I think it is exactly what he says. A prototype as stated in point 1 of his post?
BPA-200 gives some 225 W into 8 Ohms and 335 W into 4 Ohms. For 4 Ohms the power is current limited to 14 A nominal which is pretty much maximum you can get with the two LM3886s. If however you would use 3 of them in parallell then the maximum output current needed would be 21 A which is pretty much maximum swing 80 V divided by the load 4 Ohms giving some 20 A peak. In this case the maximum power becomes voltage limited. This is the way I was aiming to run my BPA-200 (BPA-300?) which would then be perfectly able to run full throttle into 4 Ohms with 21*21*4/2 W output which would be some 750-800 W output power in real life I expect.
BPA-200 gives some 225 W into 8 Ohms and 335 W into 4 Ohms. For 4 Ohms the power is current limited to 14 A nominal which is pretty much maximum you can get with the two LM3886s. If however you would use 3 of them in parallell then the maximum output current needed would be 21 A which is pretty much maximum swing 80 V divided by the load 4 Ohms giving some 20 A peak. In this case the maximum power becomes voltage limited. This is the way I was aiming to run my BPA-200 (BPA-300?) which would then be perfectly able to run full throttle into 4 Ohms with 21*21*4/2 W output which would be some 750-800 W output power in real life I expect.
Follow up
Hi
I think it is time for another ”update”.
First all al, thanks for the warm welcome to this forum. Very nice.
I can see that since my first reply, other issues have “popped-up”. A major part of thise is dealt with in the original article. Seeing the interest that my “Das Modul” has coursed will make me “push” High Fidelity a little harder, so I can put the article on the net (in English) as soon as possible. I think that we all would benefit from that.
I will let you know if/when I succeed.
Since the article at present is only in Danish/Swedish and not on the net I will try to put some light on the issues brought up.
No, I am not related to Michael Madsen (editor on High Fidelity). There are about 100.000 people in Denmark called Madsen.
Actually I don’t work directly for High Fidelity. I did “Das Modul” for myself. One of my friends who do free-lance work for High Fidelity saw it and asked if I would be interested in doing a couple of articles on it.
The Schematic for “Das Modul” and BPA-200 are not exactly the same, but there are not fare of. However when you buy the PCB for “Das Modul” you also get the schematic for it.
If you want to buy the magazine you have to contact High Fidelity (see www.hifi.dk).
I very much appreciate peoples nice word about my CAD work……and YES it does take time and a couple of prototypes to get it right.
“peranders” talk about non-optimal cool when you use the flat-mounting of the LM3886. It is clear that you loose a small amount of heat transfer in the L-profile compared to standing LM3886 that are mounted directly onto the heatsink. Otherwise I can’t see where you loose all the cooling. Maybe “peranders” would be kind enough to explain a little bit more……………….
Why I used the LF411 and LF412 ??.
No specific reason. It is in the application note and I just thought it would be a good starting point. They can without any problems be exchanged with similar devices. For the LF412, just make sure it has a low off-set, since it is used in the servo’s (if you use them).
If I where to change them, I would stay with the LF412 and then maybe change the LF411 to newer and more “sexier” op-amp like the BB OPA-134 or similar device.
“UrSv” does some power calculations based on BPA-200.
In my experience, BPA-200 is fare to optimistic regarding heatsink requirement and therefor also to soem extent power output(like most application notes).
The original article deals to some extent with this.
The bottomline solution is simply to parallel as many “Das Modul” as you like to get the current you need. By adding more modules you are very effectively sharing the current (and heat) between them all. You don’t get more voltage swing, but since this is a balanced design running on up to +/- 42 V you have almost 80 V across your speaker………………..al you need now is the current to go with it.
In theory you should be able to get about 1000 W in 1 ohm, if you use enough module and your PSU is up to the task.
Before you try to get 500W in 8 ohm out of ONE module, I should say that normally is wise not to use a higher supply voltage than about +/- 30 to 35 V DC.
When you supply more than 30 to 35 volt to the LM3886 it tends to get very hot and when you then ask it to do some serious work it seems to have a hard time to get rid of the heat. You will then need more modules.
This issue is dealt with in the original article to some extent.
The picture in my first post is a “Das modul” prototype as stated. I just used 6 x LM3886 instead of 4 x LM3886.
Why did I go back to only 4 x LM3886 ??.
First of al because most people will be fine with just 4 x LM3886. For people who need more power; just stack (parallel) as many “Das Modul” as you need.
Secondly. Even through is might look almost the same on the pictures, the 6 x LM3886 are mounted another way that the 4 x LM3886 on “Das Modul”.
On the “Hex-Das Modul” the PCB goes al the way UNDER the L-profile. This makes it quit difficult to mount the LM3886, because the L-profile can now not be very thick, because then the legs on the LM3886 will not be able to get into the PCB !!……………..there really is no free ride.
Mounting 6 x LM3886 that have to “short” legs are not very DIY-friendly IMHO.
However if there is sufficient demand for it, I don’t mind designing a “Das grosse Modul”, but maybe you should just start out trying “Das Modul” as it. (“grosse” means big).
All for now.
Thomas
Hi
I think it is time for another ”update”.
First all al, thanks for the warm welcome to this forum. Very nice.
I can see that since my first reply, other issues have “popped-up”. A major part of thise is dealt with in the original article. Seeing the interest that my “Das Modul” has coursed will make me “push” High Fidelity a little harder, so I can put the article on the net (in English) as soon as possible. I think that we all would benefit from that.
I will let you know if/when I succeed.
Since the article at present is only in Danish/Swedish and not on the net I will try to put some light on the issues brought up.
No, I am not related to Michael Madsen (editor on High Fidelity). There are about 100.000 people in Denmark called Madsen.
Actually I don’t work directly for High Fidelity. I did “Das Modul” for myself. One of my friends who do free-lance work for High Fidelity saw it and asked if I would be interested in doing a couple of articles on it.
The Schematic for “Das Modul” and BPA-200 are not exactly the same, but there are not fare of. However when you buy the PCB for “Das Modul” you also get the schematic for it.
If you want to buy the magazine you have to contact High Fidelity (see www.hifi.dk).
I very much appreciate peoples nice word about my CAD work……and YES it does take time and a couple of prototypes to get it right.
“peranders” talk about non-optimal cool when you use the flat-mounting of the LM3886. It is clear that you loose a small amount of heat transfer in the L-profile compared to standing LM3886 that are mounted directly onto the heatsink. Otherwise I can’t see where you loose all the cooling. Maybe “peranders” would be kind enough to explain a little bit more……………….
Why I used the LF411 and LF412 ??.
No specific reason. It is in the application note and I just thought it would be a good starting point. They can without any problems be exchanged with similar devices. For the LF412, just make sure it has a low off-set, since it is used in the servo’s (if you use them).
If I where to change them, I would stay with the LF412 and then maybe change the LF411 to newer and more “sexier” op-amp like the BB OPA-134 or similar device.
“UrSv” does some power calculations based on BPA-200.
In my experience, BPA-200 is fare to optimistic regarding heatsink requirement and therefor also to soem extent power output(like most application notes).
The original article deals to some extent with this.
The bottomline solution is simply to parallel as many “Das Modul” as you like to get the current you need. By adding more modules you are very effectively sharing the current (and heat) between them all. You don’t get more voltage swing, but since this is a balanced design running on up to +/- 42 V you have almost 80 V across your speaker………………..al you need now is the current to go with it.
In theory you should be able to get about 1000 W in 1 ohm, if you use enough module and your PSU is up to the task.
Before you try to get 500W in 8 ohm out of ONE module, I should say that normally is wise not to use a higher supply voltage than about +/- 30 to 35 V DC.
When you supply more than 30 to 35 volt to the LM3886 it tends to get very hot and when you then ask it to do some serious work it seems to have a hard time to get rid of the heat. You will then need more modules.
This issue is dealt with in the original article to some extent.
The picture in my first post is a “Das modul” prototype as stated. I just used 6 x LM3886 instead of 4 x LM3886.
Why did I go back to only 4 x LM3886 ??.
First of al because most people will be fine with just 4 x LM3886. For people who need more power; just stack (parallel) as many “Das Modul” as you need.
Secondly. Even through is might look almost the same on the pictures, the 6 x LM3886 are mounted another way that the 4 x LM3886 on “Das Modul”.
On the “Hex-Das Modul” the PCB goes al the way UNDER the L-profile. This makes it quit difficult to mount the LM3886, because the L-profile can now not be very thick, because then the legs on the LM3886 will not be able to get into the PCB !!……………..there really is no free ride.
Mounting 6 x LM3886 that have to “short” legs are not very DIY-friendly IMHO.
However if there is sufficient demand for it, I don’t mind designing a “Das grosse Modul”, but maybe you should just start out trying “Das Modul” as it. (“grosse” means big).
All for now.
Thomas
3 years, just one word...
Great, after suffering through three years of German language lessons that was the one word I remembered. Its not the language; I'm denser than lead.
The PC boards seem to indicate "XLR in" for the audio input, does this mean that it is set up for true balanced inputs?
The BB op amp substitution sounds interesting. I think I'll try to get hold of the magazine and scratch my head about his some more.
I hope the good people at ( www.hifi.dk ) speak enough English...
Thanks for joining in and chatting with us. I hope we start placing lots of board orders with you soon.
-Dave
Great, after suffering through three years of German language lessons that was the one word I remembered. Its not the language; I'm denser than lead.
The PC boards seem to indicate "XLR in" for the audio input, does this mean that it is set up for true balanced inputs?
The BB op amp substitution sounds interesting. I think I'll try to get hold of the magazine and scratch my head about his some more.
I hope the good people at ( www.hifi.dk ) speak enough English...
Thanks for joining in and chatting with us. I hope we start placing lots of board orders with you soon.
-Dave
XLR and more
Despite the german name of "Das Modul" my german skills seems to be in line with David.
Yes. "Das Modul" have a true balanced input. It it made via the use of a BB INA137 chip, so we can to some extend discus how "true" it is.
However, a very large amount of audio-equipment with balanced in/out are made this way.
On the PCB ther is a small switch that alows you to change betwen balanced and singel-ended input.
I am sure that the people at High Fidelity understand English.
Thomas
Despite the german name of "Das Modul" my german skills seems to be in line with David.
Yes. "Das Modul" have a true balanced input. It it made via the use of a BB INA137 chip, so we can to some extend discus how "true" it is.
However, a very large amount of audio-equipment with balanced in/out are made this way.
On the PCB ther is a small switch that alows you to change betwen balanced and singel-ended input.
I am sure that the people at High Fidelity understand English.
Thomas
Re: Follow up
I did some experimenting with different thickness of the moutning plates. The original was a 7 mm X 70 x 70 L-profile. I did make a 20 mm thick plate and the temperature difference was noticable. But as I stated before, for home use it's overkilll to have maximium cooling capacity.tlmadsen said:
An externally hosted image should be here but it was not working when we last tested it.
It's a INA??? from TI (BB) which converts the signal.
XLR and cooling follow-up
Yes, it is a INA137 and it is actually very nice (see http://focus.ti.com/lit/ds/symlink/ina137.pdf).
Thanks to "peranders" for putting some ligth on the cooling thing. Now I undertand what you mean.
However I can't really see that the way you mount your LM3886 (flat or horizontal) have anything to do with the thicknes of the L-profil. The L-profile shown in High Fidelity is only 3 mm thick. As I also state in the articel, this is an absolut minimum and the way the LM3886 are mounted on the Module you can use any thicknes for the L-profile (as long as there is room for it in your case !).
I think that "peranders" and I share the same view on this issue, but that we are talking slightly "pass" each other...............or maybe not.
Looking forward to some more feedback.
Thomas
Yes, it is a INA137 and it is actually very nice (see http://focus.ti.com/lit/ds/symlink/ina137.pdf).
Thanks to "peranders" for putting some ligth on the cooling thing. Now I undertand what you mean.
However I can't really see that the way you mount your LM3886 (flat or horizontal) have anything to do with the thicknes of the L-profil. The L-profile shown in High Fidelity is only 3 mm thick. As I also state in the articel, this is an absolut minimum and the way the LM3886 are mounted on the Module you can use any thicknes for the L-profile (as long as there is room for it in your case !).
I think that "peranders" and I share the same view on this issue, but that we are talking slightly "pass" each other...............or maybe not.
Looking forward to some more feedback.
Thomas
Hi Thomas, it looks like you have designed a fascinating project for many of us.
To get round your publishing problems, is it worth suggesting to the magazine that they host the English article on their website, once the current issue is sold out. That way, they have visible control of it, and links could be made to sell the board directly on line.
To get round your publishing problems, is it worth suggesting to the magazine that they host the English article on their website, once the current issue is sold out. That way, they have visible control of it, and links could be made to sell the board directly on line.
Cooling
Hi
No, I have not measured the temperature of the LM3886 or the L-profile when runnning at full throttle. ( but it seems like a good idear) .
But I can fore sure say that they both get HOT and this is the very reason that
A. I recommend to use more modules if needed.
B. The BPA-200 is overoptemistic
C. Don't go over 30-35 V DC unless you really need that level of voltage (x2) over your speakers.
I think that the heattransfer with the L-profile is OK, but "peranders" have a very good point;
You can't really be to careful with the heattransfer away from the LM3886.
The "problem" is that the LM3886 can devliver an extrem amount of power compared to it's size and it then "run into lach of cooling".
The cooling issue is not so much how big your main heatsink is, but rather the tranfer of heat from LM3886 to the main heatsink. This is where the "bottleneck" is.
Thomas
Hi
No, I have not measured the temperature of the LM3886 or the L-profile when runnning at full throttle. ( but it seems like a good idear) .
But I can fore sure say that they both get HOT and this is the very reason that
A. I recommend to use more modules if needed.
B. The BPA-200 is overoptemistic
C. Don't go over 30-35 V DC unless you really need that level of voltage (x2) over your speakers.
I think that the heattransfer with the L-profile is OK, but "peranders" have a very good point;
You can't really be to careful with the heattransfer away from the LM3886.
The "problem" is that the LM3886 can devliver an extrem amount of power compared to it's size and it then "run into lach of cooling".
The cooling issue is not so much how big your main heatsink is, but rather the tranfer of heat from LM3886 to the main heatsink. This is where the "bottleneck" is.
Thomas
One idea to increase the cooling is to use standing LM3886 and mount them on a 30-50 x 20 mm aluminium rod with square section. This rod can then be bolted into a big heatsink. With this arrangement it's easy to mount and test the pcb. You will also get pcb space behind the IC pins, at least 20 mm.
Let me also say that the height of the rod can be anything, higher is better. 100 x 20-30 mm.
Let me also say that the height of the rod can be anything, higher is better. 100 x 20-30 mm.
Those of you who wonders, the "Das Modul" is more less a blueprint of the AN-1192 application note. One the few things Thomas has added is balanced input with INA137 . The application note tells you a great deal about cooling and output power. Overall, (as allways from National) a very well-written application note.
I will modify my post a bit. With no servos at all _and_ a cap to create gain of 1 you _may_ get up to 20 mV or more if you have input current offsets also. This creates still rather high currents flowing between the amps. It's possible to get 150 mA or more. This cretaes power of max 10-15 W in pure waste. If you read the application and the posts here you notice that cooling is not trivial. It's not a good idea to warm the chips in vane.I think you gain more with servos rather buying expensive resistors and caps. As the servo opamp you can choose a precision type like the OP07, OP97 or similar. You can also make the servo slow so it won't influence on the sound.PaulHilgeman said:
Cooling, DC-offset, killer amp and more
Per. I where seriously looking at the ”rod-solution” before I decided to go the “L-profile”. To me the “rod-solution” seems to have slightly more cons than pros. However, I very much encourage you or anyone else to go that way. We might all be wiser at the end of the day.
I have a “nasty” plan about a killer amp made of 12 x LM3886 (for one channel) mounted onto some form of cooling tunnel, together with some SMD-components to keep the PCB area down. It would solve a lot of problems, but also create some new interesting ones……………………
Per is also right about running without the DC-servos.
It is very easily done, just don’t plug in the op-amps for that.
The example by Per shows you what kind of trouble you run into if you don’t us the servos + you also get a nasty amount of DC into your speaker (….but they will most likely survive).
If you insist on using it without the servos, you should for each LM3886 place a 1 M variable resistor between + and – supply. The “arm” should then go to input on the LM3886 where the input from the servo would normally go. This should make it possible for you to “balance” the DC-offset when in idle. When you are playing music the DC-offset will most likely be present again in some form.
I would like to stress that I have never tried this in practise, but in theory it should work to some extent. If you try it please let us know how it worked.
I don’t fully understand why one don’t want to use the DC-servos. I an aware of that badly implemented DC-servos can make the low-freq part of a amplifier sound bad, but I seriously think that we by this date are more or less pass such thing.
Regarding “Das Modul”, I can for sure say that one of the very strong points of “Das Modul” is the handling of the low freq. It holds the woofer in a very “strong grip”.
I also understand and respect when people say that they want a more “minimalistic” approach, but when you are talking abut something (the LM3886) that already is build with a LARGE amount of transistors, resistors and capacitors inside the chip, I have a hard time understanding why an external DC-servo made up of an op-amp and a few passive components, scare people of ??
About “High quality” and larger resistors. I have used 1% metal resistors witch I think is a “de facto standard” component today. I can’t really see how you get it better than this.
OK, you could go to 0.1% resistors. I tried this for the resistors that set the gain. I could not measure or hear any difference so……………………….
By lager resistors, I take it that you mean “output” resistors on each LM3886. They are 0.1 ohm and 3 W. You can make them bigger, but why would you ? The 3 W power rating allows you to push more than 5 amp through them before you get anywhere near max.
You can by-pass closer to the LM3886 by mount some caps underneath the PCB. It will not be too difficult. On the first of my prototypes I also bypassed each LM3886, but found it to be unnecessary, but normally it can never hurt (…….actually it can, if you start mounting big electrolytes……you then have to be careful where you place the ground connection of your E-lyt.).
And YES, “Das Modul” is just a practical realisation of AN-1192 together with some practical observation and tips. This is also stated in the article.
All for now…………have to take the kids swimming
Thomas
Per. I where seriously looking at the ”rod-solution” before I decided to go the “L-profile”. To me the “rod-solution” seems to have slightly more cons than pros. However, I very much encourage you or anyone else to go that way. We might all be wiser at the end of the day.
I have a “nasty” plan about a killer amp made of 12 x LM3886 (for one channel) mounted onto some form of cooling tunnel, together with some SMD-components to keep the PCB area down. It would solve a lot of problems, but also create some new interesting ones……………………
Per is also right about running without the DC-servos.
It is very easily done, just don’t plug in the op-amps for that.
The example by Per shows you what kind of trouble you run into if you don’t us the servos + you also get a nasty amount of DC into your speaker (….but they will most likely survive).
If you insist on using it without the servos, you should for each LM3886 place a 1 M variable resistor between + and – supply. The “arm” should then go to input on the LM3886 where the input from the servo would normally go. This should make it possible for you to “balance” the DC-offset when in idle. When you are playing music the DC-offset will most likely be present again in some form.
I would like to stress that I have never tried this in practise, but in theory it should work to some extent. If you try it please let us know how it worked.
I don’t fully understand why one don’t want to use the DC-servos. I an aware of that badly implemented DC-servos can make the low-freq part of a amplifier sound bad, but I seriously think that we by this date are more or less pass such thing.
Regarding “Das Modul”, I can for sure say that one of the very strong points of “Das Modul” is the handling of the low freq. It holds the woofer in a very “strong grip”.
I also understand and respect when people say that they want a more “minimalistic” approach, but when you are talking abut something (the LM3886) that already is build with a LARGE amount of transistors, resistors and capacitors inside the chip, I have a hard time understanding why an external DC-servo made up of an op-amp and a few passive components, scare people of ??
About “High quality” and larger resistors. I have used 1% metal resistors witch I think is a “de facto standard” component today. I can’t really see how you get it better than this.
OK, you could go to 0.1% resistors. I tried this for the resistors that set the gain. I could not measure or hear any difference so……………………….
By lager resistors, I take it that you mean “output” resistors on each LM3886. They are 0.1 ohm and 3 W. You can make them bigger, but why would you ? The 3 W power rating allows you to push more than 5 amp through them before you get anywhere near max.
You can by-pass closer to the LM3886 by mount some caps underneath the PCB. It will not be too difficult. On the first of my prototypes I also bypassed each LM3886, but found it to be unnecessary, but normally it can never hurt (…….actually it can, if you start mounting big electrolytes……you then have to be careful where you place the ground connection of your E-lyt.).
And YES, “Das Modul” is just a practical realisation of AN-1192 together with some practical observation and tips. This is also stated in the article.
All for now…………have to take the kids swimming
Thomas
Thomas, you seem surprised over the DC-servo issue. Just do a search over the subject, lot's have been said! You can't really argue about some things 
I think a DC-servo is good but others thinks it's 
A small comment about ypur pcb layout: I try to get a "whole" groundplane as much as possible. I avoid traces in the groundplane and also "inverted" islands (no copper areas) . You never know when it's really important to have have a good groundplane so I have it all the time. It's good for RF immunity and emission for one thing.
I think a DC-servo is good but others thinks it's A small comment about ypur pcb layout: I try to get a "whole" groundplane as much as possible. I avoid traces in the groundplane and also "inverted" islands (no copper areas) . You never know when it's really important to have have a good groundplane so I have it all the time. It's good for RF immunity and emission for one thing.
Das Modul articel on the net.
It now seems that I will be allowed to put the original “Das Modul” on the internet. However, since it has to be translated (by me) I need to know a couple of things:
A. Is there sufficient interest in this ??
I don’t mind spending the time translating, but if only two people are going to read it, I would properly prefer to spend my time on something else.
B. If there is sufficient interest I would very much like that someone who is native speaking English (British properly the best, but “American” could definitely also be OK) would read my translation AND correct it before it goes out.
I should say that my English in general is OK (know different words and phrases), but my spelling/grammar is really poor (and the spellchecker in word don’t get it all).
C. How to make it available??.
People could drop me an Email and I could then send the article back I pdf-format.
The other thing is to place it on some form of homepage.
Good ideas are very welcome.
Thomas
It now seems that I will be allowed to put the original “Das Modul” on the internet. However, since it has to be translated (by me) I need to know a couple of things:
A. Is there sufficient interest in this ??
I don’t mind spending the time translating, but if only two people are going to read it, I would properly prefer to spend my time on something else.
B. If there is sufficient interest I would very much like that someone who is native speaking English (British properly the best, but “American” could definitely also be OK) would read my translation AND correct it before it goes out.
I should say that my English in general is OK (know different words and phrases), but my spelling/grammar is really poor (and the spellchecker in word don’t get it all).
C. How to make it available??.
People could drop me an Email and I could then send the article back I pdf-format.
The other thing is to place it on some form of homepage.
Good ideas are very welcome.
Thomas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.