900W LM4780 Gainclone

[rgrayton]

Thank you to all who have helped me over the years. It has been so long since I started this project that I cannot even recall all of you, but thanks.

Motivation: 8 Channels for SoundEasy Speaker Design DF
Design Goals: Drive 4ohms ea., re-usable for HT.
Specs:
8x115W/ch rms into 4ohms
2x800VA toroid
172000uF total

Being my first amp, I learned quite a bit!

-Bob

[sivan_and]

Congrats..first i thought it was PMPO....

[Groundloops]

Bigger picture, please!

[rgrayton]

how do you do that? they kept making me squeeze the thing down to nothing.

[xaudiox]

upload it on Photobucket or similar site.. and post the link here.

[rgrayton]

let me know how this goes -- never did it and don't know if I set permissions and have link correct. -bob

http://s576.photobucket.com/albums/s...ayton/Gargan8/

[xaudiox]

Congrats Bob.

[Groundloops]

Really nice! No hum, no noise? How did you hook up rectifier boards - 4 per transformer? What is that on top of the transformers? Already familiar with SoundEasy?

Please do a little description of your decision process. THX!

[rgrayton]

Yes, I am familiar w/SE, and now I should be able to perform digital filtering – you need an amp per driver, so a 4-way spkr will require 8 ch for stereo.

Yes, Rectifier boards are 4 per toriod. On top of them is a 30ohm R to slow the I surge on pwr up – not needed with my prototype (2ch, one toroid). When I added the second toroid (or maybe that was second w/added caps), I started flipping the breaker. I did not want any more design than required, so I opted for the less elegant solution of simple slowing the current thru the input R. The relay shorts this R, once the surge is over and the R is out of the picture. There are other (better?) solutions out there, but this thing was dragging out and I needed to get it done.
I did have a hum, before I went to a star ground. Now, all 8 pcb have pretty much the same amount of copper between them and the star. I don’t think I am overstating myself when I say it was as much of a mechanical challenge as electrical one – fitting all that stuff in a single box and not looking like a rats nest.. I was too lazy to make 2 separate boxes. I really could not see any other solution than a (actually 2) huge heat sink on the box rear. Once that is defined, basically that sets the size of the box, at least in height. The toroids and caps set the size in w and d. My proto only would output about 60W/ch into 4 ohms – that was with the 2x1500uF on the pcb for the chipamp. After some direction from the board, it became apparent I needed to add much more, if I was looking for anything close to the 120W rating of the LM4780, so I added 2x10K per ch. What you wind up with is a nice power supply with a couple chips hanging off. Now, I measure about 21.5Vrms across an output load on each ch, driven w/1KHz signal. -bob

[pjpoes]

I don't believe that the capacitance is doing anything to increase the amplifiers output. It's really just a filter. It's the transformer which should be offering the continuous voltage and current necessary to drive the amps to their peak output. The capacitor filter directly after the rectifier does offer a voltage increase, but that voltage increase is limited to no more than 1.5 volts regardless of how much capacitance there is. The only way I could see the capacitors increasing output would be over sustained peaks which were previously draining the onboard caps before you took the measurement, and thus lowering the voltage.

None the less I have always agreed that the onboard capacitance wasn't even close to enough for those, and was never able to get my amplifiers to sound right with just the 1500uf's onboard. I personally changed mine out with 100uf film caps which have a much lower esr mixed with 1000uf panasonic fc's (I think thats what I did, been a while), all hooked to a CLC power supply I made from some extra parts I had from my big amp project. This uses 48,000 uf's of Hitachi brand capacitors per rail, then 11mh of inductance, followed by 96,000 uf's of capacitance. While others have argued that their is no scientific reason why such a quiet power supply is necassary, given the amps ripple rejection, besides hearing a difference in the noise floor, I can measure a difference in the noise floor, as well as distortion levels at very lower power (<1 watt), indicating reduced noise from the power supply in the output signal.