Hi Sebastian, Andrew
I have the FRD files for all the drivers I have used. In fact I used these files to design my crossovers using Isaac's Crossover Simulator. I also have WinIsd installed which I used to design my enclosures. Let me see how to use WinIsd to calculate the impedance.
Hari
I have the FRD files for all the drivers I have used. In fact I used these files to design my crossovers using Isaac's Crossover Simulator. I also have WinIsd installed which I used to design my enclosures. Let me see how to use WinIsd to calculate the impedance.
Hari
Ummm,
@Andrew: I might have mixed this little detail up in my memory. Sorry about that. But OTOH, it really doesn't? I couldn't imagine this...
@Hari: Well, the Crossover Simulator already does what I implied: it simulates the impedance response of your modeled crossover on your drivers. How accurate and complete were your FRD files to begin with? Did they contain the measured drivers' impedance data? Care to post a picture of the simulated crossover's response?
Cheers,
Sebastian.
@Andrew: I might have mixed this little detail up in my memory. Sorry about that. But OTOH, it really doesn't? I couldn't imagine this...
@Hari: Well, the Crossover Simulator already does what I implied: it simulates the impedance response of your modeled crossover on your drivers. How accurate and complete were your FRD files to begin with? Did they contain the measured drivers' impedance data? Care to post a picture of the simulated crossover's response?
Cheers,
Sebastian.
jethari said:keypunch said:
Hari,
Are you driving a 2 or three way speaker system and would it be passive or active crossover? If you are driving a passive crossover based speaker system then I would suggest you have am amplifier of 60-100W RMS and of the MOSFETS with a Pd of 190W you will likely need at least 2 pairs of such rated devices. Passive crossovers are well known to add load complexity to an amplifier, but if you are driving a 2 way speaker system it will be less comples.
John
I'm using a three way passive crossover speaker system of equivalent impedance of 8 ohms.
Take a look at the STW26NM50 at http://www.rapidonline.com/netalogue/specs/47-0442.pdf This looks more suitable for a single pair version (sorry for being so dogged about the single pair)
Hari
Hi Hari,
I am just curious why you like to use only a single pair of MOSFET output devices?
WRT the STW26NM50 it is similar to the APT4025BN as turns out with the exception of having a higher Ciss than the APT4025BN and much higher Coss, but the On/off and rise/fall times are very good. Compared to the IRFP450 a toss which Coss is better, Ciss for IRFP450 not as good and IRFP450 rise/fall on/of much slower and of couse the IRFP450 doe snot have as much power handling.
Some math for 60V rails, assume 8V loss for net 52V rails which is about 340W P-P for 8 ohms resistive. The STW26NM50 @ 70C 200W P-P and about 5 amps DC SOA and 6.5 Amps for 8 Ohm resistive.
Frankly speaking for a 3 way passive crossover speaker system you have to use 2 pairs of these very good STW26NM50 devices, but you will be very safe with them for domestic use. A single pair for domestic use is going to be tight at 60V rails. Now if you go for 65W then one pair of STW26NM50 will do just fine. 31V rails will get you 65W RMS or 130 P-P.
I agree with all, you really need to know and understand the speaker system using with the amp and the speaker sensitivity.
Regards,
John L. Males
Willowdale, Ontario
Canada
Official Quasi Thread Researcher
keypunch said:
jethari said:
jethari said:
Hari,
Math based on a 3 ohm load to the amplifier with 60V rails assuming a 8V loss for net 52V rails is 900W P-P (or 450 WRMS). The STW26NM50 @ 70C 200W P-P and about 5 amps DC SOA @25C with implied SOA no higher than 3.8A at 70C and 17.3 Amps P-P for 3 Ohm resistive. You would need 4.5 pairs of output devices, so 5 pairs. VA per channel RMS would be 734VA for a 1038VA P-P.
I do not know how to calculate for 100ms SOA like quasi uses for the IRFP450. I have not have spent the time to figure out how to model above DC operation. So I will run the math using headroom allowance which means the MOSFETs will heat up past their SOA, but I will not be able to calculate how long this time will be.
If one allows for 6 dB peak headroom the math based on a 3 ohm load to the amplifier with 60V rails assuming a 8V loss for net 52V rails is 225W P-P (or 113 WRMS). The STW26NM50 @ 70C 200W P-P and about 5 amps DC SOA @25C with implied SOA no higher than 3.8A at 70C and 4.3 Amps P-P for 3 Ohm resistive. You would need 1.13 pairs of output devices, so 2 pairs. VA per channel RMS would be 183VA for a 258VA P-P.
As you can see my previously posted comment that one pair would be tight for a 8 ohm resistive using the rated impedance/2 rule of thumb is a helpful and makes it easy to do simple ballpark SOA calculations. Andrew will of course be able to provide a more percise assessment.
Regards,
John L. Males
Willowdale, Ontario
Canada
21 July 2007 (0748 -) 08:43
Official Quasi Thread Researcher
keypunch said:
jethari said:
Hari,
It appears from your speaker system SPL data your speaker system efficiency is basically about 92dB. A 65WRMS (130W P-P) would have add 21dB P-P to the base reference efficiency which means 113dB. In a real room the reference efficiency will be higher, so I would think you would be fine with 65WRMS module.
Regards,
John L. Males
Willowdale, Ontario
Canada
21 July 2007 (08:45 -) 08:58
Official Quasi Thread Researcher
Hari Latest PCBs
Hi Hari,
A few comments re your latest PCBs of Post #2074 and Post #2075:
1) Are you using a different method of creating the PDF now? If one zooms in just a bit on the single pair version the tracks show lots of black speckle in them, some of the polygon outlines show incomplete even though the polygons will not be part of the final artwork, the text shows wonderfully clear for most part (some connecting lines, as well as characters, are shaved vertically). Also, the image is right against the top and right margins which can present printing problems as most printers do not have the ability to print edge to edge on a page physically. It appears the PDF page size created was just the size of the PCB dimensioning inclusive and a document page size of 4.4 inches x 5.2 inches. In contrast the two pair version the tracks are a consistant grey inner and a slightly darker grey outside border, the text is fuzzy as has been the case from the start (the same as I sadly had in the conversion process I eMailed you and remainded as you did the same conversion process) and no shaved lines nor text. The PCB image has excellent margin clearance for any printer and the PCB image has been placed on a standard letter size paper size of 8.5 inches x 11 inches. Personally I think overall the way you created the single pair version has much better quality and if you can figure out the speckling and shaving cause, and have the image placed with at least a margin of 0.5 inches (13mm) would be the method of choice.
2) A number of the parts seem to not connect with their respect PCB pads, examples include L1, R24, R26, and R27.
3) I would think T8 should be centered between T9 and T10 for thermal sensing reasons.
4) What are the pros and cons of wraping L1 about R35? I know some designers like to keep them seperate others like to wrap the coil about the resistor. Mounting L1 horizontially to save some space even when it is seperate from R35 might be a consideration, but I do not know if there is a reason, if any, why a horizontal vs vertical zobel network coil mounting might be preferred?
5) I still think taking a some of the extra wide (wider than the narrowist part) of Vrail width and giving that extra width to C11 and C13 rail returns makes sense as mentioned in the last part of (1) of Post #2054
Regards,
John L. Males
Willowdale, Ontario
Canada
22 July 2007 (08:45 -)10:37
Official Quasi Thread Researcher
Hi Hari,
A few comments re your latest PCBs of Post #2074 and Post #2075:
1) Are you using a different method of creating the PDF now? If one zooms in just a bit on the single pair version the tracks show lots of black speckle in them, some of the polygon outlines show incomplete even though the polygons will not be part of the final artwork, the text shows wonderfully clear for most part (some connecting lines, as well as characters, are shaved vertically). Also, the image is right against the top and right margins which can present printing problems as most printers do not have the ability to print edge to edge on a page physically. It appears the PDF page size created was just the size of the PCB dimensioning inclusive and a document page size of 4.4 inches x 5.2 inches. In contrast the two pair version the tracks are a consistant grey inner and a slightly darker grey outside border, the text is fuzzy as has been the case from the start (the same as I sadly had in the conversion process I eMailed you and remainded as you did the same conversion process) and no shaved lines nor text. The PCB image has excellent margin clearance for any printer and the PCB image has been placed on a standard letter size paper size of 8.5 inches x 11 inches. Personally I think overall the way you created the single pair version has much better quality and if you can figure out the speckling and shaving cause, and have the image placed with at least a margin of 0.5 inches (13mm) would be the method of choice.
2) A number of the parts seem to not connect with their respect PCB pads, examples include L1, R24, R26, and R27.
3) I would think T8 should be centered between T9 and T10 for thermal sensing reasons.
4) What are the pros and cons of wraping L1 about R35? I know some designers like to keep them seperate others like to wrap the coil about the resistor. Mounting L1 horizontially to save some space even when it is seperate from R35 might be a consideration, but I do not know if there is a reason, if any, why a horizontal vs vertical zobel network coil mounting might be preferred?
5) I still think taking a some of the extra wide (wider than the narrowist part) of Vrail width and giving that extra width to C11 and C13 rail returns makes sense as mentioned in the last part of (1) of Post #2054
Regards,
John L. Males
Willowdale, Ontario
Canada
22 July 2007 (08:45 -)10:37
Official Quasi Thread Researcher
John
You do have a keen eye
I did use different methods to create the pdf files. The first difference is the use of "Zealous crop" for the two pair as against "Autocrop" for the single pair. The second difference was in the actual pdf creation. I inserted the two pair png file into a word document and used the Adobe print option to create the pdf file. For the single pair version I directly created the pdf from Adobe out of the png image file. As for the outlines, as I mentioned earlier, they will not be there in the track side artwork which will be used for making the pcb. The fonts look different because I forgot to check mark the "TT fonts" option in Protel while generating the output.
I have used a coil wrapped around the resistor. In fact that is the design in my present amp (Mini Crescendo). But then the schematic had the coil and the resistor connected in parallel.
Hari
You do have a keen eye
I did use different methods to create the pdf files. The first difference is the use of "Zealous crop" for the two pair as against "Autocrop" for the single pair. The second difference was in the actual pdf creation. I inserted the two pair png file into a word document and used the Adobe print option to create the pdf file. For the single pair version I directly created the pdf from Adobe out of the png image file. As for the outlines, as I mentioned earlier, they will not be there in the track side artwork which will be used for making the pcb. The fonts look different because I forgot to check mark the "TT fonts" option in Protel while generating the output.
I have used a coil wrapped around the resistor. In fact that is the design in my present amp (Mini Crescendo). But then the schematic had the coil and the resistor connected in parallel.
Hari
Hi,
a coil with the width of the turns about the same as the height of the turns and the bore about double the width maximises the inductance for a fixed length of wire. i.e. minimum resistance. (3m=9b=10c, where m=mean diameter, b=coil width, c=coil height)
These long single layers coils have a higher resistance than the minimum.
a coil with the width of the turns about the same as the height of the turns and the bore about double the width maximises the inductance for a fixed length of wire. i.e. minimum resistance. (3m=9b=10c, where m=mean diameter, b=coil width, c=coil height)
These long single layers coils have a higher resistance than the minimum.