BrianGT said:
Actually Brian, 1.5kVa is not much of an overkill as 40V rails get you about 24W per channel which translates to about 600W dissipation per channel. Now given that a transformer shouldnt carry a constant load more than 50% its rated VA, it doesnt allow all that much headroom.....
I guess I just did not realize that heat dissipation was so high for each channel. I thought that it was 600W for both channels combined. If it is actually 600W per channel... then he is going to have problems with the size case that he is building.
--
Brian
gte619j@prism.gatech.edu
--
Brian
gte619j@prism.gatech.edu
I think we need Nelson to resolve this one!
Its one of those situations where Nelsons otherwise comprehensive notes are just a bit ambiguous. Do his descriptions of power supply requirements refer to supply requirements for one or two channels? I've presumed one and have organised supply requirements accordingly.
OK, I might be wrong. Nelsons supply may refer to stereo rather than mono requirements - but even if I am putting a stereo supply into a mono implementation, the end result? Better, not worse supply regulation and current delivery. (Better to play safe, eh!)
Oh, one minor caveat though - the space I have availiable in which to fit a decent (low flux) transformer means that I've had to revise my transformer specifications. These are now 2.5x Nelson's stated total dissipation figures. So, for a 25w output, (500w total dissipation in Nelson's notes) I'm specifying a 1.25KVA transformer with 32V secondaries per channel.
However, my heat dissipation calculations for 25w monoblocks give the following:
8R resistor dissipation = 480w (4 x 120w)
1R resistor dissipation = 45w (3 x 15w)
Output device dissipation = 110w (2 x 55w)
TOTAL = 635w
A 1.25KVA transformer is, in fact, just under double this figure. (x1.97, or thereabouts). Which figure should we be using Nelson, 500w or 625w? More importantly, do these figures apply to mono or stereo operation?
On the heat dissipation front, OK, OK, I know it seems unlikely at best (laughable at worst) that I could fit a 25w SoZ monoblock into a chassis half the size of a CD player, but I've done the math and it should be OK.
As the calculation above shows, total heat dissipation requirements for a 25w monoblock are 625w (my figures).
given that I'm using 2 x 0.12 deg C/W heatsinks, the temperature rise breaks down as follows:
625w x 0.12 deg C/W = 75.0 deg C
Shared between 2 heatsinks = 37.5 deg C
Now, on the forum, I've seen people describing ambient temperature (somewhat hopefully, in my view - isn't the ambient temperature in most homes closer to 20 deg C?) as anything from 17 - 20 deg C. That would give a heatsink temperature of between 54.5 - 57.5 deg C.
As I've explained elsewhere in this thread, I relly feel that the delights of class-A amplification should be in the realm of the masses, rather than in the hands of people like us, seen as audio 'freakies'. Unfortunately, what this means (for those commited to this idea) is getting away from the belief that a prerequsite for class-A projects such as SoZ is a huge case and mammoth temperatures. It is precisely this attitude that has helped to ensure class A amplification remains a minority interest.
The key prerequisite for Class-A amplification is that such a topology requres prodigious heat dissipation, not huge, scalding hot cases.
I've considered the 'traditional' route, huge, heavy, hot heatsinks, bolted either side of a chassis. Fortunately, I also considered the alternative - svelte, half-width cases, approx 4' x 9' x 14' (h x w x d). I couldn't see any advantage in the former approach and despair at the attitude that seems to prevail (even reflected in this thread) that as far as huge cases go, there is no other alternative. Even Nelson's notes make the point,
"...this is a project for those who like their hardware big..."
Anyhow, what say you, comrades?
Its one of those situations where Nelsons otherwise comprehensive notes are just a bit ambiguous. Do his descriptions of power supply requirements refer to supply requirements for one or two channels? I've presumed one and have organised supply requirements accordingly.
OK, I might be wrong. Nelsons supply may refer to stereo rather than mono requirements - but even if I am putting a stereo supply into a mono implementation, the end result? Better, not worse supply regulation and current delivery. (Better to play safe, eh!)
Oh, one minor caveat though - the space I have availiable in which to fit a decent (low flux) transformer means that I've had to revise my transformer specifications. These are now 2.5x Nelson's stated total dissipation figures. So, for a 25w output, (500w total dissipation in Nelson's notes) I'm specifying a 1.25KVA transformer with 32V secondaries per channel.
However, my heat dissipation calculations for 25w monoblocks give the following:
8R resistor dissipation = 480w (4 x 120w)
1R resistor dissipation = 45w (3 x 15w)
Output device dissipation = 110w (2 x 55w)
TOTAL = 635w
A 1.25KVA transformer is, in fact, just under double this figure. (x1.97, or thereabouts). Which figure should we be using Nelson, 500w or 625w? More importantly, do these figures apply to mono or stereo operation?
On the heat dissipation front, OK, OK, I know it seems unlikely at best (laughable at worst) that I could fit a 25w SoZ monoblock into a chassis half the size of a CD player, but I've done the math and it should be OK.
As the calculation above shows, total heat dissipation requirements for a 25w monoblock are 625w (my figures).
given that I'm using 2 x 0.12 deg C/W heatsinks, the temperature rise breaks down as follows:
625w x 0.12 deg C/W = 75.0 deg C
Shared between 2 heatsinks = 37.5 deg C
Now, on the forum, I've seen people describing ambient temperature (somewhat hopefully, in my view - isn't the ambient temperature in most homes closer to 20 deg C?) as anything from 17 - 20 deg C. That would give a heatsink temperature of between 54.5 - 57.5 deg C.
As I've explained elsewhere in this thread, I relly feel that the delights of class-A amplification should be in the realm of the masses, rather than in the hands of people like us, seen as audio 'freakies'. Unfortunately, what this means (for those commited to this idea) is getting away from the belief that a prerequsite for class-A projects such as SoZ is a huge case and mammoth temperatures. It is precisely this attitude that has helped to ensure class A amplification remains a minority interest.
The key prerequisite for Class-A amplification is that such a topology requres prodigious heat dissipation, not huge, scalding hot cases.
I've considered the 'traditional' route, huge, heavy, hot heatsinks, bolted either side of a chassis. Fortunately, I also considered the alternative - svelte, half-width cases, approx 4' x 9' x 14' (h x w x d). I couldn't see any advantage in the former approach and despair at the attitude that seems to prevail (even reflected in this thread) that as far as huge cases go, there is no other alternative. Even Nelson's notes make the point,
"...this is a project for those who like their hardware big..."
Anyhow, what say you, comrades?
Re: I think we need Nelson to resolve this one!
Originally posted by Lenin
Its one of those situations where Nelsons otherwise comprehensive notes are just a bit ambiguous. Do his descriptions of power supply requirements refer to supply requirements for one or two channels? I've presumed one and have organised supply requirements accordingly.
You are correct, the description given by Nelson is for one channel.
Oh, one minor caveat though - the space I have availiable in which to fit a decent (low flux) transformer means that I've had to revise my transformer specifications. These are now 2.5x Nelson's stated total dissipation figures. So, for a 25w output, (500w total dissipation in Nelson's notes) I'm specifying a 1.25KVA transformer with 32V secondaries per channel.
According to Nelson's note, 40V supply gets you about 24W, total dissipation is 600W not 500W, 500W only gets you 20W output. So, 1.25kVa is not much headroom @ all.
So, see below.
Total heat dissipation requirements for a 24w monoblock are 600w.
given that you're using 2 x 0.12 deg C/W heatsinks, the temperature rise breaks down as follows:
600w x 0.12 deg C/W = 72.0 deg C
Shared between 2 heatsinks = 36 deg C
Now, on the forum, I've seen people describing ambient temperature (somewhat hopefully, in my view - isn't the ambient temperature in most homes closer to 20 deg C?) as anything from 17 - 20 deg C.
That would give a heatsink temperature of between 53 - 56 deg C.
You then have to determine the junction temperature for the output devices you will be using....
Mica + grease is a little less than 0.5deg C/W and Rth(JC) is about 0.83 assuming TO-247 devices so if each fet is dissipating 52.5W, then device case temp is approx. 80deg .... and junction temp is approx. 125deg. So, you now have to find a suitable device for this given derating factors, inductive loading etc.
audio 'freakies'.
Hehehe, gosh i dont know what to say, thanx LOL.
Originally posted by Lenin
Its one of those situations where Nelsons otherwise comprehensive notes are just a bit ambiguous. Do his descriptions of power supply requirements refer to supply requirements for one or two channels? I've presumed one and have organised supply requirements accordingly.
You are correct, the description given by Nelson is for one channel.
Oh, one minor caveat though - the space I have availiable in which to fit a decent (low flux) transformer means that I've had to revise my transformer specifications. These are now 2.5x Nelson's stated total dissipation figures. So, for a 25w output, (500w total dissipation in Nelson's notes) I'm specifying a 1.25KVA transformer with 32V secondaries per channel.
According to Nelson's note, 40V supply gets you about 24W, total dissipation is 600W not 500W, 500W only gets you 20W output. So, 1.25kVa is not much headroom @ all.
So, see below.
Total heat dissipation requirements for a 24w monoblock are 600w.
given that you're using 2 x 0.12 deg C/W heatsinks, the temperature rise breaks down as follows:
600w x 0.12 deg C/W = 72.0 deg C
Shared between 2 heatsinks = 36 deg C
Now, on the forum, I've seen people describing ambient temperature (somewhat hopefully, in my view - isn't the ambient temperature in most homes closer to 20 deg C?) as anything from 17 - 20 deg C.
That would give a heatsink temperature of between 53 - 56 deg C.
You then have to determine the junction temperature for the output devices you will be using....
Mica + grease is a little less than 0.5deg C/W and Rth(JC) is about 0.83 assuming TO-247 devices so if each fet is dissipating 52.5W, then device case temp is approx. 80deg .... and junction temp is approx. 125deg. So, you now have to find a suitable device for this given derating factors, inductive loading etc.
audio 'freakies'.
Hehehe, gosh i dont know what to say, thanx LOL.
The operating range for the 240 is quoted at -55 to 150C ...... so 125 will be tolerated, just won't last that long
The thermal gradient from junction to case is one of the major factors in the large number of devices used in his commercial designs.
In my original 50W version I used 4 FETs per channel for this reason.
cheers, mark
PS: You may need more heasink. Better still, separate the FETs and bias resistors, you can run the later much hotter and this is exactly why Nelson suggested this
The thermal gradient from junction to case is one of the major factors in the large number of devices used in his commercial designs.
In my original 50W version I used 4 FETs per channel for this reason.
cheers, mark
PS: You may need more heasink. Better still, separate the FETs and bias resistors, you can run the later much hotter and this is exactly why Nelson suggested this
@80deg C case temp, IRFP240 is only good for about 80w.
Find a bigger mosfet cause that one is gonna die real quick.
Also for the 1ohm resistors, use 2 x 0.5ohm resistors in series or 2 x 2 ohm resistors in parallel so that you can spread the heat evenly across the heatsinks.
As suggested above, it would be good if you put the resistors on a seperate heatsink WRT the mosfets but given your design criteria I doubt you will do this.
Find a bigger mosfet cause that one is gonna die real quick.
Also for the 1ohm resistors, use 2 x 0.5ohm resistors in series or 2 x 2 ohm resistors in parallel so that you can spread the heat evenly across the heatsinks.
As suggested above, it would be good if you put the resistors on a seperate heatsink WRT the mosfets but given your design criteria I doubt you will do this.
Why don't you parallel mosfets. Here is a link:
http://members.ozemail.com.au/~dkfinnis/passlabs/images/soz_modified_circuit.gif
I beleve Nelson Pass sugested that for higher power SOZ.
http://members.ozemail.com.au/~dkfinnis/passlabs/images/soz_modified_circuit.gif
I beleve Nelson Pass sugested that for higher power SOZ.
As the owner of the parallelled SoZ circuit above, I would just clarify a few things.
In the example at hand, if the heatsink is 80C and the FETj 130C, then if I use 2 devices the FETj will fall to half the difference, ie 105C as Nelson states. The total W dissipation will be the same, irrespective of the number of FETs, as determined by the rail voltage selected for your SoZ.
As my first SoZ attempt was at 50W (the limit of Nelsons design) I thought 2 devices would be more robust, so wrote to NP who guided me toward the listed schematic. You will only need to consider this if (1) you run the heatsinks near the thermal limit for the FETs, and/or (2) build the amp at higher power levels, say 40+ W
Like any multiple FET circuit, if you choose 2 devices it pays to match them so they share the load equally.
cheers, mark
You do not gain higher power by doing this, the only gain is to share the FET junction-to-heatsink temperature differential between multiple devices. So for a given total W dissipation, the device junction temperatures will be lower, hence the devices will last longer.
In the example at hand, if the heatsink is 80C and the FETj 130C, then if I use 2 devices the FETj will fall to half the difference, ie 105C as Nelson states. The total W dissipation will be the same, irrespective of the number of FETs, as determined by the rail voltage selected for your SoZ.
As my first SoZ attempt was at 50W (the limit of Nelsons design) I thought 2 devices would be more robust, so wrote to NP who guided me toward the listed schematic. You will only need to consider this if (1) you run the heatsinks near the thermal limit for the FETs, and/or (2) build the amp at higher power levels, say 40+ W
Like any multiple FET circuit, if you choose 2 devices it pays to match them so they share the load equally.
cheers, mark
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.