i have this amp also which is the current one that im using
http://www.connexelectronic.com/documents/LM4702_BPL_Audio_Amplifier_Module.pdf
which has the same output RC RL networks.. so expect the same from both the modules..
http://www.connexelectronic.com/documents/LM4702_BPL_Audio_Amplifier_Module.pdf
which has the same output RC RL networks.. so expect the same from both the modules..
What are you using for a signal source during your frequency response test? It would be much more likely that a cheap computer sound interface rolls off at 14k than a power amplifier. I don't see anything in the amp schem that would do it. Did you do a loopback test from source to mic input?
no R31 is not getting hot...
I cant see the values on the capacitors here..
is there any problem in the circuitry? you can see the image of that here..
http://connexelectronic.com/images/2x100W LM4702 BPL.jpg
I cant see the values on the capacitors here..
is there any problem in the circuitry? you can see the image of that here..
http://connexelectronic.com/images/2x100W LM4702 BPL.jpg
the test was done by a professional sound engineer i have seen the RTA on his mac which was showing clear rolloff one the curves after 14k...
well actually If the speaker out is connected before the inductor it sound fabulous... but after the inductor it rolls off..
well actually If the speaker out is connected before the inductor it sound fabulous... but after the inductor it rolls off..
You'd need something like a .2 ohm load for the 1uH inductor to cause audio band rolloff. You have something unaccounted for going on. If you have an RTA, why ask? Measure.
yeah i first wanted to see the quality and if it really worth the effort and then work on the own design knowing what are all the mistakes and any imp tips that I know before go for a decent design... but see found a bug kinda thing so just to know these first..
my friend has RTA i dont have but it was clear even by listening
so you say that using .5ohm added to the inductor will actually make it rolloff?
so you say that using .5ohm added to the inductor will actually make it rolloff?
0.1 to 0.5 should do the trick.
What are the physics on the coil? .18uH seemed a little too, eh, little.
(Wich should be good, since the rolloff probably woult be worse if the coils induction gets higher. )
0.1uH to 0.5uH?
if so what should be the resistor in parallel with it? it should be same?
i used this calculator
Air Core Inductor Coil Inductance Calculator
it gave as
.25 inch dia
.5 inch length
turns 7
resultant L = .125uH
Air Core Inductor Coil Inductance Calculator
it gave as
.25 inch dia
.5 inch length
turns 7
resultant L = .125uH
tried some small trick just removed the insulation on the existing coil with sand paper and connected the speaker output directly on the inductor so now at 3 turns im able to get clear highs so shall i finalize on that? would such low value would generally be sufficient?
Amplifiers can use up to 10 uH before affecting rolloff in the 20-20kHz range. If this damping network is wired up correctly,
it will have zilch to do with roll-off, whatever you think may be happening.
The resistor is a damping element, to lower the Q of the resonant circuit formed with the load and so prevent ringing.
So, it doesn't need to be altered for different values of L and the resistance is not critical - typical values from 4.7 -12R.
it will have zilch to do with roll-off, whatever you think may be happening.
The resistor is a damping element, to lower the Q of the resonant circuit formed with the load and so prevent ringing.
So, it doesn't need to be altered for different values of L and the resistance is not critical - typical values from 4.7 -12R.
0.1 Ohm to 0.5 Ohm
I see Ian talks about 4.7 to 12 Ohms.
I have found that the rolloff gets a bit too steep with such values.
BUT, this is a bit dependent of what rolloff already lies in the construction as a whole.
Steeper rolloff, greater the phaseshift at the output.
By the way, is the resistor in paralell with the inductor getting hot?
Last edited:
The inductance is there to do a specific job - prevent ringing affects when driving capacitive loads. It also aids in reducing RFI, received on the speaker cables. Loudspeakers become capacitive as the frequency drops toward the resonance frequency and become inductive above. However, any crossover networks complicate the actual reactance seen by the amplifier.
In multiple speaker loads, the reactive components in the load can be severe and cause overloading as well as ringing effects. This will be an important factor for successful PA work where loads can be anything at all. In Hi-fi, the wiring is short and often simple so there may be few problems in not using any output coil at all; just relying on a stable design with particular cable inductance and some extra series resistance, as do Naim products.
As you are now at the point of having effectively no coil anyway, that's where you are too but I suggest that if you use the amplifier in typical PA work, you really will find out why a proper coil (~3uH) is necessary, soon enough.
As other posters have said, there is something else affecting performance here, but if you can't measure and are reporting results that technically, cannot be, you can see that we can't help much.
Just so It is understood a little better than usually seen on the forum, this is also called a Thiele network but some seem to think that means it's part of the merchandising strategy of certain "Thiel" commercial products. The designer, part of the Theile-Small loudspeaker paramaters team and author of AES papers on stabilising networks for amplifier output stages many years ago, developed a series of networks and this is one of the simplest. It is designed around typical loudspeaker characteristics, not the amplifier, so it will always be a compromise in practical applications. That's also why you do need overkill since the amplifier was advised as being for PA, not Hi-fi.
That brings us to the question of what you are using for a load when you listen to the response and the professional tested with his sound card. A high impedance or no load implies that the network will no longer be suited, so it could be that adding small inductances radically affect bandwidth. See post #12 for another view on suitable values.
Normally, a different arrangement is used where the Zobel network is placed as close as possible to the amplifier output node for maximum snubbing efficiency and the output coil is placed anywhere after, but preferably away from the output stage to avoid EMI feedback - possibly at the output terminals. Though impractical, it could even be in the loudspeaker itself.
By way of comment, soundcards need to be capable and configured for at least 24bit/192 kHz resolution to reach 100 Khz and begin to display roll-off usefully. 14 KHz roll-off is consistent with the best 16 bit soundcards. Of course, these make distortion and noise measurements look wonderful, until you remeasure on proper instruments.
With reference to the schematic, the designer has shown the zobel after the network. This can be done but will be less effective. I suggest you look at practical designs on the forum and in amplifier design texts like Douglas Self's, Bob Cordell's or Randy Slone's. If books are just too hard, at least read the appropriate articles at the ESP website. Note the values those guys recommend or show as part of their designs. Randy, for one, was experienced with PA and musical instrument amps and shows several high power designs. You might find the experience of others with PA work worthwhile to read. Commercial amps by Crown, QSC etc. may well show low inductances compared to textbook designs according to their use and configuration but DIY amplifiers, modified and unsupported by adequate test data, are no place to experiment.
There are many crackpot and unproven schema in Hi-fi, The threads here are full of put-downs to amplifiers that have only worked with certain light loads and use conditions, whatever the claims for sound quality. In high power equipment, you don't often have the luxury of light, experimental use. Amplifiers must work reliably at full power on nasty loads.
I've said this a few times, that just reading opinions here, which could be wrong, right or irrelevant, even supported by miles of mathematics, may not bring about much more than confusion. You need a consistent course of study to bring you up to a base level of understanding of analog electronics. Then you will at least have the tools to find the information or derive it for yourself without stumbling into logical deadends. We all do this to some degree but learning reduces this to a confident independence and helps focus any queries we may still have, to gain more useful replies.
That brings us to the question of what you are using for a load when you listen to the response and the professional tested with his sound card. A high impedance or no load implies that the network will no longer be suited, so it could be that adding small inductances radically affect bandwidth. See post #12 for another view on suitable values.
Normally, a different arrangement is used where the Zobel network is placed as close as possible to the amplifier output node for maximum snubbing efficiency and the output coil is placed anywhere after, but preferably away from the output stage to avoid EMI feedback - possibly at the output terminals. Though impractical, it could even be in the loudspeaker itself.
By way of comment, soundcards need to be capable and configured for at least 24bit/192 kHz resolution to reach 100 Khz and begin to display roll-off usefully. 14 KHz roll-off is consistent with the best 16 bit soundcards. Of course, these make distortion and noise measurements look wonderful, until you remeasure on proper instruments.
With reference to the schematic, the designer has shown the zobel after the network. This can be done but will be less effective. I suggest you look at practical designs on the forum and in amplifier design texts like Douglas Self's, Bob Cordell's or Randy Slone's. If books are just too hard, at least read the appropriate articles at the ESP website. Note the values those guys recommend or show as part of their designs. Randy, for one, was experienced with PA and musical instrument amps and shows several high power designs. You might find the experience of others with PA work worthwhile to read. Commercial amps by Crown, QSC etc. may well show low inductances compared to textbook designs according to their use and configuration but DIY amplifiers, modified and unsupported by adequate test data, are no place to experiment.
There are many crackpot and unproven schema in Hi-fi, The threads here are full of put-downs to amplifiers that have only worked with certain light loads and use conditions, whatever the claims for sound quality. In high power equipment, you don't often have the luxury of light, experimental use. Amplifiers must work reliably at full power on nasty loads.
I've said this a few times, that just reading opinions here, which could be wrong, right or irrelevant, even supported by miles of mathematics, may not bring about much more than confusion. You need a consistent course of study to bring you up to a base level of understanding of analog electronics. Then you will at least have the tools to find the information or derive it for yourself without stumbling into logical deadends. We all do this to some degree but learning reduces this to a confident independence and helps focus any queries we may still have, to gain more useful replies.
I meant to say, measure the output frequency response with a known good meter and sinewaves. With no load, resistor load, speaker load and if necessary on the tweeter terminals to see where are you loosing the output.
if you don't have a good meter there are some free applications that can use your sound card with a few resistors. Be carefull, you can smoke your sound card very easily.
I always test before I listen. If something is not as expected during the tests, I fix it, test again and when I am satisfied I listen and enjoy.
if you don't have a good meter there are some free applications that can use your sound card with a few resistors. Be carefull, you can smoke your sound card very easily.
I always test before I listen. If something is not as expected during the tests, I fix it, test again and when I am satisfied I listen and enjoy.
yeah i need to buy a CRO I have a doubt that these days we are getting only Digital CRO is it worth enough to buy them Tektronix always calls me on that regard. I love CRT based but what about the Digital ones are they ok?
CROs
Digital ones, ('scopes) are fine and do much more, like FFT analysis and some even have USB connections to your PC which can be impressive. The problem is the resolution of cheap ones that most people can afford, as they don't look too nice on high resolution/frequency. The trace tends to look "jaggy". Just look at threads here and check any digital 'scope trace images. People are sometimes surprised to find these are basically only 8-bit devices and whatever the sampling tricks, the problems still show up.
I've bought good analog TEK scopes on the bay for less money and they can work fine too and look superb in the higher grade/screen sizes but if something does go wrong and old, special parts for the EHT sections are needed, you may be in trouble. Old TEK and HP etc. are high quality gear and in large, heavy cases, though. At the right price and in properly working condition, they can still be a better audio choice.
I wouldn't think about old (or new) Cheap Asian analog scopes to save money, though. Few can even be focused well enough to be really useful.
A good solution for clear high res. images is to use a high res. sound card in your PC, of at least 24/192 resolution and good reputation for lowest noise and being fitted with appropriate audio connectors. Some are for sound studios and stage work and come fitted with strange cables and connectors that are no good here. M-Audio have reasonable products as do Creative Audigy etc. If your PC bus allows separation of the sound card from other noisy cards like the video card, system noise may not be too noticeable but external cards and their adaptor cases are not cheap nor such a good idea for several reasons.
Note though, you need to construct a high quality attenuator and add overload protection diodes at the input to prevent frying your card input with signals over ~2V P-P. The common signal ground may be the PC ground and so, noisy without precautions. It's also a good place to fit correct BNC connectors to match CRO probes and give it that pro. convenience.
Digital ones, ('scopes) are fine and do much more, like FFT analysis and some even have USB connections to your PC which can be impressive. The problem is the resolution of cheap ones that most people can afford, as they don't look too nice on high resolution/frequency. The trace tends to look "jaggy". Just look at threads here and check any digital 'scope trace images. People are sometimes surprised to find these are basically only 8-bit devices and whatever the sampling tricks, the problems still show up.
I've bought good analog TEK scopes on the bay for less money and they can work fine too and look superb in the higher grade/screen sizes but if something does go wrong and old, special parts for the EHT sections are needed, you may be in trouble. Old TEK and HP etc. are high quality gear and in large, heavy cases, though. At the right price and in properly working condition, they can still be a better audio choice.
I wouldn't think about old (or new) Cheap Asian analog scopes to save money, though. Few can even be focused well enough to be really useful.
A good solution for clear high res. images is to use a high res. sound card in your PC, of at least 24/192 resolution and good reputation for lowest noise and being fitted with appropriate audio connectors. Some are for sound studios and stage work and come fitted with strange cables and connectors that are no good here. M-Audio have reasonable products as do Creative Audigy etc. If your PC bus allows separation of the sound card from other noisy cards like the video card, system noise may not be too noticeable but external cards and their adaptor cases are not cheap nor such a good idea for several reasons.
Note though, you need to construct a high quality attenuator and add overload protection diodes at the input to prevent frying your card input with signals over ~2V P-P. The common signal ground may be the PC ground and so, noisy without precautions. It's also a good place to fit correct BNC connectors to match CRO probes and give it that pro. convenience.
- Status
- Not open for further replies.