48v slow regulator question...

Hi folks,

been planning on using a LM317 based slow-regulator as in the datasheets for a 48v supply. However I've found this one in another post which Kev has at the DIY Factory.



This one is fed with AC, rectifried etc. Could this be adapted to be work with an already existing 48vDC regulated source? Using just the soft-switch part after C2 and feeding R4 to gound and R3 to 48v?

I don't quite get it.

Cheers Tom

You could do it that way, but I wouldn't do it that way. Why not just make the 317-based regulator ramp up slowly by itself? Adding the extra stuff just makes the regulation worse.

What you need is a means of making the voltage-setting R of the 317 circuit look small at turn-on, then effectively ramp up to its 48-volt programming value. This could be done by pulling current out of the ref pin to ground initially, then gradually ramping the magnitude down. Or, you could use an FET as a variable resistor. If you have a handy negative voltage around a JFET could be used; if not, a DMOS part, which only turns on with positive gate-source voltage (the JFET will be on normally and requires a negative gate-source V to turn off).

In either case the current goes away or the FET open-circuits at the end of rampup, so the regulator's performance is uncompromised. PM me with an email addy and I'll send you a picture of what I'm talking about.

Thanks guys....

Brad PM'd you.

Cheers Tom

This is the slow start zener regulated PSU from the Wellborn labs site. I have not built it for 48v, but I have for a 250v valve supply & it works very well. Takes about 15-20sec for the voltage to reach the regulated value.

The ramp time is set by the 5Meg/1uF res/cap combo.

Simple but effective.



The original is here:

http://www.welbornelabs.com/ps3.htm

Peter

Hi
Strictly speaking the input voltage is above that which a 317 will stand unless it is the HV version and even that is getting close. There is a '583 (I forget the number) which can handle 125 volts and 700mA. Same pinout as a 317. You could add the extra transistor for slow switch on from the LM317 datasheets, uprating the transistor voltage spec if necessary.
Matt S

[quote author="Matt Syson"]Hi
Strictly speaking the input voltage is above that which a 317 will stand unless it is the HV version and even that is getting close. There is a '583 (I forget the number) which can handle 125 volts and 700mA. Same pinout as a 317. You could add the extra transistor for slow switch on from the LM317 datasheets, uprating the transistor voltage spec if necessary.
Matt S[/quote]

I believe you are thinking of the TL783.

You are right---there is a slow turnon circuit using an emitter-follower in the TI databook notes for the 317. It's not quite what I conceived but it looks like it works fine and "gets out of the way" after turnon for normal operation. I will still develop the alternatives and transmit to the requestors (Tom and Samuel so far).

Actually, just a big-enough cap from the ADJ pin to ground (and, especially in this case, include the protect diode from ADJ to OUT) would also do the job. The other app note circuit with the transistor simply buffers a smaller cap so that it need not be so large.

Do a delay mesurement on your existing 48 supply if you have the equipment.

I am beginning to think that with the massive filteing everybody uses nowdays with these cool caps that there is a ton of delay aready built into a phantom supply.
You turn on the switch and that transformer has to charge up three to five caps with resistor in between.

Power down, same thing. Caps take long time to bleed...

Even with a delay supply, your forget its on, plug the mic in anyway, and the point is mute.
.
.
. And so is your custom Marik Z1 Ribbon! :grin:

Check Ouncer p/n's at surplus whenever you get chance and thanks!
0-12, 0-14?

Hi
Thanks Bcarso, I use the TL783 but just couldn't remember it's number.
I don't know why Tom wants a slow turn on / off usually none or perhaps a couple of hundred ohms in series and about 47uF cap is enough to take the edge off switching. It also depends on how much current and voltage the mics want. Some are fussy at wanting 48 or even 50 but most seem to be happy with 'some volts' on them (9 - 52).
If tom wants to come back and fill us in with info. By the way I am in UK. TL783 is available from CPC in Preston among other places.
Matt S

I had heard there were supply issues for the 783 in anything other than SMD package, but that by popular demand TI was going to reissue the TO220.

I guess if some mics are really going to be zapped by a quick application of phantom power, then what one really wants is a transient current limiter on a given connector, one that senses a change in demand and initially limits/ramps and then gets out of the way of the nominal 6.8k pullups. But I would question the design of any mic that couldn't handle hot-plugging, especially when there is (or damn well should be) 6.8k's in series with pins 2 and 3!

The circuit I just sent to Tom and Samuel is interesting in that it gives linear ramp as opposed to the typical settling time of an R-C lowpass, that is Vo [ 1 - exp(t/tau)]. Whether that would be any advantage or not remains to have pot shots taken at :razz:

Hi
Yes there should be 6K8's which limit the current to about 7mA.
The BBC often used mic lines with premanent phantom in TV studios for all mics.
Matt S

Yes Matt, that's true, but it caused all sorts of problems!
Phantom power on a jackfield is a dangerous thing, even with the 6K8 resistors; that's not going to protect your ribbon if one leg shorts to ground as you plug it in :?
The other horror lurking behind the innocence of a jackfield with phantom power is the possibility of accidentally plugging an OUTPUT from something expensive with solid state outputs, into the jackfield.... even for the fraction of a second while you shout 'oops!' It's got nothing to do with current.... there's a fair chance that the 48V will punch a hole in the output device and everything goes strangely quiet...with an underlying hum.....
No, keep phantom on the XLRs and make sure it goes off when you switch it off!

Also in case Matt's comments regarding the voltage rating of the 317 are not fully appreciated---the slow start of any variety (unless done via the input ahead of the regulator) will definitely be well outside the rating of the 317 input-output spec. So the 783 part is mandatory. Sorry I didn't label that in the schematics I emailed, and thanks to Samuel G. for reminding me.

Thanks everyone.

Got the email Brad - thanks, very interesting. I can host the PDF if thats OK with you so everyone here can have a gander at it?

I don't like loud pops associated with turning 48v on and off, and thought as its DIY - why not add a slow-ramp? I see what you guys are saying about the LM317 and its voltage rating - I didn't consider the fact that the in/out difference is huge when ramping.

So dilemma - looked for the TC783, Farnell, Rapid, RS and CPC (which seems to be a subdivision of Farnell) all out of stock....so maybe there is a supply issue. I'd rather not use that part if its going to be hard to get...

What to do, the idea of doing the ramp before the regulator is interesting if possible....or just sticking with the transistor soft-switch. How badly will that effect the regulation and how?

Maybe its not worth it?
I wonder how JoeM is doing it on his new baby animal pres, I see a big transistor on the boards...BC6xx

Cheers Tom

Hi
Moving R3 suggested above may not tirn on Q2 and 3 well enough to saturate them which you need if you want the supply to follow the regulator. Either swap the 2 sub circuits around , so ramp then regulate or fit a fet perhaps as the Q2/3 combination although you may well have a couple of volts drop. Alternatively put Q2/3 as a fet and use much bigger resistors 2M2 would probably be OK. You could reduce the cap value and you should fit a zener to protect the gate.
Phantom powered mics will thump and bang as they power up anyway to some extent so you could abandon the ramp thing anyway.
Alternatively you can use a 317 if you fit a 39V 5 watt zener across it to protect it in case of a shorted load. Put the ramp circuit before it and have a series resistor (fusible type) to that nothing really bad happens if you have a sustained short. Alternatively again look out for a discrete regulayor design by Lindsley Hood in the Everyday Electronics I think, Quite a good use of components, mostly cheap, the only expensive bit being a PNP darlington although with not much load it may not even need to be darlington.
Try also NIKKO electronics for TL783 (Surrey area although it is web based).
Matt S

Tom, I don't mind the circuit's being more accessible, but mediatechnology's comments are astute: the (load) regulation is less critical than the noise/hum. I would remark, though, that we get help, usually, by the fact that our phantom power is at least being used in a balanced fashion, so in many cases noise and hum are common-mode. And if the mic's signal-current demands are also really differential, then the net drain will be more-or-less constant.

But before you show my first effort let me send you a corrected version with a TL783 so at least people who skip over the cautionary comments don't sue me when their 317 parts pop.

I think what's really needed (or at least desirable) is a proper hybrid circuit that fulfills all the demands: exemplary intrinsic low noise, good line regulation, and adequate load regulation with (if desired) slow turnon behavior. This will not be achieved with a 783 alone---there is only so much gilding of the lily possible---although using the 783 as a pass element might not be too bad an idea.

As well, it might make sense to have a system that prevented application of full voltage to a given output until it was sensed that both XLR pins were being comparably loaded. The trouble with too-clever schemes like this: they have to be intelligent enough to reject high-level shifts arising from normal use, and not false-trigger. You don't want the engine of your airplane shutting down in midair because it detects that your oil might need changing soon.

Hi
It would be handy to hear from Tom what he has already and what the aim is.
If it is a unit that already has the 48 volt supply and you simply want to add a slow ramp circuit to it then either the Welborne or the 2 transistor addition would be OK excepting that you will lose a volt or two across the FET or darlington connected transistors.
Using PNP transistors may be beneficial as you could then power the base from the ground and get them to saturate.
If the project is to build a supply from scratch, is it for 1 channel or something like a mixer with many channels?
These questions have a lot of bearing on the 'best' way of doing it.
1 channel from a blank paper could use transformer, rectifier, cap ,resistor, big cap+zener (62volt) resistor, Cap + zener (48 volt). Since you would only be needing about 20mA total (max load plus some zener current) this may be easiest as it would ramp up itself.
Matt S

Gentlemen,

The point about moving R3 is probably going to work out ok, but my experience is that its unnecessary. If you don't like the zener reg, you can sub in a 3-term and use the remainder of the switch. If you want the zener reg to be as quiet as possible, only use one diode and you might want to hand pick it.

CJ's point about self ramping is right on, but I asked for a circuit that was load independant so that the timing is always consistant. Absolute regulation is not very important, otherwise I would have worked on a different scheme.

I challange you to build the ckt and critique it after you have tried it...

Peace!
Charlie

Hi
We have lots of ideas here but we still don't know Toms requirements, where the source is from or the load expected.
For a linear ramp how about the FET follower shown earlier preceeded by constant current source / sink to charge the capacitor.
Maybe instead of the FET use a high voltage power amp chip with a few ohms and a cap to reduce any noise on it's output.
Matt S

You could use a pic chip and a motorized pot if you really wanna get slick.
Or leave everything on like the big boys do.
But that uses juice which uses up my kayak water due to dam releases, so we don't want that.

CJ,

I think you just said exactly what I was thinking, except for the fact that I don't speak kayak-ian. Though I believe that I was able to translate.

Peace!
Charlie