Filterless Class D

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
a friend of mine (pitch254 i think his nick is) has, a tas50xx based class d digital amp.

Tweeters managed perfectly, no probs there. Only thing was that it gave so much hf junk on the speaker lines & the internal wiring that it had trouble locking on the spdif signal when turning up te volume. even at normal listening volumes it would sometimes lose the signal.

That, and the tv & radio were completely useless when the amp was playing music :smash:
 
I've been measure a 2W stereo filterless class-D chip in the EMI anechonic chamber, a pair of speaker cable is only 30cm, the EMI radiation result is really bad even in the quiescence and supply voltage is only 5V, don't even think about rated power output.

Filterless type class-d is only suite for "no speaker cable outside" applications , You needs to install all componenents in a shielded case, such like a laytop PC or mobilephone.
 
Filterless really is a low-cost solution for speakers next to the Class D chip, like in mobile phones, playstation portable etc.. Still required are suitable ferrrite beads to control the electrical noise.

As others write, if you send the PWM down several metres of cable, you'll be wiping out AM and FM broadcasts in the area!

Also important is the fact that the voice coil of the speaker is the filter inductor. If its magnetic properties don't properly extend up to the PWM frequency, then the losses become huge and it will get very hot. Putting the PWM into a crossover would be quite entertaining for the same reasons.

So best is to have a top-quality inductor-based filter next to the PWM chip. That's not as easy as it sounds and there are plenty of examples of how not to do it out there!
 
I'm working on an active speaker.
I plan to wire a filterless output Tripath based amp (TA2020) directly to my tweeter using about 50cm twisted-pair shielded cable.
Same configuration for the woofer (filterless TA2020 connected with 50 cm twisted-pair shielded cable).

I don't want to use output filter in order to optimize sound quality.

First question is : will this amp fail without filter ?
Second : will the EMI radiation be acceptable ?

Thanks !
 
Interesting that you bring this up. I was reading this document recently. It states that tweeters specifically may have issues with a filterless output due to low inductance.

I took the zobel out on a TA2024 Trends 10.1 and changed the caps with some Vishay MKT caps and Sonicap Gen2 caps. The Gen2's did not sound very good to me, so I kept the MKT caps in.

I'm a little worried about the zobel since a separate document raised issues with damage to the chip if no speaker load is present. I guess I'll have to be more careful. I basically cut down the output filter from 6 caps and a resistor per channel to just 3 caps (and 2 inductors). I want to try a half filter so that I can cut it down even further to just 1 inductor and 1 cap per channel at some point, but I won't be going filterless.

The Audiodigit amp installation manual also states that you can cut out the diodes for a minimalist installation. It doesn't seem like a very good idea after reading Tripath documentation, but I suppose somebody has tried it with some luck.
 
In my experience, anything over about 5W should be filtered from an audio Point of View (EMI says otherwise) if it is intended to be a full range amplification channel. The sound starts to get really harsh due to the high frequency switching signal present. That being said, I would still recommend a BTL filterless amp with a filter, before any single ended emp. Due to the PWM (with no audio signal) being in phase, a lot of the high frequency is common mode and won't be seen across the load. This would give you better signal to noise performance. Also the supply pumping effect of a single ended amp makes it nessessary to make sure that either the output MOSFETs are selected with a derating guidline so the supply pumping will not damage them, add protection, or make sure that your power supply can run in all four quandants (can sink current as well as source). Think about it, if you have one PWM half bridge, that is the same as a switching power supply. The only difference is that instead of referencing a constant voltage it is referencing a changing voltage (the audio signal). When using a BTL, one half bridge is supplying the current and the other is sinking so that it is balanced so linear power supplies will work fine if you amp has PSRR. Since a PWM amplifier naturally has 0dB PSRR, feed forward and feedback compensation should be used (most analog input PWM single chip amps do this internally). If not and your supply sags you will get a compression type sound. But back to the filtering... make sure you design the filter for the correct load impedance. If mismatched, then you could have a Low Pass Filter with a high Q which will also make it sound harsh. If you know the speakers that you are going to use, you should do an impedance plot to see the Z at 20kHz. Remember a 4 ohm speaker is 4 ohms at Z min. As Freq increases so will Z.

Dave
 
Some months ago, I went to a friend's apartment which was having problems with his brand new high def TV. The picture would blank out randomly. The service tech had already inspected the TV twice, and found no trouble with it.

But the problem kept reappearing.

So I went to his home. When I arrived, the picture was flawless. However some time later, it started breaking up. Fortunately, I noticed a slight thump-thump-thump coming fom the apartment upstairs as this ocurred.

I had a hunch..could it be EMI?
Fortunately, I had brought my oscilloscope with me. Using the headphone output of a common table radio connected to the scope, I could see the thump-thump-thump being synchronized with the received noise.

We went to see the upstairs neighbor. He was an EE student on a tight budget. He had actually built a class-D amp, but because of budget constraints, eliminated the output inductors, figuring out that since it was only driving the subwoofer, it would not be audible.
He was right on that count, but he had forgotten about the EMI. Fortunately, I had some spare inductors back home, and over beers and pizza we fixed the problem.

The bottom line: EMI is very real, and it causes very real interference.
 
I would still use a filterless "type" amp but just add the filters. If you are worried about the EMI (and you may need to if you have trouble with radio reception) you should put some ferrites on the outputs as well. There will become a frequency at which the interwinding capacitance will start allowing the very high frequency to come through again. As a hobbyist it may not be needed as the carrier will be attenuated enough to really not cause issues, but when I've had to pass FCC and European emissions on designs before, I needed to put them. Generally you will be able to see a good amount of broadband emissions up to 250-300MHz and start to go down after that. Carrier based designs may be a little worse than self oscillating because the carrier is constant. Self - oscillating will change frequencies. If you are doing more than one channel of amplification then self oscillating will be very dependant on PCB layout. As the frequencies change and differ from one channel to the other you can start to hear beat frequencies between the two where it is less likely with syncronized carrier designs. Same rule applies if using a SMPS. Analyze your switching frequencies of your SMPS and amp to see how the work together.

Generally you should watch out for conducted emissions as they could be pretty high with the high current switching (more emissions as cross conduction is increased too). AC mains filtering may be needed so your amp doesn't interfere with other equipment if you are at high output levels. 5W - 10W should not be an issue, but if you try high power amp projects keep it in mind

Dave
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.