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.
That's about the voltages I'm looking at. +- 24 or so, +/-18 or so, +48 and maybe one more for LEDs, etc.  My worry is that, if I cascade them as you describe, I'll be limited by the first regulator's max current (if I take the 2nd input from the output of the 1st). The way JLM does their 5-rail, they run the inputs of each regulator to the bridge and I think that, in my case, that would make my limit the rectifier limit (and trafo of course).

My problem with the LM338 as negative regulator is, can I run two (as negative regs) off one secondary/rectifier/ground?  I'd have to tie both their outs to ground to be able to use them as negatives, that's what I'm not sure will work (well), and why I'm considering sticking to the 317/337 with extra transistor(s) for higher current. I just thought the 338 way would be simpler due to lower parts count...
 
mitsos said:
My problem with the LM338 as negative regulator is, can I run two (as negative regs) off one secondary/rectifier/ground?  I'd have to tie both their outs to ground ...
Will work, but the output of the -18V part does not connect to common ground and will dissipate lots of heat.
If the 1st regulator is set for 24V output and +output tied to common ground, the 2nd regulator with input in parallel (not series) to the 24V one needs setting for 6V output to get -18V relative to common ground.
 
mitsos said:
My problem with the LM338 as negative regulator is, can I run two (as negative regs) off one secondary/rectifier/ground?  I'd have to tie both their outs to ground to be able to use them as negatives, that's what I'm not sure will work (well), and why I'm considering sticking to the 317/337 with extra transistor(s) for higher current. I just thought the 338 way would be simpler due to lower parts count...

considering sticking to the 317/337 with extra transistor(s) for higher current is a good way to having a simple regulator. I've designed a dual voltage regulator with power bjts , It's a simple way to having a very
reliable high current regulator. I don't love the LM338 regulator for 5A out current, because 5A is its max current and it suffer too much with this current.
 
this the schematic of the 16V 5A that have designed for 500 multiple rack
 

Attachments

  • APP  16V 5A reg n.4.jpg
    APP 16V 5A reg n.4.jpg
    98.7 KB · Views: 74
Thanks you both Harpo and PPA.  After Harpo's clarification I think the LM338 would be easiest, but I'd still have to go TO-3 to be able to dissipate heat more efficiently... So size-wise I guess it could be either way. 
Questions on the schemo: 
1. The transistor base resistor has no unit.. is it R or K? 
2. I've seen similar schemos but with an NPN transistor above the PNP (NPN B to PNP C with series 500R to output, NPN C to PNP E, and NPN E to output. This is from page 12 of the TI 317 datasheet, but is listed only "high current."  I am curious because of your warning "BIG HEATSINK". Would using two transistors divide the current between the two and perhaps lessen the heat put out by each one (so I could use smaller heatsinks or maybe smaller package transistor TO-39 or something?)? 
3. If we paralleled another LM317 to your schemo would it work or would I have to "adjust" the second one the way harpo recommends with the 338?

thanks again!
 
mitsos said:
Thanks you both Harpo and PPA.   After Harpo's clarification I think the LM338 would be easiest, but I'd still have to go TO-3 to be able to dissipate heat more efficiently... So size-wise I guess it could be either way. 
Questions on the schemo: 
1. The transistor base resistor has no unit.. is it R or K?   
2. I've seen similar schemos but with an NPN transistor above the PNP (NPN B to PNP C with series 500R to output, NPN C to PNP E, and NPN E to output. This is from page 12 of the TI 317 datasheet, but is listed only "high current."  I am curious because of your warning "BIG HEATSINK". Would using two transistors divide the current between the two and perhaps lessen the heat put out by each one (so I could use smaller heatsinks or maybe smaller package transistor TO-39 or something?)? 
3. If we paralleled another LM317 to your schemo would it work or would I have to "adjust" the second one the way harpo recommends with the 338?

thanks again!
the base resistor is a 2.2 ohm 5% 1W , it's optional, you can avoid it so you can reduce the components added.

Regarding the big heatsink , it's always necessary because at 5A on out the regulator must dissipate 25-30W.
Moreover the LM338 required a bigger haetsink with this high current than my schematic.....................................
It's necessary a big heatsink (not so big however.........) for the MJ15025 and a little heatsink for the LM317.
This schematic has been simulated with a cad. 

For the negative rail I will post the schematic very soon.






 
mitsos said:
.. After Harpo's clarification I think the LM338 would be easiest, ..
.. the way harpo recommends with the 338?
Huh? It will work with 2x LM338, but I did not recommend it. Depending on current draw you'll need a river to cool it down.
For your 24V part you want at least 27V raw DC in front of your reg. Calculating the needed transformer for 10% line undervoltage, your reg might see about 32V in front for a 10% high mains scenario. For 6V regulated output coming from this 32V raw DC you'll have to burn away 26V times current. Seems too much to be usable.

The national circuit with negative voltage regulator suggested by burdij seems to fit much better.
With ppas' schematic, using a positive regulator you'd be in the same situation as before (wanting -24V and -18V higher current from a single transformer winding). You'll need this for a negative voltage regulator.
-Harpo

uups, pier seems already working on it.
 
negative regulator 16V 5A out
 

Attachments

  • APP  16V 5A neg reg n.4.jpg
    APP 16V 5A neg reg n.4.jpg
    155 KB · Views: 48
regarding these regulators of mine, they use the MJ15024/25 transistors.
Respect the LM338 solution , these transistors offer very very bigger reliability,
MJ15024/25 are 250W 15A max BJT with TO3 case . They have a junction temperature of 250°C.
With 5A on out and a good heatsink they run with high margin.

 
Thank you very much Pier. I've mashed up your schem with JLMs and Keiths, to show you what I had in mind. It's just scrapped together, no negative rail parts (I'm lazy and wanted to post this quickly), and no labels on the second rail parts. Besides that, I mayy have made mistakes. Let me know if you think this will work. I found some MUR1560, they are good for 15A, so I guess I could use one bridge for the whole thing (ideally I'd find something capable of 20A, so that all rails would be capable of 5A (besides phantom).

As is, this is a 5 rail supply, which is mostly what I need. 2x positive, 2x negative, and phantom. But let's say I needed 12V for LEDs relays, etc, could I put a voltage divider on the 24V (and take 24V from the top and 12V from the resistor junction). I assume that the divider is "burning off" the 12V, so it's not very efficient?

 

Attachments

  • 5A_PSU_draft_schem.jpg
    5A_PSU_draft_schem.jpg
    51.2 KB · Views: 33
do you want use the +-24V outs and the +-18V outs togeder?

If yes:
- for having a power transformer only, you should do a series of regs, the 24V regulator that supply the 18V reg , but the limitation is that, however, the total max current, given adding 24V reg's out current and 18V reg's out current,  must not exceed 5A. 

- using two power transformer, one for 24V regs and one for 18V regs, to permit to have 5A max in both two outs.

Your shematic is good but some things should be changed, like the leveling caps.
I will post the complete schematic, let me know about the outs' currents. 

 
Hi Pier,
Ideally I'd like 5A availble on each of the four rails. That's why I'm trying to avoid the series regs. I got the regulator configuration from JLMs 5 rail PSU. He uses them this way but without the transistors. Although, since his schem has 1n400x diodes for rectification, so his max current for all rails is determined by that, right? But if one put higher currrent rectifiers (say, 8A), could you get the full 1.5A out of each +/- rail?

Anyway, back to this, maybe I won't be using all that amperage, but it's good to have available.  The final PSU is going to be running 24 channels of preamplification.  All I know so far is that 8 channels will be API-type (I'm doing my own board...it's 95% finished) and then I'll get started on a Neve board (1272 or 1290), not sure which one yet.  It's the Neve pres that worry me, since they consume quite a bit of current.  There are people here building 1290s who budget 0.5A per channel. That may be way overboard (Neve specs say 120mA) but even if I do 250mA, that's still 2A per 8 channels.

Does the way I drew them (I guess thats in parallel) pose problems?

thanks for all the help so far!
 
You know the 2520 draws around 1mA on each leg...... the 40 input legacy console in the truck I used to run had one Power One module for each rail, at something like 2.5 amps each
I do find the discussion of interest, but I think it is often easier to have the power supply give unregulated DC and use local regulation of the standard LM variety, at 1.5 amps (or 2 with the ST's) each it goes a long way and you can use as many little regulator boards as you need
 
Actually it's interesting that you bring this up. That's not a lot of current for a 40 input console (lots more than 40 opamps actually in that thing, right?) I was under the impression they were 25mA per rail, so I double checked the 2520 datasheet just now.  I don't know where I got the 25mA from. The datasheet at http://www.waltzingbear.com/Schematics/API/API_2520.htm
says quiessent current 15mA at +/-15V and 26mA at +/-20V. This is per rail and not under load, right? Just with the power on but not necessarily any signal going through?

I ask because below that it lists "current at rated output 600 Ohm" and then 75 Ohm. 600 is not my case here I think, cuz I'd almost always have a 75 Ohm trafo on the back of the pre (except for the THAT 1646 version I'm working on, which has a 5K input impedance).

For a 75 Ohm load, it lists 60mA at +/-15V and 85mA at +/-20V. The question is: is this 60-85mA per rail?? If so, we're at 120-170mA, which is a LOT more than I thought they used... Is that right? How do we test this in the real world?

You may be right about distributing raw DC and regulating locally. I read that minimizes crosstalk.. I wonder how large consoles deal with crosstalk? I started this thread to get the best possible solution to my current problem (pun intended), and to be able to offer this to the lab since I see people running multiple PSU boards for projects like Neve stuff, where I thought a simpler, high current solution may be better/easier. 
 
here the schematic.

ATTENTION! : Since a PSU has high voltage inside, this schematic is only for who knows all electrical safety rules. Moreover, pay attention with dangerous voltages! DANGER!
I do not assume any responsability on this circuit and all uses to which will be done with it.
Moreover, this circuit is only for DIY (Any commercial use is denied) and it's only for people very expert on power supply units.

 

Attachments

  • PSU APP dual 24V 48V.PNG
    PSU APP dual 24V 48V.PNG
    42.5 KB · Views: 55
The 1uf 50V cap is way too close to 48V. pointless to skimp in parts here, use a 1uF 100V or something..

I'd actually use a 10-100uF 63-100V electrolytic with a 100nf 100V bypass cap there instead.
 
Kingston said:
The 1uf 50V cap is way too close to 48V. pointless to skimp in parts here, use a 1uF 100V or something..

I'd actually use a 10-100uF 63-100V electrolytic with a 100nf 100V bypass cap there instead.
I've written the necessary minimum values, but people can use greater value. However a polyester cap of 50V for 48V has a very long life, so using an higher value is an optional.
However, I'm going to put this PSU with high current regulators in the market (some recording studios in Italy have requested it to me to power several 312 preamps with my APP2050)  and I will use higher voltage values for caps.
Moreover , I'm using 250W and 15A transistors for only 5A...........

 
ppa said:
I'm going to put this PSU with high current regulators in the market (some recording studios in Italy have requested it to me to power several 312 preamps with my APP2050) 
Are you going to offer a board for sale? If so, are you planning on doing this soon?  So I know if I should wait to buy your boards! :) bipolar plus phantom? I assume you won't do my crazy 5 rail idea? Are you going to do regulation on the same board or in the box with the pres?

Have you used those rectifiers before? What are your thoughts on schottkys or fast recovery diodes? if you sell a board, would you consider putting pads for some other type of rectifier?

Also, after Nielsk's comment I'm curious..  have you measured the avg/max current consumption of 2520s? I'll try with my DMM, but I wonder how accurate it is...

For extra safety, I was thinking of rigging some sort of relay to cut off the high current DC when the cable is unplugged. I'm not sure exactly how, maybe an extra rail (Low V/low I) on an extra conductor going to ground with a series resistor? So that when you unplug the cable, you break the circuit and the relay cuts off the rest of the voltages.  BUT, I'm not sure how long a (almost) permanently running relay would last. And it's wasted energy, but this way no one will be able to stick a needle or something else small and conductive into the female XLR or whatever connector is on the back of the PSU case, and get the shock of their life.

thanks for offering up all this work for DIY. 
 
Back
Top