You're asking me to unravel something that someone made up to explain something they didn't understand. The peak power dissipated by an amplifier driving a complex load can be computed we do not need a new term anymore than we needed TIM to explain full power BW and slew rate limitations.
I bow to your expertise Scott, however the fact remains that EPDR is being used by the Hi-Fi press to describe how benign or otherwise a loudspeaker is as an amplifier load.
If EPDR is a fabrication, then it is a useful one to the laypeople who read the magazine reviews.
If anyone is still interested, here is a link to the Stereophile article which describes the technique by which the Hi-Fi News assessments are carried out.
Heavy Load: How Loudspeakers Torture Amplifiers Page 2 | Stereophile.com
At least I tried to make a positive contribution to the thread!
If EPDR is a fabrication, then it is a useful one to the laypeople who read the magazine reviews.
If anyone is still interested, here is a link to the Stereophile article which describes the technique by which the Hi-Fi News assessments are carried out.
Heavy Load: How Loudspeakers Torture Amplifiers Page 2 | Stereophile.com
At least I tried to make a positive contribution to the thread!
At least I tried to make a positive contribution to the thread!
No problem, Eric is an old friend they point out a speaker with a nominal impedance can make extra demands on an amplifier due to crossover design or other issues.
Look for a textbook named "AC Circuit Analysis." (generic title, many authors have written such textbooks). Someone posted a title they liked here about a month ago (in this thread that's hundreds, maybe a couple thousand posts back, sorry). It goes through Ohm's Law with resistors, and how to use complex numbers (values with phase angles) for the values when a circuit also has capacitors and inductors (reactance).
Do they still use PRAT? I haven't even read that in the magazines, I rarely read them anyway, I've only read about it online.
Looking at the graphs, they're power vs. frequency, something that can apparently be discerned from the more common impedance vs. frequency graphs. It seems to me the impedance vs. frequency graph is more useful (ignoring the phase for simplification, when the impedance goes down by half, the power doubles). EPDR seems to be an acronym to make it seem (even) more complicated than it is.
Learning LRC circuits is a bit complicated (it was a quarter course in electrical engineering for me a few short decades ago), but it really helps cut through the crap.
Hmm, that article also uses the word modulus many times, and I can't quite figure out what it means in that context.
This is a tough crowd!At least I tried to make a positive contribution to the thread!
Say an amplifier could put out 200 watts at 8 ohms, and make the amp dissipate 100. 67% efficiency, pretty much par for the course for a regular old class AB amp right at clip. Now load it with a pure capacitor (or close to one) where it draws 200 volt-amps. So where does all that power go? Back into the amp, making it dissipate 200 actual watts. Which is what the amp would dissipate with a 4 ohm resistive load. Real world won’t be quite that bad, but it gives you an idea what the dissipation could approach. An ESL at high frequency will get you close. Dual driver systems with high Qms woofers can get pretty bad too - at frequencies between the peak and the minimum. Phase angles can get over 60 degrees with the magnitude being a few ohms off the minimum. You can get a lot of stress in a very narrow band of frequency - usually in the low midrange where you really don’t want it. Another consideration is that reactive power puts the most dissipative stress on the output transistors at full voltage, not half way up the rail as it does driving a resistor.
Last edited:
For a voltage sweep, the impedance never dips. And Ottala was simulating speakers with crossovers, it's true. If I go back to your first sentence, though, would you additionally say the impedance can never be less than DCR even under transient conditions with the cone moving?
The best book for me concerning loudspeakers is this under
Lautsprecherboxen. Aufbau - Nachbau - Umbau: Amazon.de: Gotz Schwamkrug: BA1/4cher
Schwamkrug- Lautsprecher-Boxen, Grieder Elektronik Bauteile AG
and this (both from ELEKTOR):
Lautsprecher (PDF) - Elektor
www.hornlautsprecher.de - solutions in sound
download for free.
Unfortunately not available in english at those days.
Maybe available in english In the meantime.
It was only after studying these books that I understood loudspeaker technology.
Last edited:
Douglas Self demonstrated how, with very special signals, the dynamic impedance of a loudspeaker with passive crossover can appear to be lower than its DC resistance :
"Loudspeaker undercurrents" Electronics World, February 1998-99
Keith Howard also made investigations on the topic :
Heavy Load: How Loudspeakers Torture Amplifiers | Stereophile.com
"Loudspeaker undercurrents" Electronics World, February 1998-99
Keith Howard also made investigations on the topic :
Heavy Load: How Loudspeakers Torture Amplifiers | Stereophile.com
Last edited:
Thanks benb, I have studied AC theory and didn't class myself as part of the confusion till Scott Wurcer suggested its existence! 🙂
The impedance vs. frequency graph does not tell the full story since, at certain frequencies, the impedance dips to values below the minimum impedance shown on the graph.
This behaviour is what is under discussion in this thread and that is what EPDR, rightly or wrongly, attempts to quantify for the non-technical reader.
Regards!
You are obviously knowledgeable about amplifier peak output stage power dissipation.
It would appear to me that EPDR is effectively an inverse of power dissipation, i.e. minimum EPDR occurs at peak power dissipation.
Resistance is an easier concept for the layperson to understand than peak amplifier power dissipation, hence the use of the term EPDR by the hi-fi press
Do you consider that EPDR is a reasonable way to quantify how difficult a load a speaker presents to an amplifier?
I bow to your expertise Scott, however the fact remains that EPDR is being used by the Hi-Fi press to describe how benign or otherwise a loudspeaker is as an amplifier load.
If EPDR is a fabrication, then it is a useful one to the laypeople who read the magazine reviews.
If anyone is still interested, here is a link to the Stereophile article which describes the technique by which the Hi-Fi News assessments are carried out.
Heavy Load: How Loudspeakers Torture Amplifiers Page 2 | Stereophile.com
At least I tried to make a positive contribution to the thread!
While I do not care for the terminology as it's too simplistic, I can understand where it's coming from.
A resistive load will only force dissipation in the pass elements of the polarity of the signal. IOW, for a positive voltage, the output stage on the positive rail will be doing all the dissipating with a resistive load, and the maximum voltage during dissipation will be the single rail voltage (within the straight load line of course).
When the load is reactive, the current and voltage are no longer in phase. This means that there will be times when the voltage at the load is negative yet the positive output elements are driving current. This makes the elements drive currents while they see much more than a single rail voltage.
For bipolar devices, pushing the currents at higher voltages is much more difficult (secondary breakdown). This was a big driver for SOA testing, BVceo(sus) testing of transistors back in the day. SWTPC wrote this up back in the 70's in their tigersauruses manual, they showed a curve in the load line to represent what happens to device dissipation due to reactance.
Many amps use a current fold back scheme to protect the output devices, others overkill the output capability to cover the reactive loads expected.
The difficulty with fold back is, a three way speaker will require much heavier currents when the stimulus is music as opposed to a sweep. One only needs look at the V/I plot during music to see how much dissipation occurs in quadrants 2 and 4, these regions are where the fold back will compromise the output fidelity.
I have not yet seen a write up on the complexity of driving three reactive frequency dependent parallel loads and the dynamic impact on pass element dissipation. This simplistic EPDR is a start.
I do not know if this simplistic measure is due to the limitations of the technical people trying to figure it out, or due to them attempting to explain it to a non technical audience so dumbing it down.
Edit: Motorola for example, started screening devices by applying some large voltage then high current. This dissipation at high voltage was used to screen out defects in die attach as well as current crowding due to loss of base width at the higher voltages. Some diffusion processes were not uniform enough either across the die side to side, or even across the width of an emitter finger. As voltage increased, less and less of the actual emitter would carry current due to "early", and gain runaway also happened. Back in the day, even the base resistance would factor heavily, as the pattern of fingers on the die may not have supported high base currents sufficiently, lateral base resistance causing emitter area decay.
Jn
Last edited:
On a separate note...
In my motion control work, the ratio of lead screw inertia to load dynamic inertia creates a complex frequency dependence and so throws phase margin all over the map. The simplest system is fourth order, and a sweep plot shows large resonance as well as anti resonance.
I've not seen data along the lines of the current discussion, however do indeed wonder if such behavior can occur with fourth or sixth order speakers?
Jn
In my motion control work, the ratio of lead screw inertia to load dynamic inertia creates a complex frequency dependence and so throws phase margin all over the map. The simplest system is fourth order, and a sweep plot shows large resonance as well as anti resonance.
I've not seen data along the lines of the current discussion, however do indeed wonder if such behavior can occur with fourth or sixth order speakers?
Jn
To me "dynamic impedance" implies a non-stationary system and conventional analysis does not apply.
EDIT - After reading the Stereophile article I agree with the conclusions, the concern is power dissipated in the amplifier driving complex loads and the information is there in conventional measurements and simply needed a different presentation. "Dynamic impedance" to simply mean complex load just adds confusion. Unfortunately EDPR does not treat phase margin or stability issues.
Last edited:
I strongly suggest the latter! After all, ninety-nine point how many nines percent of the population are not electronic engineers! 🙂
Most of what you have said is beyond the scope of my humble physics degree, but I value the fact that you have given of your time to contribute to the discussion.
It looks like "modulus" means the real component of the impedance.
I note with "casual interest, dismay, and silent smirking", that the entire discussion centers around simple sine wave excitation.
Really? That's where we are at after 50 years or so???
Sigh..
In this day and age, it is trivial to measure the pass currents, the pass element V drops, and compute in real time what any load does to the amplifier dissipation. Why has that not been done yet?
It is even trivial to make a system that goes between the amp and load to measure the reactive loading for any arbitrary music signal. Also, to simply provide that system the rail voltage numbers to calculate the dissipation as a total, as well as how it splits rail to rail.
This is not rocket science.
Jn
Ps. I also note their test methods do not consider the actual energy being delivered to the load vs frequency. The power required of the tweets and resulting amp dissipation is not the same as the woofs. Using real music to do the test would be more exacting.
I note with "casual interest, dismay, and silent smirking", that the entire discussion centers around simple sine wave excitation.
Really? That's where we are at after 50 years or so???
Sigh..
In this day and age, it is trivial to measure the pass currents, the pass element V drops, and compute in real time what any load does to the amplifier dissipation. Why has that not been done yet?
It is even trivial to make a system that goes between the amp and load to measure the reactive loading for any arbitrary music signal. Also, to simply provide that system the rail voltage numbers to calculate the dissipation as a total, as well as how it splits rail to rail.
This is not rocket science.
Jn
Ps. I also note their test methods do not consider the actual energy being delivered to the load vs frequency. The power required of the tweets and resulting amp dissipation is not the same as the woofs. Using real music to do the test would be more exacting.
Last edited:
I strongly suggest the latter! After all, ninety-nine point how many nines percent of the population are not electronic engineers! 🙂
Most of what you have said is beyond the scope of my humble physics degree, but I value the fact that you have given of your time to contribute to the discussion.
I would argue that point with you. It is only because some of the terminology and concepts are not familiar to you, not that my discussion is beyond your degree or intelligence, nor most here.
Jn
You sure know how to boost a guy's confidence Jn!
Thanks, and be assured I'm learning all the time from experts such as yourself. 😎
Last edited: