I am designing an amplifier for my senior project at Purdue University in Electical and Computer Engineering Technology in the USA. I worked at a highly regarded pro audio company to which I hope to return after graduation and pump out some excellent amplifiers in addition to lots of DIY work for you folks. First, however, I need to graduate 🙂
I am fairly familiar with how PWM works, with some experience working with [but not designing] an amplifier using the SG3525. However, I am looking for any alternative schemes for getting my PWM signal. I have a rather complex analog input stage designed outputting 2Vrms to my amplifier. The amp will be powered by 40Vrms single ended. Are there any decent alternatives that work well other than the SG3524, SG3525, TL494, UC3637 in an audio amp? The amp is planned to be a half-bridge using either the IR2011 or IR2110/3 gate drivers. I have not decided on FETs yet.
My only other idea was to convert the analog signal into PCM and then use one of the TASxxxx or Wolfson PCM-PWM chips to do all the hard work for me. This seems totally ridiculous.
Just to make you feel guilty for looking and not giving me any info 😀, I would love to provide the finished schematics and layout files to all interested. I have the input section and power supply nearly finished and, I think, available for viewing. The input section has both RCA and USB inputs and provides LR-4 LP xover, 0-180 phase adjust, and volume control. The power supply is a flyback using Power Integrations' TOPSwitch-GX 250 IC for control. If you are interested, post here and I will try to provide the power supply and input stage schematics shortly.
Thank you, Evan
I am fairly familiar with how PWM works, with some experience working with [but not designing] an amplifier using the SG3525. However, I am looking for any alternative schemes for getting my PWM signal. I have a rather complex analog input stage designed outputting 2Vrms to my amplifier. The amp will be powered by 40Vrms single ended. Are there any decent alternatives that work well other than the SG3524, SG3525, TL494, UC3637 in an audio amp? The amp is planned to be a half-bridge using either the IR2011 or IR2110/3 gate drivers. I have not decided on FETs yet.
My only other idea was to convert the analog signal into PCM and then use one of the TASxxxx or Wolfson PCM-PWM chips to do all the hard work for me. This seems totally ridiculous.
Just to make you feel guilty for looking and not giving me any info 😀, I would love to provide the finished schematics and layout files to all interested. I have the input section and power supply nearly finished and, I think, available for viewing. The input section has both RCA and USB inputs and provides LR-4 LP xover, 0-180 phase adjust, and volume control. The power supply is a flyback using Power Integrations' TOPSwitch-GX 250 IC for control. If you are interested, post here and I will try to provide the power supply and input stage schematics shortly.
Thank you, Evan
I don't think there is a chip solely for adding a power stage just for Class D apps. The ones that you listed are all power supply chips and are not the most accurate when it comes to audio signals.
A discrete design with high bandwidth components will give good results. I'll post my design a bit later in the clocked feedback thread if you want a starting idea (although I'm no graduate myself).
Cheers
Matt
A discrete design with high bandwidth components will give good results. I'll post my design a bit later in the clocked feedback thread if you want a starting idea (although I'm no graduate myself).
Cheers
Matt
Sounds good. If complexity is not an issue, my vote is towards discrete implementation.
If you'd like to consider Silicon Valley when you graduate, send me a resume.
If you'd like to consider Silicon Valley when you graduate, send me a resume.
Thank you. I looked at some discrete designs, and after talking with a couple people who have designed Class D amps in commerical products, they said that discrete was too much of a headache with propogation delay matching and such so I hadn't considered discrete until now.
Complexity is not an issue, I have complete faith in myself that I can deliver my project ahead of time regardless of anything.
fr0st, thank you. I have looked at the application in the IR reference amplifier but haven't found anything else that I had great faith in it's performance.
thespeakerguy, YGM
Complexity is not an issue, I have complete faith in myself that I can deliver my project ahead of time regardless of anything.
fr0st, thank you. I have looked at the application in the IR reference amplifier but haven't found anything else that I had great faith in it's performance.
thespeakerguy, YGM
Class D amps
First study what has already been done so you don't repeat bad history. There are several class-D amp schematics in the Audio circuits section of www.schematicsforfree.mattsoft.net. From indifferent to quite good.
My self I have used thousands of ICE Power modules. They use a balanced modulator with a pair of high precision comparitors (LM319) producing PWM for a pair of the IR2110 driver chips then feeding an H Bridge,
First study what has already been done so you don't repeat bad history. There are several class-D amp schematics in the Audio circuits section of www.schematicsforfree.mattsoft.net. From indifferent to quite good.
My self I have used thousands of ICE Power modules. They use a balanced modulator with a pair of high precision comparitors (LM319) producing PWM for a pair of the IR2110 driver chips then feeding an H Bridge,
EF, thanks but this project has to be built by hand or by an old oven, so that FET is out of the question. it looks nice tho 🙂
dmfraser, thank you. I have been checking through them and I think I have a good PWM section coming along.
A few more questions for you Class D wizards:
1. I need a comparator for the PWM section and I was looking at MAXIM's offerings. I want to use a push-pull type and it can't be uMAX, BGA, LLP, etc. 🙂 I also eliminated everything that wasn't rail-to-rail. I can't decide between the MAX985, MAX989, or something really fast that might give me problems like the MAX963. TI didn't have anything that looked better than the MAXIM offerings, unfortunately.
I didn't want to use the LM319 because it is open-collector and, although I haven't had any problems, I have been told to avoid open-collector comparators if possible. Are there any other comparator suggestions?
I noticed that the IR reference amp uses a LT1120. Why would they use an op amp instead of a comparator? Isn't the point of a comparator to react quickly because it doesn't have a dominant pole cap inside it?
2. For my sawtooth generation, I saw the LM556 and MAX038. The MAXIM part doesn't look like what I really want, but the DADJ pin tempts me.
3. By using this discrete design, I will have to have an additional inverter in one of the 2 PWM signals. Won't that add propogation delay? Should I add a delay somehow to even them out?
Thank you, Evan
dmfraser, thank you. I have been checking through them and I think I have a good PWM section coming along.
A few more questions for you Class D wizards:
1. I need a comparator for the PWM section and I was looking at MAXIM's offerings. I want to use a push-pull type and it can't be uMAX, BGA, LLP, etc. 🙂 I also eliminated everything that wasn't rail-to-rail. I can't decide between the MAX985, MAX989, or something really fast that might give me problems like the MAX963. TI didn't have anything that looked better than the MAXIM offerings, unfortunately.
I didn't want to use the LM319 because it is open-collector and, although I haven't had any problems, I have been told to avoid open-collector comparators if possible. Are there any other comparator suggestions?
I noticed that the IR reference amp uses a LT1120. Why would they use an op amp instead of a comparator? Isn't the point of a comparator to react quickly because it doesn't have a dominant pole cap inside it?
2. For my sawtooth generation, I saw the LM556 and MAX038. The MAXIM part doesn't look like what I really want, but the DADJ pin tempts me.
3. By using this discrete design, I will have to have an additional inverter in one of the 2 PWM signals. Won't that add propogation delay? Should I add a delay somehow to even them out?
Thank you, Evan
This looks good, but slightly complex http://www.maxim-ic.com/appnotes.cfm/appnote_number/825 I am thinking about replacing the open-collector MAX991 with the push/pull MAX987. And perhaps using the MAX987 for the main PWM comparator also. PSpice, here I come 🙂
I can't edit for some reason.
Anyway, in question #3 above, I must have been missing something. I understand what is going on now.
Anyway, in question #3 above, I must have been missing something. I understand what is going on now.
Friends, don't let your friends use sawtooths for class-d amps !
Pass the signal through two XORs. Feed one input of both gates with the same PWM signal and tie the other input of one of the gates to logic H and the other one to logic L. And voila: you have an inverted and a non inverted version of the same signal, delayed by the same amount of time.
Regards
Charles
Why not use a sawtooth? Won't that keep the frequency constant whereas a straight triangle will allow the freq to shift slightly as the input voltage varies? If this has been posted before, sorry. I was posting anyway so I put it here - I will search after I post.
Ok, what about using a MAX4295, a full bridge, filterless, 2W, Class D amp to do all my PWM. I have 2 eval boards here and they work wonderfully so far. I am listening to them on full range speakers and they do well. Anyway, that would give me the signals I need to run a half bridge driver, or a pair of them for full bridge - like below.
A and B would be the two outputs.
Ok, what about using a MAX4295, a full bridge, filterless, 2W, Class D amp to do all my PWM. I have 2 eval boards here and they work wonderfully so far. I am listening to them on full range speakers and they do well. Anyway, that would give me the signals I need to run a half bridge driver, or a pair of them for full bridge - like below.
A and B would be the two outputs.
Attachments
Here is the input section I have developed. I am contemplating a different phase shift section that uses 2 op amps, this one is from Randy Slone's Audiophile Projects book.
There are a couple notes on the schematic if anyone is interested in helping answer them. Feel free to make any other suggestions. The gain will need to be changed if I am to run the MAX4295.
The LR-4 filter values are all incorrect. They are right for a 2nd order, which I copied when I doubled the filter order.
There are a couple notes on the schematic if anyone is interested in helping answer them. Feel free to make any other suggestions. The gain will need to be changed if I am to run the MAX4295.
The LR-4 filter values are all incorrect. They are right for a 2nd order, which I copied when I doubled the filter order.
Attachments
I'm designing a classD amplifier for digital input line, like AES/EBU or SPDIF from CD drive or any digital source.
The first part of this design is to convert SPDIF to PWM, not so easy, but not so difficult using existing integrated solutions 😎
In order to make this idea "a good to hear idea", the PWM frequency will be 1 MHz, which makes a big challenge for the power bridge and its associated driver.
Anyway, I'm running some simulations to determine the constraints in terms of bridge drive ( this design will use a single power supply and a full H bridge ) according to brigde FETs and load.
The HIP4081, from Intersil, may be a friendly approach, because it allows easy to set dead-time and provides a good current drive to be fast enough.
So Evan, it may be another solution for you...
I'm not totally confident using this solution, and I'm still working in a fully discrete one.
Any feedback or experiences on this HIP4081 ?
Fabien
The first part of this design is to convert SPDIF to PWM, not so easy, but not so difficult using existing integrated solutions 😎
In order to make this idea "a good to hear idea", the PWM frequency will be 1 MHz, which makes a big challenge for the power bridge and its associated driver.
Anyway, I'm running some simulations to determine the constraints in terms of bridge drive ( this design will use a single power supply and a full H bridge ) according to brigde FETs and load.
The HIP4081, from Intersil, may be a friendly approach, because it allows easy to set dead-time and provides a good current drive to be fast enough.
So Evan, it may be another solution for you...
I'm not totally confident using this solution, and I'm still working in a fully discrete one.
Any feedback or experiences on this HIP4081 ?
Fabien
SudFab said:
Be aware direct PCM to PWM conversion brings in severe derivative distortion, the usual path is to interpolate first to compute what should be an analoge triangle or sawtooth modulator output and this usually involves DSP.
I can point to papers on the subject if you are interested.
Rodolfo
Evan Shultz said:
Ever tried to load with current sources?
I did and got nice stright edges both up and down instead of the usual exponential RC rise when resistor loaded.
A simple discrete PNP current mirror does it.
Rodolfo
Yes, I'm aware of that key point. Using a self-developped or third-party algorithm shall be the better way to proceed. But to clarify, I do not plan to design that right now.
Reason is very simple, I prefer to first focus on drive stage and power stage, H bridge topology.
However, I will be pleased to read your comments, link or any biblio you have 🙂
For your information, I plan to use an integrated PCM to PWM solution ZXCW8100.
To not distrurb this Evan post, please, switch on the PowerDAC post for next talk 😉
Fabien
SudFab,
Thank you for the suggestions. Use this post however you would like as long as it is basically on track 🙂 PCM>PWM is interesting
I have worked with an amp using the HIP4080A and it was rather noisy. I repaired another amp with the HIP4080A and it used a C to ground on each gate - apparently to slow down rise times and eliminate noise? Or perhaps that was the dead time? It was a Kicker DX700 car amplifier. Those experiences sold me on a pair of IR2110 half bridge drivers.
I have the bridge drivers ordered, and hopefully everything will prototype well. I am concerned that the MAX4295 dead time will not fit my application. However, since the MAX is an IC I imagine dead time would be low, and I can increase it externally. So hopefully everything will work out 🙂
Thank you for the suggestions. Use this post however you would like as long as it is basically on track 🙂 PCM>PWM is interesting
I have worked with an amp using the HIP4080A and it was rather noisy. I repaired another amp with the HIP4080A and it used a C to ground on each gate - apparently to slow down rise times and eliminate noise? Or perhaps that was the dead time? It was a Kicker DX700 car amplifier. Those experiences sold me on a pair of IR2110 half bridge drivers.
I have the bridge drivers ordered, and hopefully everything will prototype well. I am concerned that the MAX4295 dead time will not fit my application. However, since the MAX is an IC I imagine dead time would be low, and I can increase it externally. So hopefully everything will work out 🙂
I think that using max4295 is a great idea!
No need to trim anything! Just input the audio
and you get the pwm. It has very low THD.
So everyting depends on your power stage!
And you can order samples for free from maxim!
Defenetly better than using standalone triangle generator and comparator. And it in SO8!
way way smaller than opas and comps.
No need to trim anything! Just input the audio
and you get the pwm. It has very low THD.
So everyting depends on your power stage!
And you can order samples for free from maxim!
Defenetly better than using standalone triangle generator and comparator. And it in SO8!
way way smaller than opas and comps.
- Status
- Not open for further replies.