Hi Guys,
I'm designing a self oscillating class d amplifier, with feedback from the mosfet.
Can you guys suggest a good comparator other than LM311, LM319? and I had search for self oscillating class d amp and found thet most are half bridge. may I know how to use full bridge for self oscillating? (2 output from masfet output stage, output from full bridge is float from ground)
regards,
Bigbulb
I'm designing a self oscillating class d amplifier, with feedback from the mosfet.
Can you guys suggest a good comparator other than LM311, LM319? and I had search for self oscillating class d amp and found thet most are half bridge. may I know how to use full bridge for self oscillating? (2 output from masfet output stage, output from full bridge is float from ground)
regards,
Bigbulb
Hi I use LMV7239 by National Semiconductors. It's very fast and best of all, it has a push-pull output configuration.
I know that maxim makes a lot of fast comperators, if that can help.
www.maxim-ic.com
www.national.com
I know that maxim makes a lot of fast comperators, if that can help.
www.maxim-ic.com
www.national.com
Thanks sovadk.
I still searching for self osc type of class d in full bridge configuration. I had build the half bridge one and wish to try full bridge. Can't figure out how's the feedback connection, since the speaker output is floating. Please help!
I still searching for self osc type of class d in full bridge configuration. I had build the half bridge one and wish to try full bridge. Can't figure out how's the feedback connection, since the speaker output is floating. Please help!
If you understand German a bit, you may want to give this thread a try.
At least the user Beobachter develops a so called (by himself) SODFA as a full bridge version with pre filter feedback. He posted links to the schematics too.
Regards, Timo
At least the user Beobachter develops a so called (by himself) SODFA as a full bridge version with pre filter feedback. He posted links to the schematics too.
Regards, Timo
You can take it differentially across the load, then use a diff amp to convert to single ended (ugly) or be really slick and keep it fully differential, subtract with differential input, like UCD.
You might want to start with a half bridge though, you'll cook less mosfets getting through the ol learning curve that way.
If you go for the fully differential idea you won't even need to change much of anything to go with a full bridge later on.
I believe Linear Technologies has some insanely fast comparators and offer some of them with differential outputs as well, might want to have a look at their selection.
Cheers
You might want to start with a half bridge though, you'll cook less mosfets getting through the ol learning curve that way.
If you go for the fully differential idea you won't even need to change much of anything to go with a full bridge later on.
I believe Linear Technologies has some insanely fast comparators and offer some of them with differential outputs as well, might want to have a look at their selection.
Cheers
LM319 for double channel UcD
LM319 has two compararors per chip. Now use one chip in double channel UcD (or any other self-osc configuation), one comparator for a channel.
Would the parastic coupleing (via inside or via supply railes) mix two switching frequencies and cause the "whistling" problem?
Do I need to use one LM319 chip per channel (and waste half of them)?
LM319 has two compararors per chip. Now use one chip in double channel UcD (or any other self-osc configuation), one comparator for a channel.
Would the parastic coupleing (via inside or via supply railes) mix two switching frequencies and cause the "whistling" problem?
Do I need to use one LM319 chip per channel (and waste half of them)?
Hi Kenshin,
I would not even recommend that you put both amplifiers on the same PCB if you intend to use them for audio. I once used three LT1394 comparators on the same board, and noise level was not Hi-Fi. But for motor control (intended application) it was OK. There was no whistling since this was a clocked design.
Best regards,
Jaka Racman
I would not even recommend that you put both amplifiers on the same PCB if you intend to use them for audio. I once used three LT1394 comparators on the same board, and noise level was not Hi-Fi. But for motor control (intended application) it was OK. There was no whistling since this was a clocked design.
Best regards,
Jaka Racman
Hi,
You might want to check out Linear Technologies' offerings, they seem to have a foot above the rest.
Regards,
Chris
You might want to check out Linear Technologies' offerings, they seem to have a foot above the rest.
Regards,
Chris
Another fast one is NE521 by Philips. I doesn't have differential outputs though but two comparators in one package.
Regards
Charles
Regards
Charles
classd4sure said:
Does every board need to have seperate ground terminals
(just like some ADC have seperated AGND/DGND)?
LM160 or LM360 from National are pretty nice 20ns types. Furtheron they offer inverted outputs.
The outputs are push pull between GND and pos rail, not sloping to neg rail.
The outputs are push pull between GND and pos rail, not sloping to neg rail.
I read these quote :
Is it really not necessary to use comparator faster than LM319 (<80nS)? say, for 400khz switching frequency?
http://www.diyaudio.com/forums/showthread.php?postid=747891#post747891
http://www.diyaudio.com/forums/showthread.php?postid=1362335#post1362335
Is it really not necessary to use comparator faster than LM319 (<80nS)? say, for 400khz switching frequency?
I have used LT1713 for some applications. If you generate the dead time your self (diodes, caps and resistors) the complementary outputs are very handy. It is fed by +/-5V and the digital output is fed by +5V. The only problem with it as I can remember is some inflexibility with the output. I think I tried to connect the output to the minus rail and feed the entire thing with +/-2,5V but that did not work. So, the output stage needs to be GND referenced but that is usually not a problem.
The LM361 is an easy-to use part with complementary outputs.
If you also want the amp to run off a unipolar supply rail then I've used the TL712 succesfully. This runs off a single +5V supply and also has complementary outputs.
If you also want the amp to run off a unipolar supply rail then I've used the TL712 succesfully. This runs off a single +5V supply and also has complementary outputs.
For a self oscillating design that I'm doing for my employer, I'm using a TLV7239 which is fast, very low power, and also has a low input offset voltage.
If you are using a UCD design with dc coupling (like the Hypex designs have), then it is often forgotten that the input offset voltage of the comparator will be multiplied by the gain of the amplifier. It is all too easy to find over 100mV of dc offset at the output of the amplifier. This is negligable when driving loudspeakers, but my designs are for driving step-up output transformers for 100V line loudspeakers. These transformers have very low primary dc resistance, so 100mV of output offset can cause almost 1A of dc to flow in the primary, which can cause problems with toroidal transformers. A slight redesign of the basic UCD circuit with ac coupling can drop the dc gain to unity, while keeping the desired gain at the audio frequencies. In this case the output offset will only be slightly greater than the comparator offset.
Hypex have an app-note showing how a dc-servo can be used with their amplifiers to reduce the dc offset when driving an output transformer.
If you are using a UCD design with dc coupling (like the Hypex designs have), then it is often forgotten that the input offset voltage of the comparator will be multiplied by the gain of the amplifier. It is all too easy to find over 100mV of dc offset at the output of the amplifier. This is negligable when driving loudspeakers, but my designs are for driving step-up output transformers for 100V line loudspeakers. These transformers have very low primary dc resistance, so 100mV of output offset can cause almost 1A of dc to flow in the primary, which can cause problems with toroidal transformers. A slight redesign of the basic UCD circuit with ac coupling can drop the dc gain to unity, while keeping the desired gain at the audio frequencies. In this case the output offset will only be slightly greater than the comparator offset.
Hypex have an app-note showing how a dc-servo can be used with their amplifiers to reduce the dc offset when driving an output transformer.
- Status
- Not open for further replies.