Folks:
I'd like to build a chip amp for my daughter. I have constructed four LM3875-based integrated amps to date, but all have been mid-powered (about 45 watts/channel into 8 ohms). On 86 dB speakers, it doesn't take much volume before clipping is audible on those amps.
For this next project, I'd like to build something a bit more powerful. The audio inputs into this new amp will all be single-ended and the speakers will probably have a nominal impedence of 8 ohms. I'd like to shoot for an integrated stereo amp capable of producing 90 to 100 watts/channel.
Can anyone offer a recommendation? I'll look for PCBs once the requisite chip has been determined.
Many thanks,
Scott
I'd like to build a chip amp for my daughter. I have constructed four LM3875-based integrated amps to date, but all have been mid-powered (about 45 watts/channel into 8 ohms). On 86 dB speakers, it doesn't take much volume before clipping is audible on those amps.
For this next project, I'd like to build something a bit more powerful. The audio inputs into this new amp will all be single-ended and the speakers will probably have a nominal impedence of 8 ohms. I'd like to shoot for an integrated stereo amp capable of producing 90 to 100 watts/channel.
Can anyone offer a recommendation? I'll look for PCBs once the requisite chip has been determined.
Many thanks,
Scott
An externally hosted image should be here but it was not working when we last tested it.
128W into 8ohm at +/-30V according to Nationals Overture Design guide. First board is up and playing now. Sound is excellent
CJ900RR said:An externally hosted image should be here but it was not working when we last tested it.
128W into 8ohm at +/-30V according to Nationals Overture Design guide. First board is up and playing now. Sound is excellent
Is this three in parallel? If so, how did you figure 128W in to 8 ohms at +-30v?
Assuming ~28v peak
28v peak = 19.8v rms?
19.8v rms in to 8 ohms = 49W rms?
28v peak in to 8 ohms = 98W peak?
That's how I understand it anyway. It's may be wrong..
markiemrboo said:
Yes it's 3 in parallel. I filled in the values I have used on the board in Nationals XLS-dokument (http://www.national.com/appinfo/audio/files/Overture_Design_Guide13.xls) and didnt bother to calculate by my self at all. Dont know if their calculation is correct.
For 8 Ohms speakers bridged topology might be a better choice (although it serves me great with my 4 Ohms speakers).
You can see the details here:
http://jrexton.ecv.ms/GC3.htm
http://jrexton.ecv.ms/JFET-Buffered-opa1632-Bridg.gif
You can see the details here:
http://jrexton.ecv.ms/GC3.htm
http://jrexton.ecv.ms/JFET-Buffered-opa1632-Bridg.gif
markiemrboo said:
Your understanding is pretty much on the spot. Extensive paralleling of chips improves current availability <=> more power on low impedance loads (2-4 Ohms).
Higher impedance loads are better served with more voltage - bridge topology (but there still has to be enough current to support the demand - Ohm's Law can not be cheated)
SRMcGee said:
Hi Scott!
I'm wondering if your daughter is a musician? Otherwise 107db is a little bit intense (ear damage) for home use. Anyway, about 120 watts will do it.
Easier: Your current amplifiers with 91db speakers (quite a lot of them at Parts Express) will also do that same 107db. Just a thought. Personally I like that option best because "lively" speakers tend to do more intense dynamics, which is somewhat more helpful to modern recordings.
For a few amplifiers worth of production, PCBs are somewhat unnecessary due to the innovation of "Experimenter's Board." That's just like veroboard (phenolic) except with circular solder pads at every hole (avoids the usual tangle). Its nice because you can make adjustments to the layout. Its purpose is actually testing before ordering PCB boards, but it works well for home amplifiers too.
That's all I can think up this late in the evening. Have fun!
EDIT: The reason that I was thinking about speakers is. . . the next step after exposing most speakers to more than 100 watts, is a purchase of replacement speakers, so might as well just skip the speaker cooking step and get 91db speakers instead.
markiemrboo said:
That figure is for a 2.6 ohm load, not 8 ohm
But then again... isnt that the purpose with paralleling them? That each LM3886 see's only some-what-ohm, in this case 2,67ohm/IC. They share the power dissapation equally. The result is a higher output. In the case of 8omh it might be more efficient to use a bridged or a bridged/parallel config...
If authenic, the TDA7293 puts out a lot of power; however, if its fake (not manufacturered at ST Thompson), then it smokes. Oops.
There's also prebuilt Hypex modules available--with still plenty of DIY to do on power supply, amp enclosures, etc. . .
The sturdy answer to an extreme power chipamp is a bridge+parallel LM3886.
Here's a little chart about what the power actually does:
Watts, 86db, 92db speakers
1w, 86db, 92db
2w, 89db, 95db
4w, 92db, 98db
8w, 95db, 101db
16w, 98db, 104db
32w, 101db, 107db*
64w, 104db, 110db
128w, 107db*, 113db
256w, 110db, 116db
512w, 113db, 119db
There's also prebuilt Hypex modules available--with still plenty of DIY to do on power supply, amp enclosures, etc. . .
The sturdy answer to an extreme power chipamp is a bridge+parallel LM3886.
Here's a little chart about what the power actually does:
Watts, 86db, 92db speakers
1w, 86db, 92db
2w, 89db, 95db
4w, 92db, 98db
8w, 95db, 101db
16w, 98db, 104db
32w, 101db, 107db*
64w, 104db, 110db
128w, 107db*, 113db
256w, 110db, 116db
512w, 113db, 119db
CJ900RR said:
Sort of. In parallel the current will be split between each IC, and lower the pd of each chip. This means that in parallel, you get more current handling ability.
However, at a basic level power output can be limited by either current or voltage. I say at a basic level, as it's limited by heat and over such things. But in to 8 ohms, you'll be limited by voltage and not so much current. With +-30v supplies, the maximum output voltage will probably be something like 25 - 28v (depends on how good the power supply is under load). No matter how many chips you parallel, the supply voltage will remain constant.
As the load impedance gets lower, you start to need more current and less voltage to get the same power output, and so it slowly starts to become the current handling ability and not the supply voltage which will be the basic limit.
Speakers aren't pure resistance though. An 8 ohm 'nominal' speaker might drop to 4 ohms at certain frequencies. It might also have a 'phase' of higher than 0 degree's. This would cause the voltage and current to be 'out of sync', which will increase power dissipation in the output transistors (see http://sound.westhost.com/soa.htm for a bit more info on that).
So, it is quite possible that an 8 ohm speaker may indeed benefit from paralleled ICs, but you still don't really get any more watts *at 8 ohms* from paralleling as you're mainly limited by the supply voltage in this case.
Nice looking boards by the way.
And sorry to the OP for going off topic. I believe there is a TDA chip that will do about 100W in to 8 ohms.
In my opinion it may be worth looking at discrete output stages for higher powers. National have a few "driver" chips that you can use for the front end, which simplifies things. I'm in the middle of building something based on the LME49810 (see also LM4702 and LME49830). I still don't understand the front end stuff really, but it has allowed me to learn a bit more about output stages, and it's not too difficult to understand the basic ideas / the basics on how it works!
The down side is that it will have no protection like the ICs. With an emitter-follower output stage using bipolar transistors, current limiting can be added fairly easily, in that the circuit is fairly simple to add. I'm still having difficulty understanding it enough to pick out appropriate resistor values. Thermal limiting and DC offset protection might be done with a microcontroller, and if you include this I believe it might not be strictly necessary to have over current protection, though it is obviously useful to help prevent blowing of transistors.
I say this because the DC offset protection would save the speakers should the output transistors blow. This is something which most, if not all ICs don't have in built protection against. They'll attempt to save themselves from overheating or overdrive, but if their output transistors do blow for whatever reason, I believe they'll quite happily take the speakers with them!
It's an option
And sorry to the OP for going off topic. I believe there is a TDA chip that will do about 100W in to 8 ohms.
In my opinion it may be worth looking at discrete output stages for higher powers. National have a few "driver" chips that you can use for the front end, which simplifies things. I'm in the middle of building something based on the LME49810 (see also LM4702 and LME49830). I still don't understand the front end stuff really, but it has allowed me to learn a bit more about output stages, and it's not too difficult to understand the basic ideas / the basics on how it works!
The down side is that it will have no protection like the ICs. With an emitter-follower output stage using bipolar transistors, current limiting can be added fairly easily, in that the circuit is fairly simple to add. I'm still having difficulty understanding it enough to pick out appropriate resistor values. Thermal limiting and DC offset protection might be done with a microcontroller, and if you include this I believe it might not be strictly necessary to have over current protection, though it is obviously useful to help prevent blowing of transistors.
I say this because the DC offset protection would save the speakers should the output transistors blow. This is something which most, if not all ICs don't have in built protection against. They'll attempt to save themselves from overheating or overdrive, but if their output transistors do blow for whatever reason, I believe they'll quite happily take the speakers with them!
It's an option
Thanks for the info on the LME49810! Good luck with your amp project, keep us posted!
I just DL'd the datasheet for the LME49810. Seems like a nice easy way to build an amp without having to figure out the front end part. Even so, I imagine it takes less space than a descrete front end. I may consider this instead of descrete LTP and VAS for a mini 4 channel amp. Does the chip require a heatsink at lower voltages? (+/-30V)
Good luck with your amp project, keep us posted! I just DL'd the datasheet for the LME49810. Seems like a nice easy way to build an amp without having to figure out the front end part. Even so, I imagine it takes less space than a descrete front end. I may consider this instead of descrete LTP and VAS for a mini 4 channel amp. Does the chip require a heatsink at lower voltages? (+/-30V)
EWorkshop1708 said:Thanks for the info on the LME49810!
No problem!
It'll probably be a while before the PCBs are stuffed and then tested. It'll probably then be forever till I actually completely finish it lol. I never did get around to making a case for my LM3886 parallel amp. I just got a job though, so I might be able to afford some tools and stuff now!
I think it's a good stepping stone from all-in-one IC amp to completely discrete amp to be honest
I don't believe it requires a heatsink at lower voltages. At the moment I have just cut up some scrap PCB as "heatsinks"!
An externally hosted image should be here but it was not working when we last tested it.
Sweet! This may be a good IC chip to use for SRMcGee's 100W/ch amp. That way it has the simplicity of a chip, with the power and drive of descrete. Try the MJE15034/15035 drivers and NJL4281/4302 outputs, they are awesome transistors!!!
I am familiar with descrete amp circuits, but if I can save space and need less resistors and such, I may just go the IC route. It claims to have good sound and low %THD. It may just do well and save MUCH board space for a 4x 60W amp. That would ensure all 4 amps could fit on one board.
I am familiar with descrete amp circuits, but if I can save space and need less resistors and such, I may just go the IC route. It claims to have good sound and low %THD. It may just do well and save MUCH board space for a 4x 60W amp. That would ensure all 4 amps could fit on one board.
SRMcGee said:
Two flavors -- parallel or bridged -- and this is an unashamed plug:
http://www.tech-diy.com/chipamps.htm
Well, you can bridge TDA7294's with a single 22k resistor pin 2 on one to pin 14 on the other. That orientation is really convenient, because its just two amplifiers side-by side. So, two ordinary amp boards can bridge pretty quickly, and that distance is very short. Given about 220uF onboard instead of 2200uF, that makes a nearly cold-running 100 watt bridge amp. No fans needed, but you'll need a good seperate power supply board.
It can run quite nicely on a 40v, 4 amper, center tap transformer, like the White Rogers for $38, if you need a lot of power, or you could use the Allied store brand 36v, 4 amper, center tap transformer which is about $28 (to run the amplifier cooler, or use 4ohm speakers). Either case, one transformer runs 2 chips (takes 2 transformers for 2 bridge amps--monobloc format).
There's three little caveats.
One, you need to change the output zobels to half value for the caps and double value for the resistors (because now there's two of them).
Two, its necessary to disconnect the mute and standby circuitry from one board and jumper from the trace that operates those two pins (9 and 10) to the identical trace on the other board (prevent sudden xmax by having the chips go on and off at exactly the same time).
Three, for power, you'll need, 3, super-short, identical length wires (-g+) to simultaneously feed power to both boards--with the power supply connected centerpoint. That's not any different from an optimal stereo pair, though.
Power supply board:
The easy way is to use a pair of 10uF, a pair of 10,000uF, and a pair of 100nF with a KBPC 25a rectifier. Hey, you can change those 10uF electro for some 4.7uF poly caps if you like. Each power supply runs 2 chips (takes 2 power supplies for 2 bridge amps).
Bridge Connection:
The NFB's are connected together by adding an additional resistor, probably 22k.
Identify the chip with the 22k resistor (bridges the two boards) on pin 14 (speaker output). That chip has the positive polary for signal in+ and speaker out+. Signal in+ connects to the input filter cap on this chip. Speaker+ connects to speaker out+.
Next,
Identify the chip with the 22k resistor (bridges the two boards) on pin 2 (inverting input). That chip has the negative polarity for signal in- and speaker out-. Signal in- connects to the input filter cap on this chip (backwards of normal). The connection for the negative terminal of your speaker is the speaker+ connection on this chip.
Tip, on this bridged amp, cables for speaker and signal won't connect to any negative terminals on the boards.
The above was written for 100w monoblocs (use two for stereo), because if its going to be that loud, I think it might as well be as dynamic as possible. It comes in pretty economical at 50 cents (or so) per watt.
I think that a bandaxall (bass-n-treble) preamp (with gain) would be a good addition because it can let you do component selection based on clarity rather than compromising clarity by obsessing over frequency response (during component selection). Hey, its a 100w per channel amp, so it may as well sound good.
Have fun!
It can run quite nicely on a 40v, 4 amper, center tap transformer, like the White Rogers for $38, if you need a lot of power, or you could use the Allied store brand 36v, 4 amper, center tap transformer which is about $28 (to run the amplifier cooler, or use 4ohm speakers). Either case, one transformer runs 2 chips (takes 2 transformers for 2 bridge amps--monobloc format).
There's three little caveats.
One, you need to change the output zobels to half value for the caps and double value for the resistors (because now there's two of them).
Two, its necessary to disconnect the mute and standby circuitry from one board and jumper from the trace that operates those two pins (9 and 10) to the identical trace on the other board (prevent sudden xmax by having the chips go on and off at exactly the same time).
Three, for power, you'll need, 3, super-short, identical length wires (-g+) to simultaneously feed power to both boards--with the power supply connected centerpoint. That's not any different from an optimal stereo pair, though.
Power supply board:
The easy way is to use a pair of 10uF, a pair of 10,000uF, and a pair of 100nF with a KBPC 25a rectifier. Hey, you can change those 10uF electro for some 4.7uF poly caps if you like. Each power supply runs 2 chips (takes 2 power supplies for 2 bridge amps).
Bridge Connection:
The NFB's are connected together by adding an additional resistor, probably 22k.
Identify the chip with the 22k resistor (bridges the two boards) on pin 14 (speaker output). That chip has the positive polary for signal in+ and speaker out+. Signal in+ connects to the input filter cap on this chip. Speaker+ connects to speaker out+.
Next,
Identify the chip with the 22k resistor (bridges the two boards) on pin 2 (inverting input). That chip has the negative polarity for signal in- and speaker out-. Signal in- connects to the input filter cap on this chip (backwards of normal). The connection for the negative terminal of your speaker is the speaker+ connection on this chip.
Tip, on this bridged amp, cables for speaker and signal won't connect to any negative terminals on the boards.
The above was written for 100w monoblocs (use two for stereo), because if its going to be that loud, I think it might as well be as dynamic as possible.
It comes in pretty economical at 50 cents (or so) per watt. I think that a bandaxall (bass-n-treble) preamp (with gain) would be a good addition because it can let you do component selection based on clarity rather than compromising clarity by obsessing over frequency response (during component selection). Hey, its a 100w per channel amp, so it may as well sound good.
Have fun!
EWorkshop1708 said:Thanks for the info on the LME49810!
I just DL'd the datasheet for the LME49810. Seems like a nice easy way to build an amp without having to figure out the front end part. Even so, I imagine it takes less space than a descrete front end. I may consider this instead of descrete LTP and VAS for a mini 4 channel amp. Does the chip require a heatsink at lower voltages? (+/-30V)
The footprint 110 x 60 mm of ProJet Power Module which uses LME49810 front-end driver is really small. So I can only recomend this chip for universal amp applications ranging from 100-1000W/4ohm with easy way of building it.
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.