Have a look at....
http://focus.ti.com/lit/an/slua172/slua172.pdf
On pages 3 and 4 they suggest a method of getting a 'regulated' auxilliary supply off the PFC boost inductor.
If you replace the words PFC Boost Inductor with Output Inductor then that might work better......
I only mention it because this seems to be one of those perennial(?) problems and you are in a position to try it unlike dreamers like wot I am.
DNA
http://focus.ti.com/lit/an/slua172/slua172.pdf
On pages 3 and 4 they suggest a method of getting a 'regulated' auxilliary supply off the PFC boost inductor.
If you replace the words PFC Boost Inductor with Output Inductor then that might work better......
I only mention it because this seems to be one of those perennial(?) problems and you are in a position to try it unlike dreamers like wot I am.
DNA
Oh...., and while I have my thinking cap on.
How about trying a complementary source connected pair of N and P channel mosfets driven with the floating driver of the IR2010 or whatever it is.
Now, my thinking cap is not fully on but you would probably need some logic level devices since the swing is limited to 15V or +/-7.5V......
And you would have to figure out how to reference the output of the IC to produce that positive and negative swing.
However, then you would have no (well ideally no) problem with crossconduction or a need to set any dead time.
Does that make sense?
DNA
How about trying a complementary source connected pair of N and P channel mosfets driven with the floating driver of the IR2010 or whatever it is.
Now, my thinking cap is not fully on but you would probably need some logic level devices since the swing is limited to 15V or +/-7.5V......
And you would have to figure out how to reference the output of the IC to produce that positive and negative swing.
However, then you would have no (well ideally no) problem with crossconduction or a need to set any dead time.
Does that make sense?
DNA
After some thinking, I suspected that the highside driver was working, or we wouldn't have a perfectly good output during the negative half of the input signal.
so I got a storage scope today and the negative side has those fine sawtooth. This is confirmed also with probing Vgs - just over 10v during the negative cycle of the input signal and about 4v otherwise.
Now, something, somewhere the highside driver stopped working during the positive half of the input signal.
I am going to increase the voltage on the driver chip and if that doesn't work, I will try a driver without shootthrough protection.
so I got a storage scope today and the negative side has those fine sawtooth. This is confirmed also with probing Vgs - just over 10v during the negative cycle of the input signal and about 4v otherwise.
Now, something, somewhere the highside driver stopped working during the positive half of the input signal.
I am going to increase the voltage on the driver chip and if that doesn't work, I will try a driver without shootthrough protection.
UCD/simu question
Hi,
First post here
I am trying to make a UCD design and while doing my searches i found this post.
I always like to have a good overview and feeling of how the signals should look so i first started running the LTspice simulation of what kartino posted (post #52).
Then i wanted to try with only one stage comparator(like in the philips ucd design) keeping the same gate drivers. But if the operating point seems about the same, even playing a bit with the values i cant get this thing to work.
So I wanted to know if i am missing something specific the kartino design does that i am not getting.
Or is it just a simulation problem ?
Any help appreciated. thanks!
Hi,
First post here
I am trying to make a UCD design and while doing my searches i found this post.
I always like to have a good overview and feeling of how the signals should look so i first started running the LTspice simulation of what kartino posted (post #52).
Then i wanted to try with only one stage comparator(like in the philips ucd design) keeping the same gate drivers. But if the operating point seems about the same, even playing a bit with the values i cant get this thing to work.
So I wanted to know if i am missing something specific the kartino design does that i am not getting.
Or is it just a simulation problem ?
Any help appreciated. thanks!
Re: UCD/simu question
Hm, you're not yet really messing with UCD designs but more so people's interpretations of them... everyone tries to add their own twist. You can find closer designs if you keep on digging.
There's several errors in your schematic, and you're going to have a real hard time figuring them when you're just starting off.
My suggestion to you is start off by studying the real thing. That means look up the patent, study it, look up the AES paper on UCD available at Hypex.nl , that alone should solve most of your issues.
You'll then be armed to start figuring this stuff out, along with some practice on the simulator. I don't know how long you played around with this but I'm thinking, not nearly enough.
I see no model statement for your pass transistor.
I see a run time of only 100us.
I see your input signal is 0Vac
I see you taking feedback differentially (not made obvious by the schematic but what do you suppose that network is connected to the non inverting input your signal goes to.... ) and driving it single ended..
I see your feedback networks are improperly terminated.... c4 for instance... there goes your proportional gain.
If you manage to solve all that, you'll just need different resistor values for your driver turn on transistors.... try 300ohm. Then you'll get some waves. Also notice these are Kartino's tweaked drivers... not the real thing, you'll note that from the schematics in the proposed writings.
You've much work to do. I hope that's some help, I know it's not much, but it's clear to me you haven't put alot of time into it so far.
If you have to, dig up a simpler version, idealize what you're uncertain of, and read up on how it's supposed to work.
I'll be happy to help out more later. Welcome to the forum.
Cheers,
Chris
lnbig said:Hi,
First post here
I am trying to make a UCD design and while doing my searches i found this post.
I always like to have a good overview and feeling of how the signals should look so i first started running the LTspice simulation of what kartino posted (post #52).
Then i wanted to try with only one stage comparator(like in the philips ucd design) keeping the same gate drivers. But if the operating point seems about the same, even playing a bit with the values i cant get this thing to work.
So I wanted to know if i am missing something specific the kartino design does that i am not getting.
Or is it just a simulation problem ?
Any help appreciated. thanks!
Hm, you're not yet really messing with UCD designs but more so people's interpretations of them... everyone tries to add their own twist. You can find closer designs if you keep on digging.
There's several errors in your schematic, and you're going to have a real hard time figuring them when you're just starting off.
My suggestion to you is start off by studying the real thing. That means look up the patent, study it, look up the AES paper on UCD available at Hypex.nl , that alone should solve most of your issues.
You'll then be armed to start figuring this stuff out, along with some practice on the simulator. I don't know how long you played around with this but I'm thinking, not nearly enough.
I see no model statement for your pass transistor.
I see a run time of only 100us.
I see your input signal is 0Vac
I see you taking feedback differentially (not made obvious by the schematic but what do you suppose that network is connected to the non inverting input your signal goes to.... ) and driving it single ended..
I see your feedback networks are improperly terminated.... c4 for instance... there goes your proportional gain.
If you manage to solve all that, you'll just need different resistor values for your driver turn on transistors.... try 300ohm. Then you'll get some waves. Also notice these are Kartino's tweaked drivers... not the real thing, you'll note that from the schematics in the proposed writings.
You've much work to do. I hope that's some help, I know it's not much, but it's clear to me you haven't put alot of time into it so far.
If you have to, dig up a simpler version, idealize what you're uncertain of, and read up on how it's supposed to work.
I'll be happy to help out more later. Welcome to the forum.
Cheers,
Chris
well thanks for taking the time to answer.
This is true i am new to class D amplifier design and just starting learning how it works. The thing is i started with a working design and maybe i wanted to add my changes to it too quickly heh.
As for LT Spice, i am not an expert but i did some simulation with it for some other class AB amplifier with no problem before.
Anyway it just tells me i need to get in it more deeply as it is surely not something simple.
I ll get back here eventualy when i have some progress.
This is true i am new to class D amplifier design and just starting learning how it works. The thing is i started with a working design and maybe i wanted to add my changes to it too quickly heh.
As for LT Spice, i am not an expert but i did some simulation with it for some other class AB amplifier with no problem before.
Anyway it just tells me i need to get in it more deeply as it is surely not something simple.
I ll get back here eventualy when i have some progress.
Yah, I just saw more than one problem with it.
I'd go for DC coupling the inputs BTW, one less source of confusion.
I managed to get that sim going but making all the changes mentioned. Maybe they're not all required.
Feel free to ask whatever should you run into a wall though. It's a great circuit to learn from.
I'd go for DC coupling the inputs BTW, one less source of confusion.
I managed to get that sim going but making all the changes mentioned. Maybe they're not all required.
Feel free to ask whatever should you run into a wall though. It's a great circuit to learn from.
yes, i did do some of the modification you suggested. I also changed the gate drivers to their simplest version, like the one presented at the hypex.nl page. Also changed the feedback for a DC coupling and put a little less gain. (also simulation time because capacitors takes more than a few periods to load heh.)
Now the circuit is a little less confusing and i think i understand better how it works and what each part does.
there are still some points i need to clarify a bit more (for instance i am not sure of why the 2 diodes in the comparator are needed)and then i can go on and try with more optimised design i guess.
Also tried to change MOS models but LTSpice doesnt like it too much
. but thats another problem...
Anyway i ll probably spend some more time on the forum the next days to learn more .
Now the circuit is a little less confusing and i think i understand better how it works and what each part does.
there are still some points i need to clarify a bit more (for instance i am not sure of why the 2 diodes in the comparator are needed)and then i can go on and try with more optimised design i guess.
Also tried to change MOS models but LTSpice doesnt like it too much
. but thats another problem...Anyway i ll probably spend some more time on the forum the next days to learn more
.
Hi Fokker
Are you sure the 1N4148 is really of for your use?
I think the high side supply draws quite high peek currents through the charging diode. I havent seen any construction with diodes smaler than 1A in this place!
Try something like 330nF and try using a fast 1A diode like MURS120, UF4001, BYW98-100, ES1D or one like these.
I think 470nF for the high side supply is more than enough and will not be your problem.
I actually tryed changing my BYW98-100 with a 1N4148, and it still worked, nut I'm only running something like a single watt from my IR2110/IF640 thing at the moment.
Also like Chris said, maybe the IR2011 is a bit on the low side for running the IRF540s, which can be helped by a buffer stage on both high and low side (think this is actually how Eva is running her multi kW construction).
Are you sure your input/comparator stage is running as it should?
Like your idea of trying out 3 different drivers, to see which sounds the best. Would really like to get an answer for that
But I guess you'll have to make an optimised version for each of the drivers to really judge the result.
What voltage are you using for the driver? I think it should be 12V to make sure there is enough juice to switch the FETs.
What did you mean by running the LM5104 through an adapter?
Are you sure the 1N4148 is really of for your use?
I think the high side supply draws quite high peek currents through the charging diode. I havent seen any construction with diodes smaler than 1A in this place!
Try something like 330nF and try using a fast 1A diode like MURS120, UF4001, BYW98-100, ES1D or one like these.
I think 470nF for the high side supply is more than enough and will not be your problem.
I actually tryed changing my BYW98-100 with a 1N4148, and it still worked, nut I'm only running something like a single watt from my IR2110/IF640 thing at the moment.
Also like Chris said, maybe the IR2011 is a bit on the low side for running the IRF540s, which can be helped by a buffer stage on both high and low side (think this is actually how Eva is running her multi kW construction).
Are you sure your input/comparator stage is running as it should?
Like your idea of trying out 3 different drivers, to see which sounds the best. Would really like to get an answer for that
But I guess you'll have to make an optimised version for each of the drivers to really judge the result.
What voltage are you using for the driver? I think it should be 12V to make sure there is enough juice to switch the FETs.
What did you mean by running the LM5104 through an adapter?
Baldin said:
I am reasonably sure. I have used 1n4148 in discrete drivers, I have seen 1n4148 spec'd for such applications, and I have used beefier shotteky diodes here without any sucess. I think the issue is somewhere else. But again, I am not 100% sure.
the charging current appears to be about 150-200ma in this particular circuit with irf540 type output devices. and just slightly lower with irf510.
Baldin said:
Yeah. the calculation suggests about .1uf - and it is frequency-indepdent, contrary to popular believe.
Baldin said:
I am pondering that for now. the ir2011, if I remember correctly, does a couple amps output. and if you look at the type of transistors used for the driver stage in the Philips UcD application note, they are no where near that capable, current-wise. so I suspect that the ir2011 is more than enough. but that is always one possibility that I cannot completely rule out at this point.
Baldin said:
I thought I was sure but I am not. I tried to look at the output from the comparator with the feedback disconnected and it gave nice square waveform. I have also built it twice with brand new devices and it still functioned the same. so if I have to bet, it would be somewhere else.
but to your point, I am building a pwm generator to test the driver / output stages independently and we shall see.
Baldin said:
that's a great question. I initially ran the set-up at +/- 8v and it didn't work. so I suspected that the ULVO circuitry in the driver was at fault. so I ran it at +/- 32v and I got the same thing. so I am now back at +/-8v.
Baldin said:
the 5104 came in soic form. so it is hard to use it directly on my breadboard. I soldered leads to it, rearranged the pin-outs so that it mimiced that of ir2011 (5104 doesn't have the L-in pin), and fixed it to a dip socket. That way, i can plug it right into a ir2011 circuit and it works right away without any need to rearrange the breadboard.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.