High Current PSU question

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

mitsos

Well-known member
Joined
May 4, 2007
Messages
2,886
OK... I am trying to get a PSU together to power 8 (or more) pres. I started searching for high-current regulators and found the LM150 and LM338 (3A and 5A respectively) but I cannot find a negative equivalent.  I think you could use a positive wired "backwards" to produce a negative rail (ground the output and take your output from the ground pin) but is this the only way? Why would national (or whoever) not produce a higher current negative regulator?

The reason I am looking at these is they are easy to implement and have a small parts count.  Any ideas as to how to build a (relatively simple) bipolar +phantom PSU that could put out 5A? Judging from all the Neve builds using multiple PSU boards, I think this is something missing here, so if someone can help me put it together, I'd be more than willing to share.

TIA
 
They just had a discussion about this with the Group DIY rack project in several fronts, with schematics available, and a working prototype.

here: http://www.groupdiy.com/index.php?topic=33849.240

and here: http://www.groupdiy.com/index.php?topic=35154.0

I'll quote burdij here because it contained the suggestion to use the higher rated LM338, and notes about using positive regulators in a negative configuration.

burdij said:
The rectifier diodes in high current areas should be at least fast recovery. This prevents the very narrow spikes of current at zero crossings when both rectifiers are conducting. We used MR851s in the 5 volt and memory drive power supplies in an instrument we built at Tracor Northern.

I am curious why you selected an LM350 instead of an LM338 for your voltage regulator. The LM338 goes up to 5A.

The diodes across the rectifiers themselves are okay but your values are a bit high. More is not better here. These diodes are to suppress line transients from destroying the diodes. You should also have one across each secondary.

You should have a small, low inductance noise suppression cap across each electrolytic. I use .1uf monolythics usually.

I don't think using a positive regulator to build a negative supply, the way you are, is optimal. This will have some detrimental effects on the servo loop. One of the problems with packaged regulators for negative supplies is that they don't have as high a rating as their positive voltage counterparts. You do have enough margin to use a negative regulator and pass transistor, though. There are some circuits in the National datasheets.

I would consider using pre-regulators to bring the +24V supply down to feed to the +16V regulators but that is just because I'm allergic to paying for a power transformer with five secondaries. Also, you could feed the +48v supply from a full wave voltage doubler, eliminating the need for another secondary. The +48 power consumption is going to be less than one amp for an 11 slot chassis even with mike pres in every slot.

Maybe someone more knowledgeable should comment on why positive regulators in negative configuration are not so optimal. Detrimental effects, how?
 
I suggest you start out by first making sure how much current you *really* need. 5 A for phantom power is way too much--a typical mic draws 5 mA, and max. current through two 6k8 resistors is 14 mA. So your 5 A would power at least 354, and typically 1000 mics.

Why would national (or whoever) not produce a higher current negative regulator?

Because no one's gone buy them (in large quantities). High current is typically only needed for low-voltage digital stuff, which runs on single positive supply. LT1185 comes to my mind though.

Samuel
 
Kingston said:
Maybe someone more knowledgeable should comment on why positive regulators in negative configuration are not so optimal. Detrimental effects, how?
hey thanks.. missed those discussions completely. As for the positive regulators in negative config, I read a Q/A on National's site where the only precaution their engineer gave was to use separate secondaries for positive and negative.

I too would like to know more about the "detrimental effects" Burdij talks about.

also, when he says
The diodes across the rectifiers themselves
he means caps, right?

To Mr Sam G: 
mitsos said:
  Any ideas as to how to build a (relatively simple) bipolar +phantom PSU that could put out 5A?
Sorry, I could have been clearer. I meant a bipolar AND phantom PSU (3 rails). I know I don't need 5A for Phantom.  I don't know how much I need for the bipolar stuff but 5A should give me enough headroom for 24 APIs, for example, or an 8-pack of Neves (using only a +24 rail). 
btw, I had started another thread a while back (if you remember) about parallel (redundant) PSUs, and you had mentioned adding DC fuses before each mic pre, but I have never been able to find them. Any clue?


 
Here is a circuit adapted from the National datasheet. Transistor selections are optional (I have a lot of the MJ15025s). Minimum load current should be >50Ma. Mount the LM337 and pass transistor on the same heatsink, as close as possible. Although not marked in the diagram, the Vin and Vout are obviously negative.

neg_voltreg_booster.jpg
 
Hey burdij, thanks for stopping by as well. Would this be able to put out 5 (or so) Amps?  I saw something similar (but +12V, using the LM317), the difference being that there was a resistor in series with the emitter of the (2N4401 in this case) and no resistor at the base.  Says it could do 4.3 Amps.

How reliable is something like this vs. an LM338-based supply?  Could anything bad happen to whatever connects to it if either the transistors or regulator go down?

EDIT: About diodes: I've used 1N400x without too much trouble, also used MUR860 but don't really like the TO220 package. Tried some fast version of the UF400x and those worked OK as well.  I saw in the other thread that people like the BYV2X diodes. where do you get those? No one seems to stock them?  Any other schottky suggestions?
 

Attachments

  • highcurrentLM317.jpg
    highcurrentLM317.jpg
    24.6 KB · Views: 57
All these circuits that combine a 3pin reg with a bunch of high-power transistors are not as well protected as a fully integrated solution, but, if you take the pain to fit a well-suited heatsink, they are good enough. Regarding your fear of frying the units that are powered by it, neither configuration offers absolute protection; you should install overvoltage protections. There are several possibilities:
Crowbar protection: an overvoltage detector triggers a thyristor, which shorts out the output of the regulator. A fuse must be fitted.
Individual overvoltage protection: The regulated voltage is distributed via individual fuses to each unit, each one being fitted with a power Zener of suitable voltage
Timed relay:  The relay is powered via a ramp generator if and only if the output voltage of the regulator is within a suitable window
 
abbey road d enfer said:
Individual overvoltage protection: The regulated voltage is distributed via individual fuses to each unit, each one being fitted with a power Zener of suitable voltage
I had been planning individual fuses per rail per channel. But, what kind of fuse do I use and what is the zener's job? Wasn't considering a zener at all. thanks!
 
You have to fit a Zener across the PSU input on the module. If the unit is powered with 24v, I would put a 27V Zener. Then, if the PSU goes south and delivers too much voltage (that happens when the ballast transistor melts), the Zener prevents that and blows the fuse.
 

Attachments

  • Zener ovr volt protect.jpg
    Zener ovr volt protect.jpg
    8.8 KB · Views: 56
Thanks for the drawing... always helps. If I understand, the PSU is on the left.. So if we have 24V , no problem, everything works fine. If we go above 27V, the zener sends it to ground and since it sees a route direct to ground, the fuse blows.. is that close?

I'm still not sure what kind of fuses I should be looking at.  I've only used mains fuses so far, is there a specific DC type fuse, or do I grab normal fuses rated for the max current draw (+ a bit of headroom) of the specific module?

thanks again!

 
just checked and they do of course have one: LT1185 is a 3A, adjustable negative reg. The mention a 25V limit. It's not reall clear, is this the max Voltage output or the max difference between input and output (like the LM317/337 specs)?  25V is not a problem, per se, the most I would want is +/-24V which I could further regulate to +-18 or 16 for use with certain opamp (API) pres. 

I'm gonna try to work up a quick schematic to have something concrete to discuss.
 
With those Lm388k etc it is worth noting that the more voltage the reg is actually dropping the less current they can deliver.  The 388 maybe be rated at 5A, but it won't deliver 5A if it is dropping loads of volts.  There is a graph on the data sheet, which will give you the info required.

 
mitsos said:
Thanks for the drawing... always helps. If I understand, the PSU is on the left.. So if we have 24V , no problem, everything works fine. If we go above 27V, the zener sends it to ground and since it sees a route direct to ground, the fuse blows.. is that close?
That's exactly it, on a human POV; if you looked at it from an atomic POV, with a nanosecond time-scale, that would be different ;)
I'm still not sure what kind of fuses I should be looking at.  I've only used mains fuses so far, is there a specific DC type fuse, or do I grab normal fuses rated for the max current draw (+ a bit of headroom) of the specific module?
You have to know the maximum current draw of the module and use a fuse rated at ca. 2x this value. In case of overvoltage, the Zener will be almost like a short-circuit, so the fuse WILL blow. You may select a fast-action fuse ("F" marking).
 
Rob Flinn said:
With those Lm388k etc it is worth noting that the more voltage the reg is actually dropping the less current they can deliver.  The 388 maybe be rated at 5A, but it won't deliver 5A if it is dropping loads of volts.  There is a graph on the data sheet, which will give you the info required.
This is not specific to the LM388; all regulators that include their ballast transistor have the same dissipation limitation. The power dissipated in a reg is (Vin - Vout) x Iout.
This is also valid for any transistor in its linear operation area. The safe-operation area (SOA) is delimited by the hyperbolic curve where the voltage/current product is constant and the abs max current and the abs max voltage (plus secondary-breakdown area for the purists).
 
Just for something completely different (the page is unfortunately only german): http://www.thel-audioworld.de/module/spr/spr10.htm
This is a discrete 10A regulator, the company does DIY HIFI pre- and poweramp modules. One module is 88,-€ incl VAT, if you're just building one psu it might be worth a shot though it is some money because it might save quite some time - unless you particuarly enjoy diying psus of course  ;)
I am successfully running my 60 channel DDA console on a dual lab psu I bought on ebay for roughly 200,-€, if I wouldn't have found that, I'd probably have built a psu with the Thel modules. Btw, the current my whole console needs is roughly +/- 4 A, a lot less than I had expected. Maybe you don't need 5A for your pres?

If you're interested I can translate some important bits from the Thel site. The key point is that the modules are connected like a 3-legged chip (http://www.thel-audioworld.de/module/spr/SPRGES.GIF), they are good for 10 A (soft limited), I'm not totally sure about the voltage range - if the min Vout is low enough, but I guess so. Max Vin is 100V or so, way out of sight. A heatsink might of course be necessary.

Michael
 
Hi Michael,

thanks for pointing that out. While it is an option, I am trying to DIY something together that will serve for the current project I'm working on and hopefully for future use.  I don't need 5A, that would be for extra headroom, I estimate about half that as necessary.  I don't mind DIY PSUs btw, it's usually the only part of the project I DONT have to troubleshoot ;)

back to the max current vs voltage difference question.  Are you referring to the following chart (from LM338, 5A regulator). It's not completely clear, but there is a DC current limit and Peak current limit.  Let's say I used a 2 x 20V trafo.  20V x 1.414 = (more or less) 28.28V post rectification.  If I regulate that to 24V (or even 18) I am within or at the 10V point on the chart, so my max would be 8A?  What am I missing here? They guarantee 5A, so it should do more in certain situations?

EDIT: one more question on bridges: can I use a heavy duty bridge (or diodes) and send the + and - to four regulators simultaneously (two positive, two negative)? Or would I need 2 bridges for this (so that I don't regulate and re-regulate)?

EDIT 2: Looking at JLM's 5 rail, seems to be how they do it. One rectifier for 4 rails, and one to do the multiplier for phantom, so I guess that answers that question. Just need to make sure the bridge can handle the amperage.
 

Attachments

  • 00906033.pdf
    11.1 KB · Views: 13
ok, I know I said I'd have a schemo in a couple days, but life took more time than expected.  Then I opened Eagle after two months of not using it, and couldn't remember half the things I thought I knew... getting old?  I wish someone would do a board layout thread in the white market...

I'll be using MUR860s for rectification unless someone has a better idea?  They can do 8 Amps and are physically wide enough that (I think) they can be substituted with axial diodes without much trouble, if need be. 

Right now, the problem is that the LM350/lm338, can be used as negative regs but need a separate trafo winding, much like the GDIY500 rack PSU. I can see that I could do 2 (positive) LM338s off one winding, but it's not clear if I could do 2 negative regulating LM338s off one winding ( to be clear, I mean both negative regs off one winding, not positive AND negative off one winding).  If I can not do this, I could still use the LM338 for positive and use an LM337 with 1-2 extra transistors (like burdij's image above) for negative regulation. But then, why go to the extra cost of the LM338 for positive, and not do the same with an LM317 and an extra transistor.... I was never good at making decisions. The more options, the worse it gets... what I DO know is that I want this to work with a standard 2 secondary trafo.

so, sorry for the delay, and thanks for the help so far everyone.
 
mitsos said:
Right now, the problem is that the LM350/lm338, can be used as negative regs but need a separate trafo winding, much like the GDIY500 rack PSU. I can see that I could do 2 (positive) LM338s off one winding,

What were your voltages again? Do you mean +-24V and something like +-15 from one supply? If yes, then you will be perfectly fine to first regulate down to +-24V with two LM338, and cascade another similar stage after that and to drop to +-15V. No need for exotic transformers. I believe this is a quite common pre-regulator + regulator combo, although I can say how optimal this would be not knowing your exact needs.
 

Latest posts

Back
Top