|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
130.199.3.130
In Reply to: Re: Hey there Dan. posted by Dan Banquer on March 19, 2007 at 10:02:36:
db: ""
I'll disagree with some of this. If negative resistance is not going to be an issue, then exercising an amp in the 2nd and 3rd quadrant is for academic exercise only. ""Your question was about crossovers exhibiting negative resistance. The answer was no. That does not mean that a typical load does not traverse either quadrant 2 or 4. (Note, these are the quadrants you actually meant, as 1 and 3 are the quadrants which a resistive load will traverse. The distinction of quadrants 1 and 3 are as follows: if a positive signal is being applied to the load, the positive rail output devices will be the ones which current is passing through (quadrant 1), and the same with negative signals (quadrant 3). When a reactive load is applied to the amp, there will be times when the output is positive, yet the negative rail devices are the ones with current within them (quadrant 2), and a negative out with the positive devices having current (quadrant 4).
This is why reactive loads can be so difficult for the SOA of an amp...the voltage across the devices carrying current can exceed rail voltage, but yet the device must carry current.
The bottom line is, for speakers with any reactance, all 4 quadrants will be exercised. One can easily see this with a scope...run x axis as voltage, y axis as current...a resistor will plot a line from upper right to lower left...an inductor will show an ellipse or circle, same with a capacitor. Any plot which is not a simple line is showing reactance. (note, you can use a tek type 576 curve tracer(I think that was the model) set to ac mode to show this as well).
db: ""
The last time I was looking at speaker impedance plots, the max was about +/- 60 degrees for worst case. ""Any E/I phase shift requires all four quadrants operation. That is the definition of feeding a reactive load.
db: ""
High End doesn't matter anymore as the market is now so small it is nearly back to 1950's as a hobby market.""It does appear to be rather unhealthy, however, that was not within the context of our discussion..that of reactive loading of an amplifier, with the consequential issues.
db: ""
Summing it up, time/phase issues are going to be issues of the source, whether it be recording or format, or loudspeakers. I don't see amps/preamps being an issue in that regard, if I'm understanding you correctly.""I do not know what you are "summing up". While I have real issues with the fact that all the recorded source material has had imaging cues butchered beyond recognition, this discussion was about the output stage of a power amp, the fact that it exercises all four quadrants, and that a passive load will be unable to exercise the output stage adequately to allow measurement of the output parameters under such stress.
Cheers, John
Follow Ups:
"When a reactive load is applied to the amp, there will be times when the output is positive, yet the negative rail devices are the ones with current within them (quadrant 2), and a negative out with the positive devices having current (quadrant 4)."Are you sure about this? I have never seen any plots that would indicate that the above is correct.
"This is why reactive loads can be so difficult for the SOA of an amp...the voltage across the devices carrying current can exceed rail voltage, but yet the device must carry current."
I think this is more of a rarity than one might think. I am going to presume for a moment that you are talking about the peaks of music which are not very fast transients, so the rate of change is generally pretty manageable, and they are still well within SOA of the amp. Well er, uh, most reasonably well designed amps.
My general response to the above is that crossover/speaker design in general has realized that high reactive/inefficent loads limit their sales so designers are now more careful about this than in the past. I'ts not the major issue it was 10 or more years ago.
John: These days I don't think that much about this stuff, I'm tending mostly to other things, like ripping close to 500 CD's (uncompressed) and putting them on my hard drive, and trying to find the time to repaint the kiitchen.
d.b.
db: ""
Are you sure about this? I have never seen any plots that would indicate that the above is correct.""Guess you don't get out much, eh?? Don't you remember Time Tunnel?? That huge lissajous? :-)
The equations are easy. When you plot the voltage and current of a reactive load, it doesn't go through the origion. It's also easy to see on a scope.
The easiest way is on a 576 curve tracer in AC mode..but it's still easy with a speaker and music. After all, it's just phasor analysis.. Make a simple current viewing resistor, say 100 milliohms, and use it's drop for the scope vertical, the amp output voltage as the horizontal, and any speaker as the load..if it's pure resistive, it'll be a straight line through the origin..regardless of the musical content.. But any reactance will cause the line to form an ellipse..that means operation in 2 and 4 as well as 1 and 3..but be careful of the .1 ohm resistor, it'll also have a non trivial b dot error component at typical audio frequencies, so any measurements you get will be useless unless you make the CVR correctly.
As for speakers and reactance?? Guess you don't listen to those resonators much, eh? One note wonders made for cars.
Sheesh, even the SWTPC 250 had SOA protection which extended into quadrant 2 and 4..waaay back then..
Amp design/system reactance is important in two applications. First, when the user is beatin the livin hell outta the amp...which I tend to do...:-( ..second, when one is nitpicking the daylights outta the image reconstruction parameters. (which I can't be bothered with given today's recording/delivery technology).
Cheers, John
Cheers, John
O.K., if it does go into the 2nd and third quadrants that implies negative resistance. How can that be?
Curve tracers? I haven't seen one of those in at least twenty years!
d.b.
db: ""
O.K., if it does go into the 2nd and third quadrants that implies negative resistance. How can that be?""
It doesn't imply negative resistance. It states reactive phase shift, stored energy. Going into quads 2 and 4 is a simple reaction of the load to changes in the output that the load cannot follow.Think of charging an inductor.. When the voltage is positive, the inductor current increases..this continues until the sine lobe hits zero volts..at that time, the inductive current will be at a maximum. Then (here's the important part), when the amp voltage goes negative, the current within the inductor will go down..but being an inductor, it's current cannot change instantly, but will ramp down based on voltage and inductance..so when the amp voltage goes negative, the inductor current is POSITIVE...going down, but positive nonetheless..
So, negative voltage, positive current...that be quadrant 2..
db: ""Curve tracers? I haven't seen one of those in at least twenty years!""
I'd take a pic of the one here, but this site doesn't support jpegs. I still use it on occasion to test diodes and the such. It works great for my 3 inch diameter, 11 pound diodes (I'd send a pic (one next to the monitor), but this site...sigh.... Guy in the next room might be buying 1500 cree silicon carbide schottky's for a repair of lots of switcher supplies..I man the 576 for him..
Whow Nellie! If your talking a near pure inductive load you are correct. But a loudspeaker driver will have the voice coil inductance and resistance so you should never see the full effect of what you describe. To take it even further, the Back EMF of the driver will have a good deal of that dissapated by the series choke in the crossover.
d.b.
No, I am not talking about just a pure inductance.Any reactance whatsoever will do it.
VC inductance is one part, another is the energy stored in the acoustic system, like the cabinet.
The series choke has both inductance and resistance. Series inductors are not normally designed to dissipate excess energy coming back from the driver. It is the amplifier which has to absorb that.
An amp is designed to do one thing. Provide a desired voltage in time.
It has to do this regardless of what the load current is. Again, at all times.
The biggest problem with load reactance is, it can drive the current in wonderful ways that the amp has to be able to absorb.
Supply regulation and loop stability are all that I can offer at this time. To tell you the truth, it may be all anyone can offer.
d.b.
db: ""
Supply regulation and loop stability are all that I can offer at this time. To tell you the truth, it may be all anyone can offer.""
Nah, there's more. Heck, even SWTPC figured it in back when velociraptors were chasin me.At the immediate level, load reactance can seriously impinge on the secondary breakdown region of bipolars..it can really go bad when the devices are made with a thin base region for gain, as that impacts the zistor's ability to turn off in the heat of the moment.. Many drive circuits cannot supply sufficient alternate direction current to shut off a zistor that enters heavily into second breakdown. It's really neat to watch this effect, I can detail it if you wish..
At the listening level, it can impact the mids/highs if the bass driver causes the output zistors to become unruly, the drive circuit can become overwhelmed short term..
And of course, the transient thermal excursions are a reliabilty concern.
Are you saying that back EMF can then prevent the transistor from shutting off properly or suffer from secondary breakdown? Would this be due to the extra heating from current being pumped into the transistor going "the wrong way"? The phase of this current would be something approaching 180 degrees out of phase, would it not?
morricab: ""
Are you saying that back EMF can then prevent the transistor from shutting off properly or suffer from secondary breakdown?""Yes. For inductors, it's the stored energy which manifests as a pseduo constant current, and it will continue to supply that current even if the output goes to the opposite polarity. Perhaps it's easier to envision a large capacitor as load..have the amp ramp slowly to the negative rail, then once the load cap is charged, change the output to the positive rail..the cap doesn't move quickly, so the output stage draws lots of current in an attempt to move the cap. The pass devices trying to pull the cap up are from the positive rail, so the devices are dissipating both rail voltages and whatever current being forced..in the case of an inductive load, it's quite easy for the pass devices to move the out to the other rail, however, since an inductor cannot change it's current as quickly, it continues to have current in the same direction for a while. A more controlled current, but the passes still have to hold it for double rail voltage momentarily. (it's easier for most to envision caps rather than inductors, but they are both reactive storage mechanisms.)
If this brings the outputs close to the secondary breakdown, poof. (bipolar discussion). As devices approach this dangerous region, several mechanisms happen..as you raise the C-B voltage, the neutral base region gets thinner (early effect), increasing the gain of the device. Also, the leakage collector to base increases, this leakage is amplified by the device gain.. And thermally, gain is also going up..these conspire to make the output devices "cramp up"..this is secondary breakdown.
..Consider a darlington config output..if nothing is put in to pull the output bases down when this "cramping" happens, the device will clamp on. As this starts, the only way to stop it is to pull the base down hard. If the only thing the base sees is an emitter, then that driver has to go reverse breakdown base-emitter to get any current into the base. (note, I've been using current nomenclature loosely for ease of discussion). That'll probably toast the driver e-b junction.
This seems to be primarily a problem with bipolars because if I am not mistaken what you are describing is a form of thermal runaway. FETs tend to have the opposite effect under thermal stress and reduce gain if I am not mistaken. Sorry if I am not talking in quite a technical language, I understand the issues conceptually quite well but the actual jargon I am not so experienced in using. Keep talking away though because I appreciate the lessons. Now if I am also not mistaken many (if not all) BJT amps require extensive thermal tracking to prevent such a situation from getting out of hand. A Class A mosfet amp would not require this thermal tracking as they are more or less self clamping.I find it a relevant topic because I use a big pair of full-range electrostatic speakers. These are about as reactive (mainly capacitive) as a speaker is likely to get. A very large percentage of the signal fed into the speakers is sent right back into the output stage of the amp. Now I am using a tube hybrid amp (with the tubes on the output unusually) without negative feedback. I think the tubes (or the output transformers) are able to dissipate this back EMF without trouble and of course there is no breakdown possible. What I have found though is that a lot of SS amps sound very anemic on these speakers, with my modest 30 watts sounding much more powerful. I am wondering if the speaker is regularly putting the output stage under a lot of stress or if a significant portion of the back EMF is making its way back into the input and being reamplified (as described by Otala as IIM distortion).
morricab: ""
This seems to be primarily a problem with bipolars because if I am not mistaken what you are describing is a form of thermal runaway""While I did say "bipolar discussion", I did neglect to say the word thermal runaway...sorry bout that. I was describing thermal runaway as a result of secondary breakdown in a bjt...
The thermal tracking for most bjt amps has a far slower time constant. The silicon die and the thermal structure have several time constants, ranging from the 100 uSec range for the junction, to the millisecond range for junction to case, to the seconds from case to heatsink. So the thermal tracking schemes typically correct in the second or so timescale. Some schemes can be faster, like the sensefets, or that new device with an onboard sense device.
I could guess as to your amp's reaction to it's load, but that'd be a guess. I'm talking with Dan (and anybody else who's interested) about a test scheme for amps which cover reactive loads. Sent him a first pass jpg (then of course, modded it into a better scheme) for a setup. Rev 2 has a programmable resistive load, as well as a completely independent reactive load control, from capacitive through to inductive. That type of testing regimen is what is needed to get to the bottom of why you hear differences between amps..It may also be the root cause of tube vs solid state sound...who knows..
Hopefully, it can get posted here in a new thread..I fear that I will not be able to do it at AH, as I am "this close" to being banned there because I posted my displeasure with having some of my analysis misrepresented as part of an AH article. Apparently it is allowed to misrepresent the words of another, and there is no rebuttal allowed on an open forum. In my humble opinion, that is an abuse of the term "open discussion".
I'm a bad boy..:-)
Well as long as DB can remain civil I am more than willing to dive in and learn (hopefully contribute a bit as well but I have to think more deeply on these things first). I appreciate a sane discussion for once on this forum.
Ah, yes...now I recall some kind of animosity between you two. I honestly do not know why, nor am I really interested to know why.I agree, I hope civil discourse continues..
Can you post diagrams here?
If that's your only complaint then Motorola has been making TO3 pkg transistors for years that will take that stuff. Go to On Semiconductor and look up MJ15003, MJ15004, MJ15024, MJ15025, etc. etc. Parallel a couple or more of these and you can satisfy any audio requirement. Heatsink them right, and they will last longer than you and I will.
Sheesh; and I thought for a minute you where you were going to ask more hard stuff.
d.b.
There's a difference between just tossing more silicon mass at the problem, and engineering it. Anybody can put a dozen 15003's in to do the job required of one if engineered correctly..sheesh.But the question is, what's the damping factor at mid/hf while the device is dissipating reactive energy?? Does the drive accurately keep the outputs in line at all frequencies during sweeps through quads 2 and 4?
That's the question..not can the outputs survive..
First of all Damping Factor is really a misnomer, it should be output impedance, that is if you're Dick Pierce.
I've measured into a resistive load over power and frequency and got a minimum of 12 milliohms to a maximum of 25 millohms.
I am not convinced just yet that output impedance would change into a reactive load, or at least change significantly.
Cheer up, if it doesn't I won't ask you to eat your shirt collar.
d.b.
shirt collar..he he..bet some here have no clue what that refers to..How did you measure the hf damping factor with an inductive load being driven at a low freq?? Yah I know, you just used single tone, didn't ya...so old school there dude..you gotta get into the future..
Best way I know of is a second amp, with a load resistor between them.
And two signal sources and a set of summers..
I can detail ya the setup if your really interested.. Unfortunately, it does start to define new standards of test and performance..
Yeah, I'm old school. Now my amp is a sliding bias and the output transistor bias will vary with signal voltage. The higher the signal voltage the higher the DC Bias. Now, as us circuits guys know, the higher DC bias will reduce output impedance. So your theory may well be more appropriate to AB designs with a "static" DC bias is my guess.
What do ya think there John;
d.b.
P.S. Post your technique, I will archive it.
nt
db: ""
So your theory may well be more appropriate to AB designs with a "static" DC bias is my guess.
What do ya think there John;""I think you are correct...you are just guessing...:-)
(sorry, couldn't pass that one up...)
Honestly, the discussion was without regard to topology. I do not know which topology will be better or worse.
db: ""
P.S. Post your technique, I will archive it.""Which part of "I can't post pics here" did you not understand???
Duh...
(boy, your given me such material today)...:-)
btw, I'm just preppin ya for your test..
I could put a drawing together for ya, might take a day or two..but I'd hafta e-mail it to ya..
I am not understanding how the current could go into the 2nd or 3rd quadrant if the resistance is always positive.
Am I missing something?
d.b.
db: ""
I am not understanding how the current could go into the 2nd or 3rd quadrant if the resistance is always positive.
Am I missing something?""Yes, big time.
You have to consider the return of stored energy within the load back to the amp.
You are forgetting that.
O.K. but that would oppose the energy going out, the feedback would further oppose the energy coming back in, as it would feedback the output voltage,and since when do amplifiers like this sink current? and I still don't see how we could ever go to negative resistance in this application.
I don't get it John:
d.b.
nt
This post is made possible by the generous support of people like you and our sponsors: