THD is always going to be low with a dummy resistor as a load, a laboratory regulated supply and a single channel driven by a continuous test tone. Power supply may be indeed sagging a lot, but that voltage drop is constant and fully proportional to the own signal thus not contributing to THD nor allowing for PSU-sagging output modulation.
Make several channels share the same real-world unregulated PSU and the same PCB, add a reactive non-linear load, fed the circuit with periodic tone bursts (more like music), and the system will show all its weakness at a time.
Make several channels share the same real-world unregulated PSU and the same PCB, add a reactive non-linear load, fed the circuit with periodic tone bursts (more like music), and the system will show all its weakness at a time.
-TI chips are meant to be used with regulated SMPS. They do even provide variable SMPS reference voltage since they partly use supply voltage to regulate volume which is a good thing, given lousy idle efficiency of analog class D.
-google PEDEC before you discuss feedback in digitally controlled classD
-google PEDEC before you discuss feedback in digitally controlled classD
Open loop digital power amps
Kims:
Good results!
Eva:
PSURR=0dB. Correct. I say:So what?
Since when has it been mandatory for audio amplifiers to use unregulated power supplies? Watching the continued use of bulky 1920's vintage technology, mains frequency transformer-rectifier PSUs with hostile power factor is for me somewhere between painful and comical.
PSURR=0dB poses an enjoyable challenge for the designer of open loopers, more so if it's required to have variable Vout. It forces the amplifier designer to consider the PSU-OPS holistically as a single system. Because I know you are a fellow competent 'switchmode artist' I 'm surprised you don't share this view!
The argument that a properly regulated PSU costs more and is more complex has long since been discarded. Switchmode PSU's (even split rail ones) are now well entrenched in high-end amplifiers. The regulator I'm presently developing for each channel of PowerDAC 3 is a 600W average-current-mode synchronous reg with fs=352,8kHz, Vout from 2-60V at 10A max. It requires no heatsink because power stage efficiency = 99%.
The Vout of the regulator is controlled from a digital input command, and the control loop gain is optimised for slope matching at all Vout by digitally controlled gain elements. Control bandwidth exceeds 20kHz. The high efficiency is due to the adoption of similar SMD technology to that which I used on the amplifier output stage. I'll take a photo of the proto and post it in the PSU threads in a day or two.
HEY HO NOW THIS IS WHAT I CALL FUN!
You simply can't have this MASSIVE amount of fun with a boring old toroid and caps the size of baked bean cans!
Cheers
Kims:
Good results!
Eva:
PSURR=0dB. Correct. I say:So what?
Since when has it been mandatory for audio amplifiers to use unregulated power supplies? Watching the continued use of bulky 1920's vintage technology, mains frequency transformer-rectifier PSUs with hostile power factor is for me somewhere between painful and comical.
PSURR=0dB poses an enjoyable challenge for the designer of open loopers, more so if it's required to have variable Vout. It forces the amplifier designer to consider the PSU-OPS holistically as a single system. Because I know you are a fellow competent 'switchmode artist' I 'm surprised you don't share this view!
The argument that a properly regulated PSU costs more and is more complex has long since been discarded. Switchmode PSU's (even split rail ones) are now well entrenched in high-end amplifiers. The regulator I'm presently developing for each channel of PowerDAC 3 is a 600W average-current-mode synchronous reg with fs=352,8kHz, Vout from 2-60V at 10A max. It requires no heatsink because power stage efficiency = 99%.
The Vout of the regulator is controlled from a digital input command, and the control loop gain is optimised for slope matching at all Vout by digitally controlled gain elements. Control bandwidth exceeds 20kHz. The high efficiency is due to the adoption of similar SMD technology to that which I used on the amplifier output stage. I'll take a photo of the proto and post it in the PSU threads in a day or two.
HEY HO NOW THIS IS WHAT I CALL FUN!
You simply can't have this MASSIVE amount of fun with a boring old toroid and caps the size of baked bean cans!
Cheers
Ha ha good point!
But i suspect the open loop amplifier would require incredible precision from the power supply rails under load condition.
For example if you are working with a 100W amplifier you would use a rail voltage of say 60 Volts per rail.
Now just 10 mV of rail change on the supplies will add 0.02% of THD. This is 10 times the total THD level on the currently best Class D amplifiers.
John Hope: can you really keep this kind of precision, and stability, in real life?
Well, then we need a precise and low distortion closed-loop analog class D amplifier as a power supply for each open-loop digital class D channel... Furthermore, the distortion, noise and frequency response abnormalities of the analog class D power supply will appear directly on the output of the digital amplifier.
So what is the point of such a "cascoded" system? The digital control stuff is made redundant from the beginning as it's the analog power supply what defines performance!!
Hook the speaker directly to the power supply and you will get the same or better results
So what is the point of such a "cascoded" system? The digital control stuff is made redundant from the beginning as it's the analog power supply what defines performance!!
Hook the speaker directly to the power supply and you will get the same or better results
kims said:
Hi all!
In post #481 I asked about FDD5612. It seems I had a good instinct
. Is this some new development board from TI?Jaka Racman said:
This is the equation for THD in BD modulation from TI's aplication SLEA028:
THD = Rpsu*M^2 / (4*Zload)
where M is maximum modulation index. I don't know what is relation for AD modulation.
Best regards,
Dejan
PSRR and SMPS
EVA
PSRR is not 0dB! PSRR depends on the signal you are playing.
Vout=Vpsu*D - so if the audio signal is zero the PSRR should extaully be infinite, but more can be read in :
http://focus.ti.com/lit/an/slea049/slea049.pdf
Lars
The power supply used for the THD measurements I showed in the earlier post is shown in the below document.
http://focus.ti.com/lit/an/slea038/slea038.pdf
the DC/DC concverter A706 has less than 6mR output impedance over the entier audio band - when used with a amplifier with more than 6x1000uF low Z caps (less than 20mR).
For this design the inductor does hardly affect the THD at all - there is only a small rise to 0.02% at 10kHz. We use Micrometals T106-2 or the same material in a rod-inductor. 2pcs 15uH in parrallel.
Rookie
Yep you where right the FDD5612 has very low Qg and Rdson, and yes it is used in a new reference design soon to be released.
the THD as function of Rpus for AD-mode modulation is not as simple as for the BD-mode, but it will at leat always be lower than that.
rgds,
Kim
EVA
PSRR is not 0dB! PSRR depends on the signal you are playing.
Vout=Vpsu*D - so if the audio signal is zero the PSRR should extaully be infinite, but more can be read in :
http://focus.ti.com/lit/an/slea049/slea049.pdf
Lars
The power supply used for the THD measurements I showed in the earlier post is shown in the below document.
http://focus.ti.com/lit/an/slea038/slea038.pdf
the DC/DC concverter A706 has less than 6mR output impedance over the entier audio band - when used with a amplifier with more than 6x1000uF low Z caps (less than 20mR).
For this design the inductor does hardly affect the THD at all - there is only a small rise to 0.02% at 10kHz. We use Micrometals T106-2 or the same material in a rod-inductor. 2pcs 15uH in parrallel.
Rookie
Yep you where right the FDD5612 has very low Qg and Rdson, and yes it is used in a new reference design soon to be released.
the THD as function of Rpus for AD-mode modulation is not as simple as for the BD-mode, but it will at leat always be lower than that.
rgds,
Kim
kims: Thanks for that explanation. It seems your measurements are quite lucky compared to the specs. Ex 100W in 4 Ohms is 5A from your PSU. 5A x 6 mR = 30mV which should give you something like 0.06% THD or maybe even modulation.
What i mean is a 50 Hz signal could modulate a 1 kHz signal through drops on the rails. Have you tested this potential weakness?
Just to keep you updated i now got the new ferrites installed, and that equalized the THD. See curve below (1W 8 Ohms vs freq)
All the best from
Lars Clausen
What i mean is a 50 Hz signal could modulate a 1 kHz signal through drops on the rails. Have you tested this potential weakness?
Just to keep you updated i now got the new ferrites installed, and that equalized the THD. See curve below (1W 8 Ohms vs freq)
All the best from
Lars Clausen
Attachments
..Unbelievable, the new ferrite better in highs, but worse for middle in four times! I never seen such effect of the output coil. In my experience, properly gaped core give almost the same THD figures for many (if not to say "any") ferrites. I can't understand what is the reason for such THD "equalization"..
IVX: Yes i was also baffled by this effect. But it is probably because the lower permeability requires a lower biasing of the core, and thus gives the higher THD at low freq but lower THD at higher freq. Thats the only explanation i can find at this point.
This new ferrite material also features lower THD at power than the ferrites i have used before. See below:
(40 Watts RMS 8 Ohms )
This new ferrite material also features lower THD at power than the ferrites i have used before. See below:
(40 Watts RMS 8 Ohms )
Attachments
Hi Lars
The THD as function of Rpsu is given by the follwing eq for BD-mode:
THD = Rpsu*M^2/(4*Rload)
for AD-mode this eq does not hold - but will at least be lower than that.
Rpsu = 6mR, M=0.94 (max modulation), Rload=8R
6mR*0.94*0.94/(4*8R)=0.016% - not far from the measured results.
With these NON-feedback designs the demand for feedback is moved from the amplifier to the power supply - some where the 50/60Hz signals from the mains need to be filtered and regulated - so for a 6ch design using a common PSU, we have one feedback system e.i. the SMPS, handling the mains frequencies, 100Hz ripple and harmonics thereoff.
Making the SMPS having low output impedance is hardly any issue today with the knolege on feedback systems used in manny class-d amps (just a SMPS with a varaible reference voltage).
Take any class-d amp with feedback, feed it with a dc-voltage, and you have a very good SMPS - and it can even sink current (that many SMPS can not).
The THD graphs you show looks very good - what material is used for the inductors (we use Micrometals -2)
rgds,
Kim
The THD as function of Rpsu is given by the follwing eq for BD-mode:
THD = Rpsu*M^2/(4*Rload)
for AD-mode this eq does not hold - but will at least be lower than that.
Rpsu = 6mR, M=0.94 (max modulation), Rload=8R
6mR*0.94*0.94/(4*8R)=0.016% - not far from the measured results.
With these NON-feedback designs the demand for feedback is moved from the amplifier to the power supply - some where the 50/60Hz signals from the mains need to be filtered and regulated - so for a 6ch design using a common PSU, we have one feedback system e.i. the SMPS, handling the mains frequencies, 100Hz ripple and harmonics thereoff.
Making the SMPS having low output impedance is hardly any issue today with the knolege on feedback systems used in manny class-d amps (just a SMPS with a varaible reference voltage).
Take any class-d amp with feedback, feed it with a dc-voltage, and you have a very good SMPS - and it can even sink current (that many SMPS can not).
The THD graphs you show looks very good - what material is used for the inductors (we use Micrometals -2)
rgds,
Kim
Good sound vs THD
Hi Mikeks
I realise I'm probably opening a can of hysterical and angry worms here, but this is always good fun, so:
I would firstly ask you, have you ever heard what you would classify as 'good sound'. Anywhere, anytime, anyplace?
Maybe you have? Surely yes?
And was that sound reproduced through loudspeakers?
Yes?
Well then the 'good sound' you heard was subject to THD far in excess of the of 0,01% you would set as a criterion for amplifiers. A typical loudspeaker has THD+N of the order of 0,5% to 1% at best, perhaps more if driven hard. Including harmonic and an-harmonic distortion.
Therefore it would certainly seem that to get 'good sound' one could safely forget about all THD unless it was above about 0.1%.
Think holistic.
Have a good weekend and Skol
Hi Mikeks
I realise I'm probably opening a can of hysterical and angry worms here, but this is always good fun, so:
I would firstly ask you, have you ever heard what you would classify as 'good sound'. Anywhere, anytime, anyplace?
Maybe you have? Surely yes?
And was that sound reproduced through loudspeakers?
Yes?
Well then the 'good sound' you heard was subject to THD far in excess of the of 0,01% you would set as a criterion for amplifiers. A typical loudspeaker has THD+N of the order of 0,5% to 1% at best, perhaps more if driven hard. Including harmonic and an-harmonic distortion.
Therefore it would certainly seem that to get 'good sound' one could safely forget about all THD unless it was above about 0.1%.
Think holistic.
Have a good weekend and Skol
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.