like I mentioned about the ground (like with 7805’s) if you take em off ground you get less dropout, wether they are functioning they way they should or not, they do what they need to do. I read the removing from ground supply on a robot forum somewhere, gosh knows where it was after a long eve of trawling and a whole load of them were fine with taking em off ground. If you haven’t figured out already I do read the data sheet but find that the data sheet is always way too on the safe side. for most things anyway. maybe I haven’t been running things for long enough. but I’ve not had many catastrophic failures and horror stories and some things are getting to the point that if they pop they would just as likely pop in normal machines. but hey ill give both of those voltage regulators a try. the midi to cv will do another run of ports with another vreg coolio!
using a to-92 package but could just change it up again
well yes like I said about the forum im still trying to find, it definitely makes the lm317 not to what its meant to do but it worked!
yep your right 1117! my inability to read really makes me facepalm sometimes.
right im sold on the lm1084, as you say the margin means anough to bring back 10 ohm protection. I’ve ordered a bunch ill plop it in. 3 quid is pricey but meh, I would rather it be technically correct than blame it on which craft with the lm317. which for functions the way it does surprisingly. but yeah. all good.
thats should be a music genre haha
I’m not entirely sure why you seem to be convinced that LM1117 is obviously wrong and LM1084 is obviously right when they have pretty much the same dropout/noise specs and topology, based on interpreting two graphs of typical values drawn in different resolutions 
I mean, if cost is no concern, there’s no shortage of newer regulators that are much better than both of them; a quick parametric search brought up this fellow:
(that’s a TPS7A24 that’s actually cheaper that the TO-220 ones, probably because it comes in a tiny package)
But on the other hand, assuming the circuit’s been using a non-regulating LM317 this far, maybe a divider and an opamp would be a better thing here? Or better, a TL431 and an opamp?
indeed! its more the package. I vowed to the electrical gods to never go near surface mount. its just another thing for people to worry about when building these things. surface mount is indeed easy as fudge to solder. but with a series thats all through hole, I don’t want to evoke anxiety over 1 component.
I may also have a completely unexplainable dislike for surface mount, surface mount has a look about it of it smelling like Johnsons baby oil and disorano, when through hole has a more regal old spice and whiskey in the armchair vibe to it.
but to be honest I have ordered both to try and if it comes to it and I have to break tradition for 1 component, I will.
if I stumble on the lm1117 in a TO92 package. id be over that like hot cakes. but it seems there isn’t one.
if I had a spare opamp I would do the opamp to be honest when I was figuring out the lm317’s secret witchcraft mode I was humming and aahing about the divider but yeah if it comes to it!!!
ill test these see if there is any joy. in all practical ways I have tested, the witchcraft mode is fine, and it could just be a troll to the eurorack gods. but ill try these ways out as it was my 2020 resolution to be more proper
I mean it may take a day but I might just plop the lm1117;s onto the boards. could be an option (edit I mean the TPS7A24)
First impressions, etc, but looks pretty good to me 
I’d consider switching out the opto-coupler circuit for one based on a 6N137, which 1) is faster (guaranteed to be within MIDI specs (*)), 2) has an actual logic output instead of a transistor, 3) is more widely available (e.g. Tayda has it), and 4) is cheaper, from what I can tell. The circuit I used for the minimachine is here. But this is optional; the 6N138 is used in a ton of DIY MIDI stuff and people would probably have noticed by now if it was all wonky.
And yeah, I still wonder if we cannot come up with a better 10.6 V supply, but that’s for a different subthread
*) Found some MIDI document that says HP’s 6N138 “have been found acceptable” (which makes me think of Lemongrab, not circuit design) but they don’t seem to make it these days and the non-HP 6N138 datasheets I’ve looked at are less convincing, so may depend on manufacturer. The 6N137 is at least 20× faster.
aaah I didn’t realise Tayda and places dont have the 6n138, I swear thats changed recently… I have been jaded by ordering 2000 138’s a couple of years ago.
welll since i am doing another round messing around with the power I may aswell change to 137’s. Im sure the LM1084 solution would be suffice as its near enough a drop in solution, and as popping more components (more opamps etc) will be making a reasonably busy module more busy, just trying to keep it all to a minimum.
Design constraints are good contraints.
I’m occasionally tempted to use SMD decoupling caps (possibly on the reverse side) and have also contemplated adding SOIC footprint to some circuits, as an option (SOIC are easy to solder and often cheaper than DIP, but they don’t fit inside a DIP footprint anyway, so then you end up needing more PCB space).
I did cave in and added some optional SMD LEDs to a couple of designs in the latest round. Will see if I still think it’s a good idea when they get here (I can always paint over the pads with a sharpie or something if I don’t like them).
Nothing jumps out to me as obviously broken.
A couple of questions though:
- The transistor outputs (gate signals?) will be limited to about 4.4V in that emiter follower configuration, is that high enough to trigger most envelope generators reliably?
Edit: 3.8V, with the 1N4148 drop. - No protection resistors on the opamp outputs?
cool! but yeah tbh I question sometimes wether it is just stubborn ness or what not. I am trying a smd version of stuff pre assembled this month to see how feasible it is, but I dunno I remember the first time I came up against a couple of smd components in a build years ago and it was enough to stop the enjoyment of building, no matter how easy it was to actually solder haha. smd easier to repair too. so I dunno its a tough one!
fair point. I should probably add 100r’s to the outputs of the voltage following opamps. as for the transistor outputs used this in many builds learnt it from a muffwiggler thread years ago. but to be honest on second look could probably ditch the diode in place of a 1k resistor (edit) after the 10k instead of before.
An option could be to plop in another mc14504 and drive the transistors with that. That would bring the outputs up to about 10V.
What’s the input impedance those outputs are driving?
An mc14504 could drive a couple mA directly, no transistor needed, if that is enough.
fair point!!! With the gates its up to anyone to plug it into anything usually very high just driving opamp buffers on other modules however who am I to tell them they cant plug it into their ear… I think ill stick to transistors for simplicity sake. Main reason is because the transistors are on the jack board away from everything else. 4volts has been driving things without much issue in wrong modular synth designs for 20+ years I am not one to judge . Plus this module has taken quite a bit of finicketing to make it all go together happily adding 6 more buffers and an mc14504 might push it over, if it isnt broke im not sure I want to completely rework it especially if those outputs are fixed to just gate functions.
this is good tho! ill share progress on these things from early especially things like midi-cv. usually the only things that need to be remotely good in a modular haha
Not to get into a religion argument here but I’m with Antoine on the ferrite bead thing — not that I know much, but from all evidence I believe Matthew Skala does. He’s nearly as pungent about 10R resistors although a lot of that is about 10R resistors without capacitors, which I’ve never seen. I tend to like to replace both with Schottkys if I think I can get away with the voltage drop.