Sneasle said:
Warning: novella follows, but I hope, one worth reading
Sorry for the confusion. I'm trying to give you a bare minimum nuts and bolts price and David is giving you the price for a complete cased amplifier.
My 'quote' is for components only, and at that, stripped to the absolute bare minimum. Power supply and amp, basically everything except the PCBs that you really need for a working amp. It is missing: 'misc' stuff like screws and standoffs, PCBs, heatsink, and enclosure - and assumes qty. 20. I linked the the BOM and the math used to get that price... Basically the absolute bare minimum for a working amp (that will immediately go into SPiKE protection). If you were going to offer a project like that, at the very least I would sort out something for the heatsink and make sure a suitable enclosure (ie. anything but a cardboard box
) is provided by the builder. I am interested though where the $50 discrepancy is coming from, do you have a complete BOM prepared David? It's scary how the little things all add up...I'll stop getting involved in the BOM planning since clearly it's just causing confusion, and instead offer some practical advice.
I'm not sure David's heatsink option is really worthwhile. It adds thermal mass, which will slow the heating, but it doesn't help dissipate the heat to the air, which is the critical step. It'll still overheat, it'll just take longer than without the aluminum sheet (that is, if the enclosure itself is insufficient). Your money is much better spent increasing exterior surface area than interior mass. I'd suggest getting some small aluminum U channel and bolting (with the 'fins' vertical) as much as you can fit onto the outside of the enclosure near the chips to form a makeshift sink. Even better, make cutouts in the chassis and mount the chips directly to the 'heatsink'. This stuff is pretty cheap, McMaster-Carr sells 8' lengths of 1/2"x3/4" U channel for about $10, you'd need maybe 12-18 inches per amp, and I think it would be fairly effective - and you can probably source it locally. Paint it with a very thin coat of matte black to improve effectiveness (and appearance
) since it's not black anodized. Still not ideal since the individual bits of U channel won't be coupled extremely well, but should still make a significant improvement.I'm not sure since I've never tried, but you may also have a large enough quantity to order heatsink extrusion in bulk. I suspect it's going to be outside your budget, but it might be worth contacting your local Aavid or Comair Rotron distributor to find out. By far the easiest and most professional choice. As a bonus, they will tell you the thermal resistance, so you can calculate how much extrusion you actually need per amp instead of the guesswork we'd otherwise be doing. Unless you were planning to measure that during prototyping, of course...
If you put any sort of sink inside the enclosure as opposed to outside, you will need to provision for air movement across the sink. That means drilling or cutting lots of holes in the case. Also, expect it to perform worse, holes or not. If you choose a case that doesn't have flat sides, you will need to cut holes and mount the chips directly to the heatsink or thermal transfer to the sink will be terrible.
The 'sink is an important part of the project if you want it to be successful. You can't afford to spend money on an overkill heatsink, so you really should calculate at the very least exactly how much heat you'll have to get rid of, and either measure the thermal resistance of the heatsink you choose, or test it empirically and see how much temperature rise there is at max power, and work back to find the die temperature. The section on this in the LM3886 datasheet is actually pretty good, I will also refer you to Rod Elliott's article on heatsinks for more than you possibly wanted to know: http://sound.westhost.com/heatsinks.htm
Good practical stuff for these students to be learning.
Cheers, and don't get discouraged. Even if you don't find a huge amount of interest at your school, I think the information gathered will be invaluable to the 'starving student' community. A complete budget chipamp kit with everything included for under $100 would be a boon for under funded DIYers everywhere.
Regarding the heat sink, I will try the simplest option first: heat spreader bolted to the aluminum enclosure. I have the appropriate measurement equipment and adequate knowledge to determine for myself if this approach will work. If it does not, then some type of external heat sink may be necessary. I feel like I have left enough wiggle room with the price quote to accomodate the worst case that an external heat sink is required without raising the price from the $80 quote.
Regarding the expressed concern over the seemingly large difference in price between the component costs on the previous BOM's and my estimate: I have been and will continue to be transparent about the costs of this group buy. Attached is the BOM which reflects the quantity discounts and combined shipping costs.
There may yet be room for discounts as firm volume prices are set, however I will not know until the orders are placed and the parts are shipped if the cost will drop. It would be unconscionable to promise a price that I cannot deliver, so I will not even attempt to estimate what the final price will be. As I've said before, the $80 apiece is what I feel confident to put in writing at this time.
I will draw attention at this point to the part of the BOM which reflects the labor, tooling, and r&d recovery costs. I intend to fully test and verify the performance of this design, and to supply a kit that is ready to assemble with normal hobby tools. As such I feel that the rear panel should be supplied with cutouts for the connectors and AC power receptacle and that the front panel should be supplied with cutouts for the power LED and Potentiometer. These will be cut with a template and the proper tools and will be of excellent quality.
The endplates can be provided uncut and the heatsinks undrilled to eliminate the $5 cost per unit if necessary. Other than that, the aformentioned costs are not negotiable. They were already estimated with very little value placed on my time, and I did not even budget the time for development and BOM planning into the cost (based on the fact that I take some form of enjoyment from the process and it is my hobby). I believe that these reasonable costs are within this forums guidelines for a "group buy".
I believe that $80 is a fair (if not optimistic) quote for the price for a ready to assemble kit. The price is not set in stone, but any price drops would have to occur on the materials end (and may yet do so).
Best Regards,
David Malphurs
Edit: Having trouble uploading the BOM, I'll try again soon.
Regarding the expressed concern over the seemingly large difference in price between the component costs on the previous BOM's and my estimate: I have been and will continue to be transparent about the costs of this group buy. Attached is the BOM which reflects the quantity discounts and combined shipping costs.
There may yet be room for discounts as firm volume prices are set, however I will not know until the orders are placed and the parts are shipped if the cost will drop. It would be unconscionable to promise a price that I cannot deliver, so I will not even attempt to estimate what the final price will be. As I've said before, the $80 apiece is what I feel confident to put in writing at this time.
I will draw attention at this point to the part of the BOM which reflects the labor, tooling, and r&d recovery costs. I intend to fully test and verify the performance of this design, and to supply a kit that is ready to assemble with normal hobby tools. As such I feel that the rear panel should be supplied with cutouts for the connectors and AC power receptacle and that the front panel should be supplied with cutouts for the power LED and Potentiometer. These will be cut with a template and the proper tools and will be of excellent quality.
The endplates can be provided uncut and the heatsinks undrilled to eliminate the $5 cost per unit if necessary. Other than that, the aformentioned costs are not negotiable. They were already estimated with very little value placed on my time, and I did not even budget the time for development and BOM planning into the cost (based on the fact that I take some form of enjoyment from the process and it is my hobby). I believe that these reasonable costs are within this forums guidelines for a "group buy".
I believe that $80 is a fair (if not optimistic) quote for the price for a ready to assemble kit. The price is not set in stone, but any price drops would have to occur on the materials end (and may yet do so).
Best Regards,
David Malphurs
Edit: Having trouble uploading the BOM, I'll try again soon.
Second attempt to upload BOM.
A ready to build chipamp kit has always been in the back of my mind. I guess your thread inspired me to give it a shot. What remains to be seen is how much demand there will be for such a kit. Any luck with selling the idea to the faculty?
If there is enough demand for this kit, I might consider offering it as a product after the group-buy. This would require a paid-for thread on the forum and would require me to really shave the materials cost down enough to make a worthwhile profit. We'll see what happens I guess.
-David
Edit: The second attempt failed as well. I may try out that newfangled googledocs thing.
A ready to build chipamp kit has always been in the back of my mind. I guess your thread inspired me to give it a shot. What remains to be seen is how much demand there will be for such a kit. Any luck with selling the idea to the faculty?
If there is enough demand for this kit, I might consider offering it as a product after the group-buy. This would require a paid-for thread on the forum and would require me to really shave the materials cost down enough to make a worthwhile profit. We'll see what happens I guess.
-David
Edit: The second attempt failed as well. I may try out that newfangled googledocs thing.
Ok, here's a link to the Google Documents Spreadsheet.
http://spreadsheets.google.com/ccc?key=pKKPjYSGJcRluWV3bS9y4Gg&hl=en
I have not allowed editing, but anyone can view it.
-David
http://spreadsheets.google.com/ccc?key=pKKPjYSGJcRluWV3bS9y4Gg&hl=en
I have not allowed editing, but anyone can view it.
-David
gtforme00 said:
Hey, if you're specifying chipamp.com boards, they do have LM1875 now. And, the recipe here http://www.diyaudio.com/forums/showthread.php?s=&threadid=123255 gives low dc offset without using the feedback capacitor.
EDIT: There's also the similar "schoolies" amplifier, popular with students in the UK.
I think the LM1875 can contribute less expense for the power supply and much ease of assembly. It can run a stereo pair on the 1.5a 36vct (2x18) $15 center-tap transformer from Triad. At voltages up to 27vdc rails, the LM1875 competes with the LM3886.
What do you think of that?
Not a bad idea really. I'll talk with Brian and see if he could supply the whole board/components kit at a discount. The only issue then is the transformer. The apexjr one could be had for as little as $10 apiece (diyaudio group discount), and would be hard to beat for price. The only problem is that the voltage on the secondaries is a bit high considering the power output and thermal dissipation capability of the design. I don't know how well regulated the 21V secondaries will be, but after diode losses and under load I would expect the rails to be +/- 25V.
The LM1875 requires +/- 25V rails for 20W into 4 or 8 ohm loads. It will dissipate about 32W of heat at that power rating with a 4 ohm load, and about 19W of heat with an 8 ohm load. For the point of discussion, we'll assume that the LM3886 dissipates the same amount of heat under the same conditions (though from the datasheets it appears to dissipate slightly less heat than the LM1875 for the same output power at +/- 25V.)
The 80VA transformer should be able to handle 20W of power output+20W of heat dissipation for two chips. The 80VA rating will become a limiting factor especially with 4 ohm loads, where increased current is demanded for the same power output. That is actually fine with us, as the 4 ohm load would otherwise require the chip to dissipate more power than we'd like.
Looking at the datasheets for these parts, the thermal resistance of the junction to case for the LM1875 is rated at 3 C/W (though in the application notes it says 2 C/W and they use that number for all of their calculations). The thermal resistance of the junction to case for the LM3886 is 1 C/W. This gives the LM3886 an advantage in that it can dissipate more heat energy for a given rise in temperature. This means that for the same power output, the LM3886 will actually require a smaller heat sink than the LM1875.
While we are calculating thermal resistances, lets go ahead and evaluate the heat sink performance required to output 20W into a 8 ohm load (assuming 20W of heat dissipation to achieve this).
The maximum junction temperature for either chip is 150 C. These amplifiers will be used in a consumer environment, so we will consider the maximum ambient temperature to be 40 C (104 F). This means that the total thermal resistance from the junction to the air has to be (150-40)C / 20W. That's 5.5 C/W total thermal resistance for the system. That number must include the sum of all resistances from the junction all the way to the air. There are three resistances that we will have to consider.
1. The thermal resistance from the junction to the case. This is given in the datasheet and we cannot change this number.
2. The thermal resistance from the case to the heatsink. This will depend on the type of insulator and heat sink paste we use, as well as what type of heat sink material is used and how well it is surface finished.
3. The thermal resistance from the heatsink to the air. This depends mostly on the surface area of the heatsink, though it also is affected by the orientation of the heatsink and the geometry of the heatsink.
If any other junctions are introduced, such as the heat spreader to aluminum housing, they will add to the series thermal resistance (but may still improve the overall thermal resistance).
Starting with our 5.5 C/W requirement we first subtract the junction to case resistance, 2C for the LM1875 and 1C for the LM3886, which leaves us with 3.5 C/W and 4.5 C/W respectively. Second we subtract the case to heatsink thermal resistance, which is estimated by National to be approximately 1.6 C/W for the LM1875 and 0.4 C/W for the LM3886. These estimates are based on a 2 mil mica insulator with heat sink paste on both sides. This leaves us with the required heatsink to air thermal resistance of 1.9 C/W for the LM1875 and 4.1 C/W for the LM3886.
For 30W of heat dissipation (20W power into 4 ohm load with +/- 25V rails), the thermal resistance requirement is .07 C/W for the LM1875 (not attainable) and 2.3 C/W for the LM3886.
The obvious alternative is to lower the voltage of the transformer for the LM1875. Doing this requires sourcing an inexpensive 18V secondary transformer.
I'm afraid with the extra heatsinking requirements and the loss of ability to use the inexpensive apexjr transformer, the cost savings of the LM1875 boards and parts will be a wash.
-David
The LM1875 requires +/- 25V rails for 20W into 4 or 8 ohm loads. It will dissipate about 32W of heat at that power rating with a 4 ohm load, and about 19W of heat with an 8 ohm load. For the point of discussion, we'll assume that the LM3886 dissipates the same amount of heat under the same conditions (though from the datasheets it appears to dissipate slightly less heat than the LM1875 for the same output power at +/- 25V.)
The 80VA transformer should be able to handle 20W of power output+20W of heat dissipation for two chips. The 80VA rating will become a limiting factor especially with 4 ohm loads, where increased current is demanded for the same power output. That is actually fine with us, as the 4 ohm load would otherwise require the chip to dissipate more power than we'd like.
Looking at the datasheets for these parts, the thermal resistance of the junction to case for the LM1875 is rated at 3 C/W (though in the application notes it says 2 C/W and they use that number for all of their calculations). The thermal resistance of the junction to case for the LM3886 is 1 C/W. This gives the LM3886 an advantage in that it can dissipate more heat energy for a given rise in temperature. This means that for the same power output, the LM3886 will actually require a smaller heat sink than the LM1875.
While we are calculating thermal resistances, lets go ahead and evaluate the heat sink performance required to output 20W into a 8 ohm load (assuming 20W of heat dissipation to achieve this).
The maximum junction temperature for either chip is 150 C. These amplifiers will be used in a consumer environment, so we will consider the maximum ambient temperature to be 40 C (104 F). This means that the total thermal resistance from the junction to the air has to be (150-40)C / 20W. That's 5.5 C/W total thermal resistance for the system. That number must include the sum of all resistances from the junction all the way to the air. There are three resistances that we will have to consider.
1. The thermal resistance from the junction to the case. This is given in the datasheet and we cannot change this number.
2. The thermal resistance from the case to the heatsink. This will depend on the type of insulator and heat sink paste we use, as well as what type of heat sink material is used and how well it is surface finished.
3. The thermal resistance from the heatsink to the air. This depends mostly on the surface area of the heatsink, though it also is affected by the orientation of the heatsink and the geometry of the heatsink.
If any other junctions are introduced, such as the heat spreader to aluminum housing, they will add to the series thermal resistance (but may still improve the overall thermal resistance).
Starting with our 5.5 C/W requirement we first subtract the junction to case resistance, 2C for the LM1875 and 1C for the LM3886, which leaves us with 3.5 C/W and 4.5 C/W respectively. Second we subtract the case to heatsink thermal resistance, which is estimated by National to be approximately 1.6 C/W for the LM1875 and 0.4 C/W for the LM3886. These estimates are based on a 2 mil mica insulator with heat sink paste on both sides. This leaves us with the required heatsink to air thermal resistance of 1.9 C/W for the LM1875 and 4.1 C/W for the LM3886.
For 30W of heat dissipation (20W power into 4 ohm load with +/- 25V rails), the thermal resistance requirement is .07 C/W for the LM1875 (not attainable) and 2.3 C/W for the LM3886.
The obvious alternative is to lower the voltage of the transformer for the LM1875. Doing this requires sourcing an inexpensive 18V secondary transformer.
I'm afraid with the extra heatsinking requirements and the loss of ability to use the inexpensive apexjr transformer, the cost savings of the LM1875 boards and parts will be a wash.
-David
I found this chart which displays the convection thermal resistance of a flat aluminum plate of two different thicknesses.
If this graph is any way accurate, then one can deduce that an aluminum enclosure having dimensions of 20cm x 14.5cm x 7cm is not precluded from having a sufficient thermal resistance for this application. Like I said before, the best way to know is to test, and test I shall.
-David
An externally hosted image should be here but it was not working when we last tested it.
If this graph is any way accurate, then one can deduce that an aluminum enclosure having dimensions of 20cm x 14.5cm x 7cm is not precluded from having a sufficient thermal resistance for this application. Like I said before, the best way to know is to test, and test I shall.
-David
Updated the spreadsheet with the newest cost calculations. Unfortunately, there was an error in the spreadsheet which incorrectly summed the cost as lower than the actual sum of parts (left some rows out of the sum function). The good news is with the new cost reductions that the price is still at/under $80 each. I am ordering the parts to prototype the design this week and will start a new thread when pictures and details are available.
-David
Meanwhile, I got some samples of the single supply automotive style chips in. The packages look similar to the LM3886 chips, and should be comparable in heat transfer capability. I don't know if or when I will get time to build up a board and evaluate them.
-David
Meanwhile, I got some samples of the single supply automotive style chips in. The packages look similar to the LM3886 chips, and should be comparable in heat transfer capability. I don't know if or when I will get time to build up a board and evaluate them.
gtforme00 said:
Howabout the TDA7265's brilliantly simple single or bridge applications and a $3 transformer: http://www.diyaudio.com/forums/showthread.php?s=&threadid=123003
Ok guys, this thread has kinda died and now I'm going to revive it.
I finally got my amp from home. It is based on the TDA7384 4x35w.
2 resistors, 8 caps, 1 volume pot.
We did not use a transformer. Instead, we used a walwart kicking out 9v AC @1A.
We ran this through a bridge rectifier (4 diodes, 4 caps) and then into a 10kuF cap along with some snub caps.
So.. whatcha think about that? From the looks of it, this looks to me like it could be built very cheap assuming we can find the wall warts.
I finally got my amp from home. It is based on the TDA7384 4x35w.
2 resistors, 8 caps, 1 volume pot.
We did not use a transformer. Instead, we used a walwart kicking out 9v AC @1A.
We ran this through a bridge rectifier (4 diodes, 4 caps) and then into a 10kuF cap along with some snub caps.
So.. whatcha think about that? From the looks of it, this looks to me like it could be built very cheap assuming we can find the wall warts.
yes, 9V AC, and it sounds great. It is more then loud enough for my living room at home, and that isn't a small room.
Have a look at the datasheet:
4 x 22W/4W @13.2V, 1KHz, 10%
so why on earth would you want 140vac transformer?
And why woudl we want to use computer power supplies when this little wall wart does an admirable job?
This guy built his as an actual car amp:
http://www.scienceprog.com/diy-4x22w-car-audio-amplifier-based-on-tda7384/
Have a look at the datasheet:
4 x 22W/4W @13.2V, 1KHz, 10%
so why on earth would you want 140vac transformer?
And why woudl we want to use computer power supplies when this little wall wart does an admirable job?
This guy built his as an actual car amp:
http://www.scienceprog.com/diy-4x22w-car-audio-amplifier-based-on-tda7384/
Sneasle said:
A computer power supply outputs clean 12vdc from the yellow and black connections of the Molex plug (hard drive plug).
Computer power supplies are frequently found at $0, when the computer is old or inoperable (salvage).
Unlike "wall plug" devices, computer power supplies are usually clean and free from hum. That's a pretty good trick for $0.
It represents a way to increase the quality while decreasing the cost.
EDIT: Further reference, see this page at Decibel Dungeon: http://myweb.tiscali.co.uk/nuukspot/decdun/gainclonesmps.html
What about something like this? http://skycraftsurplus.com/index.asp?PageAction=VIEWPROD&ProdID=1912
oooohhh... I like that.
That might be workable. Size and price seems reasonable and 3.5A @ 12v should be enough to get all 4 channels running to a reasonable level.
or maybe this guy:
http://skycraftsurplus.com/index.asp?PageAction=VIEWPROD&ProdID=1672
That might be workable. Size and price seems reasonable and 3.5A @ 12v should be enough to get all 4 channels running to a reasonable level.
or maybe this guy:
http://skycraftsurplus.com/index.asp?PageAction=VIEWPROD&ProdID=1672
Got an update for ya'll. Finally took the time to get the schematic drawn.
It is pretty much exactly what the datasheet has. Any input on if I should change/add something somewhere?
What about my caps and resistors? I guess the pull up resistors can be any ol cheap resistor, but what about the input caps and the PSU holdup/bypass/filtering caps?
Also, Can anyone recommend some decent speakers for zigma hornets that can be had at a decent price? I want to build one of these amps for mom along with a set of 4 hornets for her bedroom.
It is pretty much exactly what the datasheet has. Any input on if I should change/add something somewhere?
What about my caps and resistors? I guess the pull up resistors can be any ol cheap resistor, but what about the input caps and the PSU holdup/bypass/filtering caps?
Also, Can anyone recommend some decent speakers for zigma hornets that can be had at a decent price? I want to build one of these amps for mom along with a set of 4 hornets for her bedroom.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.