Ryan,
Did you ever build this? I'm thinking of something similar and would love to see if it worked well for you.
Did you ever build this? I'm thinking of something similar and would love to see if it worked well for you.
not yet
Not yet...but it's on my todo list. Unfortunately I think it's deeper buried now than it was when I first queued it up. ;-(
If you proceed post back with details, I'd love to check out what you come up with.
- Ryan
Not yet...but it's on my todo list. Unfortunately I think it's deeper buried now than it was when I first queued it up. ;-(
If you proceed post back with details, I'd love to check out what you come up with.
- Ryan
The power consumption is reliant on load.
Either a zero load or a 100R load has a chipamp running nearly cold.
Per each stereo pair, I'd install a switch and Rod Elliot's headphone adapter circuit (100R load), which is just inexpensive resistors. You can use this to knock out the speaker load for unused channels (considerably less power consumption), but you can so use it to test your amplifier with headphones. For me, this makes testing easier than any attempt at getting in "just the right spot" in front of a loudspeaker.
Star grounding is required that way (above); however, this means that you only need to run 3 conductors to a stereo pair of loudspeakers. Reasonably good outdoor extension cords can do it. You can use the 14ga for shorter runs or the 12ga for longer runs. Unfortunately, they are orange. 😉
Either a zero load or a 100R load has a chipamp running nearly cold.
Per each stereo pair, I'd install a switch and Rod Elliot's headphone adapter circuit (100R load), which is just inexpensive resistors. You can use this to knock out the speaker load for unused channels (considerably less power consumption), but you can so use it to test your amplifier with headphones. For me, this makes testing easier than any attempt at getting in "just the right spot" in front of a loudspeaker.
Star grounding is required that way (above); however, this means that you only need to run 3 conductors to a stereo pair of loudspeakers. Reasonably good outdoor extension cords can do it. You can use the 14ga for shorter runs or the 12ga for longer runs. Unfortunately, they are orange. 😉
danielwritesbac said:
At idle the power consumption is independent of load. Every amp has a quiescent current draw when idle unless it is completely disconnected from the supply. For the case of the LM1875 with +/-25V rails the datasheet says the chip will draw about 70mA of current. Power dissipation is just voltage multiplied by current, so the idle dissipation of the chip with +/-25V supply rails would be 50V*0.07A=3.5W. That is regardless of load because there is no voltage across the load and thus no current through the load.
That is one of the things I was struggling with in my design and I never did solve. At 12 stereo pairs, I didn't want to burn that much idle power. IIRC there is a 3 channel chip w/ mute feature, but even 8 of those muted were going to sink more power than a small incandescent lightbulb. Since this thing will be on 24/7/365, that's not desireable.
Pro amps designed to provide this feature are low-voltage triggerable by the source to turn on the amp via relay, and consume very little power until triggered (the trigger circuit would be independent with it's own PSU from the amp itself). I for one wouldn't mind the clicking on of a relay as the amp is going in the basement. That was the avenue I was going to start going down, as as long as the amp is only sinking significant power while it's playing, I could care less. I did find a website that had a circuit to do this, but off hand I can't remember it, and that bookmark isn't available on this PC. If I find it later I'll post it.
My goal was to find a simple low-voltage trigger circuit with it's own small, efficient PSU which would drive a high-load relay that could sit either on the AC or DC side of the main power supply. The downside of the AC side is the PSU caps would cause the amp power to gradually taper off (ie music would slowly fade), but that wouldn't be a problem if the source which triggered the turn off had stopped or muted the source input anyway (alternatively, if the chip amp was mutable, the trigger circuit could drive that as well I guess). If on the DC side, then there would be no concern with the fading sound, but as I understand it it's not as safe to leave the caps charged, as it could take quite some time for them to charge down. So I was leaning towards putting the relay on the AC side.
Ryan
Pro amps designed to provide this feature are low-voltage triggerable by the source to turn on the amp via relay, and consume very little power until triggered (the trigger circuit would be independent with it's own PSU from the amp itself). I for one wouldn't mind the clicking on of a relay as the amp is going in the basement. That was the avenue I was going to start going down, as as long as the amp is only sinking significant power while it's playing, I could care less. I did find a website that had a circuit to do this, but off hand I can't remember it, and that bookmark isn't available on this PC. If I find it later I'll post it.
My goal was to find a simple low-voltage trigger circuit with it's own small, efficient PSU which would drive a high-load relay that could sit either on the AC or DC side of the main power supply. The downside of the AC side is the PSU caps would cause the amp power to gradually taper off (ie music would slowly fade), but that wouldn't be a problem if the source which triggered the turn off had stopped or muted the source input anyway (alternatively, if the chip amp was mutable, the trigger circuit could drive that as well I guess). If on the DC side, then there would be no concern with the fading sound, but as I understand it it's not as safe to leave the caps charged, as it could take quite some time for them to charge down. So I was leaning towards putting the relay on the AC side.
Ryan
I'm still quite a ways from actually doing this. My ultimate goal was to use this amp along with CQC software to do my wholehouse audio. I don't need as many channels as you, 5-6 zones would be plenty so my power consumption would be slightly less but still a big concern.
I want to be able to control the zones serially thru CQC, so I was thinking of trying to use a PIC to switch relays which would turn off the zones when not in use. But admittedly I haven't thought this thru very well yet. You can pick up Nuvo systems for such a resonable amount online now, that I'm not sure that this project is really worthwhile. That's kind of the dilema I'm working thru.
I want to be able to control the zones serially thru CQC, so I was thinking of trying to use a PIC to switch relays which would turn off the zones when not in use. But admittedly I haven't thought this thru very well yet. You can pick up Nuvo systems for such a resonable amount online now, that I'm not sure that this project is really worthwhile. That's kind of the dilema I'm working thru.
My portable has a single LM1875, operating on a 3" to3 heatsink, and it has a 28va transformer (28vct 1a). This, after a lot of fine tuning, operates only a few degrees over room temperature when its going full blast. I cheated and used a small value resistor between it and the load; however, I chose a driver that sounds nice that way. 😉
This isn't much power; however, there's available drivers that don't require much power, like this: http://www.parts-express.com/pe/showdetl.cfm?Partnumber=264-852
91db 8 ohm high fidelity driver
Or, the Pioneer B20
or a highly efficient wideband/fullrange of some sort
For either of the 91db efficient drivers, you can use a VERY high crossover point to the tweeter. The small inductor involved in the crossover doesn't drop the efficiency much. Its also an inexpensive crossover. That's typical of better quality drivers, like the Tang 6-1/2 in the link above. More spendy on the driver, less spendy on the crossover. If you use a series type crossover (tie the tweeter + and - to the inductor--just one inductor, but you can add an additional cap to the tweeter), the amplifier can run a bit cooler. That takes a bit of tinkering, but I like the idea of "whatever doesn't come out of one driver must come out of the other" and of course purchasing only 1 little inductor is really nice too. 😉
And now. . .
My current home project is to use pairs of the Tang driver for 16 ohm speakers. There will be a slight efficiency drop, but my amplifiers here at home will run cold temperatures. The $90 price tag per pair is well compensated for me via so little expense for crossover and tweeter, and these speakers will have paid for themselves several times over their service life.
You can often find older 16 ohm drivers to restore, or some that don't need restoration. Its 16 ohms that'll chill out this big multi-channel project.
This isn't much power; however, there's available drivers that don't require much power, like this: http://www.parts-express.com/pe/showdetl.cfm?Partnumber=264-852
91db 8 ohm high fidelity driver
Or, the Pioneer B20
or a highly efficient wideband/fullrange of some sort
For either of the 91db efficient drivers, you can use a VERY high crossover point to the tweeter. The small inductor involved in the crossover doesn't drop the efficiency much. Its also an inexpensive crossover. That's typical of better quality drivers, like the Tang 6-1/2 in the link above. More spendy on the driver, less spendy on the crossover. If you use a series type crossover (tie the tweeter + and - to the inductor--just one inductor, but you can add an additional cap to the tweeter), the amplifier can run a bit cooler. That takes a bit of tinkering, but I like the idea of "whatever doesn't come out of one driver must come out of the other" and of course purchasing only 1 little inductor is really nice too. 😉
And now. . .
My current home project is to use pairs of the Tang driver for 16 ohm speakers. There will be a slight efficiency drop, but my amplifiers here at home will run cold temperatures. The $90 price tag per pair is well compensated for me via so little expense for crossover and tweeter, and these speakers will have paid for themselves several times over their service life.
You can often find older 16 ohm drivers to restore, or some that don't need restoration. Its 16 ohms that'll chill out this big multi-channel project.
Creepy.danielwritesbac said:
.. or get lost in the cross-over like in any other cross-over as well.danielwritesbac said:
Contradiction alert.danielwritesbac said:
pacificblue said:Creepy.
.. or get lost in the cross-over like in any other cross-over as well.
Contradiction alert.
I thought I'd never meet the man who could possibly win an argument with a signpost.
What a relief that the long wait is over.
EDIT: Stubbornness is an excellent trait, but, at this time, I'd like to see it directed towards answering the power consumption, heat output, and other concerns involved in making a 24 channel "chip amp" amplifier, because that is the topic of this thread. Thanks.
pacificblue said:
If I read all that, and then made a guess from there. . .
Its 4 LM4782 per each transformer. And, that's making a lot of assumptions, because there was no clear directive. 😉
I think that could use some "L" channel aluminum rail across the chip fronts to help speed up the heat dissipation, what do you think?
And, will about 2 of these do it http://www.alliedelec.com/Search/ProductDetail.aspx?SKU=6190079 or is that an overdo?
Well, that's what I thought of the comments when I read them.
danielwritesbac said:
If you read all that, you would not need to guess and assume, because then you would know the directive.
Is it a good idea to give recommendations, when you do not even know exactly what about?
And is it a good idea to participate in a forum, when you easily tire of reading?
For 4 of LM4782, what transformer VA size do you recommend?
Do you think 8 ampers is enough per 4 LM4782? If so, there's an economical solution. If not, well, 12a+ will cost more.
Well, that was one question.
Here's another:
With the proposed 12 left channels and 12 right channels, the input impedance is in question, if this is from one single source. What's the most seemly way to deal with this?
Do you think 8 ampers is enough per 4 LM4782? If so, there's an economical solution. If not, well, 12a+ will cost more.
Well, that was one question.
Here's another:
With the proposed 12 left channels and 12 right channels, the input impedance is in question, if this is from one single source. What's the most seemly way to deal with this?
Hey PacificBlue, you put some great power supply facts here: http://www.diyaudio.com/forums/showthread.php?postid=1675923#post1675923
This bit is fantastic (quote from PacificBlue):
"The advantages of separate rectifiers for positive and negative rails are less losses in the transformer, better power distribution across the secondary windings, better regulation and better suppression of noise (hum) during unbalanced load situations. Disadvantages are higher component count = higher price and double the voltage drop. And of course two separate bridges only make sense with double secondaries, not with center tapped transformers."
And, I believe that "better regulation" results in "more durable" as well.
This bit is fantastic (quote from PacificBlue):
"The advantages of separate rectifiers for positive and negative rails are less losses in the transformer, better power distribution across the secondary windings, better regulation and better suppression of noise (hum) during unbalanced load situations. Disadvantages are higher component count = higher price and double the voltage drop. And of course two separate bridges only make sense with double secondaries, not with center tapped transformers."
And, I believe that "better regulation" results in "more durable" as well.
Depends on the voltage and the speaker load you use, which determines the output power. Smallest possible choice is ±10 V and 8 Ohm loads resulting in 12 x 3,6 W -> 45-135 VA. Take into acount AndrewT's recommendation not to use transformers with less than 100 VA. That makes more than 10 A from a 2 x 9 V transformer with two rectifier bridges. Connect 4 Ohm speakers to that and you get 12 x 5,6 W -> 70-210 VA. Still a 100 VA type will do.danielwritesbac said:
Trouble with that configuration is that mains voltage tolerances may lead to undervoltage at the ICs. So rather go for a transformer with more than 2 x 11 V.
Biggest possible choice is ±27 V and 6 Ohm loads, resulting in 12 x 36 W -> 450-900 VA. That would be more than 20 A from a 2 x 22 V tranformer with two rectifier bridges. More than ±24 V however bring serious heatsinking problems with 6 Ohm loads.
Don't use more than ±20 V with 4 Ohm speakers. Then the transformer would be for 12 x 25 W -> 300-600 VA. More than 17 A from a 2 x 17 V transformer with two rectifier bridges.
Conclusion. Use a 2 x 15 V transformer with 300 VA. That gives 12 W/ch into 8 Ohm, 14 W/ch into 6 Ohm, 18W/ch into 4 Ohm and does not take heatsinking to its limits. 1 K/W per IC for 25 °C, and temperature controlled fans for higher ambient temperatures should work fine even with 4 Ohm loads. With only 8 Ohm loads you could get by without fans.
The same way you deal with any parallel configuration. One blocking resistor per chip, and one Rin for all. Add an input buffer between Rin and the blocking resistors to be on the safe side with so many ICs in parallel, although it will probably work without one as well.danielwritesbac said:
A bit more on this heatsink: http://www.alliedelec.com/Search/ProductDetail.aspx?SKU=6190079
It is 0.72 C/W per 3" of height.
The product (Tabor Extrusions) is in the bulk category, and it is 6" height.
You could say that there are two of 0.72 C/W heatsinks in the one item (except that they are not cut apart).
Speed of transfer is a different matter, but there's some good news:
The surface isn't totally slick on either the chip or the heatsink. I have one, and raked a fingernail across it to discover that its not perfect surface. Although the surface of the heatsink is shiny--which is very good news!
And the bad news. . .
The chip lists as "TA" model, not "TF" model. I think this means that we're going to need some Kapton (insulator sheet) and some of either Arctic Ceramique or GC Waldham Type 44. IE: a non-conductive thermal paste that does have "filler" properties to help the unlevel surfaces.
Here's a puzzle:
In order to avoid heat buildup within the chip, we can also use the face of the chip, which is plastic. And, we can relieve the heat from it. An available, successful, and non-adhesive "plastic to metal thermal compound" is Artic Silver. This, and a bit of metal held tight to the front (printed side face of the chip) via the chip's mounting screws (termed "clamp"), will remove (level out) the heat from the plastic (actually, it will prevent "hot spots"), and that will help postpone the SPIKE system.
Bit of metal. . .
It needs to be something a). Metal, b). Stiff, and c), Available easily to the diy builder.
What bit of metal to use for clamp?
An externally hosted image should be here but it was not working when we last tested it.
It is 0.72 C/W per 3" of height.
The product (Tabor Extrusions) is in the bulk category, and it is 6" height.
You could say that there are two of 0.72 C/W heatsinks in the one item (except that they are not cut apart).
Speed of transfer is a different matter, but there's some good news:
The surface isn't totally slick on either the chip or the heatsink. I have one, and raked a fingernail across it to discover that its not perfect surface. Although the surface of the heatsink is shiny--which is very good news!
And the bad news. . .
The chip lists as "TA" model, not "TF" model. I think this means that we're going to need some Kapton (insulator sheet) and some of either Arctic Ceramique or GC Waldham Type 44. IE: a non-conductive thermal paste that does have "filler" properties to help the unlevel surfaces.
Here's a puzzle:
In order to avoid heat buildup within the chip, we can also use the face of the chip, which is plastic. And, we can relieve the heat from it. An available, successful, and non-adhesive "plastic to metal thermal compound" is Artic Silver. This, and a bit of metal held tight to the front (printed side face of the chip) via the chip's mounting screws (termed "clamp"), will remove (level out) the heat from the plastic (actually, it will prevent "hot spots"), and that will help postpone the SPIKE system.
Bit of metal. . .
It needs to be something a). Metal, b). Stiff, and c), Available easily to the diy builder.
What bit of metal to use for clamp?
pacificblue said:
This is very wise. And, it is excellent.
Unfortunately, it won't do. I'm so sorry to have to mention this, but the useful power is limited with the Overture chips, because of the Spike system's really unseemly clipping noise. We must use more voltage.
That is approximately 25v rails before you can have a useful 14w (average) with 6 ohm (average spec) loads.
Therefore, howabout nearly inaudible fans: http://www.newegg.com/Product/Product.aspx?Item=N82E16835103022
*At the hardware store is a passive thermostat (bimetallic spring), known as "attic ventilator control" that can turn these fans on and off as needed.
**Specify air intake vent hole in amplifier enclosure as directly underneath the heatsink, when these fans are used for exhausting hot air from the amplifier enclosure.
danielwritesbac said:
Distortion due to clipping is NOT related to the spike protection. The LM1875 doesn't even have spike protection. If it does it is not stated anywhere in the datasheet... It does have internal current limiting and thermal shutdown protection which is supposedly not the same as the spike protection circuitry.
Look at page 3 of the LM1875 datasheet. There is a power output versus supply voltage graph. With an 8 ohm load and +/-20V rails you can get 17 watts @ 1% THD. With the LM3875 you will have similar power into 8 ohms but with 0.1% THD. The LM3875 will also run cooler because it has less quiescent current and a physically larger package.
BWRX said:
LM1875 wasn't included in the most recent discussion. . .
We were using LM1876 and/or its 3 channel version.
Limited in power, it would be prone to clip if run on insufficient voltage.
The audible effects of that on an Overture chip are documented.
BWRX said:
THANK YOU for dropping a hint about LM3875. This has a few advantages in that it does run cool and its available in the super-convenient "TF" insulated package (as in one screw and Artic Silver), which decreases complexities. Yet another advantage are the many point-to-point wiring guides, that "could" eliminate the expense of circuit boards, as well as the PCB availability issues.
What do you think about it?
- Status
- Not open for further replies.