I was going through my National databook and noticed a striking similarity between their switching regulators and digital amplifiers. Both of them have an integrator upfront, an oscilator, an comparator. Now, their drive mechanism isn't exactly the same: those switching regulators tend to drive just one mosfet, vs. half / full bridge for digital amps. sure, there are other differences but I think there is a chance we can use those switching regulators in a digital amp.
Here is my thinking: those regulators look very much like the front end of the IRAudamp. The output from those regulators can be used to drive a gate driver, like the ir2011.
Some of the latest switching regulators go as high as 1mhz so that got to be good for audio,
.
I know this is a completely crazy idea but it may work. Your thoughts?
Here is my thinking: those regulators look very much like the front end of the IRAudamp. The output from those regulators can be used to drive a gate driver, like the ir2011.
Some of the latest switching regulators go as high as 1mhz so that got to be good for audio,
.I know this is a completely crazy idea but it may work. Your thoughts?
here is my preliminary thought:
a) we can keep the front end of the IRAudamp, and replace the logic gates and gate driver with a simpler solution, like LM5104 - it requires just one signal, and has shoot-through protection, low-voltage lock out and soft start, not to mention programmable adaptive dead time.
b) we can also replace the entire front end of the iraudamp, with a pwm controller. here we have a lot of choices. my favorite is lm5030: high switching frequency, high voltage, etc.
The difficulty is to feed it with an audio signal,.
Those pwm controllers have an error amp where the feedback (from output) is compared to a reference voltage (usually 1.23-1.25v). obiviously, that doesn't work for us unless you want to listen to DC.
so my suggestion is to use an enternal error amp / summing amp that calculates the difference between the output and audio input and feed it to the feedback pin on the pwm controller.
c) this solution is also quite interesting. There are switching mode regulators that integrate the pwm controller and gate driver - lm2655 for example. so with it, you need to add a mosfet (for asynchronous amplification) or a couple of mosfets (for synchronoous amplification) and you are ready to go.
The downside is that they usually handle low voltages so that is kind of limited to low power amplifier. but nonetheless interesting to try.
One final point: all those solutions assume single rail, and due to their regulator nature, they can only source current. As such, there is a DC element in this type of output. I assume that you can use a DC-blocking capacitor after the low-pass filter to keep the DC out of the load.
what do you think?
a) we can keep the front end of the IRAudamp, and replace the logic gates and gate driver with a simpler solution, like LM5104 - it requires just one signal, and has shoot-through protection, low-voltage lock out and soft start, not to mention programmable adaptive dead time.
b) we can also replace the entire front end of the iraudamp, with a pwm controller. here we have a lot of choices. my favorite is lm5030: high switching frequency, high voltage, etc.
The difficulty is to feed it with an audio signal,
.Those pwm controllers have an error amp where the feedback (from output) is compared to a reference voltage (usually 1.23-1.25v). obiviously, that doesn't work for us unless you want to listen to DC.
so my suggestion is to use an enternal error amp / summing amp that calculates the difference between the output and audio input and feed it to the feedback pin on the pwm controller.
c) this solution is also quite interesting. There are switching mode regulators that integrate the pwm controller and gate driver - lm2655 for example. so with it, you need to add a mosfet (for asynchronous amplification) or a couple of mosfets (for synchronoous amplification) and you are ready to go.
The downside is that they usually handle low voltages so that is kind of limited to low power amplifier. but nonetheless interesting to try.
One final point: all those solutions assume single rail, and due to their regulator nature, they can only source current. As such, there is a DC element in this type of output. I assume that you can use a DC-blocking capacitor after the low-pass filter to keep the DC out of the load.
what do you think?
it seems to work!
here is a digital amp based on linear's LTC1735 - the first one I tried out of the SwCad download.
Vosense is the feedback point, and I added the signal source, V1, to the bottom end of R4.
The output is a reasonably shaped sine wave - the bottom end needs more work.
maybe it can be further improved?
here is a digital amp based on linear's LTC1735 - the first one I tried out of the SwCad download.
Vosense is the feedback point, and I added the signal source, V1, to the bottom end of R4.
The output is a reasonably shaped sine wave - the bottom end needs more work.
maybe it can be further improved?
Attachments
I simulated this against quite a few Linear SMPS controllers and it does seem to work, at least at low frequencies (1khz), due to their somewhat low switching frequencies (usually around 150khz).
so I think the concept appears to be sound. We just need to find the right chip to make it work.
I will check it out. And along the same line, if we used a transformer-based configuration, we can hook up the smps right to the mains and be done with it! No DC to worry about, no PS to build, just a controler and a mosfet!
Wow! That would be interesting to build.
so I think the concept appears to be sound. We just need to find the right chip to make it work.
soongsc said:
I will check it out. And along the same line, if we used a transformer-based configuration, we can hook up the smps right to the mains and be done with it! No DC to worry about, no PS to build, just a controler and a mosfet!
Wow! That would be interesting to build.
soongsc said:
that's something I am finding out. the responses from those pwm controllers don't seem to be particularly fast. Not sure if the circuitry needs tweaking or the chips are foundamentally slow.
soongsc said:
so far, they seem to be pretty immune to that, in simulation at least.
soongsc said:
no. I want to stay analog. otherwise, i would have gone the TI route.
soongsc said:
not sure why that's related here?
That line really tickled me
"Just need the right chip" (been there!) but none are ever 100%.
I think the chip you're waiting for is the Mueta and I think you''ll be waiting a looooooooooooooooooooong time.
I just started playing with discretes and amps which are fully discrete and am no longer a slave to the requirment of that seemingly non existant perfect chip.
Still, you can do this, but I don't know about high quality.
Regards,
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.