Modular Power Supplies

Speaking as a designer, not someone troubleshooting specific power issues, I wonder if DSP firmware can be designed for a more gradual power up.

The ARM Cortex-M chips that I’ve developed on have individual power enables for each internal peripheral. Of course, it’s common to immediately turn on every peripheral interface that’s needed, but it seems like it would be possible to have a time delay in the firmware startup such that the inrush current would not be so drastic.

I guess the real question is: How much of a problem are DSP modules and initial power draws? If it is a significant problem, then it might be worth pushing manufacturers to update their firmware. We humans won’t notice a few extra milliseconds - even hundreds of milliseconds - at startup, but the power supply would!

Hardware makers: throw more hardware at it!

3 Likes

That works for analog modules. However, I’m fairly certain that a digital module that needs a lot of current at the start will freak out with soft start. Instead, it’s necessary for the digital firmware to control the current draw on the “load” side rather than have soft start on the “supply” side.

Something like this was happening with the Teensy 3.6 when I put it in the Teensy 3.2 slot in my TXo module. In order to use the 3.6 (and get the extra horsepower), I had to make a little daughterboard with a voltage supervisor that delayed the start of the Teensy using the reset pin. Not all Eurorack power supplies caused this, but enough did that it was critical to implement the workaround. (It was maddening until I found the thread over at the teensy forum and the solution suggested by user neutron7 - J.Matheson, creator of the Orgone Accumulator.)

Here is the silly little board I made for the workaround:

sorry, I didn’t fully explain the concept. since most modules already have a 2x5 male connector, each 1x5 row could accept a 1x5 cable - only one 1x5 row is needed to power Module A, so a second cable could connect the unused 1x5 row to Module B to carry power down the line in a marginally more graceful and easy to rearrange manner.

I’ve made plenty of my own daisy chain cables and am using 2 or 3 now, but in many situations it is hard to connect various modules together due to the lack of consistency in power header placement/orientation and I end up needing to make new cables to connect different placements of modules when I swap out a module or rearrange my case. sometimes the modules that I want to chain aren’t next to each other - I try to isolate my digital modules from analog to avoid noise issues, for example, so my daisy chains sometimes have to leapfrog other modules on a separate power cable due to where I want them ergonomically.

granted, some completely reasonable solutions are to simply use such daisy chain cables, passive bus boards, or flying bus boards to split power when one has plenty of power but not enough sockets, but I could imagine a use for a lower-profile solution to try to make a case as slim as possible.

also, there is something that irks me from a purely conceptual standpoint about the design decision to use the 2 row ribbon cable for Eurorack power and this felt like an idea which could, at least in concept, be a more logical solution (removing the redundancy of two separate, side-by-side wires carrying the same power/ground/signal to the same place) while remaining backwards compatible…but I also acknowledge that there likely isn’t much appetite for a widespread change at this point.

in any case, using a single 1x5 row to power 1U and small 3U modules (2hp and similar) does offer the advantage of taking up much less space on a PCB.

I readily admit this is a nuanced and sort of stupidly specific issue! I’m also thinking of this in part as I would like my next case (if I end up building or buying another at some point) to be as low profile as possible, and minimizing excess volume taken up by power cables and power supply PCBs would help with that.

It’s certainly a todays reality but unless your are running at the limit of your PSU - I don’t think it’s a huge problem. Only few modules are known to be really problematic in this regard (I’m looking at you, Roland Aira range). People are hacking their power supplies with additional capacitors to ease the peaks. But sure, if you are hardware designer and can come up with better power rails that survive whatever you can throw at them - that would be awesome. :ok_hand::wave::clap::yum:

I was thinking more along the lines of designing the modules themselves to draw power more gradually during startup. After startup, if the power draw is too great for the power supply, then you simply need a bigger power supply. I might have gotten the wrong impression, but I thought that one of the problems mentioned was modules that draw a huge, but temporary, in-rush current.

1 Like

Jumping in on this bit of the thread, I am also interested in solving a power / ground problem.

I have a Rene in a DIY skiff which is not responding very well to touch unless I actively hold on to one of the rails with my other hand (one of the workarounds recommended by MN as I understand it).

While that does work for interaction, the other issue I see is that there appears to be ghost triggers occurring, as if I were touching the key pad, even when I’m not - a built in source of uncertainty right there I guess but not quite what I’m liking for :slight_smile:

The skiff is powered by a Synthrotek PWR module (not the Deluxe Power, but the earlier DIY model), which came to me with a bunch of other power parts, including a power brick that seemed a bit questionable.

I got a Meanwell GSM60A12-PJ1 to replace the random power brick, and I can see on the specs list that this as has -ve connected to the AC frame.

The new power brick hasn’t improved the Rene’s behaviour on it’s own. Trying other power outlets hasn’t improved it either.

I could try grounding a rail with a wire to something large, instead of having to hold on to it myself.

I’m up for replacing the PSU or replacing flying bus cables with a better bus board but Rene seems to be a fussy enough module that I could go through a few things that may not make any difference, so I guess I’m trying to understand the issue better to start with. It seems to be s common enough issue, but I haven’t yet found much that helps me understand the situation.

If there’s any good resources that people here can point me to, I’m all ears.

keep eliminating the variables. can you try the module in another case? i’d suspect your power module itself.

For sure. I have tested with a 4MS RowPower 40 in place of the Synthrotek and had similar issues with Rene, even though other folks report no problems with a RowPower. My next step is to replace the flying bus cable with a proper bus board, though I have learnt on this thread that not all bus boards are equal. I’ll also try some kind of grounding arrangement from the rail in place of myself, and will report back.

At the end of the day there seems to be lots of variables. I feel like I already have a functional “workaround” - it’s just not very ergonomic. But then again neither is a cello and I learnt to play that back in the day :slight_smile:

It’s all a process of learning so I don’t mind encountering a few challenges along the way.

I would connect it with ground of the PSU. Would that improve things? Not sure about this with switching PSU…

In the Eurorack world there seems to be quite a few options for powering 5V modules from PSUs that don’t provide it natively.

I have the opposite problem though. My new case has a pair of Mean Well RT65Bs for power. Here’s where I’m at according to ModularGrid:

+12V: need 2311mA, PSU provides 2800mA x2
-12V: need 958mA, PSU provides 500mA x2
+5V: need 20mA, PSU provides 2000mA x2

I also have one VCA with unpublished power requirements, which worries me.

Very little margin on the -12V rail, and yet I’ve got almost 4A of +5V power doing nothing. It’d be nice to convert that and power a couple of modules off it to ease the burden. That doesn’t seem like it should be expensive or difficult, and yet…

Current and voltage are not remotely the same thing. What you have is excess power capacity on the 5V rail (4A @ 5V = 20W). You would need a second power supply to utilize this excess capacity and “convert” it to 12V, which would get you ~1500mA minus the conversion losses, which can be significant depending on the design. It would also increase the load on the main PSU and you’d be running it really close to full-bore at that point. This is not optimal - most power supplies last longest and provide best efficiency around 75-80% total load.

Stepping 5V up to 12V would also require a boost-mode DC-DC switcher, which can be electrically noisy. While I’m not intimately familiar with the options available for modular, I would suggest it might be a better idea to get either a more capable primary supply or a second 12V supply rather than try to utilize the 5V rail as the input to another supply.

I don’t know much about electronics but I do know that much at least. :grin:

I have an $8 cable that provides a max of 800mA of 9V power from 5V USB power, which I use for my Volca Modular. So I assumed this sort of thing must be fairly easy to do for small amounts of current. But the couple of options I see out there for powering one or two Euro modules from USB aren’t so cheap.

I’ve got my old Synthrotek PSU which I can keep using if I have to, I suppose.

1 Like

I would suggest, in all honestly, that that cable is massively overspecified compared to what it can actually provide.

My experience with cheap chinese power supplies is they’re good for ~1/2 their rating at best, and well under that in reality if you care about true regulation on the output rails. They might hit their rating briefly before vaporizing themselves or they may massively exceed it but sag horribly - all depends on the model.

Plus, that must assume a high current USB power supply, because normal USB cannot provide more than 2.5W (500mA @ 5V nominal) and the spec for that is 7.2W (800mA @9V)… you can’t produce more power from less, no matter what magic you do with the volts and amps individually!

Those little units are tiny boost-mode DC switching supplies, rather inefficient (usually ~50% or so max, in reality, again disregard the published specs and measure it yourself!), and not capable of providing regulated power with any significant fraction of their output current. I have a few myself that are likely similar to yours - they don’t test out very well on the bench! But for tiny loads like the Volca which have their own down-regulator internally (I think they run at 3.3V inside) they work just fine since sagging 9V is still well within the tolerance for a 3.3V supply rail. This doesn’t work out well for modular stuff that requires clean, non-sagging 12V and -12V rails.

Boost-mode switchers require a set of capacitors that are charged in parallel by chopping the DC into a rough AC square wave and discharged in series. In order to get any serious amount of current through them with high efficiency you need a carefully balanced switching sequence and large, high quality capacitors. This is why you don’t see a lot of cheap DC-DC boost converters with higher spec ratings that actually meet their ratings.

4 Likes

Ok, this explains why it seemed like there was a gap in available Eurorack accessories then :slightly_smiling_face:

I think I can probably reduce the -12V power draw enough to work with the RT65Bs. I can drop the Crossfold and make some of the changes to VCAs and pedal interfaces that I wanted to make anyway.

1 Like

Do people have experience daisy-chaining power from power supplies that aren’t necessarily designed for that purpose (as opposed to, say, 4ms Row Power, which is)? I have a Doepfer 3U suitcase with a PSU2, which seems like it should have enough juice for a small control skiff in addition to those 3U.

I’ve been eyeing the Koma Strom passive module, which appears to just plug in with a ribbon cable like any other module and outputs/receives power through a P4 connector. Seems like all I’d need would be two of those, one in the suitcase and one in the skiff, plus a cable between them (finding or making a good quality cable may be the hardest part). Is this a bad idea for any reason? Or are there other, better solutions?

What do you look for in a power supply, besides being able to output enough power for your modules and having enough headers? Are there any other, less obvious characteristics that are important?

Also, I know about leaving a bit of headroom (as in using around 80% of what the power supply can put out), but do you calculate that per rail or per total power supply output? For example, if PSU is rated for 2000mA on +12V and 500mA on -12V, do I realistically can use 1600mA and 400mA, or is using whole 500mA on -12V rail OK as long as I do not exceed 2000mA in total?

Some people prefer linear PSUs over switched, believing they are less noisy. (Personally, I think it’s a question more of well made vs. poorly made PSUs instead.)

I’m not sure about the headroom question. Overall, I think it’s more that a PSU running at full capacity runs hotter and is more likely to fail earlier. I exceeded the rated -12V on a uZeus for a while without noticing any problems, and didn’t realize it until I happened to see it on ModularGrid. Maybe I was relatively okay because the +12V was well below the threshold? I don’t know.

My current case is 12U 114HP and has a pair of RT65Bs, which have 500mA capacity on the -12V rail. I’ve run them at about 430mA each, and have to choose my modules carefully and work up a spreadsheet on which modules get plugged into which busboards to keep it balanced – but so far it’s been doing okay :fingers crossed:

1 Like

Looking for a (probably obvious) answer from any 4ms pod owners out there. Just received my 48x and one thing that was slightly unexpected was the lack of any kind of on/off switch. As far as I can see you have to just pull the plug out to turn it off?

Just wanted to see if anyone was doing anything differently? Is it safe for modules to just pull the plug out?

I’m daisy chaining it from a row power so was assuming the row power switch would turn the pod on and off, but the pod turns on once plugged in regardless of if the rest of my system on the row power is on or not.