OT - PSU question re; regulator voltage drop.

[MartyMart]

EDIT : schem for this PSU : http://fucanay.fischerworks.com/diy/FETBOY_PSU_Schem.gif

Just built this PSU to supply 48v dc / 24v dc / 0v using a Traffo with Primary of 230/0 v and dual
secondaries of 15v/0v.
It's for powering a studio pre-amp and for Phantom power also.

Secondaries are wired is series, rectified ( 4 x 1N4002 bridge ) then split / filtered for two rails.
Lower rail after LM317 is at 23.4v dc ( close )
Upper rail seems to have been "trippled" using three more forward biased 1N4002's, filtering between and
hitting a second separate LM317 for the 48v rail.
This is outputting 49.3 v dc ( also close, though can be as much as 53v dc here before the standard 220ohm and pair
of matched 6k8's for the phantom rail connection to pins 2&3 of the mic )

This is all working, no mains "shocks" etc, but the problem is that the dc voltage BEFORE the LM317 for 48v dc is
sitting at 96v and before the LM317 for the 24v rail, it's sitting at 56 v
This is a LOT of drop for the LM317's , current use for the mics is low, 10Ma typical per mic.
should I be worried about the "life" of the 317's ? ( specs say they can take a 40v drop max ? )
I have the 48v rail 317 with a big heat sink, but neither that nor the other got more than "warm" after 15 minutes
run-time.
I'll see if I can get the schem or a link up, the original called for the traffo to be a wired in series dual 12v, so my
wired in series dual 15v is already:  30x 1.414 = 42.42v ! doubling again is causing the problem I guess.
( bit green with rectification calculations  )
Caps are fine, several were over - rated (160v) as that's what I had around, none are under these higher dc voltages
on the board.
Insight most welcome !!
MM

[gez]

The absolute max input voltage quoted in data sheets is exactly that: absolute!

I didn't check your link (sorry), but how much current is being drawn from the 24V regulator?  There's a little trick you can do to make things 'safe' and drop some of that input voltage.  Outlined in this thread (scroll down):

http://www.diystompboxes.com/smfforum/index.php?topic=61546.0

[gez]

Upper rail seems to have been "trippled" using three more forward biased 1N4002's, filtering between and
hitting a second separate LM317 for the 48v rail.

How do you expect to provide 48V when, according to the LM317 data sheet, the max output range only extends to 37V??

Perhaps if the regulator's ground pin was tied to the output of the 24V regulator it might work.  I wonder about stability, though.  Plus, not all regulators like to sink current (would have to check the data sheet).

[MartyMart]

Hmm, I'm not quite understanding this particular setup.
I have approx 95v dc after D4 to the top LM317 and approx 65v dc before D4.
So removed & jumpered D4, thinking the voltage would be more like required.
Now 48v rail output is at 39v and the 24v rail output increased to 37v !!!
Perhaps I just need a 2w resistor after D4 to drop the voltage and some possible "heat" ....

[MartyMart]

Gez, thanks for the tip, if you have a look at the schem perhaps it will be clearer ( than it is for me ! )
There are several similar 48v dc supplies @ prodigypro forum, used for various pre-amps such
as the "Green pre" and API clones etc.
Many are using TO220 style LM317's for this purpose !
They are not getting hot and I have no load on them yet, perhaps I should "mimic" some load with a resistor
at the output.
I only need 20Ma for two mics ( they dont draw much ) and 300Ma for the preamp circuit.
Traffo's secondaries are 1.15VA each and are used in series.

[gez]

Many are using TO220 style LM317's for this purpose !
They are not getting hot and I have no load on them yet, perhaps I should "mimic" some load with a resistor
at the output.

Leave the load off until you have the basics sorted.  OK, forgot that the LM317 is 'floating', so it's the differential voltage between output and input that is the key thing. 

I only need 20Ma for two mics ( they dont draw much ) and 300Ma for the preamp circuit.

If you keep within the input-output voltage differential, then you shouldn't have any problems.  The resistor trick can drop some voltage for you.

By the look of things, that diode/cap network supplying the 48Vreg is supposed to be a voltage multiplier (probably a tripler).  The configuration is not one I'm familiar with.  Are you sure it's drawn up correctly?

Is the preamp powered from the 24V circuit? 

[R.G.]

The LM317 stuff cannot hold off much more than 40V reliably. Since the 317 itself floats, that's the maximum differential it can support. Even if a higher differential happens not to kill it, the extra voltage stress will shorten its life.

At power on, the output caps are 0V. So the entire incoming DC voltage appears across the 317.

I have not seen the examples at "prodigy pro" but I suspect they are using the self-cascode trick. Imagine that I want 100V out of a 200V raw DC supply. Impossible with the LM317, right? No, it can be done.

The LM317 floats, but can only hold off 40V. If you put a high voltage transistor between the LM317 and the raw DC voltage, and use a zener from the output of the 317 to the base of the transistor, you get an "emitter follower" which is always supplying the zener's worth of voltage more than the output voltage to the LM317. The 317 lets through enough current to make the output voltage be correct, as it only sees the 1.25V of its current reference. The zener sets up the follower which supplies just enough voltage to the LM317, and the follower holds off the excess voltage.

That CAN be made to work. There are a lot of details to prevent transients and startup/shutdown from killing the LM317.

The 317 can also be used as a purely floating current source feeding zener diodes to do the voltage regulation for voltages higher than 40V. This setup requires that the raw DC not be more than 40V above the output voltage.

Design of these floating 317 supplies is modestly tricky because the transients have to be controlled for reliable operation.

[MartyMart]

Thanks for the information guys - I'm learning !!
Gez - The preamps have actually got a supply of 24v in their rack, so I dont really need the 24v rail as yet.
The schem is a "hack" but was discussed and given the "OK" bt a number of prodigypro guru's.
Perhaps it's not the most "pro" solution, that top diode string has me very confused.
My 24v rail is suddenly 37v ? nearly blew a cap, but have now fitted a 63v 1000uf there.
I have a 2w 2k7 resistor before the top LM317 and that's helping the voltage drop some.
Perhaps I just need to get some 48v zeners and go that route !
MM.

[gez]

nearly blew a cap, but have now fitted a 63v 1000uf there.

When you shorted D4, you might have damaged C12 as it will have been reverse biased by a couple of diode drops (assuming that multiplier is right).

Gez - The preamps have actually got a supply of 24v in their rack, so I dont really need the 24v rail as yet.

The reason I asked is that multipliers can only source so much current before loading kicks in and the voltage starts to drop.  With a moderately heavy load, things might not have looked too bad at the input of the 48V regulator (had it been supplying a fair wodge of current).

[gez]

I have a 2w 2k7 resistor before the top LM317 and that's helping the voltage drop some.

Bear in mind that the amount of voltage dropped by the resistor is dependent on current.  The value of the resistor required is determined by the max current expeted to be supplied.  Do some calculations using Ohm's law based on the max current estimates you gave, then try your loads.

[MartyMart]

Gez, wanted to say a big "Thanks mate" for setting me on the right path with this one
I studied a few other setups and realized that voltage "trippling" was going on here, so ditched a pair
of the 1N4002's on the top "rail" and associated caps, ended up with 54v pre- 317 and regs down to
49.5v ... PERFECT for feeding the matched 6k8 r's to the mic socket for phantom psu
Lower rail just needed a large 2w 1k2 for an output of 24.5v
It's super clean/quiet and not even getting warm having been run for 16 hrs !!

marty.

[gez]