48V PSU schematic attempt

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Potato Cakes

Well-known member
Joined
Jul 1, 2014
Messages
2,363
Location
Nashville, TN
Hello, everyone,

A long standing issue I have been having with some custom builds is trying to supply power to single source and bipolar circuits with a single power supply. So I decided to design my own to suit my purpose and hopefully others as well. I've attached my chicken scratch schematic based on what I have experienced, read about, and seen on many other builds. I'm taking a single source SMPS and filtering it before it goes anywhere else. Then it's split between the single source regulator and the rail splitter that feeds the bipolar regulators. There is also a tap for phantom power. I'm trying to keep the board as compact as possible but still allow for a wide range of capacitors to suit each build's need by making the lead spacing 5mm. I am using a LM317HV for the 48V regulation as I have been running my Neve builds anywhere from 36-40V since the transistors used can handle anywhere from 60-80V rails and I just need to change a few cap values to handle the extra voltage.

Right now I have two capacitors parallel for per rail power reservoir to allow for higher capacitance with needing large expense caps. This might be more overkill that what is usually desired. I may scratch that if needed to save space as I'm wanting fit this in as compact of PCB as possible.

The one thing I am not sure of is the phantom power. I am purposely separating the 0V and ground as I have found doing so for my Neve type builds eliminates the hum I experience with most mics I use, especially dynamics. I've gone through various way to route the grounding and even caused personal injury to myself in the process and the one thing that I found to have worked is keeping the 0V and ground separate. But condenser mics use the ground for +48V reference unless I am mistaken.
 

Attachments

  • 48V PSU .pdf
    493.8 KB
Here is the BOM currently. There is a range for some of the values as I don't know what is best or if any of the values in the ranges listed will be fine.

C1,2 - 470uF/63V
R1,4,5 - 240R
R2,3 - 10k
C3-8 - 470-1000uF/63V
L1 - 200uH - 1mH/1.5-2A
VR1 - 5k

I also think I need a power reservoir for the phantom power as well.

I'd love hear people's thoughts, especially from the more experienced PSU guys.

Thanks!

Paul
 
Potato Cakes said:
A long standing issue I have been having with some custom builds is trying to supply power to single source and bipolar circuits with a single power supply. 
This is just not gonna work.
One could design something starting with a 65-70V supply, and a rather inelegant assemblage of regulators, but that would be a silly arrangement, more complicated than the traditional solutions.

But condenser mics use the ground for +48V reference unless I am mistaken.
That is correct.
 
abbey road d enfer said:
This is just not gonna work.
One could design something starting with a 65-70V supply, and a rather inelegant assemblage of regulators, but that would be a silly arrangement, more complicated than the traditional solutions.
That is correct.

Thanks for getting back to me. I have been banging my head against wall exploring more traditional, elegant approaches but I have not found one that has worked well trying to accomplish what I'm after. Could you point out the obvious problem(s)? As you can see, I'm essentially using blocks of what is published by the manufacturer of these ICs which I have used and seen on many designs. If I have to initially go without phantom power that is fine, but I'm not sure why this won't work. I have also said that same thing many a time right before something didn't work.

Thanks!

Paul
 
CurtZHP said:
This is what I use.  Seems to work very well.

Thanks for schematic. Unfortunately for what I'm doing it has to be SMPS as I've tried a linear approach and it caused too much noise since everything has to fit in the case.

Essentially, I'm doing an expanded version of what Joe Malone is doing with his TREK PSU which takes a single ended 48V SMPS and regulates it down to 24V and then has the 48V as through for phantom power. My thought is if it works for that then why couldn't I also take that same SMPS, split the rail and regulated for stability? This is a genuine question.

Thanks!

Paul
 
You can't take a positive supply and make a negative voltage with a linear regulator: full stop.  You can make a 'relative' supply of perhaps 24v, 12v, and ground, but that doesn't seem to be what you want to do.

Just stay with SMPS, and get a module to buck from 48v to 24v, and another to invert from 48v to -24v.
 
CurtZHP said:
This is what I use.  Seems to work very well.
This is to take with a pinch of caveat.
In oder to produce a clean 48V, the unregulated voltage at the input of the TL783 must be at least 52V; taking into account ripple and possible mains voltage variations, the design target should be about 56V. This implies that the secondary delivers at least 2x21Vac, which in turn results in 28V unregulated at the input of the 18V regulators. The regulators have to dissipate 10V, which is not without creating issues.
A better alternative consists in referencing the negative input of the bridge to the unregulated +. Then , the circuit becomes a tripler instead of a doubler, and a 2x15V xfmr can produce 56V and 2x19V unregulated, enough for 2x15V and 48V regulated, but not enough for 2x18 regulated.
A 2x18Vac xfmr produces 2x23 + 72V unregulated in the same conditions.
 
Potato Cakes said:
Could you point out the obvious problem(s)?
The 0v for the +/-18V rails cannot be at a different potential than that for 48V. Your circuit cannot produce -18V from 0V.

SMPS.
I use these:
https://fr.aliexpress.com/item/4000832991856.html?spm=a2g0s.12269583.0.0.40281324ylk1r3
I use the 24V version, followed by CLC filters and .linear regulators.
For the 48V rail, you'll need a converter, as Matador already mentioned.
However, unless you are willing to roll your own boost converter, DC/DC converters that do 24 to 48 are quite expensive. You have to use an isolated converter with dual 24V outputs, such as this one:
https://fr.rs-online.com/web/p/convertisseurs-dc-dc-isoles/1226940/
And you'll need CLC filter again.
 
Matador said:
You can't take a positive supply and make a negative voltage with a linear regulator: full stop.  You can make a 'relative' supply of perhaps 24v, 12v, and ground, but that doesn't seem to be what you want to do.

Just stay with SMPS, and get a module to buck from 48v to 24v, and another to invert from 48v to -24v.

I mislabeled my crudely drawn schematic. What I am trying to accomplish for the rail splitter is in the attached image. Not exactly like it but similar with adjustable voltages.

Thanks!

Paul
 

Attachments

  • Regulated rail splitter.jpg
    Regulated rail splitter.jpg
    61.9 KB
abbey road d enfer said:
The 0v for the +/-18V rails cannot be at a different potential than that for 48V. Your circuit cannot produce -18V from 0V.

SMPS.
I use these:
https://fr.aliexpress.com/item/4000832991856.html?spm=a2g0s.12269583.0.0.40281324ylk1r3
I use the 24V version, followed by CLC filters and .linear regulators.
For the 48V rail, you'll need a converter, as Matador already mentioned.
However, unless you are willing to roll your own boost converter, DC/DC converters that do 24 to 48 are quite expensive. You have to use an isolated converter with dual 24V outputs, such as this one:
https://fr.rs-online.com/web/p/convertisseurs-dc-dc-isoles/1226940/
And you'll need CLC filter again.

I did end up with a separate SMPS for the phantom power like the ones which you so kindly provided a link. Thanks for everyone's help. I hate dealing with figuring out how to power up gear I build or want to build, but I don't have enough money yet to pay someone to do it for me and correctly the first time. Maybe some day....

Thanks!

Paul
 
abbey road d enfer said:
So, you found a 48V version?

I did. They are about the same as the link you provided. I kept it about 100mA which should be plenty for one mic that needs phantom.

I'm going to do some bread boarding and testing with the other part. I know I won't be doing anything that is new but it will be mostly for my own edification.

Thanks!

Paul
 
Finally breadboarded the circuit and it works great! The only thing I didn't put in was a reservoir cap for the 48V section, but I know that the circuit downstream does have one at the VDC input. Also, R1 had to be changed to 130R as the Vout formula is slightly different for the HV version of the LM317. I do need to order a smaller style inductor for next time. The layout also needs to be more efficient.

My last question deals with protective diodes and bypass caps. I have seen schematics with what seems to be a diode for short circuit protection and some with a ceramic cap between Vin and Vout, while others do not have either. I'm not sure what is the best practice regarding these items or if there is one. Should I add both just to be safe or not worry about them?

Thanks!

Paul
 

Attachments

  • 48V PSU.jpg
    48V PSU.jpg
    328.3 KB
Potato Cakes said:
I have seen schematics with what seems to be a diode for short circuit protection
Call me adventurous, but I never used any ot these anti backlash diodes and never had any issue, but I must say I never pushed the limits of max input voltage. Many designers do not hesitate to run floating regulators (LM117/137/783 family) with input voltages much higher than the max voltage, considering it's only the difference of voltage between input and output that counts, which is true in permanent regime, but not at start-up and shut-down.

.and some with a ceramic cap between Vin and Vout,
I'm not sure... Do you mean caps to ground at input and output? I would not recommend them at outputs, since the regulators are more stable with some ESR. And at the input, they can do no harm, but don't provide any sensible improvement.
 
Not sure what circuit you're working with at this point but you cannot connect your virtual ground to a chassis because phantom power returns on the cable shield which is pin 1 of the XLR which is connected to the chassis. So you would be connecting virtual ground to the negative output of your SMPS which would just short out the negative rail. So you have to leave your virtual ground floating like this:

Screenshot%2Bfrom%2B2019-01-16%2B01-41-03.png


The problem with this design however is that you cannot have any DC currents on the virtual ground at all. For very simple circuits it so happens that this is usually not a problem so this circuit works great for a simple single minimalist pre. But for more elaborate or general purpose designs you will need an active virtual ground maybe using a high voltage op amp like OPA551.
 
I am no longer using phantom power with this circuit. I am using a separate SMPS for that particular function all together. The virtual ground side is being used for a couple of NE5532's and four INA134's. I have used a virtual ground scheme successfully for these same components in the same configuration with just a simple rail splitter, so I should be good with this regulated version.

Thanks!

Paul
 
Back
Top