+/- 16V symetric smps?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

arjepsen

Member
Joined
Sep 12, 2018
Messages
18
Hey.
I'm have this project, of wanting to replace an old linear +/- 16V supply from an old desk (mackie d8b) - mostly because I'd like to move the power supplies inside the desk, instead of having it in a rather large rack box next to the mixer.
I've been looking at all the symetric smps's I could find, but unfortunately I haven't been able to find anything larger that +/- 15V. (Someone actually did find a 22V, but it was rather expensive, and isn't available through any of the european suppliers that I could find).
I did try out a +/- 15V from tracopower. It does have a small trim pot, that let's me adjust it up to 16V - but that only works on the positive side... the negative side doesn't go up, and unfortunately the mixer gives me an evil eye, when I try to get to work with that supply....
I wonder if there might be a simple way to "hack" the supply, so that I could adjust both the negative and positive rail?

Here's a schematic of the original supply that I'm trying to replace:
1630257411433.png


Regards
Anders
 
If you put the power supply in the mixer itself, you will make the whole thing noisier, and now you need to make the entire device safe to contain 110V AC. Also, why would you change the power supply voltages? The circuitry will blow up if you go too far, and the extra headroom you buy going from ±16V to ±18V (the typical ragged edge absolute maximum for most analog ICs) is just over 1dB.
 
Thanks for the replies.
Unfortunately the desk doesn't work, when supplied with +/-15V only.
I'm not trying to change voltages - I'm trying to figure out how to get a compact +/- 16V supply.
 
Are you sure you have room to put it (them) inside?

There are a bunch of important requirements that must be satisfied to implement an SMPS for audio correctly. I usually link to one of my posts with details but I can't seem to get Google to find it at the moment.

The path of least resistance in this particular case would probably be to get two 24V isolated MeanWell SMPS and "stack" them to make +-24V, a +72V MeanWell for the +48V line (or anything above ~52V but 72 is a standard voltage for the MW SMPS), put those into some kind of floor enclosure and then reuse the cable so that it can just plug into the mixer without modification. It's not quite as efficient but that db8 doesn't look like a particularly large mixer so feeding raw SMPS into the existing regulators will work fine.

But you do need to pick the right size of SMPS. That's one of the important requirements.
 
Are you sure you have room to put it (them) inside?

There are a bunch of important requirements that must be satisfied to implement an SMPS for audio correctly. I usually link to one of my posts with details but I can't seem to get Google to find it at the moment.

The path of least resistance in this particular case would probably be to get two 24V isolated MeanWell SMPS and "stack" them to make +-24V, a +72V MeanWell for the +48V line (or anything above ~52V but 72 is a standard voltage for the MW SMPS), put those into some kind of floor enclosure and then reuse the cable so that it can just plug into the mixer without modification. It's not quite as efficient but that db8 doesn't look like a particularly large mixer so feeding raw SMPS into the existing regulators will work fine.

But you do need to pick the right size of SMPS. That's one of the important requirements.
Yes, I did read some of your messages on smps' and their requirements. THANKS for all that info! :)
I would prefer not to use the original "umbilical" cable that supplied the 5, 12, +/-16 and 48V - the connector at the desk (lovingly called the "BFC" - Big F... Connector) is a well-known source of trouble, so much so, that people reguarly suggest people to do a bit of "ye-olde BFC-wiggle", if their desk is acting up... This is one of the main reasons I'm trying to get it inside the desk. There is space enough inside the desk - they originally tried to stuff everything in there - INCLUDING a pentium2 motherboard with hard- and floppy drive.... I'm leaving the PC components outside, and are just working on getting the power supplies inside.

Anyways, if I understand you correctly, I could do something like this quick mockup:
1630270597612.png

And just make a small board with the same regulation circuit as from the original supply?

Just a few questions:
I'm not that well into smps theory - are there any trouble in connecting the minus from one to plus from the other?
Also - would there be any grounding issues, if these are not electrically isolated from the chassis of the desk?
 
If you put the power supply in the mixer itself, you will make the whole thing noisier, and now you need to make the entire device safe to contain 110V AC. Also, why would you change the power supply voltages? The circuitry will blow up if you go too far, and the extra headroom you buy going from ±16V to ±18V (the typical ragged edge absolute maximum for most analog ICs) is just over 1dB.
Yes. But the OP isn't trying to go to +/-18V.
Right about the safety/compliance issues though.
Seems something of a solution in search of a problem imho. The cable connection seems to be the issue but it has to be less hassle to sort that than re-engineer the power arrangement ???
 
CAPI and Collective Cases both have +16/-16/+48 SMPS. They are external supplies, but with some modification you could mount them internally if need be. (External is preferred though).
 
And just make a small board with the same regulation circuit as from the original supply?
Actually I was thinking you should just reuse the regulation circuitry. Otherwise no, I would do things a little differently if I were designing a completely new supply.

Just a few questions:
I'm not that well into smps theory - are there any trouble in connecting the minus from one to plus from the other?
If the SMPS is isolated, you can stack them like that yes. Not sure if the output protection will work correctly on the negative rail but whatev. Look at the datasheet and see if there's a transformer in the diagram or if it says it's isolated. Find out what sort of current is being used, find a MeanWell LED enclosed SMPS on Mouser, post the part number and I'll tell you if it looks good. It's probably going to be a 50W beige plastic thing for $17 USD.

Also - would there be any grounding issues, if these are not electrically isolated from the chassis of the desk?
Grounding is important. That has been discussed here many times. The short version in this particular case is that you want to have a fat wire from the 0V of the downstream end of the supply (meaning the regulator board), preferably right next to the filter caps running to earth ground. And if the chassis is metal (?) you should just connect it to a bolt in it and then have another fat wire from it to the earth ground of the mains cable in.

Are you just going to run the cable in with a cable gland or are you going to put in a mains socket?

What are you going to do about 48V?
 
Back last century while still at Peavey we put a universal input SMPS inside a 32x8 flight case mixer... The flight case was blow molded plastic so RF was not a big concern. I didn't do it but it was not rocket science even for 20 years ago. Of course that doesn't mean it would be easy to retro SMPS into an existing design.

JR
 
Actually I was thinking you should just reuse the regulation circuitry. Otherwise no, I would do things a little differently if I were designing a completely new supply.


If the SMPS is isolated, you can stack them like that yes. Not sure if the output protection will work correctly on the negative rail but whatev. Look at the datasheet and see if there's a transformer in the diagram or if it says it's isolated. Find out what sort of current is being used, find a MeanWell LED enclosed SMPS on Mouser, post the part number and I'll tell you if it looks good. It's probably going to be a 50W beige plastic thing for $17 USD.


Grounding is important. That has been discussed here many times. The short version in this particular case is that you want to have a fat wire from the 0V of the downstream end of the supply (meaning the regulator board), preferably right next to the filter caps running to earth ground. And if the chassis is metal (?) you should just connect it to a bolt in it and then have another fat wire from it to the earth ground of the mains cable in.

Are you just going to run the cable in with a cable gland or are you going to put in a mains socket?

What are you going to do about 48V?
I did think about reusing the regulation circuit board, but it's got too much stuff other stuff there that isn't needed, so I wanted to trim it down to just the needed circuitry - to save space.
I put in a mains socket - don't like the cable glands.
Regarding the 48V, I read a post in here on smps for phantom power - I'll dig into that when I get the +/- 16V sorted.

One thing I started thinking about: since getting two smps's, and since it seems most of them have a bit of adjustment to them, I could maybe just get two 15V supplies, and adjust them up to 16V, and "stack" them?

You mention you would do things differently if designing a new supply - could you elaborate?

Anyways, thanks for your input thus far!
 
One thing I started thinking about: since getting two smps's, and since it seems most of them have a bit of adjustment to them, I could maybe just get two 15V supplies, and adjust them up to 16V, and "stack" them?
You could do that, but be aware that some smps are quite noisy; that's why Bo advocates using 24V smps followed by linear regulators.
When I made a PSU for my lunch box, I used 15V smps that I tweaked up to 16V; they were followed by CLC filters and it worked just fine.
On a recent design, though, I employed the same solution but laid it out on a PCB wjih PC mount encapsulated smps. The noise was awful, so I had to redesign with 24V smps (tweaked down to 22) and linear regs.
 
ok, that makes sense.
I read some other threads, where Bo suggested using a capacitance multiplier instead.
I wonder if it would be just as effecient to use the 15V supplies - adjust them up to about 16.6V, and run them through a capacitance multiplier?
If I understand correctly, this might reduce the component count a bit.
 
ok, that makes sense.
I read some other threads, where Bo suggested using a capacitance multiplier instead.
I wonder if it would be just as effecient to use the 15V supplies - adjust them up to about 16.6V, and run them through a capacitance multiplier?
If I understand correctly, this might reduce the component count a bit.
That is the currently the method that I recommend at this point but bear in mind that for more than a few 100mA supply (what is the current of each rail?) you will need to use the two transistor version of the CM which incurs a 2 diode drop or ~1.2V. So if you want 15V, you're going to need a supply that can be adjusted up to 16.2V to get 15V out.

However, I don't know if you're going to find a supply like that. First, 15V is a little less common. The most common voltages are 12V and then it jumps to 24V. Another thing is that the adjustable ones tend to be the higher wattage which you cannot use. Again, the power has to be right (what is the current of each rail?) for reasons described in my "SMPS rules". So depending on the current of each rail (what is the current of each rail?) you might need to get 24V supplies in which case you cannot use a CM alone because the output would be 24 - 1.2 = 22.8V which is far to high. So then you might as well use regulators. Therefore, unless you find a small adjustable 15V SMPS) I would just remove the parts from your existing supply that you don't need like rectifier and replace the 6,800uF with like 100uF (See SMPS rules again) and then just keep the regulators and everything downstream from there. I think that could make for a fairly clean mod actually.

Although I would keep an eye open for excess heat. Dropping 24 to 17 is 7V * 2A? = 14W so you'll need decent heat sinks if they're aren't any already.
 
That is the currently the method that I recommend at this point but bear in mind that for more than a few 100mA supply (what is the current of each rail?) you will need to use the two transistor version of the CM which incurs a 2 diode drop or ~1.2V. So if you want 15V, you're going to need a supply that can be adjusted up to 16.2V to get 15V out.

However, I don't know if you're going to find a supply like that. First, 15V is a little less common. The most common voltages are 12V and then it jumps to 24V. Another thing is that the adjustable ones tend to be the higher wattage which you cannot use. Again, the power has to be right (what is the current of each rail?) for reasons described in my "SMPS rules". So depending on the current of each rail (what is the current of each rail?) you might need to get 24V supplies in which case you cannot use a CM alone because the output would be 24 - 1.2 = 22.8V which is far to high. So then you might as well use regulators. Therefore, unless you find a small adjustable 15V SMPS) I would just remove the parts from your existing supply that you don't need like rectifier and replace the 6,800uF with like 100uF (See SMPS rules again) and then just keep the regulators and everything downstream from there. I think that could make for a fairly clean mod actually.

Although I would keep an eye open for excess heat. Dropping 24 to 17 is 7V * 2A? = 14W so you'll need decent heat sinks if they're aren't any already.
Since aiming at +/- 16V, I presume I'm aiming for adjusting it up to 17.2V.
The meanwell LRS 15V supplies' datasheet describes it being adjustable up to 18V, so I suppose that should work?
I've measured the current draw of the rails. At idle, they are around 0.92A each. They can vary a bit , but I haven't seen them go above 0.95A.
(I've tried hooking up microphones and shouting, but it seems pretty stable in that area).
 
Yup. LRS-35-15 sounds like it would be a pretty good fit actually.

But if you're going to go through making a board, don't you want to get 48V sorted out? Technically you should measure that current too but you would have to turn on and load all phantoms to see what it really was. Maybe just calculate 14mA times the number of phantoms. LRS-35-48 might be good. The CM for that could be a little trickier though. You should integrate a shunt regulator so that when only one phantom is on, the SMPS is loaded enough (See SMPS rules again).
 
Back
Top