That would be Madsen and Soerensen, TI Applications Report SLEA049, June 2005. Plucked from Cordell, page 600.
No, I haven't read it. It may be distantly relevant here, but based on the magnitude of the effect with non-switching amplifiers, I suspect it will be a very small effect. But it is intriguing. I take IM distortion very seriously.
TI almost hired me to do a hybrid higher-power digital-modulator-based switchmode amplifier, so they, while working in Denmark on integrated solutions, could displace evil D2 Audio from Harman products (the adoption of D2 Audio amps being one of the stupidest decisions for a receiver ever made in the hallowed halls of pre-Paliwal Harman). I would have probably done, by and large, what TI eventually had to do to improve---wait for it---Power supply rejection!
I would have had my chance, but my cost estimate was a factor of about four over what their undisclosed but hinted-at budget was. The result was another waste of time.
I do this a lot. I would say I did this a lot, but I am, sadly, still likely to do it again. But life goes on.Dilbert strikes again.
Is this sufficiently high enough, listening-wise, and this is CFA?
Attachments
http://www.ti.com.cn/cn/lit/an/slea049/slea049.pdf
But I think abraxalito is talking about this one (which happens to be in line with your way of thinking about using THD to access PSRR) :
http://www.eetasia.com/STATIC/PDF/200909/EEOL_2009SEP04_ACC_TA_01.pdf?SOURCES=DOWNLOAD
George
Sorry for the late reply. Let me approach this from two different angles to explain why it is extremely unlikely that a driver's impedance dives below its Rdc.
First of all, take a look at the physics involved. If a driver moves in the same direction as it is forced by a current through the VC, the VC generates a voltages that is of opposite sign as the voltage which initially forces the current through the VC. The result is an apparent increase in VC resistance.
In order to generate an apparent decrease in VC resistance, the VC has to move in the opposite direction as the current through the VC forces it to move. There is no external force which can cause this to happen. Closest in physical terms is the tuning point in a bass reflex enclosure, where a maximum breaking force is exerted on the driver because the air pressure inside the enclosure is in counter phase with the driver. This is the point where the drivers impedance approaches its Rdc.
Another way of looking (but it boils down to the same) is that it requires a lot of energy to virtually decrease the impedance of the driver. Take an 8 Ohm speaker fed by 8V, which at its Rdc will draw 1A for a total of 8W. Now suppose that through back EMF, the impedance of the driver dives to 4 Ohm. The driver will now draw 2A, and the amp will deliver 16W into the driver. Of these 16W, only 8 will be dissipated by the Rdc. The other 8 will have to be dissipated in another way, and it is by expending the energy required to generate the back EMF. In other words, in order to lower the Rdc, the mechanism causing this to happen would cost as much energy as it would draw additionally from the amp. There is no mechanism that can input that much energy into a driver.
Last edited:
I worked with the 'S' company for a year in Tokyo to design in some 100W RMS class D amp IC's.
The cost pressure on those 'boom box' products was incredible. Lots of corners cut on things like transformers - a 150 VA unit cost them about $3.50 and filter caps (2 x 2200uF35V) around 12c a piece.
I went up to the 23rd floor in the Oval building in Shinagawa where they had their audio design group to do a 'sound taste' test on one of their units fitted out with our chips.
The room was full of esoterica - ML power amps, B&W Nautilus speakers, 802's plus an assortment of Dali's etc and some very nice looking high end stuff of their own.
Anyway, they pulled out some speaker stands, put the boom box speakers up on them and connected the receiver to them and sat down for a serious listening test.
We had all sorts of comments back about the bass not firm enough, treble ok, mid range too recessed etc. They played some real dance music crap - terrible stuff compressed to hell.
I thought to myself 'how can you hear this on lo-fi single drive unit speakers and a 150VA transformer with a totally sub-optimal filter bank?' My overwhelming feeling was that they were just going through the motions.
Anyway, we got the business. About 6 months later we get a call in our Malaysia office - there's a problem with your chips - a whole lot of them are not performing.
The app engineer soon found the issue. The speakers were supposed to be 3 ohms and a whole batch of 4 ohm units has been mixed up in the production batch. Problem solved.
These products by the way were targeted at the South Amercian market - big business over there for them at the time.
These kinds of products are absolutely designed to a price - and the biggest area of skimping was the PSU in my view . . . you see the same thing at the 'P company, Samsung, LG.
Later on we exited that business and focused on mobile.
(to prevent the rail pumping problem, the input of one channel is inverted and then the speaker outputs on that channel reversed)
The cost pressure on those 'boom box' products was incredible. Lots of corners cut on things like transformers - a 150 VA unit cost them about $3.50 and filter caps (2 x 2200uF35V) around 12c a piece.
I went up to the 23rd floor in the Oval building in Shinagawa where they had their audio design group to do a 'sound taste' test on one of their units fitted out with our chips.
The room was full of esoterica - ML power amps, B&W Nautilus speakers, 802's plus an assortment of Dali's etc and some very nice looking high end stuff of their own.
Anyway, they pulled out some speaker stands, put the boom box speakers up on them and connected the receiver to them and sat down for a serious listening test.
We had all sorts of comments back about the bass not firm enough, treble ok, mid range too recessed etc. They played some real dance music crap - terrible stuff compressed to hell.
I thought to myself 'how can you hear this on lo-fi single drive unit speakers and a 150VA transformer with a totally sub-optimal filter bank?' My overwhelming feeling was that they were just going through the motions.
Anyway, we got the business. About 6 months later we get a call in our Malaysia office - there's a problem with your chips - a whole lot of them are not performing.
The app engineer soon found the issue. The speakers were supposed to be 3 ohms and a whole batch of 4 ohm units has been mixed up in the production batch. Problem solved.
These products by the way were targeted at the South Amercian market - big business over there for them at the time.
These kinds of products are absolutely designed to a price - and the biggest area of skimping was the PSU in my view . . . you see the same thing at the 'P company, Samsung, LG.
Later on we exited that business and focused on mobile.
(to prevent the rail pumping problem, the input of one channel is inverted and then the speaker outputs on that channel reversed)
Last edited:
No cap multipliers in that CFA whose PSRR I presented, just RC filtering for the IPS.
This is PSRR with no power filter caps at all.
Attachments
Last edited:
Not trying to split hairs, but a CM doesn't change the PSRR. It decreases the power supply noise/ripple. And thus, less of it gets to the amp output. But the amp PSRR stays as it is.
A minor point but worth making IMHO.
Jan
Nice, thanks.
Now, recall that this is a steady state single frequency excitation plot.
If music is forcing heavy bass, midrange, and highs, The resultant load will be the paralleling of the three drivers. If you see 8 ohms bass, 8 ohms mid, 8 ohms high simultaneously, the amp will see a load less than 3 ohms. The only saving grace then is the fact that the mids and highs are not very large voltages.
We have this problem at work as well. The users tend to go through turn on sequences which are guaranteed to latch component inputs via the scr mechanism (digital), or damage analog ins.
Turning on the input signal to a differential input pair that is unprotected will basically guarantee E-B reverse bias breakdown conduction if the input stage is not designed with appropriate protection.
An employ a long time ago would use NPN's upside down, much better sat voltage. They'd require BVebo, the test system never seemed to damage the componenets that I was aware of. But it was a very controlled test.
I hilited the part. It's important to speak clearly on this, as it does indeed depend on how one looks at the schematic, that of an amp paralleled to the VC, or in series with it. Nonetheless, the EMF generated reduces the drawn current in steady state operation. Which is what you've stated.
Steady state, you are correct.
When the amplifier is trying to brake the system, the math changes.
John
Last edited:
Vacy,
As already mentioned you are looking at impedance as a constant for any given frequency. If we consider it to be influenced dynamically the issue changes.
In the case of an impulse into a transducer, the transducer with adequate damping factor may stop moving with the cessation of the current, but the air in the enclosure will not. This is sort of like tapping on the loudspeaker cone while measuring impedance. Now as most consumer drivers are low efficiency I can see this impulse mechanism dropping the impedance a bit below the DCR.
However where we really disagree is " There is no mechanism that can input that much energy into a driver." As you know we work in quite different environments. In my world there is a mechanism that can impart more energy into a transducer than the audio power amplifier! We refer to it as "Wind." I did a stadium where the amplifiers were only rated down to 4 ohms. I had to change them to ones that could drive much lower even though the rated impedances were above that. When the wind kicked up things sounded quite off.
As already mentioned you are looking at impedance as a constant for any given frequency. If we consider it to be influenced dynamically the issue changes.
In the case of an impulse into a transducer, the transducer with adequate damping factor may stop moving with the cessation of the current, but the air in the enclosure will not. This is sort of like tapping on the loudspeaker cone while measuring impedance. Now as most consumer drivers are low efficiency I can see this impulse mechanism dropping the impedance a bit below the DCR.
However where we really disagree is " There is no mechanism that can input that much energy into a driver." As you know we work in quite different environments. In my world there is a mechanism that can impart more energy into a transducer than the audio power amplifier! We refer to it as "Wind." I did a stadium where the amplifiers were only rated down to 4 ohms. I had to change them to ones that could drive much lower even though the rated impedances were above that. When the wind kicked up things sounded quite off.
- Status
- Not open for further replies.