Phantom power circuit for experimenting

[Andy Millar]

I noticed browsing up that there were a few questions on generating 48V phantom, here's an idea I had one lunchtime many years ago if you have a 15-0-15 transformer powering the rest of your equipment, I think it also got used in one of the SSL rackmounts...

 

[pucho812]

correct me if I am wrong which I think I might be but would that just be doubling the voltage and giving you 30 V VS 48 V?

 

[Andy Millar]

Sorry, I left a stage off! Just goes to show that you should always get someone else to check your drawings...

 

[SSLtech]

Andy, -do you know who did the phantom power design in the SSL9000 console? (it's in the little 1RU system supervisor...)

Anyhow, I ask because there's an age-related failure of a 1KΩ power resistor (it bakes and dies, and it's the series resistor in a series R/shunt Zener regulator that powers the opamp that drives the regulator).

When the resistor dies, the regulator sends the full, unregulated voltage to the phantom rail. (about 90Volts! :shock: )

We discovered this when it failed on one console and tore up EVERY SINGLE ONE of the input phantom-blocking caps on the console mic-pres. (they're 0.1% selected, matched pairs, so they're not cheap, either!) they're rated at 63V, and after they go, you lose the MAT-02s. When they go, you usually have a 50% mortality rate in the first pait of 5534s... -That was a pricey little failure... however on the bright side, it's when I started to look closely at the workings of the 9K mic preamp, I was VERY impressed with it, and one thing led to another... you can guess the rest! :wink:

-Just curious who I should send the bill to! :wink:

Seriously though, I have 3 x 9000 J series here, and after that happened, I pulled both the other supervisors and found the same failure about to happen. In each supervisor, I replaced the 2-Watt resistors with 10-Watters and haven't had a failure since. I then told all of my buddies who look after 9K's to do the same, and they all reported heavily scorched-looking PCBs under the resistors...

Pucho, if you read this, perhaps make a note and perhaps you or Brian should take a look at your phantom rail PSU. the resistor is inside, towards the back, toeards the left. While you're at it, take a look at the row of axial electrolytic caps, make sure none of them look dodgy, the heat from the resistor usually partially cooks the closest one, as reported to me by a couple of other 9k-obligated techs.

Keith

 

[Andy Millar]

Errrr...no...but it's probably me :sad: - fortunately I haven't got any money, so you can't sue me :wink:

Again I don't remember it at all, I'm really wishing I'd taken a set of 9k circuits with me when I left, but since the desk wasn't even released then it wouldn't have gone down well if it had been found out!

I'm a bit surprised, I usually get power ratings right even if I leave a third of the circuit off sometimes (see above)!

 

[pucho812]

[quote author="SSLtech"]

Pucho, if you read this, perhaps make a note and perhaps you or Brian should take a look at your phantom rail PSU. the resistor is inside, towards the back, toeards the left. While you're at it, take a look at the row of axial electrolytic caps, make sure none of them look dodgy, the heat from the resistor usually partially cooks the closest one, as reported to me by a couple of other 9k-obligated techs.

Keith[/quote]

alrighty. Will make note... :thumb:

 

[SSLtech]

:green:

No prob. The real reason that everything died was that a confused engineer couldn't get a Neumann to work on an input channel that had just worked with a dynamic mic. Then he tried another channel... still nothing.. then another... still nothing. Finaly, he enabled phantom on every single channel and went throught all of them with a patch cord. Then he tried a dynamic mic again, and when NOTHING worked, he got very confused!

The few minutes that all the phantom buttons were down was enough to kill everything...

I did alert Ernie at SSL NY, and later also passed the info on to Dave M, but I don't know if it was ever acted upon... An interesting thing was that the value used wasn't the same as the value in the schematic... -In hindsight, I can't remember if the schematic value was 1K and the actual value was 520 or vice-versa, but all three of our supervisors had the same different value installed.

It's actually rather possible that the value was changed without a re-computing of the effect on component power rating... sounds feasible, though it's a guess rather than being based on any knowlege.

Anyhow, this thread least prompted me to remember that episode, so I can at least merntion it for other 9K techs to look out for.

Keith

 

[pucho812]

Keef,
you just reminded me of a session we had over at work right before are VR's came. A "famous engineer" was in doing a clinic type deal where people watch him mix and ask questions. His assistant showed up a few hours early and managed to mess everything up by changing the master word clock without resetting the SSL 9K and Alsihad. So you know what happens next, Console Crash/freakout. Got them back and running but swore to me he never touched the master wordclock and later admitted to doing it saying at every other studio he goes to he can do that without having to reboot the SSL computer. Anyway the next day same assistant calls up to tell me about the problems with our studios and how they need to be fixed right away. As part of his story He goes I made a new pro tools session and omported all the audio in to start at one hour. HE goes then I switched the session start time to 59:40 as to have reverse coutoff. So I am like reverse countoff you mean pre roll. And he goes yeah anyway When I did that it crashed. I did it a second and third time and crashed 2 more times. So I go if it crashed once and doing the same steps again it crashed what made you think that the third time it would work? It would have been easier to 1. just add a 30 sec pre roll or B have the session start time be 59:30 and move all the audio over to 1 hour. He got very upset and was like "well whatever just fix it cause we are back in there next week" so I was like sure we will get right on it but was completely going it was human error...

 

[PRR]

> correct me if I am wrong...

You are simply wrong. Andy is dangerously wrong. Don't trust anybody over 30.

You've simply forgotten 1.414. 15V RMS is 21.21V peak.

And (I wasn't clear until I simmed it) the first (2-D 2-C) plan is a Voltage One-And-A-Halfer, giving 30V*1.414*1.5. Or more accurately: it gives (30V*1.414)+(15V*1.414). The "half" comes from one part taking the whole 30V and the other taking just a 15V share.

The danger is that, at 25K load (a small Phantom load), the DC voltage is +61V, NOT 48V as the drawing is labeled. At 5K load (one maximum legal Phantom load) it gets big ripple, with peaks near +60V and the dips around +53V.

The second plan, 3-D 3-C, gives 97V peaks, with dips to 90V at 10mA load. With 0.1mA load it rises toward 104V: this is a voltage quintupler (X5). Or rather: 2.5 times the peak of the full 30V secondary.

As Keith's elegant experiment proves: calling something "48V" when it is really 90 some Volts can have exciting results.

These are indeed the most elegant ways I ever saw (or expect to see) to get over 48V from the same windings that make our +/-15V supply. But they need regulation! And Keith's experiment shows the risk of marginal design or ignorant re-design of regulators.

The cap values are reasonable, for one mike; for multiple mikes, multiply the size. If you want a 48-in board, use 5,000u, 2,700u, 1,000u. (At this scale, it makes more sense to just buy a proper 50V winding; dubblers are for one/few-mike needs.)

The 100U cap gets 42V; a 50V is not really big enough, use at least 63V or 75V.

The 47U cap gets 84V: go for 150V rating.

The output cap gets 104V at light load; again, use 150V.

Using 100V and 200V caps is not over-kill, especially when it is your mikes and inputs on the line. A cap failure would probably just reduce or kill the output, but it could put AC on the Phantom line, and not all mikes are protected for reverse voltage.

 

[Andy Millar]

I did say that this was for experimenting! I should perhaps have made clear that this is not a finished circuit, only a concept for part of it.

THIS IS NOT A FINISHED CIRCUIT.

There, that feels better!

You're quite right, it should be followed by some simple regulation. And which version you use would depend on the level of regulation and how smooth you want the supply to be. Keeping version 1 ripple free without having large capacitors and hence large surge currents is harder, but yes it is the safer circuit.

P.S. perhaps submitting my first post just before my evening meal was ready and my second just before I went to bed was a bad idea, I'll wait until I've got more time next time