Phantom power horrific-ness. (Is that word? It is now!)

So, the adventure continues....

Bench testing my "Heavy" preamp, I plugged a headphone amp into one of the outputs and a tone generator into the input.  Played a bunch of sine waves into my ears on both channels and all was well.  It sounded quiet and clean.  (But we all know that's not how this will end....)

Next I plugged in a dynamic mic (Shure SM-58).  All good.  Sounded kind of bright for my taste, but that could have been my headphones (Sony 7506's).

Next came a condenser mic.  Flipped on the phantom power and was immediately greeting by a nasty loud buzz.  Ack!

Disconnected everything.  Checked XLR pins 2 and 3 against pin 1 with a voltmeter and both show 48VDC.  Rigged up the scope, and this is where it gets entertaining.  There is definitely a very ugly square wave sort of mess riding on that rail.    I've attached the schematic for your amusement.

15VAC goes through a voltage multiplier/half-ass wave rectifier and hits a capacitor on its way to the voltage regulator.  The ugly I'm seeing at the XLR jack is also seen on the output of that regulator, as well as the cap downstream of it (C112).  My first assumption is that that capacitor, all by itself, is simply inadequate to properly smooth out the ripples, so I'm inclined to replace it with two larger ones in parallel.

Unless I'm missing a much bigger problem....

> seen on the output of that regulator

Are you brave enough to look at the input of the regulator?

I have doubts you are getting >51V at the *dips* of the raw DC. That throws down-spikes into the "48V".

And/or: I have doubts about the way you combine, on the same AC winding and the same DC common, a multiplier and a bridge.

Yep, you guessed it.  51VDC on the input.

And, for what it's worth,  about 4.7VAC on that output, if I'm reading the meter right.

The bridge diode 3-4 is shorted.

This is not doing what you think it is.

A hasty sim shows 60V raw Phantom, but the heater raw DC is half-wave and very dippy.

Don't know what the real world is doing.

Edit: sim is showing a 600 AMPS peak. Putting any realistic impedance for "transformer" gives very poor DC both outputs.

Steal a different plan.

It looks like the 12VDC only powers the 12AX7 heater, and if so, the fix is rather simple.  Remove the ground from that 12VDC circuit, connect the low side of the LED directly to the negative sides of C106/C107, and cut the line between C103 and BR2. Also, add the ground back for the HV supply at the bottom of C103.

The voltage across the heater pins will still be regulated 12VDC, but it will be "floating" at 12VAC or so with some half-ass-rectified waveshape. You may not "like" that, but it shouldn't really matter. It's only capacitively coupled to the cathode, the heater and cathode can differ by something like +/-100V or 200V (it's been a while since I looked), and this 12VACV coupled by a few picofarads between the heater and cathode will be sent straight to ground by the 1.5k in parallel with the 100uF. (R4-C4, R5-C5).

If it does make an audible buzz and you need it to stop, you'll need a THIRD power transformer to isolate all the secondaries, or a single transformer with all the right secondaries wound on it.

If there's 51V on the input, then the problem is downstream no? Like a short.

PRR said:

And/or: I have doubts about the way you combine, on the same AC winding and the same DC common, a multiplier and a bridge.

Click to expand...

I think that's where I made my biggest mistake.  On T3 (the first mains transformer), I shouldn't have paralleled the secondaries, creating one 15VAC secondary.  I should have done split secondaries, with one feeding the tube supply and the other feeding the phantom supply.  The tube supply on this was lifted from the last gadget I built, and I (wrongly) assumed that all I needed to do was add a phantom supply circuit and just tap the AC off the existing transformer setup.

I now see what you meant about the bridge being shorted.  Yikes!  That's a mess!  :-[

Would I be safe in assuming that rewiring T3 for split secondaries would solve about 90% of this issue?

Look at how I do the G9 supply - that's about the simplest you can get away with.

And don't skip on the protecting zeners around the TL783 (the 3x39V) - this limits maximum forward voltage, and keeps the TL783 from dying at power-up or no-load condition, when there's too big a voltage difference between in- and output.

Last, you'll probably want to add a 470R-1K resistor before and after the TL783 for added protection.

Jakob E.

gyraf said:

Look at how I do the G9 supply - that's about the simplest you can get away with.

And don't skip on the protecting zeners around the TL783 (the 3x39V) - this limits maximum forward voltage, and keeps the TL783 from dying at power-up or no-load condition, when there's too big a voltage difference between in- and output.

Last, you'll probably want to add a 470R-1K resistor before and after the TL783 for added protection.

Jakob E.

Click to expand...

Are you referring to the TL783 on the phantom supply, the plate supply, or both?
I thought that's what the 1N400X across the regulator was for.

The plate supply: The TL783 does not like to see an input/output differential of more than 120V. And it gets this at power-up, while charging the second reservoir cap...

Jakob E.

gyraf said:

The plate supply: The TL783 does not like to see an input/output differential of more than 120V. And it gets this at power-up, while charging the second reservoir cap...

Jakob E.

Click to expand...

Good to know.  Thanks!

Also, you phantom power supply really should be floating for best noise performance. Its 0V should go to pin 1 of the mic input XLR which in turn should be bonded directly to the chassis. Analogue 0V should be connected to a bolt beside the mains connector and the safety earth from the connector should also be connected here. There should be no other connections of the phantom or analogue 0V to the chassis or each other.

Cheers

Ian

ruffrecords said:

Also, you phantom power supply really should be floating for best noise performance. Its 0V should go to pin 1 of the mic input XLR which in turn should be bonded directly to the chassis. Analogue 0V should be connected to a bolt beside the mains connector and the safety earth from the connector should also be connected here. There should be no other connections of the phantom or analogue 0V to the chassis or each other.

Cheers

Ian

Click to expand...

This is one thing that has always confused me.  If the phantom supply ground goes to pin 1 of the XLR, and that pin is then bonded to the chassis, and the power ground is also bonded to the chassis, doesn't that put them both on the same ground?
So how does the phantom end up "floating?"

CurtZHP said:

This is one thing that has always confused me.  If the phantom supply ground goes to pin 1 of the XLR, and that pin is then bonded to the chassis, and the power ground is also bonded to the chassis, doesn't that put them both on the same ground?
So how does the phantom end up "floating?"

Click to expand...

it doesn't but it is important that you have control over where the tow 0Vs are connected together. They should be connected together in only one place - at the chassis bolt. If the phantom is already connected to analogue 0V in the power supply then you cannot do this without creating a loop.

Having said that, on a single, self powered mic pre you can almost certainly get away with it but it is a good habit to get into.

Cheers

Ian

ruffrecords said:

it doesn't but it is important that you have control over where the tow 0Vs are connected together. They should be connected together in only one place - at the chassis bolt. If the phantom is already connected to analogue 0V in the power supply then you cannot do this without creating a loop.

Having said that, on a single, self powered mic pre you can almost certainly get away with it but it is a good habit to get into.

Cheers

Ian

Click to expand...

By the way, ruff, looking at your signature, I can now see why I've been so busy lately.  ;D

CurtZHP said:

By the way, ruff, looking at your signature, I can now see why I've been so busy lately.  ;D

Click to expand...

How very true. I have decided not to take any commissions this year so I can concentrate on developing the MK III tube mixer. I am hoping to build a simple one in the Glensound frame I acquired and at least one in one of the four EELA frames I also have.

I have to say, I am enjoying myself.

Cheers

Ian

ruffrecords said:

How very true. I have decided not to take any commissions this year so I can concentrate on developing the MK III tube mixer. I am hoping to build a simple one in the Glensound frame I acquired and at least one in one of the four EELA frames I also have.

I have to say, I am enjoying myself.

Cheers

Ian

Click to expand...

That's a pretty exciting design.  I'm really looking forward to following its progress.

CurtZHP said:

That's a pretty exciting design.  I'm really looking forward to following its progress.

Click to expand...

It is moving steadily forward. The other day I ordered some prototype PCBs of the new motherboard, the new 35mm Classic mic pre and  a new 35mm version of the Twin Line Amp. Those comprise the active parts of the new mixer and the means of interconnecting them. I have also done a first version of the 3 band EQ for the Glensound - I might be able to upgrade that to four bands for the EELA. The next big step is the long skinny modules for the EELA. Once I get the basic mechanics for that sorted I can start thinking about PCBs for it. I'll be summarising progress in the blog as usual.

Cheers

Ian

Well, now I've really done it!  I'm in deep weeds now.

Turns out, the secondaries on the mains transformer were already split, so that's the good news.  I just needed to take my phantom supply from the other winding to get it away from that 12V supply rectifier.  At this point, the transformer/rectifier topology is looking exactly like Jakob's G9 supply.

Now I get no phantom power whatsoever.  I measure 15VAC from the secondary, and some really oddball voltages from different points in the multiplier stage, but nothing, and I mean nothing -- 0 volts -- makes it to the regulator.  I'm thinking I cooked a diode or something.  I've double and triple checked the wiring, and nothing has changed.

I've completely disconnected the supply section from the rest of the project so I can better troubleshoot.  I'll have to take some notes and post the results here tomorrow.

Apparently, a healthy 48VDC rail is too much to ask for. 

CurtZHP said:

Turns out, the secondaries on the mains transformer were already split, so that's the good news. 

Click to expand...

Any chance you got the wrong pairs of secondary leads?

Never mind measuring volts on open windings, capacitive coupling can be deceiving, ohm out the windings for continuity.

Gene

PS: What is the html tag to get a fixed-width font? I was going to do an old school ascii drawing. Anyone remember usenet?