# "Ghosts in the Machine"

Page 2 of 2

### Re: "Ghosts in the Machine"

Posted: Tue Nov 24, 2020 10:37 pm
It has a dilithium zenner diode in it? :D

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 12:47 am
I fitted an Orange LED in the bottom end of the Zener in the AC phantom supply and got an open circuit voltage of 47.9. So close!

I had made up an XLR socket with two load resistors, matched out of many and they measured 6k77 each on my Fluke DMM.

On load the modified AC supply gave 23.9 and 24.0 volts, happy with that.

I also got a USB powered supply on Thursday. That is not so good.

Off load 46.9V and with the test loads just 21.14V and 21.08V. The DC-DC converter device is rated at 1W so easily enough power for full spec spook and USB 2.0 can easily supply the current.
(I checked from a USB port on a desktop PC and a 13A USB charger) Yes, the unit was cheap but it really would not have cost any more to get it right, at least the OC volts could have been 49V plus a bit! The supply is all SMT chips so not easy to mod.

Dave.

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 1:56 am

The spec for the off-load voltage is 48V +/-4V, so anything between 44 and 52V is acceptable.

On load, the critical requirement is an ability to supply a maximum of 10mA in total, or 5mA per leg, to the load (although few mics need that much juice)!

Given a 48V supply : V=IR or R= V/I or 48/0.005 = 9600 Ohms total load per leg.

But as the standard calls for a 6800 Ohm feed resistor per leg the load needs to be (9600-6800) = 2800 Ohms... Call it 2k7 to be practical.

And to check, I=V/R =48/(6800+2700) =0.00505 or, near as makes no difference, 5mA

So with 5mA flowing, the voltage at the top of the 2k7 load resistor should be :

V=IR = 0.005 x 2700 = 13.5 V

If the phantom supply Volts was only 44V, the voltage at the top of the 2k7 resistor would be 12.5V (and the current 4.6mA per leg or 9.2mA ... which is almost 10mA!) And with a 52V supply the voltage would be 14.7V.

So... If you strapped a pair of matched 2k7 Ohm resistors between pin 1 and pins 2/3 of an XLR socket, and then measured the voltages across them, anything between 12.5 and 14.7V means the phantom supply meets the full spec.

In practice, of course, most mics pull much less current and so receives a much greater voltage.

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 2:16 am

The voltages show that the USB device cannot even support a nominal 6k8 load otherwise the voltage would be half 46.9 which means the regulation BEFORE the DC hits the internal 6k8s is pretty poor. The well matched loads also indicate how well the internal are matched, important for good noise rejection which will be my next test.

I shall have a hunt through my remaining resistor collection and see if I can test for full load current. I am not optimistic!

The use of ubiquitous power from USB is such a good one that it is maddening to see it done badly. I mean, who needs ANOTHER wall rat?

Dave.

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 11:40 am
ef37a wrote:The voltages show that the USB device cannot even support a nominal 6k8 load otherwise the voltage would be half 46.9 which means the regulation BEFORE the DC hits the internal 6k8s is pretty poor.

It certainly is pretty poor! A 6k8 load draws 3.5mA of current per leg, or 7mA overall which is the kind of current required by a lot of modern Far Eastern mics and few modern mics. So it's perhaps not too surprising your mic struggled to work properly. I can't find a figure for the current requirement of the ST2-C...

The well matched loads also indicate how well the internal are matched, important for good noise rejection which will be my next test.

Very true! If memory serves matching needs to be better than 0.5% to achieve a reasonable CMRR.

Phantom power is such a simple thing and so clearly specified.... it's hugely disappointing to find products where it's not being done correctly!

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 3:24 pm
A few months back, a friend of mine bought himself a shiny new Neve 1073 DPD. He hooked up the S/PDIF output to his audio interface and was disappointed to find that everything coming through had a repetitive low-frequency 'popping' sound superimposed on it.

I suspected some sort of clocking issue, so we tried hooking the preamp up through its analogue outputs... same thing.

Then we swapped the mics out -- whisper quiet.

It turned out that this problem only manifested itself with his Sontronics pencil mics, and only with the Neve. The mics were fine through multiple other preamps, and the preamp was fine with multiple other mics. But both Sontronics mics did the same thing.

We never quite got to the bottom of why this was happening, but it was odd.

### Re: "Ghosts in the Machine"

Posted: Sat Nov 28, 2020 4:29 pm
Weird indeed.

LF thumps and bumps could be related to a power supply problem in the mic, but I'd be surprised if it was because the Neve had a less than perfect phantom feed. Neve's designers have a long history in and know how to design a phantom supply... And they aren't going to cut silly corners in a premium product like the DPD.

So I think I'd be looking for ultrasonic /RF leakage from the digital electronics in the Neve as a strong contender, probably finding it's way into the Sontronics and upsetting its valve gain stage in some way. Cables with EMC XLRs might have helped, or a clip-on ferrite, perhaps. Worth reporting back to Neve as a repeatable fault condition, too, as they may need to tweak the build or design to curtail possible EMC leakage!

On a semi-related note, I was putting a test rig together in my dining room last week to check everything before a location recording session. I had Radio 3 on in the background from a DAB radio about 3 metres away.

With everything powered up I plugged my HD25 phones into the front headphone socket of my Nagra VI... and the DAB radio suddenly muted...

When I unplugged the phones the radio came back, and pugging into the Nagra's side headphone socket didn't cause the same effect -- and neither did plugging my AKG K702s into either socket!

More weirdness!

My guess would be that the internal grounding arrangements or PCB layout inside the Nagra results in some RF leakage from the digital circuits appearing on the front headphone socket, and the length of cable on the HD25 (which is shorter than the AKG cable) is probably just right to serve as an effective aerial to radiate enough RF in my dining room to block the relatively weak received DAB signal!

It's very easy to think only about the audio bandwidth we're interested in, but modern electronics are capable of huge bandwidths and it's popular to maintain huge bandwidths inside equipment in the (arguable misguided) interests of utra-fast slew rates and transient responses...sadly though, not all products manage and protect out-of-band signals as well as they really should -- not even the very best designs! And these apparently weird situations are the result.

Having said that, and to be fair, it is difficult and rarely intuitive to control out-of-band signals sometimes, especially when there are digital control and digital audio signals flying about.

### Re: "Ghosts in the Machine"

Posted: Sun Nov 29, 2020 3:46 pm
Do the Sontronics SDCs have a pad switch? If so this could be related to my original problem, that of my STC-2 not switching to max sensitivity, the pencil mics could be "hunting"?

However here are the noise figures for the two phantom power units. Each unit was plugged
into ch 1 of my NI KA6 and the gain turned to maximum. The mic input was terminated with a 220 Ohm resistor in a screened XLR plug. Samplitude SE8 24bits 44.1kHz.

Modified AC box. rms (easiest to read) -93dBfs. AC unplugged, -94dB.

USB box: rms -91dBfs . USB unplugged -92dB.

Term' plug into KA6 -100dB fs and switching on phantom power made no difference.
Note, these reading are 'blipping' about 2 or 3 dB up and down all the time, numbers are an eyeball average but I think all are perfectly usable? Also the figures are for zero current drain of the phantom power, I should have to build a special rig for that but tests with the Sontronics mic showed noise levels below -75dB and of course that included room noise. The mic was over a metre from me and wrapped in foam but it still picked up my light breathing!

Dave.