one transformer driving two bridges

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

rafafredd

Well-known member
Joined
Jun 3, 2004
Messages
2,409
Location
Rio, Brazil
OK, Here is the schem:

JLM_OLD_PSU_moded.gif


I want to have + - 28v and +48v from a single 48vac power transformer (In the schem, the signal generator is the secondary of the power transormer, 48vAC, 62.4v peak).

So, is it OK for the secondary of the power transformer to drive both bridges as drawn, one for +-28v and the other for +48v?

Or should it be needed to couple D4 bridge and the secondary of the power transformer using two 220uf (or greater?) caps?

Also, I wanna say that I like this kind of PSU. It´s an old version of JLM, but he used a 40-0-40v transformer with one side driving both +V and +48v phantom, and the other driving the -V only. So, I wanted to use a cheaper and easier to get locally 24-0-24v transformer, so I did this mod. The idea is to use the 24-0-24 as a single 48v, with the center tap floating. It´s for a quad mic preamp I´m working on.

Thanks!
 
Thanks for taking a look, PRR. It seems that this is a place where spice lies... Maybe because I´m using a signal generator instead of a transformer??? But I don´t think so.

I apologize for insisting in it, but if this somehow is possible, it will make things so much easier for me... Otherwise, I´ll have to find a small 48v transformer just for phantom, and this is almost impossible, at least locally.

So, would this (coupling 470uF caps) solve the issue?

JLM_OLD_PSU_moded_2.gif


This is the result I´m getting with the coupling caps added:

simulations.gif


The thing is that it doesn´t differs too much from what I was getting without the caps. Are these all spice lies?
 
Spice/circuitmaker is hilariously screwed up for this sim.

I suspect the problem is in the zener models. Without the upper negative regulator connected it predicts about -4.4V on the negative unreg rail. Connect the negative regulator and suddenly we have -29V!!! Lots of yucks. I'd noticed other zener problems before but this one takes the cake. It's as if the zener has a built-in power source :razz:

As a general remark: if you want even just two voltages equally plus and minus from a single-winding no-center-tap trafo, you have to have somebody determining what the midpoint volatage is. This can be a stiff passive voltage divider or something active. In power amps the output load ground return can be to the middle of the two series bulk caps, and that along with a lighter divider that sets what is considered common for the amp will do the job.

You have a bit of this enforcement with the symmetrical arrangement of resistors and zeners. But look at what happens if the + and - load currents differ in magnitude much. Things start to move around.

Or as Yeats said: "Things fall apart. The center cannot hold." :green:
 
The caps unfortunately change nothing as they are technically short circuit at 50/60Hz anyway. Also the transformer centre tap will have to go to the 0v or once you connect a load any mismatch in current draw on each rail will pull the +/- rails out of balance.
 
Thanks one more time guys. Yes, I need a center tap, now I see it. There´s no easy way around it... Now I got it. But based on this:

kps1sch.gif


I think this will do exactly what I want:

JLM_OLD_PSU_moded_3.gif


The transformer is still a 24v-0-24v...

So, in a schematic like this I guess the 470uF caps are needed. If it works for Keith, I don´t see why wouldn´t it work for me... :?

Am I thinking right this time?
 
Thanks guy. So, this version is exactly what I need. WOW, it saves a an entire transformer, or a shitty voltage doubler...

JLM, thanks for the heads up on cap polarity!

I´ll update the schem, so that other can use it, as I think it´s a really simple and good performing kind of circuit.

Get dual rail supplies + phantom, with no high volts transformers, no voltage doublers, and only transistors that costs cents. :thumb:
 
OK, the updated schem:

JLM_OLD_PSU_moded_4.gif


Joe,

I see you used a 1000uF as your phantom-bridge blocking cap. I thought about going for 470uF, but as I will be powering up to four mics at once, do you think I should go higher, like 1000uF or even 2200uF?

Thanks for all the help, guys!
 
:grin: Hate to break the bad news but the 48v rail is running off a voltage doubler made out of the second bridge rectifer and the 2 x 470uF caps.

The 470uF caps are heaps big enough so no need to up size them.
 
:?

OK, I´m going to bed now.

Tomorrow, after getting some nice rest, I´ll draw it 10 times again, in diferent forms so that I might see this voltage doubler... :oops:
 
I have used this arrangement before. Seemed to be the most efficient I've come across concerning linear supplies for mic pres.

I've been meaning to post a similar schematic for critique, and rather than start a new thread it seems fitting to piggyback here on rafa's. I hope no one minds :grin: ...

new_discrete_psu.JPG


The goal is a simple-as-possible, entirely discrete supply which yeilds bipolar 24V and 48VDC phantom from a single 24Vrms transformer secondary. The single secondary limitation means I can use only a common wall wart and get all three voltages. Let's assume that I know how to calculate all of the R and C values based on application... do you think this arrangement will work well? Multisim and I think so, but I haven't yet built one exactly as shown. I know it won't be very efficient in terms of power consumption, but should get the job done. The actual PCB is going to have precision IC voltage references instead of zeners and plentiful bypassing in combination with a careful layout and ground plane. Low noise is a must! Speaking of which, would anyone care to enlighten me on the possible benefits of using darlington configuration for the followers as rafa did? Is it to have something faster/higher gain improving the sluggish power transistors?

Thanks!,
Jens

[EDIT] Before anyone asks, C1 through 8 are all going to be 35V rated, hence the stacked caps in the phantom section. Also, I know I forgot the current limiting resistor for D5. It'll all be proper when I build the thing :wink:.
 
> Are these all spice lies?

Different SPICE implementations will lie different ways, so I'm not going to duplicate your study.

However, I would say: start simple!!! You don't need the regulators to prove (or disprove) that you get ample raw voltages of desired polarity (or not).

bcarso says his Zeners make power. I had another simulator where the vacuum tubes would make power (much more than grid current from a hot cathode). And of course the generic SPICE Idc is a current source, while in the real world "current source" usually means a current limiter. Forget this and really odd things happen.

With the coupling caps it is indeed a voltage doubler. All rectifiers suck, single-phase much worse than 3-phase. Half-wave worse than full-wave. As suckiness goes, half- and full-wave rectifiers are not that much better than a doubler. If the high voltage is a small part of the overall load, or if you don't have to optimize your butt off, nothing wrong with a doubler. Bogen CH-200 and Dynaco Mark VI both use doubler power supplies.

Jens> post a similar schematic for critique

C1 will charge to 72V, +36V on the top and -36V on the bottom, and it will burst. You need the other type of doubler, the one that starts with a capacitor.

Oh, and as drawn Zener D5 has no bias, Q1 has no over-current/heat protection. When you get the doubler wired right, you will have +65-+70V on tap, so you can probably get away with a resistor in the collector lead.
 
Rafa!

Damn you beat me to it!! I was working on a discrete zener-follower psu and was going to drop it on the lab soon!

oh well, all i can say is your (new)schemo is almost identical to mine! I have yet to prototype it yet though.. :sad:

:thumb:


(edit) When either you or I get to prototype it, I'm going to lay it out SMD...
 
[quote author="Family Hoof"]I have used this arrangement before. Seemed to be the most efficient I've come across concerning linear supplies for mic pres.
...snip...
The actual PCB is going to have precision IC voltage references instead of zeners and plentiful bypassing in combination with a careful layout and ground plane. Low noise is a must! Speaking of which, would anyone care to enlighten me on the possible benefits of using darlington configuration for the followers as rafa did? Is it to have something faster/higher gain improving the sluggish power transistors?

Thanks!,
Jens

[EDIT] Before anyone asks, C1 through 8 are all going to be 35V rated, hence the stacked caps in the phantom section. Also, I know I forgot the current limiting resistor for D5. It'll all be proper when I build the thing :wink:.[/quote]

Remember that transistors are not perfect. An emitter follower, or a compound one, will have significant modulation of the emitter voltage by what is on the collector, regardless of how stiff and low noise the base voltage is. So these will not be the lowest ripple/noise regulators. Higher performance circuits require more parts in a real error amp comparing the sampled output to the reference.

Another gotcha on such supplies: the big bypass cap on the zener, or other device setting the base voltage, will allow the base to pull big currents in the event of a short. A final series R in the base will degrade regulation and noise a bit but will protect against the potential destruction.

Finally, if there is a lot of capacitance at the output, beware of a condition in which the input voltage falls faster than the output and the transistor junctions get reverse-biased. A reverse diode between collector and emitter will protect against this. One across the base-emitter junction is also not a bad idea.

The use of darlingtons simply reduces the loading on the reference. Otherwise, it is a slower, not faster, configuration in general (with some specific exceptions, like an f sub t doubler) and also responds asymmetrically (slow turnoff). The integrated darlingtons optimize the areas of the two devices to have closer to equal current densities and alleviate some of these effects. Some may have little turnoff speedup R's internally for the bases.
 
transistors not perfect? geez, and next you'll be telling me digital isn't either

use RC, and/or LC, filtering to both the zener and pre/post transistor. better than those fancy voltage references, i think. like bcarso suggests, use plenty of diodes and prefab Darlingtons
 
AWESOME!!! This is exactly the kind of beating I was looking for and could not find in my readings. Immense thank you to you guys for the advice! So... I think I could design a more complicated supply as suggested, but granted maybe it is time to go back to the ICs? Anybody have good experiences with the LT1083/4/5 parts? I have, but only once sofar.
 

Latest posts

Back
Top