Adding digital circuitry to analog desk...

[12dbLow]

Hello guys,

I am working on a Crest X-Eight Console (Originally a live console) to build a studio mixing setup around it. I want to add a better Monitor controlling module, that includes a microcontroller and relay switching. I want to power them from the console PSU. Now I would like some advices If the digital circuitry might induce noise into the analog part. The desk already has some digital controlling, so there is a digital ground.

The power rails look like this:

+18 V
-18 V
+12 V (They make 5V from this rail for the digital part, see picture) but also used for analog Op-Amps
+48 V

Now I also need 5V, but the 5V in the desk isn't easy to access. I already brought a cable out with all the PSU Rails, so I need to make 5V from the 12V Rail. I thought a regulator might get pretty hot when powering some relays, so what do you think about these Step-Down Converters? Is that crap?
http://www.ebay.com/itm/300954920223?ru=http%3A%2F%2F

Best,
Adrian

 

[thermionic]

That converter on ebay doesn't appear to have any shielding. You'll notice that the switching converters used in most pro audio gear (RND Portico series, for example) is well shielded, which explains the high price tag of such converters. So, don't put it near any low-level audio wiring. I've found that unshielded switchers can radiate RFI a long distance.

Do you know the exact current draw of the 5v relays? One idea could be to try a normal linear Vreg, bolted to a large-ish heatsink, and simply see how hot it gets.

 

[12dbLow]

Okay, yes I can calculate the total current drawn by the relays. I could use a heatsink, but I thought the converter might be a more elegant solution. Maybe I should put it in an extra metal case? Like the ones used for DIY stompboxes.

Is there also a danger of noise being induced into the 12V rail?

 

[thermionic]

In my experience, shielding a switching converter in a common-or-garden shielding enclosure yields minimal benefits. One of the reasons the converters used by RND and the like are pricey is that the shielding can is made from an expensive alloy.

When connecting the device to a conventional audio analyser looking at noise in the 20-20K band, I've had good results from home-brew switching converters, when it's been in a separate enclosure, nowhere near the audio circuitry. I put ferrite beads on the wires coming from the switcher PCB.

However, if I turn on a LW radio I can hear all manner of HF interference emanating from the switcher... But at 20-20k the switcher's just as quiet as a good old linear psu...

 

[JohnRoberts]

Proper console design involves dealing with such interference while still playing on a black sheet of paper.

If adding digital content after the fact, you will need to tell us what happens. I can't predict.

My best advice is to try to keep all potentially digitally corrupted wires local inside a shielded module and filtered at the interface into and out of that shielded digital zone.  A switching PS could be located inside that zone with the power feed to it filtered, but you need to take care that the switcher does not corrupt the digital poop.

If dealing with too many stems to isolate, you may need to just try it and see what happens.

=====

FWIW when I helped some friends add a small digital meter to the master section of an analog console (designed by the same engineers that did the  Crest consoles), we found noise related to my LED update frequency in the console noise floor.  My remedy for that was to shift the meter's internal digital communication rate (SPI) higher to be reduced by normal path LPF and place any residual well above the audio pass band. This was mostly a cosmetic issue down in the dirt of the noise floor, but we decided to make it impossible to hear, even if they boosted up that noise floor to silly levels. 

It's not voodoo but digital noise can find it's way around, especially inside an old analog console, so good luck.. 

JR

 

[12dbLow]

Okay, then maybe its a good starting point not to mess with the switching converter and build mainly the same circuit that delivers the 5V to the digital part in the crest and use a heatsink for the regulator. Then measure and see If there are problems.

 

[thermionic]

If you're worried about heat, one crude method (I can see John rolling his eyes already...) is to choose a Vreg that can dissipate a lot of heat - say the LM338K. Think of it like an LM317 but in a T03 can. You then find a surplus heatsink that's drilled for T03 (i.e. the kind you'd see on a pwr amp) and to hell with the heat. Zero points for finesse, but it could keep your mind at ease.

 

[JohnRoberts]

The digital switcher will waste less power... Digital circuitry can coexist with digital supplies, and adding any digital circuitry will add digital noise anyhow. So I am Ok with keeping all the digital poop together and segregated from the analog poop.

The whole idea seems rich with opportunity for learning new stuff that you didn't expect.  8) 8)  ,,, have fun.

JR

 

[12dbLow]

I will try to keep you updated on that project I still have to wait some weeks for the chinese switchers  ;D