raising driver Qts - you can't tuna fish

[mondogenerator]

Haha thanks KSTR, obviously I didnt invent this! It just struck me that it could produce some, perhaps even useful, data. And would be fairly simple to try. I didnt realise it was Laplace....oops. I guess I might be able to use that within a function block in something like MatLab then....hmmmmm

I havent read that paper, so thank you for posting it.

I still cannot see why power loss occurring in the output resistance of a 'highZ' amplifier, is better than the power loss occuring in a loudspeaker cable and 'lowZ' amplifier output, if the total Z are of equal magnitude.

surely all this does really, is cause a small volt drop, and also current limit, by loading the output stage down a little. hence reducing power. The problem is when L or C is added. as many have said, either way hiZ or loZ both have to meet transients of both V and I due to reactance.

Its a little like discussing whether to make a 2 stage amp to deliver enough current, when really the more current you get better. thats what has appealed to me with some MOSFET designs, their supposed high current capability. ive found many //BJTs to work well, beefing up o/p stages with extra pairs etc. all this stuff helps. but from what i know FETs in general dont do VAS too well. i think. So maybe then we make an amplifier, with some high bandwidth high voltage switching devices. step the voltage down using a valve OPT. Which would be better? Which would have the higher o/p impedance?

Or why arent there more push pull VC dual magnet front and back type woofers?

im no expert in that either though.

[john k...]

I think you guys are going around in a circle. I tried to address this before by making the distension between the driver and the system of which the driver is a part. When you put a driver in a box the the compliance of the system decreases. This leads to fc > fs, higher Qmc > Qms and Qec > Qes, and this system has a difference impedance than the driver as well.

[mondogenerator]

ok. thanks.

so is the power lost in an output device 'better' power loss, than that lost in the speaker cables?

[DBMandrake]

Originally Posted by mondogenerator

Im also not sure a high impedance amplifier is good, OR that bad. ive had SS amps with output impedance of around an Ohm, and others with quoted impedance of 0.1-0.2 Ohms. I think in the main this was due to the former having output choke, and the latter being DC coupled. former BJT class AB and letter MOSFET class H.... the lesser bandwidth, slower ancient BJT, with the higher impedance just sounded more musical.

Ok so most here are audiophiles/DIY nuts. including myself. to me its simple.

If you have a trusty amp youre gonna be keeping forever, then its easy enough to try.
Its something to try if youre planning to make a amp and speaker system, and design the speakers for a higher Z amp.

At the end of the day, an output impedance is a gate Z, before we even deal with speaker wire Z and crossover Z. Resistance at the gate is still power loss, its totally ignored in anyone arguements.

Maybe it is worthwhile to compare gate dissipation between 2 devices with very different output Z.

Thats what id like to see.

I think you're getting confused between closed loop output impedance of an amplifier and the impedance of the actual output devices in the amplifier.

In a solid state amplifier where the output impedance is, say, 0.2 ohms, this does not mean the output devices have an impedance of this magnitude, and therefore have low losses. On the contrary.

The output devices in an amplifier have a relatively high series impedance at any given instant on the order of tens or hundreds of ohms, and their series impedance is constantly varying with both input signal and load impedance.

It's negative feedback which lowers the output impedance to this level. This can either be loop feedback around several stages, local degeneration, (emitter follower etc) or both...it doesn't really matter.

The point is the feedback varies the high series impedance of the conducting output device in just the right way to cause it's output voltage to be maintained despite variations in load impedance - and by ohms law this means the equivalent output impedance is low. The power losses in the output devices are not reduced due to the feedback.

[mondogenerator]

of course. My error. In which case i assume the nfb attenuator feeding the previous stage, or its own input, has a role in the output z, thus less nfb and higher nfb loop z in // with the output terminals. Roughly speaking. Depending on shunt/series topolmgy etc.

[speaker dave]

Here are the curves I have mentioned. They were done when I was at KEF on a 103.2 8" woofer bookshelf. The amplifier was an old Fisher 80AZ tube amp. The "AZ" part is a variable output impedance that uses a combination of voltage and current feedback to vary the output impedance for typically low to a measured 57 ohms.

The B&K oscilator had a feature called a compressor loop that allowed a measured voltage to be fed back and, within a wide range, it would vary its output to make that return path constant. This is done in a well filtered (slow) way, much like a hand on the volume control. This feedback was absolutely necessary because a 50+ ohm driving impedance would give a dramatically different frequency response than the usual low Z. Different response would mean different excursion and obviously distortion profiles that could not be fairly compared.

The curves are done with 90dB (at 1m) as the top line of the graph. That would also be the 100% distortion line. 20dB down would be 10% and the bottom of the graph is 0.3%. To make the 3rd harmonic distortion more visible I repeated that curve, from 60Hz up, with a 20dB increase. That would make the 20dB line the 0.3% level and the graph bottom 0.03%. Note that the green 3rd harmonic curve below 60 Hz is at the normal gain and approaches 100%.

Finally there are two black curves. The one that extends down to 25Hz is the total curve, that the compressor is keeping flat +-1dB, but rising at 12 dB per Octave from the lower left is the "fundamental only" curve, representing the maximum pure output. At mid frequencies where distortion is generally low they merge to become the same curve.

What you should see is that the curves are largely the same except fot the elevated green curves show significantly dropping 3rd harmonic output with higher drive Z. There is about a 10 dB drop as Z goes up. The LF distortion is generally unchanged. Interestingly the 2nd is strong but at very low frequencies the woofer is overdriven in both directions and distortion becomes largely 3rd (odd) order.

The other thing I mentioned was that each woofer had a 12dB per Octave LF assymptote of pure output. This pure output line was essentially independent of input leveland allowed easy comparison of relative woofer capibilities.

David S.

[forr]

Thanks, SpeakerDave, this kind of data is no very common, it's very interesting.

[Pano]

Thanks David, interesting stuff. Will take a while to digest.

[bolserst]

Originally Posted by speaker dave

Here are the curves I have mentioned. They were done when I was at KEF on a 103.2 8" woofer bookshelf. The amplifier was an old Fisher 80AZ tube amp. The "AZ" part is a variable output impedance that uses a combination of voltage and current feedback to vary the output impedance for typically low to a measured 57 ohms.

Thanks speaker dave for taking the time to post the distortion curves. Very interesting indeed.
It appears that at LF the 3rd harmonic is essentially unchanged by source impedance.
The 2nd harmonic, however, is affected a bit at LF.
Increasing source resistance reduced 2nd harmonic below 30Hz, but increased it just above 30Hz.
Perhaps a byproduct of by overdriven?

I created 3 visual overlays from your posted images so you can more quickly/easily see the differences in distortion products for the three different source impedances.

1) Low Impedance .vs. High Impedance
2) Low Impedance .vs. Medium Impedance
3) Medium Impedance .vs. High Impedance

[Pano]

Originally Posted by bolserst

It appears that at LF the 3rd harmonic is essentially unchanged by source impedance.

I must be reading the charts wrong. To me it looks like H3 is the thing affected most by the impedance change. No?