Meanwell Modules - AH ML5000 PSU Replacement

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Blissy

Well-known member
GDIY Supporter
Joined
Sep 5, 2020
Messages
67
Location
Australia
Hi All,

Looking to replace the PSU for my Allen and Heath ML5000 using Meanwell modules.
The Console Requires;

+17 - 7A Max
-17 - 7A Max
+12 - 5A
+48 - 500ma

The Modules I am currently looking at are;

2 LRS-100-15 for the +-17 volts rails
LRS-17-12 for the 12v Rail
RS-25-48 for +48v

The LRS-100-15 are Adjustable from +13 - +18
Looking for some advice regarding filtering the outputs of these modules. What topology would be most effective and best way to implement it? (if necessary)

Anyone one who has tried something like this it would be great to hear any advice and results!

Cheers
Blissy
 
That should work.

Here's basically what you need to do:

SMPS_console_supply1.png


So at earth there is a FAT wire to the chassis. Then, also from the earth terminal, a FAT wire to the "star" ground. In practice the "star" ground should just be a PCB ground plane so that each ground wire connected to it is surrounded by ground plane. So there are no "traces". It's all just ground pour. You can also pour it on top and add vias to stitch them together. If you're not making a PCB, you could use a small piece of bus bar from an electrical supply store.

A few details are:

1) Use a hefty power switch to handle the surge currents of the SMPS. I'm not sure if a common 20A toggle is enough. Check the datasheets. Maybe an SPDT doubled up makes 40A? Or maybe there's some sort of "circuit breaker" type of switch that doubles as a power switch? Note that you probably cannot solder the terminals because they'll be so big. You might have to crimp some spades and get a switch with screw terminals.

2) Use fat wire for everything connected to LNE. Maybe 8 or 6 AWG? Again, your iron might not be able to solder that such as the earth wire to the ground plane for example in which case you might need to crimp a spade and screw it down somehow. Or use a bus bar.

3) Capacitors need to be rated for voltage of course but they don't have to be as big as they are in a linear supply (and should not be because SMPS sometimes don't like that). They just need to be big enough to get the high frequency ripple. The best way to check would be to just load test one and look at the ripple with a scope.

4) Inductors should be the largest inductance you can find at the necessary current. So search Mouser for something rated at 100% higher. So at 14A, the inductance is going to only be so high. Maybe 10uH or something like that. Above that, you're taking tube amp chokes for $100. But again, the inductance only needs to be enough to get close to 20kHz with the cap that follows.

Post picks of the enclosure. Presumably you're just going to recycle the exisiting one with the connector and mains in an such? Maybe reuse the power switch?

Does the exisiting supply work at all? If yes, and you have a way to measure current (measure voltage across resistors and use ohms law), it would be nice to know how much the console is really drawing. An old supply is probably not nearly as efficient as the new MW supplies so it's probably over-rated and it's not pulling anywhere near 7A per rail. But it would be good to know what it really is. The 12V line will probably vary greatly depending on how many relays and LEDs are engaged at any particular moment.

I'm sure I'm leaving out something important but if you post pics and part numbers other people will chime in with good ideas.
 
squarewave said:
That should work.

Here's basically what you need to do:

ML5000_SMPS.png


So at earth there is a FAT wire to the chassis. Then, also from the earth terminal, a FAT wire to the "star" ground. In practice the "star" ground should just be a PCB ground plane so that each ground wire connected to it is surrounded by ground plane. So there are no "traces". It's all just ground pour. You can also pour it on top and add vias to stitch them together. If you're not making a PCB, you could use a small piece of bus bar from an electrical supply store.

A few details are:

1) Use a hefty power switch to handle the surge currents of the SMPS. I'm not sure if a common 20A toggle is enough. Check the datasheets. Maybe an SPDT doubled up makes 40A? Or maybe there's some sort of "circuit breaker" type of switch that doubles as a power switch? Note that you probably cannot solder the terminals because they'll be so big. You might have to crimp some spades and get a switch with screw terminals.

2) Use fat wire for everything connected to LNE. Maybe 8 or 6 AWG? Again, your iron might not be able to solder that such as the earth wire to the ground plane for example in which case you might need to crimp a spade and screw it down somehow. Or use a bus bar.

3) Capacitors need to be rated for voltage of course but they don't have to be as big as they are in a linear supply (and should not be because SMPS sometimes don't like that). They just need to be big enough to get the high frequency ripple. The best way to check would be to just load test one and look at the ripple with a scope.

4) Inductors should be the largest inductance you can find at the necessary current. So search Mouser for something rated at 100% higher. So at 14A, the inductance is going to only be so high. Maybe 10uH or something like that. Above that, you're taking tube amp chokes for $100. But again, the inductance only needs to be enough to get close to 20kHz with the cap that follows.

Post picks of the enclosure. Presumably you're just going to recycle the exisiting one with the connector and mains in an such? Maybe reuse the power switch?

Does the exisiting supply work at all? If yes, and you have a way to measure current (measure voltage across resistors and use ohms law), it would be nice to know how much the console is really drawing. An old supply is probably not nearly as efficient as the new MW supplies so it's probably over-rated and it's not pulling anywhere near 7A per rail. But it would be good to know what it really is. The 12V line will probably vary greatly depending on how many relays and LEDs are engaged at any particular moment.

I'm sure I'm leaving out something important but if you post pics and part numbers other people will chime in with good ideas.


Thank ever so much Squarewave for your detailed answer. I have attached schematic below.
The Inductors I am looking at are;
5uh rated at 14A
https://www.digikey.com.au/product-detail/en/hammond-manufacturing/1538M08/HM3350-ND/454701

So if I am correct, a 5uh with a 10uf Cap should give me at cut of frequency of about 22kHz.
Using a Smaller 10uh inductor rated at 1amp on the 48v.

I am indeed planning on reusing the existing case, Mains input and Switch. Currently trying to get existing PSUs into a working state to verify current draw under load.

Any feedback is greatly appreciated before doing a layout. (please forgive messy ground Point on schematic)
 

Attachments

  • ML5000 PSU REPLACMENT [FOR REVIEW] 200926.pdf
    22.1 KB · Views: 15
Hammond parts are kinda overpriced. Also, 2x overspec is probably a little too conservative. It seems digikey has a "current saturation" field which sounds pertinent (and the Hammond part is not clear about). So if you do 1.5x that makes it a 10A part and then you could get over 10uH easy. Then I like to sort by price and look for something that a) they have a lot of (or just something that is not obscure or exotic) and b) has a footprint that is common across multiple manufacturers (and not obscure or exotic). So if I just do a quick search, I don't see any reason why you can't use something like this:

  https://www.digikey.com.au/product-detail/en/bourns-inc/SRP1265C-220M/118-SRP1265C-220MCT-ND/11633810

So that's 22uH, shielded, smaller and 1/4 the cost. But if you want to use through-hole (are you doing a PCB?) there are similar parts (but clearly SMD rules this space).

One thing to consider is that the switching frequency of the LRS-100 is only 47Khz. So if you have a 2 pole filter at 20kHz, it's not going to attenuate the HF ripple that much. If you go with a bigger inductor like 22uH and a bigger cap like 100u, that's going to attenuate 47kHz by like 40dB+.

However, be on the lookout for resonance in the filter. The Q of these inductors isn't that high and the impedance of the SMPS and chokes is not zero so it probably isn't a problem. But ideally I would prefer to load test one of the 100W SMPS with a giant 4 ohm 100W resistor (briefly) and look at it with a scope to sanity check it. Actually what I would really do is add a cap to the load resistor and "look" at the noise floor with an audio analyzer to make sure there isn't a bump somewhere in the noise floor. I have never had a problem with the MeanWell supplies but you don't really know how one performs until you just try it and study it.
 
squarewave said:
Hammond parts are kinda overpriced. Also, 2x overspec is probably a little too conservative. It seems digikey has a "current saturation" field which sounds pertinent (and the Hammond part is not clear about). So if you do 1.5x that makes it a 10A part and then you could get over 10uH easy. Then I like to sort by price and look for something that a) they have a lot of (or just something that is not obscure or exotic) and b) has a footprint that is common across multiple manufacturers (and not obscure or exotic). So if I just do a quick search, I don't see any reason why you can't use something like this:

  https://www.digikey.com.au/product-detail/en/bourns-inc/SRP1265C-220M/118-SRP1265C-220MCT-ND/11633810

So that's 22uH, shielded, smaller and 1/4 the cost. But if you want to use through-hole (are you doing a PCB?) there are similar parts (but clearly SMD rules this space).

One thing to consider is that the switching frequency of the LRS-100 is only 47Khz. So if you have a 2 pole filter at 20kHz, it's not going to attenuate the HF ripple that much. If you go with a bigger inductor like 22uH and a bigger cap like 100u, that's going to attenuate 47kHz by like 40dB+.

However, be on the lookout for resonance in the filter. The Q of these inductors isn't that high and the impedance of the SMPS and chokes is not zero so it probably isn't a problem. But ideally I would prefer to load test one of the 100W SMPS with a giant 4 ohm 100W resistor (briefly) and look at it with a scope to sanity check it. Actually what I would really do is add a cap to the load resistor and "look" at the noise floor with an audio analyzer to make sure there isn't a bump somewhere in the noise floor. I have never had a problem with the MeanWell supplies but you don't really know how one performs until you just try it and study it.

Thanks again Squarewave, This is all great! I am doing a PCB so I suppose there is no reason not to use SMD! As you mention there is a much larger variety of Inductors for SMD.
Yes I am doing a PCB working on it now will upload a Draft for review. Would be great if you or others could have a look over it!

Thanks
Blissy
 
Quick Update
Just measured the Current Draw on the +/- 17v Rails to Find 6.5amps On each. Which is higher than expected However the Console is Massive 54 Channel....
My guess is the original PSU was designed to work with a range of consoles including the ML3000 and ML4000 Which are smaller frame sizes. The 5000 really pushing it to its limits.
Wondering If I should use Larger modules ?

Cheers
Blissy
 
Apologies for hijacking this thread but I have a question that I think is relevant to high current draw SMPS. I use them now for the 12V heaters supplies on all my tube mixers. So far with the mixers with a modest number of channels I have had excellent results. As a rule I over rate the SMPS by 100% to allow for inrush current into cold tubes.

My question is this. I am now building a particularly big tube mixer. I have been checking the noise performance of the mic pres. I initially I had just half a dozen modules fitted. The I started adding a bunch of line amps that would be used for bus amps and outputs. With these plugged in I noticed the mic pre noise got worse. Looking at its spectrum I found this was due entirely to an increase in the 100Hz component. With 6 line amps plugged in it increases by a out 10dB. At first I thought thsi was because I had not yet upgraded the HT supply but then I realised I had not connected up the HT supply to the line amps so there was no increase in HT current. So I started with no line amps and added them one at a time. There is a small increment in 100Hz component  of the mic pre output noise each time you plug in another line amp.

I have not actually got round to looking at the heater voltage itself on the scope as I have been busy doing other things. But I have been thinking of adding additional filtering to the SMPS so any inductors used would need to be capable of carrying 10 amps or so, hence my interest in this thread. What puzzles me is the effect is at 100Hz, not at lots of KHz. Obviously, the raw dc input to the switcher is just rectified mains so it will have a substantial 100Hz component. What I am unclear about is how much of this will reach the dc loutput?

Cheers

Ian
 
Blissy said:
Quick Update
Just measured the Current Draw on the +/- 17v Rails to Find 6.5amps On each. Which is higher than expected However the Console is Massive 54 Channel....
My guess is the original PSU was designed to work with a range of consoles including the ML3000 and ML4000 Which are smaller frame sizes. The 5000 really pushing it to its limits.
Wondering If I should use Larger modules ?
Mmm, I didn't expect that. Yes. You should go to LRS-150 to get 10A. Those modules are designed to run at their limit but there's no reason to walk the edge. And if you trim up to 17V it seems to me the amps must drop accordingly.
 
squarewave said:
Mmm, I didn't expect that. Yes. You should go to LRS-150 to get 10A. Those modules are designed to run at their limit but there's no reason to walk the edge. And if you trim up to 17V it seems to me the amps must drop accordingly.

Good Idea. Here is my first revision of the PCB including updated schematics. It has been a while since I've had to do a PCB and I am by NO means an expert. Any feedback and advise would be hugely appreciated.
(I have not updated module names to LRS-150)

Cheers
Blissy
 

Attachments

  • PCB and Schematic.pdf
    694.3 KB · Views: 41
ruffrecords said:
I have not actually got round to looking at the heater voltage itself on the scope as I have been busy doing other things. But I have been thinking of adding additional filtering to the SMPS so any inductors used would need to be capable of carrying 10 amps or so, hence my interest in this thread. What puzzles me is the effect is at 100Hz, not at lots of KHz. Obviously, the raw dc input to the switcher is just rectified mains so it will have a substantial 100Hz component. What I am unclear about is how much of this will reach the dc loutput?
That's interesting. I have always used relatively small supplies. The largest that I've used was 1A. I have seem very low to zero mains feed through. It is very easy to seem mains for any number of reasons the most likely being that your test rig is not isolated enough or it's injecting mains into the input such as from a USB audio interface. But maybe I'm just not seeing it because of some averaging going on. The switching transformer thuroughly isolates the output from mains. But it's possible that there's some AC coupling somewhere. Maybe the design is a little different with the bigger supplies and under load there is some mains feed through. That would be very very disappointing as filtering the mains is difficult.

If you were not so experienced my first reaction would be that it's it's EMI from the heater wires getting into the high impedance mic pres or a grounding issue. I'm sure you appreciate how easy it would be to screw that up. I would *expect* that to happen to me and require some experimenting to get right.

One thing that might help narrow down is load testing. If you have some giant power resistors you could put the thing under 80% load and look at the noise with an analyzer (cap -> USB audio interface -> software). If there's mains under high load, I would think you would see it then.
 
Blissy said:
Here is my first revision of the PCB including updated schematics.
Ha, ha ha ha ha. No! You're hilarious. Are those the little green connectors with screws? Uh, ... no.

I actually don't do high current designs so I don't know what the right parts are without researching it. But I can tell right now that you're brain is not in the right place for this. You need to be very careful and think about what you're doing.

What do you think is going to happen when you put 10A through those little thermals?
 
squarewave said:
Ha, ha ha ha ha. No! You're hilarious. Are those the little green connectors with screws? Uh, ... no.

I actually don't do high current designs so I don't know what the right parts are without researching it. But I can tell right now that you're brain is not in the right place for this. You need to be very careful and think about what you're doing.

What do you think is going to happen when you put 10A through those little thermals?

hahaha  :-[ Well I was going to solder wires straight to the Through hole. (as this is how it was done on the original PSU) However looking up the specs of those Screw Terminals 5.08mm spacing they claim to be rated at 10A DC

I guess that is not going to fly. I assume you are also talking about the trace width as-well??
 
ruffrecords said:
Apologies for hijacking this thread but I have a question that I think is relevant to high current draw SMPS. I use them now for the 12V heaters supplies on all my tube mixers. So far with the mixers with a modest number of channels I have had excellent results. As a rule I over rate the SMPS by 100% to allow for inrush current into cold tubes.

My question is this. I am now building a particularly big tube mixer. I have been checking the noise performance of the mic pres. I initially I had just half a dozen modules fitted. The I started adding a bunch of line amps that would be used for bus amps and outputs. With these plugged in I noticed the mic pre noise got worse. Looking at its spectrum I found this was due entirely to an increase in the 100Hz component. With 6 line amps plugged in it increases by a out 10dB. At first I thought thsi was because I had not yet upgraded the HT supply but then I realised I had not connected up the HT supply to the line amps so there was no increase in HT current. So I started with no line amps and added them one at a time. There is a small increment in 100Hz component  of the mic pre output noise each time you plug in another line amp.

I have not actually got round to looking at the heater voltage itself on the scope as I have been busy doing other things. But I have been thinking of adding additional filtering to the SMPS so any inductors used would need to be capable of carrying 10 amps or so, hence my interest in this thread. What puzzles me is the effect is at 100Hz, not at lots of KHz. Obviously, the raw dc input to the switcher is just rectified mains so it will have a substantial 100Hz component. What I am unclear about is how much of this will reach the dc loutput?

Cheers

Ian
Is the noise showing up in each mic pre output?

Obvious suspects are PS loading, or perhaps clean 0V corruption.

JR
 
JohnRoberts said:
Is the noise showing up in each mic pre output?
It is. Also independent of slot.
Obvious suspects are PS loading, or perhaps clean 0V corruption.

JR
PS loading is well within limits although the current rating of this SMPS is higher than I have used before.
One side of heater is connected to 0V only in the external PSU so there should be no real opportunity for 0V corruption.

Cheers

Ian
 
ruffrecords said:
It is. Also independent of slot.PS loading is well within limits although the current rating of this SMPS is higher than I have used before.
One side of heater is connected to 0V only in the external PSU so there should be no real opportunity for 0V corruption.

Cheers

Ian
Switching supplies that are very lightly loaded can skip cycles causing noise, but that does not track with your symptoms (increasing with more modules).

Have you tried externally loading the PS to say 50%

JR
 
Blissy said:
hahaha  :-[ Well I was going to solder wires straight to the Through hole. (as this is how it was done on the original PSU) However looking up the specs of those Screw Terminals 5.08mm spacing they claim to be rated at 10A DC

I guess that is not going to fly. I assume you are also talking about the trace width as-well??
You need to figure out what guage wire you're going to use. Everything depends on that. You're going to probably want 2 sizes. Everything connected to LNE upstream from the SMPS needs to handle surge currents so it should be super fat. Think wiring in your house. Look at datasheets for SMPS with wires included and look at the AWG they used. Then increase as necessary to account for all 4 SMPS. My guess is it could be 10 AWG or something like that. For downstream from the SMPS the wire can be much smaller. Maybe 14 AWG or 16 AWG. Something like that. But that's still chunky enough that you could have trouble soldering it. You will want to get ride of the thermals in your current layout. But if you do that, a little pencil tip iron could have trouble with that. Maybe 16 AWG with a little flux and patience it could work.

Just make sure everythign upstream of the SMPS is really solid. Obviously there's no protection there.

If you post pics of the inside of the existing supply with close ups of the upstream wires / connections, that would give us a really good idea of what is needed. Also post a pic of the back with all of the connections and such.
 
Thanks Again, I have attached a few images of the internals of the PSU input and Output section as-well as the connector. The wire on the output section is 16AWG for reference (Second Image Pg 1)

Stated inrush current from data sheets
15v SMPS INRUSH CURRENT (Cold Start) 60A
12v SMPS INRUSH CURRENT (Cold Start) 50A
Wow!
 

Attachments

  • Old PSU.pdf
    755 KB · Views: 40
Actually just reuse those wires. Keep the connector PCB, ground lift switch, the ferrites around the wires and the mains PCB. Use those solder spade terminals. Those are keystone parts. Or just harvest them. Reuse the power switch which isn't in your pics but maybe it has those spade terminals? Then you can just plug in your replacement board. Just copy that tech. Make the mounting holes line up. Etc.

As for inrush currents, I don't think you literally need a 60A switch. Switches can handle brief moments of higher current. You are replacing an SMPS with an SMPS so all of the exsiting stuff that you're reusing should be fine.

Are there any fans? The new SMPS might be more efficient enough that you could ditch the fans. Think about it like a couple 100W light bulbs but most of that power is dissipated outside the enclosure.
 
squarewave said:
Actually just reuse those wires. Keep the connector PCB, ground lift switch, the ferrites around the wires and the mains PCB. Use those solder spade terminals. Those are keystone parts. Or just harvest them. Reuse the power switch which isn't in your pics but maybe it has those spade terminals? Then you can just plug in your replacement board. Just copy that tech. Make the mounting holes line up. Etc.

As for inrush currents, I don't think you literally need a 60A switch. Switches can handle brief moments of higher current. You are replacing an SMPS with an SMPS so all of the exsiting stuff that you're reusing should be fine.

Are there any fans? The new SMPS might be more efficient enough that you could ditch the fans. Think about it like a couple 100W light bulbs but most of that power is dissipated outside the enclosure.

Thanks!

OK here is my second attempt taking in to account all the advise given.
The two inductors used are
BOURNS SRP2313AA-470M for +17 -17 and +12
WURTH ELEKTRONIK 7447789147 for +48
I can easily source both from my normal supplier.
I revised the PCB layout and added some LED indicators to replace the ones on the existing supply.
The traces are 5mm thick and all connections are PCB mounted spade terminals that will connect to the existing wiring loom (16AWG) and 6 Pin Hirose connector.
Looking at the schematic there is a 100 ohm resistor between AGND and DGND I assume to reduce possible noise. Should I implement something similar?

There are fans in the existing supply. I will integrate them.

included also is a rough layout for the existing case.
I am sure I have made mistakes, any advice is greatly appreciated.

Sharing Via Google Drive to avoid multiple upload posts

https://drive.google.com/drive/folders/1Dp_4JAiCEoobhb72Us4U056dN83dUNpd?usp=sharing

Cheers
Blissy 

 
That all sounds good but you need to get rid of the thermals. Thermals are the little spokes around each hole for a part that connect to the trace / plane. The help you solder. Without thermals all of that copper wicks away the heat of your soldering iron and prevents the solder from flowing into the joint properly. But in this case you really need that copper. It doesn't really matter if you have giant traces if they then turn into 4 little spokes to get to the bus wire of the part. That's like making a 4 lane interstate highway that turns into a bicycle path for 100 feet. So I would dig around in your layout software and figure out how to turn those off. It will make them hard to solder. If the solder doesn't flow into the joint properly, you will need to get a bigger soldering iron. Usually people have one cheapo thing from the 80's that's like 50 watts just for this sort of thing. But a flux pen and some patience might also be good enough.

Although, is there no way to lay all of the SMPS out flat? Or at least the two 17V and then put the 48V over the 12V. You can put them right up against the edge of the enclosure no problem. And the PCB can be mounted anywhere. Maybe make it long and narrow so that it can be mounted vertically would help?
 
Back
Top