advertisement


LM338K equivalent?

Yes that's how I envisaged you wanted to do it and yes it will work with LT1083. I wouldn't do it like that myself but maybe you should ignore that:rolleyes: It will work fine... something just doesn't appeal about it to me.

I've no idea what you mean about the CRC or the resistors.
 
Input output differential voltage is all that matters Alan and this is a power dissipation issue hence different packages being different.
 
Thanks again chaps,

0.68A on each output device will give roughly 1.4A per reg @12V.

LT1083 spec shows no issues at 15V /5A and up to about 110degC!

My traffo output voltage options are 15.56V DC or 12.73.

I might actually try these regs with a pair of std rectifiers first, just to lose 1.2V. Using the 15.56V traffo I'll have to dissipate 2.16V at the reg with the std rectifier, or 3.36V with the active item. If I don't get the hoped for /advertised 200uV of rail noise then I'll try an additional 33,000uF cap and maybe a 1R1 resistor.

I make that either 4.7w or 3w at the reg

I can't see them getting down to the 200uV noise level, but we shall see
 
Last edited:
Chaps,

I've finally got aroud to installing these regs..

I do want to run a CRC between each of them and the rectifiers, but I'm worried about the flow of current on the negative rail.

Does the negative amp/ rail need the 'R' of the CRC connected to the negative terminal of the cap?..

... given that it will be using a positive regular 'run backwards'. It looks to me like the current would then flow through the adjustment pin of the reg. Can't be right, or can it? I ask, as right now I have a 1/2 watt 1k trimmer resistor on there

Many thanks for all the help
 
No, you build them identical as per fig 3 of that link. If using C-R-C then the R goes in the positive line before each of the regs, so you might go 15,000uF>1R>15,000uF>LM338 input terminal. You then 'stack' them from pos rail of one to neg rail of other at the outputs, as shown in that sketch.
HTH
 
No, you build them identical as per fig 3 of that link. If using C-R-C then the R goes in the positive line before each of the regs, so you might go 15,000uF>1R>15,000uF>LM338 input terminal. You then 'stack' them from pos rail of one to neg rail of other at the outputs, as shown in that sketch.
HTH
I am going to use CRC. R going on the positive side of both.

Both sides will be built identical, thanks for the help
 
& for clarity - each LM338 section used like that must have its own independent secondary winding, and own bridge rectifier, as illustrated - else.. things will go pop.


(posted as much for anyone who finds this thread in future, more than you Dan!)
 
& for clarity - each LM338 section used like that must have its own independent secondary winding, and own bridge rectifier, as illustrated - else.. things will go pop.


(posted as much for anyone who finds this thread in future, more than you Dan!)
Thanks Martin, Some more points of note:

The LT1083 needs 100R across its adjustment pin and 860R to nevative to set it to 12V (the amp that I linked to up thread appears to be set to 18V!!).

A low ESR bypass cap of at least 25uF must be used on the adjustment pin, or the fantastic ripple eating performance will be reduced to nothing better than average (another mistake in the amp that I linked to).

The LT1083 must have 3V across it to get full performance. There are 3 curves on the spec sheet.. 3V, 2V and dropout (1.2 - 1.5V). You want 3V. Any more and you just generate extra heat with no further performance improvement.

The difference in ripple between a std. rectifier and an active item is 200mV vs ~112mV with no cap. The response from the active item also has no high freq noise on it. Use active rectifiers.
 
Last edited:
Good notes on simple but solid implementation, thanks for adding those.

Not played with active rectifiers... (I usu. freehand a dab of CRC or CLC to the same end for small-signal/sensitive stuff.)
 
A pair of active reg boards up front will seriously reduce that ripple. I've seen quite startling results playing with them.
 
Chaps, some advice would be appreciated.

Setup is as follows:

Active SMT regs each with a single 1000uF cap followed by a 0R5 25w resistor into a 33,000uF cap (the 1000uF cap is a bodge until the 2nd pair of 33,000uF caps arrives)

Regs have 33uF polyprop bypass on the adjust pin and a local 470uF blackgate F on the output legs.

Amp boards have 220uF Silmic 2's on each rail.

The results I'm getting are a fantastic <50uV on the +ve rail, but 5mV on the negative! Is this to be expected given the way the negative reg is being run? I've another pair of 33,000uF caps on order for the 1st part of my CRC, perhaps that's all that's needed. I can also try a 47uF tant or MLCC on the outputs and on the adjust pins, and perhaps even better Mosfets in the regs

 
Last edited:
The active reg MOSFETs that I'm using have an RDSON of 3.9mOhm. I've realised that I could get items with 1mOhm!

I think that might chop another 25mV off the 110-112mV that I'm getting at the regs right now (without a cap)
 
The results I'm getting are a fantastic <50uV on the +ve rail, but 5mV on the negative! Is this to be expected given the way the negative reg is being run?

If the regs are set-up the same way (and the bypass and output caps all have secure solder joints) - I wonder if the difference in output ripple is down to a phasing of the two secondary windings.
If its easy to do so, I'd reverse the ac inputs to the active bridge on the poorer measuring reg, and see what if anything happens to both rail measurements.
 
If the regs are set-up the same way (and the bypass and output caps all have secure solder joints) - I wonder if the difference in output ripple is down to a phasing of the two secondary windings.
If its easy to do so, I'd reverse the ac inputs to the active bridge on the poorer measuring reg, and see what if anything happens to both rail measurements.
Will involve soldering, but I'll try it and report back
 
Still not as good as I would like on the neg rail, but I shouldn't complain, given that we just moved the noise an entire decimal place.

Massive thanks and kudos to Martin C!

Listening to it now. Immediate impression is that the amp is not quite as loud.. That classic illusion you get when the noise floor drops away.

:)
 


advertisement


Back
Top