Franz, " voltages flowing back from the speaker " are reduced by output impedance of amp ( by damping factor ), not by voltage divider 😀 .
It would appear to be so.
Look at the scope ( connected to the output of the amp ) when you tap the speaker cone very briskly. If you can , also look at the current flowing when you do this.
There may be a few surprises.
Cheers,
Ashok.
I know, I am not an EE and my english is bad. But I try to answer:
Upupa: in a classical IGC this voltages and currents are flowing direct to the feedbackpath to the inverting input.
The gain is balanced with the input-r and the feedback-r. The problem is: this balance is just related to the input signal and not the voltages coming back from the speaker. They are not influenced or reduced by the feedbackpath and are fed to the inverting input.
Wrong?
With the t-net for feedback, the voltages are first reduced by the voltage divider and then fed back to the inverting input.
In case of 10K/100R/10K the reduction is 1:100 (for the signal AND the backflowing voltages).
In fact: the voltage divider is a simple key wich could be variated to adapt different speakers optimally.
Franz
If you want NFB to work correctly you want to feed the signal at the output node back to the respective input node as accurately as possible, even the errors caused by the load ! The only thing the network is allowed to do is exact scaling of the signal. *
Since both are linear "networks" - single resistors and resistive voltage dividers - there should be no difference theoretically in terms of damping the different parts of the voltage seen at the output. It simply cannot distinguish between signal components !
Where it differs is the impedance in question. For the same gain a lower impedance feedback path can be used with the voltage divider network. This leads to reduced susceptibility to parasitic capacitances (including the amps's input node which is not even a linear capacitor !) and therefore better accuracy of the signal fed back to the input, including errors caused by the load.
Regards
Charles
* There are of course also cases where one intentionally wants to include frequency-dependant behaviour into the feedback network. I just wanted to keep it as general as possible.
Since both are linear "networks" - single resistors and resistive voltage dividers - there should be no difference theoretically in terms of damping the different parts of the voltage seen at the output. It simply cannot distinguish between signal components !
Where it differs is the impedance in question. For the same gain a lower impedance feedback path can be used with the voltage divider network. This leads to reduced susceptibility to parasitic capacitances (including the amps's input node which is not even a linear capacitor !) and therefore better accuracy of the signal fed back to the input, including errors caused by the load.
Regards
Charles
* There are of course also cases where one intentionally wants to include frequency-dependant behaviour into the feedback network. I just wanted to keep it as general as possible.
Franz G said:
Exactly. You're right Franz, and ashok too. It's the high ratio that matters here and 100:1 is pretty good.
May I make a slightly heretical statement: Low O/P Z a la High Damping Factor means Nought.
Have a look at a speaker's amplitude response, every peak no matter how small have a corresponding back EMF - so even a speaker with a benign looking impedance versus phase response will have these multitudes of low level back currents flowing back into the amp. Here a high output Z might actually be an advantage. Why? Simply because it stabilises the output Z and that value now becomes.... [sound a trumpet call] a voltage divider. It helps attenuate back EMF.
Show me a dynamic speaker with flat amplitude response and zero phase angle, and only then will you have a speaker with zero back EMF. Fat chance. Ever tried listening to resistor?
If I may be allowed to further blow my trumpet (pun intended), this is what I wrote in a paper on my Isolated Loop Amplifier theory (which might end up on a different forum), the following:
This amplifier has stable output impedance into the loudspeaker meaning it is largely immune to damaging back-EMF aberrations. Feedback may lower the Output Z but also may make it dynamically unstable... the notion that Low Output Impedance as a necessity needs to be rejected, what is needed is a stable Output Impedance.
If you can get your hands on two articles in Glass Audio (2000) by Lynn Olson, where he discusses the priciple of 'hard amps' - he actually proposes that so-called damping factor should be one (so I am not the only heretic) or 4 to 8 Ohm.
At VSE we have developed Differential (and the related ILA) amp topologies, zero feedback. Allen built a Diff power amp that was reviewed by www.enjoythemusic.com late last year (got one of their annual 2003 Awards), it was 18 Watts/channel. Yet the reviewer could not believe that this 18 Watt amp had better bass slam than a bank 100 Watt plus Class A SS amps.
Scroll down almost half way down the page of this for the DPA-300 review:
http://www.enjoythemusic.com/magazine/viewpoint/0104/aachapter52.htm
Here is a part quote: The Wright push-pull two-tube amplifier putting out 18 watts had actually done tighter and deeper bass than four channels of 100 watts of "Class A" solid-state power. This amazed me.
The point I am making is that the amp reviewed has an Output Z of 3-4 Ohm.
So Damping Factor and low O/P Z means nothing. It's all about immunity to back EMF.
OK, I admit this topic is a bit of a long held hobby horse for me. So I'll get off my soap box now. 😀
Seriously, if an SS amp has low O/P Z as a result of f/b (or relies on it) and back EMF gets into the feedback loop, the O/P Z gets modulated - it becomes even more dynamically unstable. Which makes it even harder to cope with back EMF. I now believe that when this happens even on a very low level and is dramatically reduced, then we hear very clearly any improvement. I believe this is the case with the T-Network solution. It works.
Then by all means have your low O/P Z, just as long it's stable.
Joe R.
Joe R.
Zout = 3-4 Ohm
Every amps with similar Zout, which I have heard, had bass like " snowball on tant ", but maybe for somebody is this correct.
Every amps with similar Zout, which I have heard, had bass like " snowball on tant ", but maybe for somebody is this correct.
Thanks Joe, for your explanation!
BTW: Bassresponse with the VBTnIGC
I just listen to Frank Zappa, Yellow Shark, song #4 "Outrage at Valdez".
This piece begins with some swotting impacts. Here, you just feel 3-5 seconds, how the skin from the swott is swinging. More feel than listen. Really great.
Thats what I like!
Franz
BTW: Bassresponse with the VBTnIGC
I just listen to Frank Zappa, Yellow Shark, song #4 "Outrage at Valdez".
This piece begins with some swotting impacts. Here, you just feel 3-5 seconds, how the skin from the swott is swinging. More feel than listen. Really great.
Thats what I like!
Franz
Re: Zout = 3-4 Ohm
Not all, and that's the point. I have heard bass from amps with quite low Z out and high Z out, both capable of bass of the highest order. What you've heard is not a great amp irrespective of its 3-4 Ohm out. Maybe you heard a 300B SET amp or two. Don't use them to judge this topic. They are not even in the race, not any I've heard.
The ILA and Diff amps have bass quality, slam, tightness and extention, sheer ability to shape even very deep notes and no softness or soggyness, highly agile midbass (great on rolling timps). Or listen to Niels-Henning Oersted with his 'singing' style bass playing, sublime. Most amps are boring with N-H O. They can't trrack the notes.
Read what the reviewer said re the Diff power amp's, bass click on the link I supplied above. These amps have better bass than any SS amp I know off - that's the honest truth.
Joe R.
Upupa Epops said:
Not all, and that's the point. I have heard bass from amps with quite low Z out and high Z out, both capable of bass of the highest order. What you've heard is not a great amp irrespective of its 3-4 Ohm out. Maybe you heard a 300B SET amp or two. Don't use them to judge this topic. They are not even in the race, not any I've heard.
The ILA and Diff amps have bass quality, slam, tightness and extention, sheer ability to shape even very deep notes and no softness or soggyness, highly agile midbass (great on rolling timps). Or listen to Niels-Henning Oersted with his 'singing' style bass playing, sublime. Most amps are boring with N-H O. They can't trrack the notes.
Read what the reviewer said re the Diff power amp's, bass click on the link I supplied above. These amps have better bass than any SS amp I know off - that's the honest truth.
Joe R.
Realy good bass speaker must have solid membrane. Solid membrane will be not probably light, so will have big inertia. If you can good drive big inertia, you must have low Zout, with high Zout you get overshots. Overshots and " lazy basses " is mandatory in this case, independent on claims of some recenzents 😉 .
I can admit that a given amp toplology with more gain and therefore less NFB might be less susceptible to what happens at the output but not necessarily.
But the explanation given is totally wrong IMO. If you increase gain by reducing feedback it doesn't matter whether you do this in one step or two (i.e. resistor or divider) basically. The difference comes from the IMPEDANCES in question and their interaction wirth unwanted parasitic elements. You can't have a low impedance feedback path with just a single resistor in an inverting topology for a given gain, without reducing the amp's input impedance to very low values. That's where the T-network comes to the rescue.
I am also convinced, like Joe R , that there are things that are more important than a high damping factor by itself. One requirement is that the output resistance is independant of frequency, output -voltage and -current.
Apart from that: There are much more elegant versions to increase output resistance than reducing NFB IMO. Even the use of a simple series resistor would be more elegant !!!!
Regards
Charles
But the explanation given is totally wrong IMO. If you increase gain by reducing feedback it doesn't matter whether you do this in one step or two (i.e. resistor or divider) basically. The difference comes from the IMPEDANCES in question and their interaction wirth unwanted parasitic elements. You can't have a low impedance feedback path with just a single resistor in an inverting topology for a given gain, without reducing the amp's input impedance to very low values. That's where the T-network comes to the rescue.
I am also convinced, like Joe R , that there are things that are more important than a high damping factor by itself. One requirement is that the output resistance is independant of frequency, output -voltage and -current.
Apart from that: There are much more elegant versions to increase output resistance than reducing NFB IMO. Even the use of a simple series resistor would be more elegant !!!!
Regards
Charles
Upupa Epops said:
I have been down this road many times: Any series R value erodes the Qe of the driver, but the raw Qe is modified by the box, it goes up. It could then be said that the box ruins the damping of the driver? Well, no. For the reason that the raw Qe of drivers are actually a good deal lower than is needed, or in fact it is needed to be low as one has to make an allowance that the box alignment raises the Qe.
What am I getting at? Simply this, it is the overall alignment that controls the driver mass and its inertia. In fact, the raw Qe will cause enough friction that the bass will so OVER-damped, so that it kills the bass. So the Re (series R including O/P Z) does a similar job to what the box does. So it's the total effect of series R and the box that sets the parameters that control bass damping. It's the total alignment that matters.
That damping factor and low O/P Z is a fallacy was first pointed out to me personally by non other than Richard Small. Don't know if that name rings a bell, but have you heard of Thiele-Small parameters?
Yes, THAT Richard Small. In fact, I, acting the fool (and not for the first time) disagreed with him. Only years later I realised he was right and I was wrong. Maybe that is why he IS Richard Small and I'm comparatively speaking a nobody.
Joe R.
Speaker Q's
Coming to a very interesting discussion on speaker Q.
We have several Q's. In my opinion , maybe we should also look very closely at Qm. This is also dependant on the mechanical damping of the suspension. Many speakers have high Qm and a few have low Qm.
I remember seeing ( and hearing ) the 12 inch woofers from the Wharfedale Dovedale speaker unit. From the 60's I think. It had a very low quoted resonance ( 19Hz ?) and I expected to see it flop around when tapped. However when I taped it it never moved. I then pushed it and it still did not move . It was VERY hard to push it in. I thought that it had a stuck VC . However it was played subsequently and it was the first time I heard really tight deep bass. It hit us in the stomach ! Never heard that kind of sound from such a small enclosure since.
The amp was a Leak 70 and Garrard turntable with shure M75 cartridge. One of the albums was Silver Convention but I'm not sure which album produced that incredible bass.
I 've been searching the web to see the T/S parameters of that driver but I can't find it .
So this driver would probably sound great even with a high Z out of the amp.
Do we start a new thread about speakers with low Qm and how they sound with high Zo?
Cheers,
Ashok.
Coming to a very interesting discussion on speaker Q.
We have several Q's. In my opinion , maybe we should also look very closely at Qm. This is also dependant on the mechanical damping of the suspension. Many speakers have high Qm and a few have low Qm.
I remember seeing ( and hearing ) the 12 inch woofers from the Wharfedale Dovedale speaker unit. From the 60's I think. It had a very low quoted resonance ( 19Hz ?) and I expected to see it flop around when tapped. However when I taped it it never moved. I then pushed it and it still did not move . It was VERY hard to push it in. I thought that it had a stuck VC . However it was played subsequently and it was the first time I heard really tight deep bass. It hit us in the stomach ! Never heard that kind of sound from such a small enclosure since.
The amp was a Leak 70 and Garrard turntable with shure M75 cartridge. One of the albums was Silver Convention but I'm not sure which album produced that incredible bass.
I 've been searching the web to see the T/S parameters of that driver but I can't find it .
So this driver would probably sound great even with a high Z out of the amp.
Do we start a new thread about speakers with low Qm and how they sound with high Zo?
Cheers,
Ashok.
There are already some threads dealing with current drive that would cover this as well.
Regards
Charles
Hi Charles,
That would be the extreme case. I'd rather restrict it to usual amps ( voltage sources ) with higher than usual Z out . So anything from say 0.5 to 5 ohms or so should qualify.
Current sources are a different species altogether . They do need a separate thread.
Cheers,
Ashok.
That would be the extreme case. I'd rather restrict it to usual amps ( voltage sources ) with higher than usual Z out . So anything from say 0.5 to 5 ohms or so should qualify.
Current sources are a different species altogether . They do need a separate thread.
Cheers,
Ashok.
Just a call of attention.
When I tested (and now use it) a zobel (or call it a snubber) on the output of the amp - 2.7R + 100nf between out and ground - I had an almost as big effect as regulating the PSU.
Bass got tighter, defined like there's no tomorrow.😎
When I tested (and now use it) a zobel (or call it a snubber) on the output of the amp - 2.7R + 100nf between out and ground - I had an almost as big effect as regulating the PSU.
Bass got tighter, defined like there's no tomorrow.😎
T with reg., buffer, 4780
I posted earlier in this thread regarding a 3876 amp with T. I fiddled with the grounds, and even with helpful suggestions from Carlos and others, couldn't get the hum as low as I wanted it for some fairly high eff. speakers. I had modified some of Brian GT's boards for the inverting amp, and the grounding issues may be partly because of the board, which was designed for NI. Anyway, I removed those and replaced them with a single 4780, and was careful with grounding arrangements. Now, I can only hear hum only if i stick my ear right on the speaker, less than 1ma measured AC. Even a few inches away it's inaudible. The whole setup is this:
4780 set up as stereo amp. 100uf Panasonic VFK on power pins, with 0.1uf poly for bypass. Zobel on the outputs - 2R7,0.1uf.
Pejda's jfet buffers, with 2sk 369 instead of 2sk170. LPF with 33nf and 3.3nf. Input resistors are 725R between the caps, and 20k input to the 4780.
Pedja's discreet reg, mpf 102 for current source, and 3055/2955
for output power at +/-26 volts.
Separate discreet reg. for buffers at +/-16v. Same as above, but the without the 3055/2955. I figured that with the low current draw, the BD130/140's would be enough. 2x100uf Panasonic FC per rail for the ps. On the buffer are 3.3uf on each rail, bypassed with 0.12uf poly. Transformer is Hammond 164D28.
T-network is: R1-20.7, R2-11k, R3-115, R4-9.09, Rcomp-11k. The DC offset is 26mv one channel, and -20mv the other.
I have no power on thump, as the buffer is connected direct to the mains and is always on. No need to use the mute function for this purpose. The main power supply (330VA) is switched and has thermistors in series to limit inrush current.
Thanks everyone for all the ideas,
Sheldon
I posted earlier in this thread regarding a 3876 amp with T. I fiddled with the grounds, and even with helpful suggestions from Carlos and others, couldn't get the hum as low as I wanted it for some fairly high eff. speakers. I had modified some of Brian GT's boards for the inverting amp, and the grounding issues may be partly because of the board, which was designed for NI. Anyway, I removed those and replaced them with a single 4780, and was careful with grounding arrangements. Now, I can only hear hum only if i stick my ear right on the speaker, less than 1ma measured AC. Even a few inches away it's inaudible. The whole setup is this:
4780 set up as stereo amp. 100uf Panasonic VFK on power pins, with 0.1uf poly for bypass. Zobel on the outputs - 2R7,0.1uf.
Pejda's jfet buffers, with 2sk 369 instead of 2sk170. LPF with 33nf and 3.3nf. Input resistors are 725R between the caps, and 20k input to the 4780.
Pedja's discreet reg, mpf 102 for current source, and 3055/2955
for output power at +/-26 volts.
Separate discreet reg. for buffers at +/-16v. Same as above, but the without the 3055/2955. I figured that with the low current draw, the BD130/140's would be enough. 2x100uf Panasonic FC per rail for the ps. On the buffer are 3.3uf on each rail, bypassed with 0.12uf poly. Transformer is Hammond 164D28.
T-network is: R1-20.7, R2-11k, R3-115, R4-9.09, Rcomp-11k. The DC offset is 26mv one channel, and -20mv the other.
I have no power on thump, as the buffer is connected direct to the mains and is always on. No need to use the mute function for this purpose. The main power supply (330VA) is switched and has thermistors in series to limit inrush current.
Thanks everyone for all the ideas,
Sheldon
Good work Sheldon!
You may care to try the single cap LPF as both Pedja and I prefer it to the two cap version!
I have built two TNIGC's and they are both dead silent. If I stick my ear to the (high efficiency) speaker cones all I can hear is the hum from Carlos FM's amp in Portugual!
(OK - a little exaggeration there but you know what I mean 😉 )
Did you hand pick your T-network resistor values? I went with the 10K-100R-10K and then adjusted the NI to ground value. I got under 3 mV on all four channels (NB two amps).
I am yet to try my regulated supplies with the discrete buffers and T-network. (Too
)
You may care to try the single cap LPF as both Pedja and I prefer it to the two cap version!
I have built two TNIGC's and they are both dead silent. If I stick my ear to the (high efficiency) speaker cones all I can hear is the hum from Carlos FM's amp in Portugual!
(OK - a little exaggeration there but you know what I mean 😉 )Did you hand pick your T-network resistor values? I went with the 10K-100R-10K and then adjusted the NI to ground value. I got under 3 mV on all four channels (NB two amps).
I am yet to try my regulated supplies with the discrete buffers and T-network. (Too
)That is the good idea for the supply, just don’t forget to remove the resistors from the emitters of the second BD139/140.Sheldon said:
To clarify my opinion... I do like 75R-47nF better from the tonal point of view, but the shown second order filter had better soundstage.Nuuk said:
Pedja
There is no problem with it, Uncle Nick, one of the most important benefits of this filter is that it makes the soundstage better (than it is without it). And the soundstage further grows up a bit with 2nd order in my case.
Pedja
Pedja
- Status
- Not open for further replies.