Don
Who do YOU think you are to state that I do not understand the current mirror aspect here. Of course I understand what you are attempting, yet you are going on as if I am wrong whilst it is you who has not got anything to show.
You go on about component costs, but I told you I don't want to have to buy and use another (current mirror) power amplifier in order to use the method you are promoting here - in the T-bass thread. Thus you are most ill mannered to persist as if I (and other readers) would be wise to !
The amplifier driving my T-bass/ low drivers also powers the mid/ hi drivers.
Please respect the different horses for different courses aspect here !!!!!
It wouldn't be so bad if you had a circuit to present, so why not work on your design, and if you end up using a line level T-bass component arrangement, then you know where you read it first.
PS: Parallel, series and shunt tuned arrangements are possible, but the one used here is the only one which sufficiently damped voice coil drive with respect to amplifier drive. Also T-bass works best with lo-Z drive, so your pre-amp would need low Z output too, even if you mimic T-bass with higher inductances and smaller capacitors.
.
Who do YOU think you are to state that I do not understand the current mirror aspect here. Of course I understand what you are attempting, yet you are going on as if I am wrong whilst it is you who has not got anything to show.
You go on about component costs, but I told you I don't want to have to buy and use another (current mirror) power amplifier in order to use the method you are promoting here - in the T-bass thread. Thus you are most ill mannered to persist as if I (and other readers) would be wise to !
The amplifier driving my T-bass/ low drivers also powers the mid/ hi drivers.
Please respect the different horses for different courses aspect here !!!!!
It wouldn't be so bad if you had a circuit to present, so why not work on your design, and if you end up using a line level T-bass component arrangement, then you know where you read it first.
PS: Parallel, series and shunt tuned arrangements are possible, but the one used here is the only one which sufficiently damped voice coil drive with respect to amplifier drive. Also T-bass works best with lo-Z drive, so your pre-amp would need low Z output too, even if you mimic T-bass with higher inductances and smaller capacitors.
.
Last edited:
Graham,
sorry, I did not meant to swamp your thread. I just wanted to suggest what could be a possible solution to a problem you mentioned before. This will be my last post on this subject, I swear!
Then yes, you're right, you would need more than one amp. I did not thought about that.
BTW: you can move also the cross-over before the mirror and use one mirror per driver (loudspeaker). Of course that would not be cheap and may require even more "mirror amps", but it may prove to be be quite convenient.
sorry, I did not meant to swamp your thread. I just wanted to suggest what could be a possible solution to a problem you mentioned before. This will be my last post on this subject, I swear!
well, not quite conventional, unfortunately. It needs to be able to swing the full output voltage, which is likely quite more than what a conventional preamp can do.
Oh! and the T-Bass is connected after the cross-over?
Then yes, you're right, you would need more than one amp. I did not thought about that.
BTW: you can move also the cross-over before the mirror and use one mirror per driver (loudspeaker). Of course that would not be cheap and may require even more "mirror amps", but it may prove to be be quite convenient.
low Z, yes... but definitely it needs not be as low as with the direct connection. The Z can be higher by a factor equal to the current gain of the mirror.
Hi Don,
You asked >> Oh! and the T-Bass is connected after the cross-over? <<
The T-bass is intended as part of the low frequency driver circuit, and should be connected directly to the amplifier output terminals; any lead resistance induced losses out to the rest of the LF crossover/ driver (on T-bass output) will then have much less impact upon reproduction. Resistance losses in the T-bass primary circuit negate dynamic benefits at its output.
With good components I can hear the effect of as little as 0.1 ohm lead resistance between a low-Z SS amplifier output and the T-bass circuit, so if you are thinking of a 100x impedance ratio at line level, your pre-amp source will still need to have a Z-out << 10 ohm at full amplitude, and your choke circa 1 Henry.
The thing is, this T-bass is a reative network, and to do the same at line level would still require some specialised inductors - unless - say op-amp inversion techniques are used to simplify construction via use of capacitors and resistors only, thereby implementing a tuned stage.
Cheers .......... Graham.
You asked >> Oh! and the T-Bass is connected after the cross-over? <<
The T-bass is intended as part of the low frequency driver circuit, and should be connected directly to the amplifier output terminals; any lead resistance induced losses out to the rest of the LF crossover/ driver (on T-bass output) will then have much less impact upon reproduction. Resistance losses in the T-bass primary circuit negate dynamic benefits at its output.
With good components I can hear the effect of as little as 0.1 ohm lead resistance between a low-Z SS amplifier output and the T-bass circuit, so if you are thinking of a 100x impedance ratio at line level, your pre-amp source will still need to have a Z-out << 10 ohm at full amplitude, and your choke circa 1 Henry.
The thing is, this T-bass is a reative network, and to do the same at line level would still require some specialised inductors - unless - say op-amp inversion techniques are used to simplify construction via use of capacitors and resistors only, thereby implementing a tuned stage.
Cheers .......... Graham.
No, I didn't ask. But while I'm here...
Did you mean to say 1 Henry? Or was that a typo?
You have quoted several wildly different inductance values over the course of this thread. You recently berated me for not having actually built a T-bass and having nothing to show. But since you have also not produced evidence that you have built one, I feel under no obligation to do so either. We can assume that you used it with the OB speakers that you posted details and a photo of some time back. We can assume you used a GEM amplifier. All that is missing is the T-bass.
Getting back to the discussion, the suggested buffer at the input to the T-bass circuit doesn't have to be unity gain. There are op-amps available that can swing more than 40 volts at a usefully low output impedance. Likewise, the driving impedance does not have to be less than 10 ohms. It can be several hundred ohms. The trick is to adjust the feedback of the power amplifier so that the reflected driver impedance at the input of the power amplifier is in the tens of kilohms range, and scale the T-bass network values accordingly. It would be just like building a "standard" T-bass to drive a driver with a 10,000 ohm voice coil.
I'm glad you brought that up. I was planning to use real R-L-C, at line level, to remove one possible objection to the method. If you can accept a gyrator as being equivalent to a real choke in this application, it will make adjustment of the inductance much easier.
Regards,
Don.
Apologies Don; of course that was UnixMan.
In my last post I was calmly discussing the possibility of a line level T-bass like implementation.
So is 100x 10mH = 1000mH = 1H wild ?
You even appear to be discussing a 1000x impedance - 10H - option !
You make too many attacks on me - before you have even tried your hypothetical circuit !
If you read this thread you should be able to find links to my other forum thread where discussion seems to be more considered, and considerate !
Also have I stated anywhere that your proposals are not acceptable ?
No; so why are you so critical of me ?
I am not trying to lame your horse - so go tame and ride it !
Please keep this thread for the original T-bass, and start a new one to report any findings you have in relation to any DIFFERENT implementation !
I see no harm in you reporting any success here, but to deride and criticise my work, and attempt to make me accept your design before you have even tried it, is very arrogant !
In my last post I was calmly discussing the possibility of a line level T-bass like implementation.
So is 100x 10mH = 1000mH = 1H wild ?
You even appear to be discussing a 1000x impedance - 10H - option !
You make too many attacks on me - before you have even tried your hypothetical circuit !
If you read this thread you should be able to find links to my other forum thread where discussion seems to be more considered, and considerate !
Also have I stated anywhere that your proposals are not acceptable ?
No; so why are you so critical of me ?
I am not trying to lame your horse - so go tame and ride it !
Please keep this thread for the original T-bass, and start a new one to report any findings you have in relation to any DIFFERENT implementation !
I see no harm in you reporting any success here, but to deride and criticise my work, and attempt to make me accept your design before you have even tried it, is very arrogant !
Graham,
I don't see it as any more arrogant than you wanting me to build a T-bass, working with the most random of values and no method of matching them to the specific driver used, and assuring me that I will be enlightened listening to the results of such a process. Those few people who have actually quantified and measured the effect of various T-bass attempts have shown that it doesn't perform as described, to which your answer has invariably been that they have used incorrect component values. This may in fact be true, but until they report success (or failure after trying multiple component changes), we won't know if it is in fact possible to significantly alter, in isolation, the first half cycle behaviour of a driver.
I've set myself the task of building a model that accurately shows the effect of T-bass component values on an accurate model (in the time domain) of a real world speaker driver. I'll validate that model against a physical driver and components. I'll then put all the results in another thread and invite others to use and critique it.
Over and out,
Don.
I don't see it as any more arrogant than you wanting me to build a T-bass, working with the most random of values and no method of matching them to the specific driver used, and assuring me that I will be enlightened listening to the results of such a process. Those few people who have actually quantified and measured the effect of various T-bass attempts have shown that it doesn't perform as described, to which your answer has invariably been that they have used incorrect component values. This may in fact be true, but until they report success (or failure after trying multiple component changes), we won't know if it is in fact possible to significantly alter, in isolation, the first half cycle behaviour of a driver.
I've set myself the task of building a model that accurately shows the effect of T-bass component values on an accurate model (in the time domain) of a real world speaker driver. I'll validate that model against a physical driver and components. I'll then put all the results in another thread and invite others to use and critique it.
Over and out,
Don.
After reading this thread, I have a few questions myself. Hopefully I'm not making too big a fool of myself!
I am familiar with Graham's view on 'first cycle distortion'. While I do not agree it is an issue, I'd just ask about the T-concept.
Assuming that there will be a xover involved, it seems to me that the high-frequency components of the first cycle - roughly those that make up the starting edge - are routed to the mid/high freq drivers. As long as those mid/high freq drivers are up to the task, (and as far as that first cycle is low-passed within a sensible audio bandwidth), the combined soundfield should correctly reproduce that first cycle.
If I am wrong, if there still is an issue, how can manipulating the woofer driver compensate for something that is essentially a high(er) freq band issue? Might that not be the reason that it is difficult to find component values that do solve it?
jd
I am familiar with Graham's view on 'first cycle distortion'. While I do not agree it is an issue, I'd just ask about the T-concept.
Assuming that there will be a xover involved, it seems to me that the high-frequency components of the first cycle - roughly those that make up the starting edge - are routed to the mid/high freq drivers. As long as those mid/high freq drivers are up to the task, (and as far as that first cycle is low-passed within a sensible audio bandwidth), the combined soundfield should correctly reproduce that first cycle.
If I am wrong, if there still is an issue, how can manipulating the woofer driver compensate for something that is essentially a high(er) freq band issue? Might that not be the reason that it is difficult to find component values that do solve it?
jd
Hi Don,
Did you check out the other Forum ?
The only reason I suggested you try the T-bass is because you are criticising the circuit and me, yet you have (still) not even tried it or read all my notes !
Does my asking you to do so before you start criticising make me arrogant ?
Thus I look forwards to seeing your published *results*, and I should appreciate an e-mail, link or entry here, because I am presently not active with audio, and no longer look out for new threads.
Did you check out the other Forum ?
The only reason I suggested you try the T-bass is because you are criticising the circuit and me, yet you have (still) not even tried it or read all my notes !
Does my asking you to do so before you start criticising make me arrogant ?
Thus I look forwards to seeing your published *results*, and I should appreciate an e-mail, link or entry here, because I am presently not active with audio, and no longer look out for new threads.
Hmmm. Doesn't the 'sectioned lf' waveform just consist of the fundamental in an increasing amplitude? Why would a woofer have a problem with that?
Did anybody run a simmed first cycle through a simmed xover yet? Didn't see it in this thread but may have overlooked it.
jd
That is precisely the situation.
Some time ago Graham referred to Linkwitz's analysis of the effects of mounting a driver to a baffle, see section N on page:
Issues in speaker design - 2
He used it as an example of excursion requirements for LF drivers mounted in an OB. He then went on to describe the requirement to provide extra drive at frequencies around the primary resonance of the driver, to compensate for the first half cycle or so of the acoustic output waveform being diminished in amplitude due to energy storage in the driver.
Ironically, the previous section (M) on the Linkwitz page discusses energy storage in the driver. It shows how the use of simple sine tone bursts, as used by Graham to demonstrate the effect, gives a spurious result. Linkwitz then goes on to show that the correct waveform is one with a ramp up to full amplitude and a corresponding ramp down at the end. The modified waveform is actually much "purer", containing more of the fundamental and fewer other frequencies.
In short, the "standard" sine tone burst contains other frequencies too. The driver is a bandpass filter, so the LF and HF components outside its bandwidth are not reproduced. The resulting output waveform looks very much like that produced by a filter using component values transformed from their mechanical equivalents in the driver. Note that this is visible both in frequency (continuous) analysis, and impulse (time) analysis. (Graham says that the problem can not be shown in frequency analysis, it must be analysed in time.)
This raises another question: do real world signals exhibit the same attack and decay of the classic sine tone burst? Graham says that the attack of bass drum and plucked bass sounds are rendered more accurately using T-bass. I'll have to find some in my sample library and see what their waveforms actually look like.
None of this is intended to show that the T-bass does not have an effect. It certainly does. It alters (Graham would say corrects) the amplitude response of the driver in a way that appears to be audibly pleasing. Whether this is more accurate or not is open to debate.
Graham,
Yes, I checked out the other forums. To be thorough, I have just re-read the complete audiocircle and fullrangedriver threads. I couldn't sleep, so I was hoping some dry reading material would help.
I did find a probable answer to my question about the T-bass circuit that you used - you said it was the one posted here earlier with the 6.4 mH inductor. You developed it for a pair of Beyma 3015LF drivers in an OB. But I regret to report that it all has put me in no different mind than I was before I started. And I'm still going to make sure I have the component values as correct as possible for my chosen drivers before I heat up the soldering iron. Over the years I have learned the folly of trusting what I hear without a reference. Thus, I don't consider it a valid test to listen to speakers with and without T-bass - I may prefer one over the other, but which one is the "real" sound?
Fortunately I listen to a lot of live music - choral, organ and "rock" instruments because I run the audio systems, and record performances, at the local church. Running OB speakers in the same acoustic space as the original performances should be a fairly thorough test.
I'll do that. Good luck with your situation. One of my brothers went through a similar multi-year battle to obtain recognition of, and compensation for, the ongoing effects of a back injury.
Case File H1307764274
Regards,
Don.
A first cycle as depicted has a waveform that starts up in zero time and thus has an infinite dV/dT. Musical instruments are mechanical systems and thus cannot have infinite dV/dT. If one would accept an upper audio bandwidth limit of say 20kHz, filtering the first cycle waveform with a 20kHz low pass would strip off the higher harmonics and would slowly and gradually start up. It would still have a lf component though that would not be flat, but I don't see that as a shortcoming; it's just the way the signal is.
An exception could be electronic music which presumably could have very sudden start-stop waveforms. But again, the cross-over, whether passive or active, would make sure that those very high freq components would be routed to the tweeter. I don't see any difference with the lf waveform here.
And in either case, the final sound would be the combined sound emitted by woofer, midrange and tweeter. 'Correcting' the woofer only is likely to upset that delicate combination. It would sound different probably, but not necessarily more accurate.
jd
Yes, I agree. But a tone burst, or a sine wave that starts at zero time does have infinite dV/dT. It is not a realistic musical signal.
I'm trying to make a sine burst in my sim to show it, but I didn't manage it so far.
If you band-limit a 'first cycle' to say 20kHz, it looks quite different. And that IS realistic.
What you then end up with is a spectrum of signals within the audio band, just normal spectrum, and the components are divided to the drivers by the cross-over.
It's the same when people talk about 'fast bass'. Bass drivers never get the fast attack components, because the xover routes those to the tweeter. Bass drivers never have to start and stop fast. What people call 'fast bass' is a combination of accurate bass and accurate high frequency reproduction.
jd
No, it's not the point in this thread. T-bass is for bass, let's say under 200Hz.
A tone burst signal of 200Hz (or lower) shall not have a dV/dT as high as the ones in higher frequencies. Such 'slower' motions are woofers' duty no doubt. The main goal of T-bass circuit is to maintain the amplitude right at the start of the tone burst, i.e., the first half cycle of the waveform.
An ordinary xover'ed woofer has a common problem -- not having enough amplitude at the first half cycle, even in the first few cycles. SL's site has some diagrams showing such behavior. The amplituds of such tone bursts build up slowly.
This is not the issue of xover - splitting the spectrum and mixing them back together. This is to retain what a woofer should do within its own spectrum. We may discuss how T-bass is successful (or not) in doing this and why, but let's not mix up the concepts.
A tone burst signal of 200Hz (or lower) shall not have a dV/dT as high as the ones in higher frequencies. Such 'slower' motions are woofers' duty no doubt. The main goal of T-bass circuit is to maintain the amplitude right at the start of the tone burst, i.e., the first half cycle of the waveform.
An ordinary xover'ed woofer has a common problem -- not having enough amplitude at the first half cycle, even in the first few cycles. SL's site has some diagrams showing such behavior. The amplituds of such tone bursts build up slowly.
This is not the issue of xover - splitting the spectrum and mixing them back together. This is to retain what a woofer should do within its own spectrum. We may discuss how T-bass is successful (or not) in doing this and why, but let's not mix up the concepts.
Thanks CLS.
If we pick a bass guitar string, it does not leave the plectrum with infinite speed.
The harmonics might be correctly reproduced by mid/ tweeter, but the LF OB driver(esp) starts from zero momentum and then resonates/increases in amplitude within the first few 1/2 cycles.
This is why bass can sound much more realistic through good headphones - tiny mass/ momentum, whilst putting your head closer to an uncompensated LF driver still does not make LF reproduction sound any more realistic !
Plain amplitude compensation cannot compensate for what the driver does either because the driver has already distorted the waveform cycle in real-time. It takes an inverse distortion in real time to counter the driver error. Once the driver is compensated, then further amplitude compensation becomes possible, as long as phase distortion does not then result.
(Hi Don. Thanks for your wishes. I am not trying to claim compensation. The docs would not provide evidence anyway. I'm just trying to live through b-awful symptoms from injury the docs say I've never had, and becoming worse due to lack of appropriate treatment.)
If we pick a bass guitar string, it does not leave the plectrum with infinite speed.
The harmonics might be correctly reproduced by mid/ tweeter, but the LF OB driver(esp) starts from zero momentum and then resonates/increases in amplitude within the first few 1/2 cycles.
This is why bass can sound much more realistic through good headphones - tiny mass/ momentum, whilst putting your head closer to an uncompensated LF driver still does not make LF reproduction sound any more realistic !
Plain amplitude compensation cannot compensate for what the driver does either because the driver has already distorted the waveform cycle in real-time. It takes an inverse distortion in real time to counter the driver error. Once the driver is compensated, then further amplitude compensation becomes possible, as long as phase distortion does not then result.
(Hi Don. Thanks for your wishes. I am not trying to claim compensation. The docs would not provide evidence anyway. I'm just trying to live through b-awful symptoms from injury the docs say I've never had, and becoming worse due to lack of appropriate treatment.)
Last edited:
No it doesn't, that was what I was saying. But if you look at the 'first cycle' graph, it does look as if it starts from zero time, and that is an infinitely fast start.
That's why I am wondering if such a first cycle is a realistic test signal. One can always show that any driver has a problem if you hit it with a non-realistic test signal, of course.
jd
I look at Linkwitz' site and what I found in this context is his cosine-shaped tone bursts. I agree that these are realistic signals; because of the gradual envelope build-up the signal is much more benign for a driver, and I doubt that the effect that T-bass tries to 'fix' is there at all with a shaped tone burst.
Is this the graph you referred to? If not, could you provide a link to what you meant?
jd