You can buy most of the parts from a vendor that sells speaker crossover components. In the U.S. PartsConnexion is a good source.
https://partsconnexion.com
https://partsconnexion.com
A long time ago I found a library of speaker models for LTspice (named SwCAD III at the time). I don't remember where I got it but I've attached it here along with a speaker symbol file in case someone else finds this useful.
Yes, thank-you - that's what I'm talking about. The "Ken Kantor loudspeaker simulator" is one named design. There should be plenty to choose from and debated over their effectiveness in illuminating amplifier anomalies.
That's quite an ambitious undertaking. Difference the effect of the amplifier being in between the VNA signal and speaker. I wonder if the same or similar look-see could be done using equipment a mere mortal might own, like REW, a PC and a DAC?
Okay. So, what kind of anomalies to go hunting for? Just oscillations? Something else? What tests to use (what stimulus, what measurement, etc.)? Stability during power-on/power-off transients? Time-smear at LF? Memory effects? Asymmetrical crosstalk during high power transients?
They say dont run a tube amp no load. So when the impedance of the load goes sky-high (maybe not "sky", but...) at a couple different frequencies in the audio spectrum, what does this do / how does that amp handle that? One would think due to feedback, if there is some, the amp wouldnt care and keep producing the same voltage, even though the load went to 20 Ohms, at, say, 1.2 kHz.
Another might be doing an A/B between the resistive load and the simulated speaker load, listening to music, with headphones and volume padded down and see - using the best analyzer - if such loading effects the sound.
According to KEF, the answer is yes, which is why they build all those "burner" networks into their crossovers, so the amp doesnt see such peaks; the goal being "easy on the amp to drive". It'd be something to hear what, if anything happens.
Yes, they do say that. Care to take a guess as to what sort of anomaly may occur? And or, in a corresponding type of way what they might tell you not to do with a SS amp?
Regarding A/B between a simulated resistive load and a simulated speaker load played into headphones, how accurate would you expect such a simulation is likely to be? Better than the economy dac/headphone amp in a typical computer? Even if the computer/reproduction stuff is especially good, there is still the issue of, how good is the sim itself? My guess would be that its probably not that great in most cases. I mean, for example, we know that sims often tend to be optimistic in terms of HD estimation.
OTHO, if you mean to test a real amp with an equivalent lumped network to model a speaker, you could do that. However, usually what people actually do is test real amps with maybe a very reactive load to to see if the amp remains stable. I suppose you could connect a real HPA input across an attenuated lumped model of a speaker driven by a real amp and listen to that. Is that what you mean to suggest? However doing something like that likely won't tell you much useful about any anomaly you might hear, unless maybe you know how to deduce the cause of a problem by ear, which most people don't. Otherwise, how would you propose to assess the result?
For real amplifiers and real speakers, then of course some people say that part of what a crossover network should be designed to do is to present a friendly load to the power amp.
Last edited:
jjasniew,
To use the Rhode & Schwarz VNA to to a test of the amplifier plus loudspeaker combination does require a lot of things.
Understand the VNA, its measurement limits, impedance limits, voltage limits, dynamic range into the noise, etc.
Then understanding the calibration using the calibration kit.
Then making a setup that includes output and input connections, attenuators if necessary.
I tested speakers, amplifiers, interstage transformers, output transformers, and speaker cables.
How to set up a test setup with a 50 Ohm VNA, 50 Ohm calibration standards, and the need to provide matches from 50 Ohms to say 1k Ohms
to drive a transformer, and then proper load impedance and back to 50 Ohms was a challenge.
Testing transformers to find the inductance, distributed capacitance, leakage inductance, insertion loss, etc. was fun.
It was too bad when I no longer had access to that VNA.
I still do many of those measurements, but with far more humble equipment, and most testing is at spot frequencies, no log amplitudes, log frequency, and manual calibration only, nothing automatic.
To use the Rhode & Schwarz VNA to to a test of the amplifier plus loudspeaker combination does require a lot of things.
Understand the VNA, its measurement limits, impedance limits, voltage limits, dynamic range into the noise, etc.
Then understanding the calibration using the calibration kit.
Then making a setup that includes output and input connections, attenuators if necessary.
I tested speakers, amplifiers, interstage transformers, output transformers, and speaker cables.
How to set up a test setup with a 50 Ohm VNA, 50 Ohm calibration standards, and the need to provide matches from 50 Ohms to say 1k Ohms
to drive a transformer, and then proper load impedance and back to 50 Ohms was a challenge.
Testing transformers to find the inductance, distributed capacitance, leakage inductance, insertion loss, etc. was fun.
It was too bad when I no longer had access to that VNA.
I still do many of those measurements, but with far more humble equipment, and most testing is at spot frequencies, no log amplitudes, log frequency, and manual calibration only, nothing automatic.
jjasniew,
There are some things which are more easily measured than other things. Human nature seems to tend towards measuring things for which there are reasonably available preexisting measurement instruments. Things that are harder to measure or which require expensive, and or custom designed test equipment tend to get less attention from designers. Thus there is some tendency for design to be skewed according to whatever measurements and or specifications happen to be available. In my view, sometimes it can be a problem. My two cents only. YMMV.
https://www.scribbr.com/research-bias/availability-heuristic/#:~:text=sub-optimal choices.-,What is the availability heuristic?,is most available to us.
https://facilethings.com/blog/en/what-you-see-is-all-there-is
There are some things which are more easily measured than other things. Human nature seems to tend towards measuring things for which there are reasonably available preexisting measurement instruments. Things that are harder to measure or which require expensive, and or custom designed test equipment tend to get less attention from designers. Thus there is some tendency for design to be skewed according to whatever measurements and or specifications happen to be available. In my view, sometimes it can be a problem. My two cents only. YMMV.
https://www.scribbr.com/research-bias/availability-heuristic/#:~:text=sub-optimal choices.-,What is the availability heuristic?,is most available to us.
https://facilethings.com/blog/en/what-you-see-is-all-there-is