fab said:
Hi fab
I looked your nice projects in your webpage. In every amplifier, the PCBs are - very correctly - vertically placed in relation with toroid xformers and in some distance from them. I think the orientation it is the key for a good noise immunity as well the gnd loops avoidance... in comparison, the star points A or B are details and you can deal with those only after resolving the first two issues... unless you can't have a clear view of these small details, included the seperate gnd track for the small signal stages so you can make experiments.
Regs
Fotios
AndrewT said:
I agree with you Andrew. This 1,2 mVrms, as well the rest values that i refer are automatic numerical indications from my DSO and not taken with a true RMS DMM. What you mean as Vac? Vpp or Vaverage? Also, i refer that this measurement taken with the amp connected in the external supply via 3 cables of 1m length each. And without any element for decoupling.
I don't know in which amplifier you refer to, but mine has a voltage gain of Av=32,9. It is some big. Every possible noise it is multiplied by 32,9. You can see in post #4 the output voltage values that i have measured. The biggest it is 115Vpp/8Ù.
Finally, maybe your amplifier it is better designed from mine and it is also better in performance. I don't claim the first prize
.I desire only to remove this damned buzz. Maybe i will be forced to re-design the pcbs, to i etching to i varnish to i assemble and to i callibrate them again.
Regs
Fotios
Hi,
it does not seem to matter which design I assemble or buy as a kit.
All measure <0.1mVac of hum and noise at the output. But I haven't measured amps with a maximum output above 56Vpk (~400W into 4r0).
Vac is the AC voltage setting on a multimeter.
I rarely use the true RMS on my bench meter.
What I really must build is a low noise wideband pre-amplifier with switchable gains of +20dB and +40dB so that I can resolve the noise etc in the tens of uVac range.
it does not seem to matter which design I assemble or buy as a kit.
All measure <0.1mVac of hum and noise at the output. But I haven't measured amps with a maximum output above 56Vpk (~400W into 4r0).
Vac is the AC voltage setting on a multimeter.
I rarely use the true RMS on my bench meter.
What I really must build is a low noise wideband pre-amplifier with switchable gains of +20dB and +40dB so that I can resolve the noise etc in the tens of uVac range.
fotios said:
I desire only to remove this damned buzz. Maybe i will be forced to re-design the pcbs, to i etching to i varnish to i assemble and to i callibrate them again.
hello.
once i had a buzzing-problem.it was caused by a coil (ind.) built in near by the powersupply.i could hear a (rectifier)buzzing sound in the speaker.
greetings...............
AndrewT said:
Hi Andrew
Look at this. I have two DMMs, one old benchtop LCD very reliable Thandar and one good Escort. Escort manual does not refer in what kind of Vac is calibrated. Instead Thandar refer that its response is: Average responding, calibrated in rms of a sinewave, as well that a DC component included in measured signal does not affect the AC reading. By measuring the output voltage sum of any noise with Thandar its indication was 0,00 Vac. With Escort in auto mode, i don't know what kind was the 0,004 Vac that showed thus i rejected. I decided to make a more reliable measurement with my DSO, not auto but with cursors enabled. I placed the volts/div rotary selector in 10mV/div range. The DSO indicates a complex signal composed from different components. Placed in auto measurement, its numerical indication changed continuously from "?" to 2 mVrms because the fast variation of measured signal in amp output. After this, i decided to push the X10 mag. button (time/div=200nsec at 100MegaSamples) to have a most clear view of signal. I did the measurement manually by holding the signal and with the cursors placed on the larger peaks. The result was 720ìV or 0,72mV of average Vac level. I did this, because the Thandar DMM indicates the value of AC voltage in the same way. Because its display is 3,1/2 digits can't display a such small value.
Summarizing, you can't do a comparison of measurements taken with DMM and with DSO. Also, a reliable measurement with DSO can be done only manually with cursors and after the holding of signal.
Regs
Fotios
Thanks for the informations "mjf".
As you can see in the picture, the output inductors are in enough distance from P.S. and does not pressent any interference (i have checked also this!).
As i said, after one week of experiments proved that, the buzz or humm or crackl noise caused from the orientation of amplifier boards as well from their place very close to xformer. If the boards turned vertically then any noise disapears.
Regs
Fotios
fotios said:
Thanks Fotios for your nice words and your response. I do not want to hijack your thread but I think that it is also related. Having taken precautions regarding immunity to transfo EMI by proper positioning of pcb and avoid ground loop, if someone wants to keep the noise as minimum (without changing the amplifier schematics), what should be the best location for speaker ground and Zobel ground connect, star-point A or B (as per the reference article)?
fab said:
Hi fab, you are welcome... no problem with hijack... everything is accepted, interesting and helpfull in this thread which deals with real life problems of our projects.
By reading the particular article, i think zobel must be connected in A because is a filtering part of amplifier module, instead speaker to B because its strong impedance. I am not so sure, this is my option from a first view and a little think.
Maybe Andrew has a better and documented view on this issue.
We may ask his lights on this.
Regs
Fotios
Ground returm connection
Hi Fotios
The reason I ask is because I see some amps with different connection point for speaker ground. One example is the old Hafler DH-200/220 where all grounds (except point B for the caps) were connected to star point - A including speaker ground(with input connected to it via a low value resistor). Later , Musical Concept described a different way of ground connect where every ground connected to a star point B (including the big reservoir caps) but the transfo ground was connected to this star point with a small wire run length. There seems to be several ways around ( I am not talking about double bridge rectifier, 2 separate grounds for transfo, etc.). I wonder if all situations are covered in the same document or if peoples would share their opinions about grounding of all usual signals: input, input cable shield, chassis, bypass caps, big reservoir caps, transfo gnd, double bridge grounds, zobel, speaker, separated regulated supply ground of the front end related to the high power unregulated section, mono amps, stereo amps.
thanks
fotios said:
Hi Fotios
The reason I ask is because I see some amps with different connection point for speaker ground. One example is the old Hafler DH-200/220 where all grounds (except point B for the caps) were connected to star point - A including speaker ground(with input connected to it via a low value resistor). Later , Musical Concept described a different way of ground connect where every ground connected to a star point B (including the big reservoir caps) but the transfo ground was connected to this star point with a small wire run length. There seems to be several ways around ( I am not talking about double bridge rectifier, 2 separate grounds for transfo, etc.). I wonder if all situations are covered in the same document or if peoples would share their opinions about grounding of all usual signals: input, input cable shield, chassis, bypass caps, big reservoir caps, transfo gnd, double bridge grounds, zobel, speaker, separated regulated supply ground of the front end related to the high power unregulated section, mono amps, stereo amps.
thanks
Re: Ground returm connection
I just found the answer to most of my questions in Randy Slone High-Power Audio Amplifier... book on page 382 fig. 12.4.
I just hope the information provided is completely accurate since not all manufacturers follow - or have followed in the past - the scheme presented.
By the way, the speaker return should connect to the star-point A and not the big reservoir caps ( point B) as per Slone book.
fab said:
I just found the answer to most of my questions in Randy Slone High-Power Audio Amplifier... book on page 382 fig. 12.4.
I just hope the information provided is completely accurate since not all manufacturers follow - or have followed in the past - the scheme presented.
By the way, the speaker return should connect to the star-point A and not the big reservoir caps ( point B) as per Slone book.
Re: Ground returm connection
Hi fab
As i said, i haven't any experience about this issue because i follow always the method of absolutely seperated power supplies per each channel. This has the sense of a seperate gnd path per each channel. I suggest you to take a look also in Bryston schematics which are very helpfull and very clear indicating the different gnd paths, mains filters and many other tricks.
Regs
Fotios
fab said:
Hi fab
As i said, i haven't any experience about this issue because i follow always the method of absolutely seperated power supplies per each channel. This has the sense of a seperate gnd path per each channel. I suggest you to take a look also in Bryston schematics which are very helpfull and very clear indicating the different gnd paths, mains filters and many other tricks.
Regs
Fotios
Zobel return and Speaker return both go to A (Audio Ground). That's what the two unlabeled arrows are for.
I would add a third spare connection. It can come from either A or B for a connection through a Disconnecting Network to Safety Earth.
Many designers/builders are now inserting a low value resistor into the connection between Signal Ground and Audio Ground, i.e. where the word "ground" appears.
Was it MJL that put a pair of inverse parallel diodes across this resistor? No matter who, I approve of these extra diodes (1n4002) to take fault current imposed on the signal side to Safety Earth without blowing the 10r resistor.
I would add a third spare connection. It can come from either A or B for a connection through a Disconnecting Network to Safety Earth.
Many designers/builders are now inserting a low value resistor into the connection between Signal Ground and Audio Ground, i.e. where the word "ground" appears.
Was it MJL that put a pair of inverse parallel diodes across this resistor? No matter who, I approve of these extra diodes (1n4002) to take fault current imposed on the signal side to Safety Earth without blowing the 10r resistor.
Conclusions...
After an exhausting effort to make everything is possible to solve the problem, and a private and usefull discussion with the very polite member of this forum "aandy", which has except a big expertise also two his own winding machines for toroids, i resulted in these conclusions:
The main guilty is the toroid xformer, because its external case is made from steel. Steel is penetrating material from magnetic flux. Because the xformer is 500VA this means a high magnetizing current, thus a heavy radiation of magnetic flux at least in a distance of 2-3cm arround its circumference. The correct material for making a screened case it is a die cast bowl from aluminium of 1,5mm thick at least. Aluminium does not absorb magnetic flux, but also does not conduct it. Its behaviour is like a mirror as it causes a reflection of flux backwards into the bowl. For this reason, when i placed the bottom cover of my unit, the noise mooved from the left channel to the right. My mistake was by 99% the wrong place of amplification boards very close and in the same level surface of xformer. The rest things mentioned in all posts from the very polite and skillfull members, are of small importance compared with this, and have to make with the reduction of THD in very small percentages (i.e. from 0,005% to 0,001%).
There is a very good solution for screen, which is the known "ì-metal" or "mumetal" (from the Greek ì=magnetic permeability) from which are constructed the cases of the small input signal xformers. Mumetal, can absorb the magnetic flux in opposition with aluminium. But is expensive. It is prefferable to redraw my PCBs so as can placed vertically in relation with toroid and far away from it. To be sincere, i hate this arrangement of amp. pcb onto the surface of heatsink, because a probable repair work will be very difficult. You can imagine, that for desoldering a small signal transistor or a resistor the pcb must removed from heatsink, included the screwed 14 transistors! In such a case it is preferable the mounting of all transistors in a seperate "L" profile, and the screwing of this "L" on the main heatsink with thermal grease etc.. what a disgust!
In any case, with the additional screening with copper and alu foils, insulating etc, i have the chance to leave the unit as is, because the noise measured with DSO has dropped in 1,2 to 2mVrms as much and it is audible only from the PA speaker that i reffer above. Instead from my Mission 780 it is heared only when i stick my ear on the woofer. But the story does not finish here. In the back plate of the unit, you can see two small left and right backboards in which there are the input sockets balanced and unbalanced, the input attenuating pots and the discrete balanced to single converters. It is amazing! When the pot is placed either at its ends (fully CW or CCW) the noise remains in the shame level as described above. When the pot rotated, say, at the middle of its track to adjust the volume, the noise caused from the interference of toroid, increased. The pots are 50KÙ log, and in some point these tuned with the input dc blocking axial MKP caps of 4,7ìF and the noise is multiplied by a big amount. I have tried a small size 22ìF elec instead the foil MKP (which presents also some inductance due to its size) but the result was the shame.
Well, guys these are the real life problems. We entrust our suppliers to draw pcbs for implementing a functional or practical project, and when the parts are mounted in their place according to the initial plan everything (which is tested of course seperatelly) it falls. The proof for this was the acceptance of deficient work - after 3 phone contacts with him - from the constructor of this supossevly noiseless toroid.
BTW, i suggest you again, to make this test with a PA speaker in the out of your diy amplifier, which is the most accurate method to examine how much audible is the noise produced. In low power projects up to 100Wrms/8Ù which means a Av between 30 to 40 i think will be very small, down to 0,5 - 0,1mVrms which is not audible from domestic speakers except than you will stick your ear on the cone of woofer. In mid to high power projects from 150 to 200Wrms/8Ù, there is problem because the Av it is very large usually from 49 to 68. As well the xformers used in high power amps, are of 500 to 1000VA and have a very high magnetizing current. Special care must be taken in such case. Many times it is prefferable to split a powerfull supply xformer in two of half power, ie instead one of 1KVA we preffer two of 500VA each.
Regards, and good luck in your diy audio adventures
Fotios
After an exhausting effort to make everything is possible to solve the problem, and a private and usefull discussion with the very polite member of this forum "aandy", which has except a big expertise also two his own winding machines for toroids, i resulted in these conclusions:
The main guilty is the toroid xformer, because its external case is made from steel. Steel is penetrating material from magnetic flux. Because the xformer is 500VA this means a high magnetizing current, thus a heavy radiation of magnetic flux at least in a distance of 2-3cm arround its circumference. The correct material for making a screened case it is a die cast bowl from aluminium of 1,5mm thick at least. Aluminium does not absorb magnetic flux, but also does not conduct it. Its behaviour is like a mirror as it causes a reflection of flux backwards into the bowl. For this reason, when i placed the bottom cover of my unit, the noise mooved from the left channel to the right. My mistake was by 99% the wrong place of amplification boards very close and in the same level surface of xformer. The rest things mentioned in all posts from the very polite and skillfull members, are of small importance compared with this, and have to make with the reduction of THD in very small percentages (i.e. from 0,005% to 0,001%).
There is a very good solution for screen, which is the known "ì-metal" or "mumetal" (from the Greek ì=magnetic permeability) from which are constructed the cases of the small input signal xformers. Mumetal, can absorb the magnetic flux in opposition with aluminium. But is expensive. It is prefferable to redraw my PCBs so as can placed vertically in relation with toroid and far away from it. To be sincere, i hate this arrangement of amp. pcb onto the surface of heatsink, because a probable repair work will be very difficult. You can imagine, that for desoldering a small signal transistor or a resistor the pcb must removed from heatsink, included the screwed 14 transistors! In such a case it is preferable the mounting of all transistors in a seperate "L" profile, and the screwing of this "L" on the main heatsink with thermal grease etc.. what a disgust!
In any case, with the additional screening with copper and alu foils, insulating etc, i have the chance to leave the unit as is, because the noise measured with DSO has dropped in 1,2 to 2mVrms as much and it is audible only from the PA speaker that i reffer above. Instead from my Mission 780 it is heared only when i stick my ear on the woofer. But the story does not finish here. In the back plate of the unit, you can see two small left and right backboards in which there are the input sockets balanced and unbalanced, the input attenuating pots and the discrete balanced to single converters. It is amazing! When the pot is placed either at its ends (fully CW or CCW) the noise remains in the shame level as described above. When the pot rotated, say, at the middle of its track to adjust the volume, the noise caused from the interference of toroid, increased. The pots are 50KÙ log, and in some point these tuned with the input dc blocking axial MKP caps of 4,7ìF and the noise is multiplied by a big amount. I have tried a small size 22ìF elec instead the foil MKP (which presents also some inductance due to its size) but the result was the shame.
Well, guys these are the real life problems. We entrust our suppliers to draw pcbs for implementing a functional or practical project, and when the parts are mounted in their place according to the initial plan everything (which is tested of course seperatelly) it falls. The proof for this was the acceptance of deficient work - after 3 phone contacts with him - from the constructor of this supossevly noiseless toroid.
BTW, i suggest you again, to make this test with a PA speaker in the out of your diy amplifier, which is the most accurate method to examine how much audible is the noise produced. In low power projects up to 100Wrms/8Ù which means a Av between 30 to 40 i think will be very small, down to 0,5 - 0,1mVrms which is not audible from domestic speakers except than you will stick your ear on the cone of woofer. In mid to high power projects from 150 to 200Wrms/8Ù, there is problem because the Av it is very large usually from 49 to 68. As well the xformers used in high power amps, are of 500 to 1000VA and have a very high magnetizing current. Special care must be taken in such case. Many times it is prefferable to split a powerfull supply xformer in two of half power, ie instead one of 1KVA we preffer two of 500VA each.
Regards, and good luck in your diy audio adventures
Fotios
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.