Modular Power Supplies


#41

I’ve never had any Erica Synth case, so I have no direct experience with them. I did build a bunch of cases using Meanwell PSUs like the R-60B and RT-65B, and some things might apply here as well. From the top of my head the things to watch out are: since Erica uses a Meanwell brick it’s likely that modules like Make Noise Pressure Points and Rene won’t work correctly, since they need a PSU that connects to earth (it’s an easy fix in case, just swap the brick with one that does connect to earth).
The other downside is that you probably have quite a bit of noise and ripple. If you don’t use any dodgy modules, do not need to boost signals either inside the modular or getting out of it and do not need to do do lots of multitrack recording you are usually fine.
I did some tests some time ago using a small skiff and a RT-65B PSU (which is this one: https://www.meanwell-web.com/en-gb/ac-dc-triple-output-enclosed-power-supply-output-rt--65b, so not the same as what Erica uses, but might be similar, but probably worse).
The test was more or less like this (not super-accurate, nor scientifically thorough, but it might give you an idea).
I did patch Braids into the Colour Palette (which I would say is a pretty decent VCA) and the output from the VCA (so from one channel of the CP) to my Presonus Firebox audio interface. I set the volume of the audio to roughly 0Db in the DAW (Reaper). Then I disconnected Braids from the VCA and looked at the peakmeter. I got a noise floor of about -58Db (peaks). Then I plugged the output of the sum of all VCAs into the audio interface and I got -57Db of noise (probably because each VCA is picking up a bit of noise from the PSU). Last thing I did was to patch the output of the CP into a Doepfer 138a linear mixer (which isn’t really intended to be an audio mixer, and even less an end-of-chain mixer), and I got about -54Db of noise.
What does this mean: well basically that even with my really less-than-ideal switching PSU noise seems to be slightly below the audible level. Actually if you make some music with lots of stuff going on I think even -30Db would be hard to hear. Of course ymmw depending on the type of music you make (very quiet ambient is probably a different thing than super-loud dubstep). What could increase the noise floor is using preamp modules, distortion, waveshapers and anything that will apply a gain that goes above unity.

Another common problem situation is multi-track recording. Let’s say you record your modular in multiple takes. I did a test with the previous config, using the output of the Doepfer mixer (just to get a kind of bad-case scenario). I did record silence (just the output from the module without anything patched into it) and copy-pasted this recording onto 10 tracks, so that all of them would play at the same time.
The noise now was -36Db and pretty audible by itself. Now of course this is not a realistic example. When doing multi-track recording you would not record everything with peaks being at 0Db, you’d probably record them much lower. With each track attenuated by 12Db, I get a noise floor of -50Db. But of course if you start to add a bit of compression and eq It might go up a bit. Still, not ideal, but also not really super bad.
I hope this gives you a better idea of what to expect from the Erica Case, even if of course that’s a different circuitry, so it might behave very differently (though as I said, I think it’s probably better than my example here).


#42

What I did forget to mention is that the noise coming from my Meanwell has a distinctive high-pitched whine (it’s not a nice white noise, which would be a lot better imho), which you can hear even if the volume is not super high. I did test by turning the volume down to a pleasant level using the recorded output from Braids (just the steady tone), then listening to the noise. Even at -58Db I could hear it. Of course if I play some music along with the noise I have a hard time hearing it at even -50Db, I did test with a field recording playing at -24Db RMS and I had to really focus a lot to be able to hear the whine.


#43

Thank you so much for such a detailed response. Case noise can be extremely frustrating and I thought given the high specs on this that it’d be fine but you have clearly articulated why that may not be the case. I don’t want to be limited by the noise as to what to put in the case although I don’t plan on any DIY etc but the multitracking is an issue. So you’re saying case noise is not always correlated with load, which I thought to be generally the case from my (one) experience with power issues. (I did know that dodgy modules can cause it and logically multitracking should increase noise.) Thanks for the help


#44

To pick your brain some more, this is the description on the 4u 104hp intellijel power supply. To a laymen like me it sounds like the issue you are saying above.
“30 Watt integrated switching power supply…Low impedance star distribution due to 4 layer PCB”

EDIT: Should this discussion go in the thread re Eurorack power issues? I think there is one.


#45

Yes, totally. I’ve moved the posts!

Now regarding your questions above: afaik load does have an impact, but I couldn’t say how much. Having a very low or a too high load can both have a negative impact on how well the PSU performs, but that probably varies a lot from design to design. I had some PSUs at the beginning that would go all bonkers below 500mA of load.
In general, for what I know, noise is first of all related to the design of the PSU, some have better values, some worse (and actually the Meanwell ones are among the better ones). The second factor is distribution (i.e. the busboards) most experts seem to agree on the fact that the power distribution matters even more than the PSU itself when it comes to noise, especially ground resistance seems to have a big impact (the less the better). Most common busboards are a “daisy chained” topology, i.e. all the connections are in series, the Intellijel uses a “star” configuration, so each connector on the busboards goes directly to the PSU. The latter is better.
The Intellijel PSU, like many of these newer ones (like the TipTop Zeus Studiobus, the Make Noise one, the Doepfer PSU3 etc.) do work with a combination of switching (the external brick) and linear regulation (the PSU on the busboards, or inside the case). These are called hybrid PSUs and to me these seem like a very good compromise in terms of performance, cost and reliability. So the Intellijel PSU seems much better than the Erica synth one because it does have a linear stage following the switching one, does have a thicker ground bus on the busboards (4 layer) and does use a star distribution (while the Erica one has the connectors in series).
For clarity, here’s more info on my setup from above: I am using a standard open frame switching PSU from Meanwell, which connects directly to the mains (and does convert 220V to +/-12V and +5V). The output from that PSU goes into a simple busboard with a 1 layer ground bus and connectors arranged in series. I did add some redundant connections between the ground bus and protective earth using thick low-resistance wire and that has improved things a bit, but it’s still not super optimal.
Also I should emphasize that the Meanwell I have used is a completely different thing than what Erica is using so it’s only to give you a rough idea for what to watch out and have something more concrete to explain things.


#46

Thanks again for the detailed response. This is one reason why this place is so great - people sharing their knowledge with learners like me. Hopefully I can share something I know with you soon.
I was aware of the benefits of star grounding from a little knowledge about guitar amps but had not put together the cumulative effects of distribution design. It makes sense when its laid out clearly.
Thanks again, I’m sure I will be going with an intellijel case at some point and now will do so with confidence!


#47

You’re welcome!
If I were you I’d ask some people about their experiences. The topic of noise&ripple seems to be a rather subjective (and polarizing) one but the more info you get the better you can decide what’s right for you :slight_smile:


Eurorack cases/racks. Ask questions, share experiences
#48

I did another quick test with the other case, a 104x9U system with 3x TipTop Zeus Studiobus PSUs (and the official Cincon brick). The case is filled with mostly Mutable modules. In general all modules I have in there, I totally trust in terms of circuit design. There’s one module whch I know sometimes to leak into the ground bus (it’s a very low volume tone that comes from the circuit). I’m using a WMD performance mixer as the end-of-chain mixer.
The TipTop PSU is a hybrid design, with a switching brick and a linear regulation stage on the busboards. The busboard is a “daisy-chain” type. The test was more or less like with the other case: use an oscillator to determine the amplification, then see how much noise I get with no sound coming out of the mixer (i.e. master volume stays up, but the channels are all muted). With the oscillator playing at -12Db, the noise floor was at about -80Db, which interestingly was only a 10Db improvement when compared to the Meanwell (which was at -70Db with the raw VCO playing at -12Db) in the other case. Still, to the ear it’s a pretty big difference since the noise is more like white noise (while the Meanwell has some spikes in the spectrum, which makes it easier to hear it) and hence is pretty much inaudible also when getting louder in a multi-track recording situation. I was actually expecting it to be better than that, but one also has to take into account that it’s a much bigger system.


#49

Not quite.

There are switching supplies on the bus boards also, which are followed by linear regulators. The tipoff that you are dealing with a switching supply of some kind inside the case is that you have a single DC voltage coming from an AC adapter: you have to generate a -12V rail from that somehow. Some supplies (like Tip Top uZeus) use a linear regulator only on the +12V rail and use a switcher to get the -12V rail.


#50

Not sure I understand what you are trying to say with this sentence:

Apart from that, there’s all sorts of power supply designs and by no means I was trying to make a complete list of them.

Btw. I did inspect the Erica Synth PSU we were discussing above a bit closer (well the photo on their online shop). There’s a small SMD component on the lower and upper part of the PCB close to the Meanwell box which looks like a regulator. So maybe they have added an additional filtering stage there to smooth things out.
Maybe that’s what you are saying?

The PSU I have is not the uZeus, but this one, just to avoid confusion: http://www.tiptopaudio.com/zeussb.php
As far as I know the regulation stage on the busboards is linear.
The external brick is a 15V DC not an AC one.


#51

Yes, I only referenced the uZeus as an illustration of an alternate design.

The board you have must have at least one switching supply on board or you would have no way to generate a -12V rail from a +15V DC input. But the Zeus boards have always had switchers for both rails AFAIK.

If you have a single DC voltage in the 15V range going into your case, you must have some kind of switching supply inside the case. If you’re not sure why this is true, I suggest you reconsider offering technical advice on this topic–and I mean this in all sincerity. :slight_smile:


#52

Yes absolutely. As I said, there’s not one design that you find everywhere, all PSUs seem to be different and calling something linear or not is often a generalisation. I guess you can either use a switcher to derive the -15V from the 15V coming from the Cincon, or you first regulate down to -12V (for example with a linear regulator) or you use an inverting switcher to get the -12V from the 12V. Do you know how TipTop solved it here? I didn’t really investigate that.


#53

Looking for opinions on a power issue with my monorocket case power. I’m pretty sure I’ve maxed it out I’m at about 1.2 amps on the + rail and according to another forum the original V1 revA power supply was rated at 1A on the + rail. So I’m left at a bit of a crossroads. Do I sell the Monorocket with the original power supply intact, and replace with a case that can support my needs (looking to go up to 2A at least just to be safe) or do I replace the power supply in the monorocket with something else. Row Power, Studio Zeus, etc? I’ve also been looking at the Trogotronic supplies as a low cost high power way, does anyone have exp with those?

Anyway looking for general thoughts on what someone else might do in my situation.


#54

question:
i’m upgrading from uzeus to row power 40 and i was hoping to just unplug my bus cables from one and plug them into the other. when i unplugged from uzeus i noticed these double-header cables, and it doesn’t look like they’ll be compatible with row power 40:

do both of these headers need to be plugged in (in which case i can’t use them)? what is the purpose of this design?


#55

is there any labelling on the uzeus that might be helpful? certainly doubling the headers can’t double the number of conductors on the ribbon cable…


#56

the back of the board is pretty vague. i had to leave for work so i’m no longer with the system, but if i remember correctly both headers were the same, but facing opposite directions.


#57

update: it looks like people on the other forum “chopped off” one of the headers without issue, but they’re pretty light on details

does anyone here have experience with this? or have an idea why this would work/be stable?


#58

I did this for my row40. I just cut off the end connector and plugged in as usual. So far no problems at all :smiley:


#59

you cut the one on the end? the manual has conflicting info:
in this example of daisy chaining the end connector is used:

but in this example of adding a third bus board it looks like the inner connector is used:

maybe i’m overthinking this? i just don’t want to fry anything!


#60

(imo this should work bc the connector itself is just allowing pins to touch the wire in the ribbon cable. since one header already covers every conductor on the cables, my hunch is that the second header is like any other header on the cable, so using it or not should make no difference. I’m not an expert by any means though)