Let me introduce you to my new amplifier based on a delta sigma converter and a full bridge stage switching at 500kHz.
The key features are:
- very low power consumption (about 1W in idle).
- closed loop design (at least two feedback pathes)
- not a ucd design to get no problem with the patent
- single supply for saving costs
So I started to simulate a few circuits and got the ******* thing working.
That's what the simulation says:
Then I tested the delta sigma converter in a first prototype:
After the successfull test I started to design and build a first prototype of the hole amp:
And finally, that's what the oscilloscope says:
... stay tuned 🙂
The key features are:
- very low power consumption (about 1W in idle).
- closed loop design (at least two feedback pathes)
- not a ucd design to get no problem with the patent
- single supply for saving costs
So I started to simulate a few circuits and got the ******* thing working.
That's what the simulation says:
An externally hosted image should be here but it was not working when we last tested it.
Then I tested the delta sigma converter in a first prototype:
An externally hosted image should be here but it was not working when we last tested it.
After the successfull test I started to design and build a first prototype of the hole amp:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
And finally, that's what the oscilloscope says:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
... stay tuned 🙂
Last edited:
Not sure yet, I need to do some more tests. I expect arount 200W/8Ohm in this configuration, maybe more. But it should be no problem to use more fets or better fets for more power. These are rated at 100Vds so a maximum of 60V supply voltage should not be used to stay in SOA.
The first tests were at 30V with a sinewave output at 55Vpp without clipping. Due to the full Bridge design you don't have to use a split power supply.
I can give you some more information in a few days.
No, two sided. I use the bottom site for GND. There are no traces at bottom, just solid copper for massive GND.
Thanks, but it's a UPL ;-)
So, I continued with my amp. The focus was the first feedback path over the delta sigma converter and the full bridge.
Unfortunately this wasn't that easy then I expected so I had to change a lot inside the delta sigma converter. Now the converter runs at 380kHz with just the first stage. I will try to redesign the second order stage the next days and I will try to come back to 500kHz.
On this screenshot you can see the bitstream, the output and a fft from the delta sigma converter. You can see how the converter works and how the noise is shaped to higher frequencies.
But now the good news 🙂 I get very good THD values, actually 0,037% at 1W so basically the amp runs. FFT plot also looks not bad but the noise could be a bit better.
The next steps are the second delta sigma stage, a feedback path in the analog part (over the output filter) and a better noise floor.
But, for the first shot, this is a very good basis.
Unfortunately this wasn't that easy then I expected so I had to change a lot inside the delta sigma converter. Now the converter runs at 380kHz with just the first stage. I will try to redesign the second order stage the next days and I will try to come back to 500kHz.
On this screenshot you can see the bitstream, the output and a fft from the delta sigma converter. You can see how the converter works and how the noise is shaped to higher frequencies.
An externally hosted image should be here but it was not working when we last tested it.
But now the good news 🙂 I get very good THD values, actually 0,037% at 1W so basically the amp runs. FFT plot also looks not bad but the noise could be a bit better.
The next steps are the second delta sigma stage, a feedback path in the analog part (over the output filter) and a better noise floor.
But, for the first shot, this is a very good basis.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
"The next steps are the second delta sigma stage, a feedback path in the analog part (over the output filter) and a better noise floor."
Any feedback before or after the output filter(s)?? Over the output filter? What does that mean?
Any feedback before or after the output filter(s)?? Over the output filter? What does that mean?
Not sure yet but I will work on it.
Thanks 🙂Nice Design...
Yes but I'm actually reworking the design and I won't use it any longer. I tried to shape the noise away throught oversampling but this gives you a lot of timing erros at the output stage. I will post a new design in the next few days.
Sorry Guys for my late reply. I had a lot to do and had some problems with my pcb manufacturing process but now it works 🙂
I tested the first amp of my rev4-board and it works great!! On the pictures you see the version with 30V supply but 50V are also no problem.
With 30V supply the amp has 60W in 8 Ohms, 0.005% THD and ~100db Noisefloor. Idle power is less then 2Watts!!
I listend to the amp the hole evening *g* I also tried it together with a 12V DC/DC converter and it worked great.
The 2W idle power and the dc/dc converter makes this amp to the perfect solution for a mobile speaker system 🙂
The next step are:
- doing all the rest of the measurments including the documentation
- 2 Ohms load test (haven't tried it yet)
- finalizing the pcb files and creating a last, final board for the professional manufacturing.
- including of some µC based monitoring of switching status, rms power and temperatur. I'm thinking about to display this values either on a lcd display or on a RS232 Terminal (maybe webinterface?!).
By the way... anybody interessted in this amp??
I tested the first amp of my rev4-board and it works great!! On the pictures you see the version with 30V supply but 50V are also no problem.
With 30V supply the amp has 60W in 8 Ohms, 0.005% THD and ~100db Noisefloor. Idle power is less then 2Watts!!
I listend to the amp the hole evening *g* I also tried it together with a 12V DC/DC converter and it worked great.
The 2W idle power and the dc/dc converter makes this amp to the perfect solution for a mobile speaker system 🙂
The next step are:
- doing all the rest of the measurments including the documentation
- 2 Ohms load test (haven't tried it yet)
- finalizing the pcb files and creating a last, final board for the professional manufacturing.
- including of some µC based monitoring of switching status, rms power and temperatur. I'm thinking about to display this values either on a lcd display or on a RS232 Terminal (maybe webinterface?!).
By the way... anybody interessted in this amp??
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I am speachless about your project and Yes. I am interested, But i am interested if it is capable to run at 100vdc beacause i have a smps power supply rated at +-100Vcc@5kw. I want to run 4 speakers of 18" in subwoofer from 32Hz till 160Hz. And also for my 2 tops made after MT122 project from poweraudio forum. The subwoofer is Goliath project from the same website. I was at the point of making the bridge version with hip4081 but than i started to read this post. My power supply also can be ajusted till +-60Vcc or i can use one of my power transformers that i have from some defekt amplifiers that i've bought from ebay. Thank you for sharing with us your project.
Good to hear that you like my project. So, my amp is based on a fullbridge and designed to work with a single supply power rail. Your smps has a split power supply so it would be better for you to use a class d amp in halfbridge design. For 100V I would have to change the mofests. This version of my design can be used with up to 50V (maybe 60V but I haven't tested it). So I guess this amp (in this configuration) is too weak for you're needs. It can handle up to 160W@8 Ohms and about 320W@4Ohms. For more I would have to change the fets and inductors.
100V supply are the point where ampdesign gets very expensive. You'll need expensive caps, big coils, high voltage fets, thicker copper on pcb... everything gets really expensive so maybe diy is not the best way to power your system.
That's a tiny bridged amp.
Mine turned out to be huge.
Mine turned out to be huge.
An externally hosted image should be here but it was not working when we last tested it.
I know and i was thinking that you will make it more like scalable Ucd style so could be adapted for any needs, but that is not a problem and this design is more than enoungh for home use. I have a smps with 50Vcc @10Amp so if you will decide to publish the schematic i will build this amp to compare it with other versions that i have build including IRS2092, Joel's non ucd, Class AB leco amp.
I am using T157-2 and T200-2 cores for all my Class d amplifiers but i like the way that fits your type of output inductor.
For the moment i am searching for an amplifier that fits and has allmost the same power like original in one of my subs peavey prosub-p beacause i have buyed it without amplifier modul(it was missing) and power supply is a toroid transformer with +-48vcc and that is why i am building any kind of amplifier in class d. Thank you.
Scaleable design? Yes and no... this amp is indeed scaleable from about 40W to 160W@8Ohms but 600W@8Ohms (what you will have with 100V) are another league (if you want to have a good and stable amp).
At the moment I don't want to publish the circuit but I think I will make a assembly kit with maybe all the smd parts pre-assembled.
And there will be a "plug-in" dsp board, based on a ADAU1701/1702 DSP from analog devices. Together with the monitoring interface this could be a very powerful and flexible amp. I will work on it the next weeks 🙂
At the moment I don't want to publish the circuit but I think I will make a assembly kit with maybe all the smd parts pre-assembled.
And there will be a "plug-in" dsp board, based on a ADAU1701/1702 DSP from analog devices. Together with the monitoring interface this could be a very powerful and flexible amp. I will work on it the next weeks 🙂
Great project and nice looking board! I'd love to see a schematic as I'm yet to understand how the delta sigma configuration is actually implemented in reality. Does it offer better performance than normal PWM?
Thanks,
Greg.
Thanks,
Greg.
Oh, I hope I don't dissapoint you but unfortunately out of my first tests with deltaSigma topology came nothing.
Real deltaSigma needs oversampling and noise shaping and this point made the most problems in my tests (I spent weeks to get it work...) so I came back to a "simple" hysteresis based 2nd order integrator solution with a very high speed diff. feedback.
This is not a typical PWM topologie with fixed frequency. The frequency in my system changes with the volume of input Signal. The range is between 300 - 500kHz (400kHz typ.).
So my design is not new but it works great and you get a very small and powerfull amp.
I hope I don't dissapoint you with this.
Real deltaSigma needs oversampling and noise shaping and this point made the most problems in my tests (I spent weeks to get it work...) so I came back to a "simple" hysteresis based 2nd order integrator solution with a very high speed diff. feedback.
This is not a typical PWM topologie with fixed frequency. The frequency in my system changes with the volume of input Signal. The range is between 300 - 500kHz (400kHz typ.).
So my design is not new but it works great and you get a very small and powerfull amp.
I hope I don't dissapoint you with this.
- Status
- Not open for further replies.