I am starting new clone with one 4780 per channel in parallel. I kinda have problems to find 22V RMS voltage output toroids here in Benelux. Conrad has only 18 and 30V for 400VA and 500VA toroids. for 100W RMS per channel, calculation shows 22V at 4 ohm. Should I go for 18V, 30V or something (somewhere) else?
After rethinking it a bit, I guess 18V is out of the question. It seems that both versions have very similar output current values, so voltage ceiling might be hit easily with 18V one. On the 30V side, 6.6 Amps RMS current capacity of 500VA 30V trafo should run down to 2.3 ohms (approx) just fine at 100W peak (not sure what are the losses exactly inside those transformers, but that should affect mostly RMS voltage drop, and 30V is lot more to drop from than 18V).
Need to check 4780 output specs too.
Need to check 4780 output specs too.
Hi,
if you want 100W into 4r0 from a parallel pair, then each chip thinks it is driving 50W into 8r0. That is equivalent to 28.3Vpk and 3.5Apk into the resistive load.
Allowing about 4.1V loss through the amplifier indicates that 28.3+4.1=+-33.5Vdc is required from the PSU when delivering full power.
Now make a decision.
Do you design for continuous (say for 1 or 2 seconds) maximum power into this load or will you accept a short term power peak into that load.
For continuous power you must allow for sag in the PSU voltage as the smoothing capacitors discharge to the fully loaded condition. This tends to give more controlled and deeper bass.
Alternatively, the peak power rating can be met from the fully charged capacitors but as the duration of the signal is increased the PSU drops to pass less continuous power to the load. This reportedly tends to favour midrange and treble frequencies.
The sag could vary between 2V for a very high capacity PSU to 6V for a cheap low capacity PSU.
I would suggest you design for +-[33.5+2V] for your first guess at quiescent load. After allowing for rectifier loss and Vdrop supplying quiescent current and for 5% transformer regulation this 35.5Vdc can be supplied by a 23+23Vac, 300VA toroid for powering 100+100W. Use at least +-20mF for the two chipamps if you want deep bass, or twin PSU each with >=+-10mF/chipamp. Personally I would fit +-15 to 20mF for every chipamp.
if you want 100W into 4r0 from a parallel pair, then each chip thinks it is driving 50W into 8r0. That is equivalent to 28.3Vpk and 3.5Apk into the resistive load.
Allowing about 4.1V loss through the amplifier indicates that 28.3+4.1=+-33.5Vdc is required from the PSU when delivering full power.
Now make a decision.
Do you design for continuous (say for 1 or 2 seconds) maximum power into this load or will you accept a short term power peak into that load.
For continuous power you must allow for sag in the PSU voltage as the smoothing capacitors discharge to the fully loaded condition. This tends to give more controlled and deeper bass.
Alternatively, the peak power rating can be met from the fully charged capacitors but as the duration of the signal is increased the PSU drops to pass less continuous power to the load. This reportedly tends to favour midrange and treble frequencies.
The sag could vary between 2V for a very high capacity PSU to 6V for a cheap low capacity PSU.
I would suggest you design for +-[33.5+2V] for your first guess at quiescent load. After allowing for rectifier loss and Vdrop supplying quiescent current and for 5% transformer regulation this 35.5Vdc can be supplied by a 23+23Vac, 300VA toroid for powering 100+100W. Use at least +-20mF for the two chipamps if you want deep bass, or twin PSU each with >=+-10mF/chipamp. Personally I would fit +-15 to 20mF for every chipamp.
Or you could parallel some LM1875's with the 18v. Sure that's a bit warm, but it'll rock the house quite nicely. 
They're about 2db behind the big chip, but with firmer driver command, they'll knock the grills off the speakers a bit more quickly.
Sort of a "personal taste" difference between dynamic power or average power.
They're about 2db behind the big chip, but with firmer driver command, they'll knock the grills off the speakers a bit more quickly. Sort of a "personal taste" difference between dynamic power or average power.
30+30vAC (60vct rated transformer) = 42+42vdc and that's too hot. This will set off the spike protection.
22+22vAC (44vct rated transformer) = 32+32vdc and that's pretty good (somewhere near optimal).
18+18vAC (36vct rated transformer) = 27+27vdc and that's going to "less hot," but make less amplifier power.
In decibels, the end result is probably little to none noticable difference between any of the three options. Remember, it takes 2x the amplifier power to make each +3db to the speakers, and it takes 10x the amplifier power for every 2x difference to the ear.
Bring in speakers and ears, then the 18v secondaries look good.
Hey don't worry, it'll still get the heatsinks plenty hot, especially if its one of the "TF" chips.
22+22vAC (44vct rated transformer) = 32+32vdc and that's pretty good (somewhere near optimal).
18+18vAC (36vct rated transformer) = 27+27vdc and that's going to "less hot," but make less amplifier power.
In decibels, the end result is probably little to none noticable difference between any of the three options. Remember, it takes 2x the amplifier power to make each +3db to the speakers, and it takes 10x the amplifier power for every 2x difference to the ear.
Bring in speakers and ears, then the 18v secondaries look good.
Hey don't worry, it'll still get the heatsinks plenty hot, especially if its one of the "TF" chips.
Locally, I can find only 500VA 2X39... They do have 300VA 2x21 though.
Hm, thinking out loud on 2x18V - why not use bridge mode instead? If I am reading specs of 4780, max current peak seems to be 7A (10ms). What is this in real life, what is the lowest impedance it could drive bridged? Even with 2X24V in parallel mode, I won't be able to reach 100W RMS per channel at 8ohm.
I would like to have versatile amp that can drive anything from 2-8 ohm at 100W RMS, but it will be much more used around 8ohm than below 4.
Hm, thinking out loud on 2x18V - why not use bridge mode instead? If I am reading specs of 4780, max current peak seems to be 7A (10ms). What is this in real life, what is the lowest impedance it could drive bridged? Even with 2X24V in parallel mode, I won't be able to reach 100W RMS per channel at 8ohm.
I would like to have versatile amp that can drive anything from 2-8 ohm at 100W RMS, but it will be much more used around 8ohm than below 4.
Ok, even with 2X24V AC on the secondary, for 8 ohm load, I won't be able to get 100W RMS per channel in parallel configuration, right? Even if we assume lossless path in between, it would be only around 70W RMS, best case scenario. Am I missing something?
Bridged is another story, it swings double of the PS voltage on the output.
Bridged is another story, it swings double of the PS voltage on the output.
My apologies that I haven't explored the topic more thoroughly. However, since Peter of Audiosector has decided to use Parallel, then there's probably a good reason. Whatever the reason, it is a good baseline. That is a great place to start. A real-life baseline is necessary in order to acquire references and assistance, so use parallel.
The 300VA 2x21 will get the amplifier (any configuration) somewhat hot. Base your heatsink size selection upon your expectation of the amplifier's performance (demand < heatsink), then it should perform well no matter of 4, 6, or 8 ohms. But, that transformer will push the limits sufficiently.
I too, am out of resources to explore those points farther, so I believe it is time to proceed with the next steps.
The next step is to concentrate on efficiency, such as amplifying only desired signals. Use an established baseline.
The step after, is voicing, to highlight desired signals, which is also a form of efficiency, although not strictly mathmatical or strict adherance to a baseline--its the "more pleasant" nearby values, also called tweaking.
However, its a good time to re-evaluate amplifier selection. If you are looking for a sustained 100 watts audio signal then you actually need a 1000 watt (one thousand watt) amplifier to do it.
Why? At endpoint, the 100 watt amplifier produces only 6db more than a 33 watt amplifier.
Why not? Most speakers lose fidelity at 55 watts, and your 4780 project will have more than enough power to cause this, when desired.
So, hey, go build your amp.
Edit: Check out what can be done with Harbeth Monitor 40, if one builds that style using Tang Band 6-1/2" (92db) as the central wideband driver (midrange), plus adds an efficient large woofer (95db) and small tweeter (93db) to help it. There, using efficiency, is another way for excellent power.
The 300VA 2x21 will get the amplifier (any configuration) somewhat hot. Base your heatsink size selection upon your expectation of the amplifier's performance (demand < heatsink), then it should perform well no matter of 4, 6, or 8 ohms. But, that transformer will push the limits sufficiently.
I too, am out of resources to explore those points farther, so I believe it is time to proceed with the next steps.
The next step is to concentrate on efficiency, such as amplifying only desired signals. Use an established baseline.
The step after, is voicing, to highlight desired signals, which is also a form of efficiency, although not strictly mathmatical or strict adherance to a baseline--its the "more pleasant" nearby values, also called tweaking.
However, its a good time to re-evaluate amplifier selection. If you are looking for a sustained 100 watts audio signal then you actually need a 1000 watt (one thousand watt) amplifier to do it.
Why? At endpoint, the 100 watt amplifier produces only 6db more than a 33 watt amplifier.
Why not? Most speakers lose fidelity at 55 watts, and your 4780 project will have more than enough power to cause this, when desired.
So, hey, go build your amp.
Edit: Check out what can be done with Harbeth Monitor 40, if one builds that style using Tang Band 6-1/2" (92db) as the central wideband driver (midrange), plus adds an efficient large woofer (95db) and small tweeter (93db) to help it. There, using efficiency, is another way for excellent power.
With all due respect to Peter (I am using his great pcbs already, and my first amp is parallel), National specs pdf for LM3886 kinda suggests bridged config for 8 ohm load. This is what Linkwitz is using too in his Plutos woofers, so it's actually not poor ignorant me who is pushing for it. Build is already on the way, I am just trying to understand on which toroid to spend the money. Moreover, Elliott from ESP suggests 400VA for 2x100W and I tend to agree, just cannot find it here (yet).
zdr said:
I've often preferred the sound of bridged, although I can't speak for that particular chip, because I don't know.
The transformer that you found 300va and 21+21vAC, sounds like plenty to run the project. Seven ampers! WOW! Some nice fast big caps on the power supply board will ensure success if there's any doubts. That is a bit much voltage, but since it stands in the centerpoint of what you want and what is wise, then just plan on buying some nice size heatsinks.
I'd love to see that amp bridged and hear how it does for you. Its just that I couldn't advise you to bridge it because I've never personally tested it that way. So, that's all there was to that. - Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.