advertisement


LM117/317 datasheet tracking pre-regulator

No; with approx 3v across it, it will need to dissipate up to 9w.

A TO-220 package without heatsink is limited to ~1w dissipation at most for reliability...
 
Then it seems that the last regulator would dissipate more heat than the first? Anyway, being about 5-6V across both and 3 on last then both would need heat sinking. I'm trying to make a pcb for myself.
My schematic is almost identical in function the the one in the link. I have an optional snubber on the output and a load resistor to keep the regulator at about 50mA. And I've split up the input capacitors to 3 and have a film one as well there. But basically it's the same circuit. Ah, and I'm going to use two parallel resistors so I can better dial in the output voltage. I only need one rail of 13V to 16V.
Since it's not that easy to source a double heatsink for the regulators (they are installed in the middle of the board, then I will try a single version and see how it goes.
This is my layout so far with a single regulator:
http://imageshack.com/a/img543/6110/rw5j.png
Any ideas for improvements are welcomed :)
 
I forgot to mention that I'm going to use this supply for a TA2020 class D amp of about 2x14W max into 8 ohm and another tda7297 class AB that is 2x15W or so into 8 ohm.
 
Nice board. Practically I would omit C7, it doesn't do any good. The place for HF bypasses is at the input to the load - not remote back at the PSU. C5 can be as large as you like. Rload (330ohm) will need to be 2 or 3W rated part - 50mA at 16v is 0.8w and you want long term reliability - it is going to run pretty hot. R28 at 1M isn't doing anything and isn't necessary for 'dial-in'. You really don't need to worry about exact DC voltage for anything (even logic only recommends +/-5%); it's a regulators AC performance that sets it apart.

I forgot to mention that I'm going to use this supply for a TA2020 class D amp of about 2x14W max into 8 ohm and another tda7297 class AB that is 2x15W or so into 8 ohm.
Click to expand...

That wil make the heatsinking rather less onerous because the peak:mean of music is reasonable, 6-10dB for even the most compressd music. The practical long term average output power of your amps is going to be down around the 1w mark at the most (that's really loud into even modest speakers!). So the reg will not be supplying 3A all the time, nothing like in fact - back of an envelope for 1w out x 4 says about 1.4A.

Note if you were driving 4 identical channels of 8 ohms then your nominal 3A output allows for about 4.5w peak into each speaker before your PSU runs into current-limiting. If you are using these amps for an active speaker, then all should be fine because all the real power will be demanded by the bass amp, and the treble amp will be practically idling all the time :)
 
You are right about C7 :)
C5 I will experiment with, I put a footprint that will allow me to try different types/sizes.
I updated the footprint of the RLoad to 3W, I didn't think about the uprating.
The one meg resistor is noted like that to represent a higher value. I got my 13.5V using two parallel resistors, I can't remember exactly the values. That I could keep just in case, I can just leave it unconnected if I don't need it.
I'll be keeping this simple version without the pre-regulator.
Thank you for the tips!
 
Bob, I'd like to get a PCB made for powering my Class-D amp board so, I'm wondering if you are OK with sharing the PCB files for what you built? If not, that's fine too. My email address is: keith(dot)correa(at)gmail(dot)com
 
I've just built the supposedly final version of Acoustica's Tracking preregulator
on Mdxs site
I've tested it with 17.5 volts input and no load and am getting 4.55 volts.
Just tried an 820r soldered underneath board in parallel with r3 but the voltage went
down to 4.47. Can anyone help. Cannot insert circuit or URL as it says my post
is SPAM like.
 
Here on pfm you need to host your own images, and post links inlne with your post.

But do feel free to email your images/schematic whatever to me at [email protected] and I'll happily take a look : )


ETA - domain-name spelling updated - thanks, Mike!
 
Last edited:
Problem solved, I initially designed circuit with regulators staggered facing each other, but as it still would have been awkward to fasten them to heatsinks I turned the one clockwise 90 degrees but turned the connections anti-clockwise.
 
You must log in or register to reply here.


advertisement


Back
Top