Hi group,
Ive just aquired some very inefficient 5ohm speakers and would like to build a simple 100watt amp. I hooked up the biggest amp I have, a MC modded Hafler and the mids and highs were horrid compared to my 3875 IGC. But of course, I have no volumn or bass.
After checking all the Bridgeclone threads, Im still confused since there never appreared to be a conclusive design. I checked the National apps and both the 100watt bridged and 100watt parallel designs look very simple. The bridged looks to have lower distortion but the parallel will drive a 4ohm load easily.
http://www.national.com/an/AN/AN-1192.pdf
What should I attempt? Im not comfortable with too much engineering but can assemble anything. Ive got 6 3386s on the way.
Insights please, and any comments on the sound of either.
amt
Ive just aquired some very inefficient 5ohm speakers and would like to build a simple 100watt amp. I hooked up the biggest amp I have, a MC modded Hafler and the mids and highs were horrid compared to my 3875 IGC. But of course, I have no volumn or bass.
After checking all the Bridgeclone threads, Im still confused since there never appreared to be a conclusive design. I checked the National apps and both the 100watt bridged and 100watt parallel designs look very simple. The bridged looks to have lower distortion but the parallel will drive a 4ohm load easily.
http://www.national.com/an/AN/AN-1192.pdf
What should I attempt? Im not comfortable with too much engineering but can assemble anything. Ive got 6 3386s on the way.
Insights please, and any comments on the sound of either.
amt
If it's a four ohm load, go with the simple paralel set up, as if you go to a birisged configuration, you will have to parallel at least four, probably six devices for each channel (if you only used two per speeker for bridging, the thermal protection would almost certinaly cut in at quite low power levels and sound nasty).
So a parallel configuration is manditory for a 4 0hm load regardless of whether its the beginning of a bridged circuit or just a standalone. Upon re-examination of the National app, it appears that the biggest difficulty with the parallel circuit is that the feedback resistors need to be matched or as close as possible. Never bought 0.5 tolerence resistors before. That should be fun.
thx
amt
thx
amt
Just to complicate things: If the speakers are two way, and you want to extend your DIY experiences, using the two ICs to bi-amp your speakers might be an alternative.
Hi, just to try and clear things up a little. If you were to run a single chip by itself, you would be fine with around 50W into an 8 ohm load, and the chip could happily drive a four ohm load to a power level of perhaps 60-75W. However, due to teh thermal limitations of the package, and things like internal resistance you are not able to get your desired 100W amp.
So, if you were to parallel two chips, each chip would only be supplying half the current to the speaker at any one time (this has the effect of each chip seeing the load as being twice the actual impedance, 16ohms for an eight ohm speaker, or eight for a four ohm speaker). So, you can have it so that each chip is supplying the same 50W of power you get into an 8 ohm load when you are in fact only driving four (which is absorbing all 100W).
The chips could drive a still lower impedance, however, the effects mention above about the limited increase in power will come into play, so you won't get double the power into a two ohm load even though connecting it to the chips won't damage them in any way, and they will be happy to give out over 100W to it.
The Bridge & Parallel option is another option for you here, although, it would probably be even more complex and you would end up needing alt least six, probable eight chips to drive each speaker which would be quite a big first project.
Andrew.
So, if you were to parallel two chips, each chip would only be supplying half the current to the speaker at any one time (this has the effect of each chip seeing the load as being twice the actual impedance, 16ohms for an eight ohm speaker, or eight for a four ohm speaker). So, you can have it so that each chip is supplying the same 50W of power you get into an 8 ohm load when you are in fact only driving four (which is absorbing all 100W).
The chips could drive a still lower impedance, however, the effects mention above about the limited increase in power will come into play, so you won't get double the power into a two ohm load even though connecting it to the chips won't damage them in any way, and they will be happy to give out over 100W to it.
The Bridge & Parallel option is another option for you here, although, it would probably be even more complex and you would end up needing alt least six, probable eight chips to drive each speaker which would be quite a big first project.
Andrew.
Good explaination. I think that even a dependable 75 watts would be sufficient since I will be eventually using a sub and active XO and be eliminating 80hz and below. This should lessen the demand for wattage in the speakers.
Is there a reason that I havent seen paralleling discussed and the focal seems to be on bridging? Is paralleling inherently inferiour or are low ohm loads uncommon?
I guess I will build one per NSCs' diagram unless someone has a successfull GC version I can use. I looked pretty closely at past thread and dont remember one but if it was discussed over several thread w/o a schematic, I may have overlooked it.
amt
Is there a reason that I havent seen paralleling discussed and the focal seems to be on bridging? Is paralleling inherently inferiour or are low ohm loads uncommon?
I guess I will build one per NSCs' diagram unless someone has a successfull GC version I can use. I looked pretty closely at past thread and dont remember one but if it was discussed over several thread w/o a schematic, I may have overlooked it.
amt
I looked over National's paper on their bridge/parallell configuration, and it looks pretty good. First, do your speakers have a 5 ohm impedence or do they have a 5 ohm DC resistance? If they have a 5 ohm imedence, you will probably want to use the parallel set-up. This will get you a little under 100 watts for a 5 ohm impedence. If your speaker has a 5 ohm DC resistance, it is probably an 8 ohm impedence and you will want to use the bridged layout to get the power you are looking for.
One thing I noticed between the two layouts is the differene in distortion figures. The bridged configuration has a much lower distortion due to cancellation of harmonics inherent to bridged designs. The parallel layout can deliver more current for a given heat dissipation, but doesn't have the distortion cancelling figures that the bridged configuration has. To get the best of both worlds, go with the bridged/paralleled configuration. The configuration isn't much more complicated than a bridged or paralleled. It just has more parts. This will give you more than enough power, and it will be able to drive a 4 ohm load just fine.
Cheers,
Zach
One thing I noticed between the two layouts is the differene in distortion figures. The bridged configuration has a much lower distortion due to cancellation of harmonics inherent to bridged designs. The parallel layout can deliver more current for a given heat dissipation, but doesn't have the distortion cancelling figures that the bridged configuration has. To get the best of both worlds, go with the bridged/paralleled configuration. The configuration isn't much more complicated than a bridged or paralleled. It just has more parts. This will give you more than enough power, and it will be able to drive a 4 ohm load just fine.
Cheers,
Zach
but which one?
I looked again at the NSC distortion plots and see that the parallel version is higher but it drops w/ output. If have the amps working at 40+ watts most of the time, will I notice the difference?
And there are three parallel/bridged schematics and Im not sure which is which. Fig 13 looks WAY too difficult w/o a PCB. Fig 14/15 looks doable P2P but doesnt appear to be 200watts. Are those variations of the 100 watt amp or merely examples of parts of Fig 13? Lastly, Fig. 17 looks like it could be handled P2P and may be the best way to go. Ive not made a PCB before so P2P would be convenient. I did see a board for sale from Europe at $45ea but since Ive not seen anyone complete one and report back, Im not sure thats a good bet yet.
amt
I looked again at the NSC distortion plots and see that the parallel version is higher but it drops w/ output. If have the amps working at 40+ watts most of the time, will I notice the difference?
And there are three parallel/bridged schematics and Im not sure which is which. Fig 13 looks WAY too difficult w/o a PCB. Fig 14/15 looks doable P2P but doesnt appear to be 200watts. Are those variations of the 100 watt amp or merely examples of parts of Fig 13? Lastly, Fig. 17 looks like it could be handled P2P and may be the best way to go. Ive not made a PCB before so P2P would be convenient. I did see a board for sale from Europe at $45ea but since Ive not seen anyone complete one and report back, Im not sure thats a good bet yet.
amt
I'm planning on building a 400W into 4 ohms subwoofer amp following the figure 13 scheme, although I think I'm going to need six chips. I chose the servo architecture insead of the "basic" precision resistior one, figure 17, becaure it has better low end response. When using precision resistors only, National claims that dc blocking caps are necessary and they cause a low freq rolloff. National omitted including a graph to illustrate this fact.
One the other hand, I have heard at least one claim the the servos are audible. So, if you are going to be using a seperate subwoofer, maybe the basic scheme would be good for you.
What I am wondering is what if any changes to the values listed figure 13 would be necessay to add two more chips.
Cheers.
One the other hand, I have heard at least one claim the the servos are audible. So, if you are going to be using a seperate subwoofer, maybe the basic scheme would be good for you.
What I am wondering is what if any changes to the values listed figure 13 would be necessay to add two more chips.
Cheers.
You shouldn't need blocking caps in the fig 17 schematic because the bridged architecture cancels out any DC that might be at the outputs of your op-amps. It's a very elegant solution, and is basically just four gainclones.........two inverting and two non-inverting. Also, there is a blocking cap at the input of the figure 17 schematic that will eliminate DC at the input. Just make this value at 0.47 uF and you will have an LF f3 point unter 10 Hz. This won't hurt your bass response, and will actually give the chip a little more headroom due to the fact that it won't have to produce subsonic frequencies (usually a good thing).
As far as using servos, I don't think they're necessary. They seem to take a little life away from the sound. They don't hurt the frequency response, but they just seem to take some of the "oomph" out. Just my opinion.
If you want to add two more chips, you shouldn't have to change any values. Just add one to each side of the balanced schematic. BTW, I don't think you'll be able to get 400 watts in to a four ohm load with this layout. The data sheets says the continuous power for the bridged/paralleled layout is 335 watts, with a peak of 450 watts. You can parallel as many more chips as you wnt, but this won't increase the output power. If you want to get more power, you will have to go down to a 2 ohm load, and double the number of chips in the bridged/paralleled layout. That should give you somewhere around 450 watts continuous in to 2 ohms.
Cheers,
Zach
As far as using servos, I don't think they're necessary. They seem to take a little life away from the sound. They don't hurt the frequency response, but they just seem to take some of the "oomph" out. Just my opinion.
If you want to add two more chips, you shouldn't have to change any values. Just add one to each side of the balanced schematic. BTW, I don't think you'll be able to get 400 watts in to a four ohm load with this layout. The data sheets says the continuous power for the bridged/paralleled layout is 335 watts, with a peak of 450 watts. You can parallel as many more chips as you wnt, but this won't increase the output power. If you want to get more power, you will have to go down to a 2 ohm load, and double the number of chips in the bridged/paralleled layout. That should give you somewhere around 450 watts continuous in to 2 ohms.
Cheers,
Zach
Hmm, I think I'm beginning to see the light. I could use six chips into 2 Ohms and lower the rail voltage a little to limit the output. I like to avoid using eight chips if possible.
Using six chips might not get you the current in to 2 ohms that your speakers will want to draw. In other words, your amp may have trouble reproducing LF transients properly. The extra 2 chips don't require that much more circuitry, and you will be running them cooler and with more headroom. I'm not saying that eight chips are required, but it will probably sound better than six chips.
Cheers,
Zach
Cheers,
Zach
That's what I'm tring to figure out. It would mean about 4.6 amps per chip. Cooling should be OK, I plan to use fans.
I've also been playing with the idea of liquid cooling, oil or water, and prehaps peltiers 🙂
Cheers,
Cyber
I've also been playing with the idea of liquid cooling, oil or water, and prehaps peltiers 🙂
Cheers,
Cyber
Put those chips on a wind tunnel heat sink, run a fan, and you should be just fine with the six chips. Sounds like a fun project. Too bad about the fan though. They're just so noisy.😉
Cheers,
Zach
Cheers,
Zach
Thanks Usekgb,
BTW, whats the U5 chip actually doing. It is a JFET input opamp that follows the signal input.
And lastly, I just read that my Maggies is a "purely resistive" 5ohm speaker. Does this have any bearing on the rail voltage I can use. I dont really know what this means in amptalk.
amt
BTW, whats the U5 chip actually doing. It is a JFET input opamp that follows the signal input.
And lastly, I just read that my Maggies is a "purely resistive" 5ohm speaker. Does this have any bearing on the rail voltage I can use. I dont really know what this means in amptalk.
amt
U5 is just an input buffer. It looks like it is there to keep the input impedence of the amplifier at 47kHz. Without it, the impedence of the four amplifiers could get too low for a source to drive them properly.
As far as you having a purely resisive speaker, that doesn't sound quite right. The impedence of any driver will change with the frequency it is reproducing. If this is not the case, I would be very interested in seeing how they were able to solve this problem. For rail voltage, just use what is stated in the data sheets. I'm not even exactly sure what rail voltage I have on my current GC that I just threw together. I made it out of salvaged parts and a toroid from a ProTools Control 24 console. The thing still sounds great!
Cheers,
Zach
As far as you having a purely resisive speaker, that doesn't sound quite right. The impedence of any driver will change with the frequency it is reproducing. If this is not the case, I would be very interested in seeing how they were able to solve this problem. For rail voltage, just use what is stated in the data sheets. I'm not even exactly sure what rail voltage I have on my current GC that I just threw together. I made it out of salvaged parts and a toroid from a ProTools Control 24 console. The thing still sounds great!
Cheers,
Zach
How precision do the resistors need to be?
It says in figure 17 that it need 21.5kohm and 20.5kohm resistors..
what happens if they arent quite that...?
It says in figure 17 that it need 21.5kohm and 20.5kohm resistors..
what happens if they arent quite that...?
- Status
- Not open for further replies.