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 22, 2017 10:01:33 GMT
For chip amps I would say as long as your regulation is tight and you have solid reserves (fast recharge) then up to 90% of max voltage is fine, over the years I have found that the closer you get to specified limits they just didn't sound as good imho.
Mine unit is not a chip amp (which can be benfical due to the board output stage), due to power requirements my output stage is pure FET.
The best I managed chip amp wise was 88Wrms in stereo and it is bloody good, however just not enough to drive the speakers I use.
|
|
|
Post by tony on Nov 27, 2017 8:19:54 GMT
I got good results running temple mono blocks-sometime four off a big car battery. Lowered the noise floor IMHO.
Never tried a really good linear supply so cant comment.
|
|
|
Post by puffin on Nov 27, 2017 9:18:43 GMT
I have used car batteries in the past to good effect, but I have been really surprised at the difference in bass heft using a good stiff linear supply.
|
|
|
Post by Clive on Nov 27, 2017 9:47:54 GMT
My feeling with Class D is that the PS is even more important than it is with A & A/B. As there's no quiescent current the PS is constantly have to go (switch) flatout from a standing start. Whatever the PS it needs to be good.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 20, 2017 12:11:07 GMT
How are you find the Bass Pump PSU effecting the amp and the over voltage trip response?
I used a common mode choke design after the bulk caps so that as for example the +V rail was getting hammered the neg rail was having volts reduced. Also I found High speed diode reversed biased from output to the power supply helped and not relying on the body diode of the FET,s.
Also not the Bass Pumping is not so bad in Bridge Mode.
|
|
|
Post by puffin on Dec 21, 2017 8:07:43 GMT
Hi Colin, I am not familiar with a Bass Pump PSU can you explain how this works. I have not experienced any over voltage trip using the linear PS.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 21, 2017 10:27:46 GMT
Yes, on class d not bridge the PWM pulse cause either the neg or pos rail to rise at low frequency audio due to non sin wave output, On all and the Tripath early chips this caused a internal protection circuit to shut the chip down. I found a way to solve this by coupling the Pos and the Neg rail after the bulk cap and into the output devices to stop it using a high frequency transformer a few turns on each winding like a common mode choke. On the 300W stereo Class T I did for B&W it worked very well and I also used it on a Class D design for Naim. As I write Larry Reed from Reed Acoustic is designing a 100W version in a PUK using my ideas. Go Larry Go Larry and I both worked for Richer Sound in there sub company designing audio me as sub contractor and Larry as chief designer, fun days I loved it hated the train form Yeovil to Waterloo and return everyday but loved the work all before TQ it now seem ages ago phew getting old here. I can bash a cct up for you after Xmas if you like to play?
|
|
|
Post by puffin on Dec 21, 2017 14:38:25 GMT
Colin, interesting stuff. I was not aware of the rail droops in Class D. Some designs I have seen incorporate a choke, now I know why. If you are not too busy then your offer of a cct would be great. You may know that I know Alan (too long since I popped over to his!) I have had a lot of Richer Sounds/Cambridge stuff that I thought was very good, but a lot of people are a bit sniffy about it. Have a good Xmas. Rob.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 21, 2017 14:52:34 GMT
Hi Rob, say hi to Alan for me I have neglected him for a while very busy and today I was told to chill out by the quack, ticker giving gip again.
OK I pop cct ASAP when I find it PC3 is on the hunt now for it only 10 1TB drives to hunt in. Then PC2 turn oh poor thing. This one PC1 not found
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 22, 2017 17:23:48 GMT
|
|
|
Post by MartinT on Dec 22, 2017 18:26:48 GMT
That's excellent. It claims that PSRR in class D is 0dB, which I interpret as 'very dependent on clean power supply'. If so, it might explain why dirty switched mode supplies are the worst thing to use and why class D responds so well to a good linear supply. Also, the power supply pumping means that the PSU needs to be extra stiff.
|
|
|
Post by puffin on Dec 23, 2017 6:31:25 GMT
Thanks for the links Colin. some Xmas reading material.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 23, 2017 13:18:33 GMT
Try and find notes Sir Clive Sinclair wrote in the early 1970 very interesting reading if I recall, he did a D Class also.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 23, 2017 20:57:15 GMT
On similar lines i always wanted to do one of Audiosector Gain clones but never got around to it. Chip Amp
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Dec 23, 2017 21:15:50 GMT
Interesting, some amazing designs out their and great concepts. I tried Class D with a Class G once, and it was crap. The SECA with a huge Dynamic PSU (tracking) it was very good but the linear PSU had a BW 10 - 20 time wider than the amp to sound good. Also did a SECA with a Mag-Amp PSU at 2MHz it was good also but I always come back to a basic SECA low power lots of heat but the sound is special.
|
|