The Mills and Hawksford paper was about the Tannoy motional feedback subwoofer, introduced 30 odd years ago, which wasn't very successful, especially after a magazine review called it "slow".
It inspired my attempt at a motional feedback subwoofer.
Their current drive, with controlled output impedance, flattened the response of the velocity feedback, which came from an additional sense coil wound on top of the voicecoil.
I thought it was an overly complicated way of implementing current feedback at the time.
It inspired my attempt at a motional feedback subwoofer.
Their current drive, with controlled output impedance, flattened the response of the velocity feedback, which came from an additional sense coil wound on top of the voicecoil.
I thought it was an overly complicated way of implementing current feedback at the time.
Why you see things so complicated in latest posts??? It takes several minutes to write a post and it takes 15min on a breadbord to assemble circuit as i showed in post #885. So why to write long posts and not to test and hear for yourself? You can get rid of that RC in parallel to the load if you are obsessed with output impedance. It will still work. Done many times. Don't need to be perfect in order to work.Now on table is this ugly current drive amp 
it is real working one, no parts missing. If you like grounded load do improved howland current pump circuit. Done that and it works perfectly based on discontinued LME49811. That was at least 10 years ago, still driving my 4way speakers. XXLS peerless woofers down to 20Hz equalised with current drive. Yes on paper not best to use current at resonance, but in reality
is better than on paper anyway thats why then i switched to mixed mode amps. But later discovered more ideas and on breadboard there are two very different output impedance amps now. And when i compare them i equalise to the same acoustic output. Without moving the mic i get identical responces. So i can compare, by switching between with one click on PC. There are two identical U baffle fullrange speakers to compare. And the sound is different. very diferent.
Resume: do it first and think about current drive problems later maybe then you will realise that it is not that hard
it is real working one, no parts missing. If you like grounded load do improved howland current pump circuit. Done that and it works perfectly based on discontinued LME49811. That was at least 10 years ago, still driving my 4way speakers. XXLS peerless woofers down to 20Hz equalised with current drive. Yes on paper not best to use current at resonance, but in realityis better than on paper
anyway thats why then i switched to mixed mode amps. But later discovered more ideas and on breadboard there are two very different output impedance amps now. And when i compare them i equalise to the same acoustic output. Without moving the mic i get identical responces. So i can compare, by switching between with one click on PC. There are two identical U baffle fullrange speakers to compare. And the sound is different. very diferent.Resume: do it first and think about current drive problems later
maybe then you will realise that it is not that hard
I've had very impressive result with small AMT tweeters under current drive, 10 to 20 dB distortion reduction.
And current drive sufficiently protects the driver from burning out which is counter-intuitive. But when an AMT heats up at some point adjacent pleats touch, shorting out part of the total meandering conductor which is printed on one side of diaphragm. With P=I²R power goes down with decreased R, whereras under voltage drive with P=U²/R power increases with smaller R, burning out the tweeter in the end.
Well, as far as I'm concerned, I really ran into these instability issues with current drive when I tested it for sealed box and open baffle, driving the woofers hard. The mentioned ringing of the system with its natural Q was unbearable, the dreaded "one-note bass" basically. At lower volumes the ringing is less severe, obviously, and I admit I found the slightly "springy" sound character often subjectively better sounding with quite a lot of music material.Why you see things so complicated in latest posts??? It takes several minutes to write a post and it takes 15min on a breadbord to assemble circuit as i showed in post #885. So why to write long posts and not to test and hear for yourself? You can get rid of that RC in parallel to the load if you are obsessed with output impedance. It will still work. Done many times. Don't need to be perfect in order to work.Now on table is this ugly current drive amp
View attachment 1212505
is better than on paper
Resume: do it first and think about current drive problems later
Du.uuh! My single remaining brain cell is obviously overloadedIt's only now I see how to modify Vanderkooy's amp to get the same results without Mills & Hawksford's BS
Anyone have a copy of John Vanderkooy's paper? I think its https://secure.aes.org/forum/pubs/ebriefs/?elib=16571A Simple Reliable Power Amplifier with Minimal Component Count - John Vanderkooy
I just saw that. Bingo! Indeed, see below, this gives you effectively the same benefits as 5:1 current drive, but using a voltage amp:
We don't need current drive, in fact we need to come up with solutions that don't require current drive. The above example with 8 Ohm driver will usually be -6dB at 3KHz. That 0.5mH is rather stable, it is the 0.1mH that is not. In fact it produces a voltage in the voice coil that leads to distortion, the much larger external inductor suppresses that.
From my practical experience I would say that the current drivers are the single-speaker-amplifiers. The crossovers are their biggest enemy.
It seems so vague and barely readable they could almost patent it and force unsuspecting independent developers to pay royalties for the amazing tech they generously published.
Edit.. the cynic in me, hehehe. It would be nice to see those colourful cross-sections with field strength, and simple impedance vs frequency graphs instead of frequencies going in some unspecified "north-north east" vector.
Last edited:
It must be because I'm sick -- the world gives me free ideas: "hey look! A speaker motor with only magnets and no iron!" and I find some way to complain about it
It has to be acknowledged that they and Mills and Hawksford and others, did some amazing work. So for us mere earthlings, it's a smorgasbord to pick and choose ideas / inspiration / topologies etc.
It has to be acknowledged that they and Mills and Hawksford and others, did some amazing work. So for us mere earthlings, it's a smorgasbord to pick and choose ideas / inspiration / topologies etc.
With DSP it is not relevant. Filters are filters just different for current. Nelson Pass has paper on filters for current drive
Don't agree at all here.
Do you mean this https://www.passdiy.com/project/articles/current-source-crossover-filters
But all speakers are connected in series. In the calculations they are treated as ideal resistances of 8 ohms. In reality they don't behave like that.
For a a given multi way loudspeaker you can linearize the impedance. If it must be easy done you can try 50 ohms in parallel to each loudspeaker, or 40/30/20 ohms depends on amp and speaker.
Or as DIY you can reconstruct the crossover using your current drive amp.
Simplest is using smartfone as sound source and use free EQ software to make the speaker linear.
Or as DIY you can reconstruct the crossover using your current drive amp.
Simplest is using smartfone as sound source and use free EQ software to make the speaker linear.
I tried connecting the speakers in series a while ago. I didn't like that sound. The resistors in parallel means more power loss. Maybe that's okay for D amplifiers, but I have class A...