So you own(ed) one and found it unsuitable for audio work or is this just a blanket statement, shooting from the hip?
What I've learned from owning and using many TM500 modules is that you shall not underestimate the skills of the then Tektronix engineers.
I well respect the skills of Tektronix and HP engineers.
No, I do not own one, but I can read data spec sheets.
And this tells me that this is a broadband device for use with several 10 amps.
Far away from let us say 1 amp audio current with a resolution of 80dB or better.
You can spend loads of money for a spectrum analyzer with GHz bandwidth and 8-12bit resulution -
useless for audio measurements.
No, I do not own one, but I can read data spec sheets.
And this tells me that this is a broadband device for use with several 10 amps.
Far away from let us say 1 amp audio current with a resolution of 80dB or better.
You can spend loads of money for a spectrum analyzer with GHz bandwidth and 8-12bit resulution -
useless for audio measurements.
I see a flux compensation transformer - a very sophisticated design.
My approach is much simpler: To minimize magnetic flux inside the toroid core, flux should be freezed by a shorted electrical coil.
To approach this goal, copper resistance of the coil should be minimized requiring a high copper fill factor.
And this winding should be shorted by a quite low resistance.
My actual probe with 100turns 1mm magnet wire is shorted by 1~10 ohms yielding satisfactory results.
At the end measuring FFT of the probe output signal shows the achievable resolution.
The official output scaling for the AM5043 is 10mV/div into 50 Ohms, but that was choosen to cope with the wide bandwidth, to avoid slewing issues.
In practice, at audio frequencies the undistorted output is very much higher and thus reasonable SNR is obtained and you can get another 6dB for free with a high-Z load. Well, you have to trick the AM503B as it moans when it doesn't see a 50 Ohms output load. The all analog AM503A is a bit better than later AM503B, noise-wise.
As I tend to use sample-synced record while playback for any kind of audio measurements, I can also apply time-domain averaging which can bring down uncorrelated noise (or mains hum/buzz) by 20..30dB easily.
In practice, at audio frequencies the undistorted output is very much higher and thus reasonable SNR is obtained and you can get another 6dB for free with a high-Z load. Well, you have to trick the AM503B as it moans when it doesn't see a 50 Ohms output load. The all analog AM503A is a bit better than later AM503B, noise-wise.
As I tend to use sample-synced record while playback for any kind of audio measurements, I can also apply time-domain averaging which can bring down uncorrelated noise (or mains hum/buzz) by 20..30dB easily.
You may compare your achievable resolution with the FFT-plots I provided with my totally different low-cost test set-up:
The donut-coil, a heavily modded MOTU-M4 sound card and REW on a laptop.
Most speakers (allegedly) are still designed with a view to using simple passive crossovers in a multiway system. At a minimum, the default / preferred behaviour for a woofer is for it to have low distortion when connected in series with at least an inductor, or a more complex RLC network.
As such, distortion tests connecting a "raw” driver directly to a voltage amplifier would only give an accurate reflection of performance if the driver will be specifically used in the same way: a purist "no crossover" full-range application, or a purist "no passive XO" DSP application.
DSP is obviously becoming a lot more common. After graduating from uni, for some years it was my view to drop those clunky old passive components from hi-fi development because I knew that everything active and digital was obviously far better... This seems very ingrained, and it's like swimming upstream to suggest that there are more subtle things going on.
For a multi-way speaker, running on a single channel, voltage drive therefore seems more versatile, as tweeters mostly use the same magnet+coil operating principle.
For small signals, yes. But at low frequencies you could get a double whammy of different effects:
-- b modulation due to current,
-- wild impedance swings across a narrow frequency band.
"Factory calibrated" EQ is not a good option when the actual sensitivity changes dynamically. Whereas velocity sensing is "throwing out the baby with the bath water".
IMO, a mixed-mode feedback system, delivering different output impedances at different frequencies, is likely a superior middle-ground. Then, if that's still not good enough, optimise by splitting the feedback between current drive at high frequencies and velocity sensing in the bass.
Modifying current drive to a mixed-mode system is a fairly simple matter of using 2 fbk lines and connecting them in a 'Y'. The 2 legs form an RC filter, so a C on the current fbk line shorts the R on the voltage fbk at high frequencies. At low frequencies the capacitively coupled current fbk is decoupled, allowing a low output impedance.
Notice that voltage feedback uses the full-scale output voltage, often divided down with resistors, 22:1 or something like that, while voltage across a current-sense resistor is much lower, so a pre-amp stage will need some kind of RC shelving filter to correct the gain. And then that has to be adjusted for the actual speaker gain.
We know what a Amprobe does.
https://www.digikey.com/en/products...c_0z8AT1d5baD6qQZ7W4iW822SzScoNYaAgGGEALw_wcB
What / how does your current sensor work? What does it measure?
Does a conductor connected to the voice coil pass through the donut hole of your donut sensor?
Does the donut sensor sit on the benchtop next to the driver magnet?
I am still thinking that a in-line 0.1R current sensing resistor is easier.
Thanks DT
@>JanRSmit<
Thank you for your question about the measurement setup and a sketch. Here you are. I hope the sketch covers your interest.
Feel free to ask.
Thank you for your question about the measurement setup and a sketch. Here you are. I hope the sketch covers your interest.
Feel free to ask.
The return path of voice coil current flows through that donut.
There is no physical connection to the driver magnet.
It does the same what the shunt resistance does.
The advantage is electrical isolation from the speaker circuit.
Change the FSAF folder so that it isn't the temp folder, then the files stay until you choose to delete them. There is a button to set the folder location.
Okay, It looks like the the donut current sensor makes less "iron noise" than the driver motor.
Show us a FFT of the driver motor current at perhaps a 80Hz sine input.
What does the FFT of the driver motor current look like with the sine input?
Thanks DT
On that topic, it's kind of funny to see crossover inductors with carefully chosen lamination thickness and special steel whereas the speaker they are driving is just a voicecoil surrounding a solid slug of unremarkable steel with some ferrite, no laminations and a hope and a prayer. Maybe crossover inductors should be made the same way?
Driver designers would love to use segmented/laminated structures that eliminate/minimize eddy current paths, it's just almost impossible to build them with reasonable effort, as of yet.
Purifi is working on that problem, AFAIK.
Powdered iron cores are readily available for PSU stuff, but the reasoning is different: essentially you get a distributed air gap that prevents hard saturation and circuit destruction due to short-circuit.
I'm not sure how well powdered iron would work in a speaker because of the extra magnetic resistance on top of the air gap that is already there.
While we're on the topic, I'm not sure what the big deal is that prevents electro-magnets / field coils from being a more popular alternative. There are so many class-A amplifiers already burning 0.5-2A quiescent current, but nobody thinks to re-use it to keep a field coil charged.
I'm not sure how well powdered iron would work in a speaker because of the extra magnetic resistance on top of the air gap that is already there.
While we're on the topic, I'm not sure what the big deal is that prevents electro-magnets / field coils from being a more popular alternative. There are so many class-A amplifiers already burning 0.5-2A quiescent current, but nobody thinks to re-use it to keep a field coil charged.