Dodgy App Note Cct.

Hi Phil. Another puzzle, cool!

1n4148 to the rails seems an odd choice for input protection, especially given that they want to limit the excursions to 0.5V.

Short the input to gound (or plug in a long mic cable) with phantom power enabled and input voltage will exceed the absolute minimum (0.5V) possibly by a volt or two. (dependant on the ESR of those capacitors)

--
  When I tried casting out nines I made a hash of it.

-------------------

** Glad somebody likes my brain teasers.

** Well, you are kinda on the right tram but cannot see the wood for the trees.

... Phil

Only one problem?

There's no series impedance, so a cable fault will discharge the 47uF caps directly into the tiny diodes. Once.

There's no RF filter.

XLR pin 1 is shown connecting directly to circuit gnd. There's no indication of how this is routed, but if not done correctly it can lead to the so called "pin 1" problem that involves RF coupling into the audio instead of (harmlessly) into the chassis.

I thought there might have been a DC offset problem, but it seems ok with their DC servo circuit *provided that the output never clips* (which seems unlikely to be true at higher gain settings).

Maximum gain is 61dB which seems a little high to me.

Allan

----------------------- Phil Allis>

** Just one "dead easy to spot " problem.

** Allan gets the Cupie doll.

That figure has been up for two decades, the sheet revised and the dumb error not corrected. Buyer beware is still a fact.

Some heftier diodes are needed, maybe 1N4001s.

However pumping the DC supply rails with 20 amp spikes is not ideal and what if some big voltage comes through the input - from say a power amp ?

** Nice, but probably not essential if the usual anti RF precaution is taken with grounding the mic cable shield immediately and relying on some lossy capacitance in the twisted pair.

** AFAIK the " pin 1" issue is all about ground loop hum so will not affect a mic input. ** Ha, you spotted an issue I missed cos DC coupled gain stages in audio are rare. ** Nope, it's typical and barely enough for a mic-pre driven by from a low output dynamic like a ribbon mic.

Most have only 0.1mV output at 74dB SPL.

.... Phil

Woot.

Figure 21 in /this/ mic preamp app note

shows 1A diodes.

This Rane app note

describes it as an RF problem, and their fix sounds the same as your "usual anti RF precaution", which I guess is commonplace today (but maybe not in 1994 when the original paper was written). I haven't worked in that industry since about that time, so I don't know.

Allan

Your brain teasers are always fun. Even if the "dead easy to spot" error wasn't so easy for me. The pin 1 problem occurs when a manufacturer connects pin 1 to the PCB. Rane note 165 explains how too many microphones act as RF antennas:

Pin 1 Revisited

The author, Jim Brown, K9YC, says this about one "rather expensive condenser mic:"

If you try to use this mic in downtown Chicago with a properly wired mic cable, the shield current induced by Channel 2, Channel 5, and a bunch of FM broadcast stations causes enough voltage drop in the black wire (which the orange wire adds to the audio circuitry) that both the video and FM signals are clearly heard!

Thank you, 73,

--
Don Kuenz KB7RPU 
There was a young lady named Bright Whose speed was far faster than light; 
She set out one day In a relative way And returned on the previous night.

They put the phantom power switch directly in line with the +48 volt supply instead of prior to an RC decoupling network.

If you use a ribbon mic with the phantom power on (not inherently a problem...) any spikes or transients on the +48 from e.g. power loss get shot right into the ribbon mic transformer and onto the ribbon. Ouch!

that current is limited by the 6.8k resistors

--------------------

** They once recommended Schottky types, then changed their mind.

A large enough signal at the input would still destroy the preamp and anything else on the same DC rails.

I have seen this happen, a near new Mackie 16 ch desk written off after a guitarist plugged his 100W Marshall into a line input.

The input had 1A diodes direct to each rail an so pumped it up to a lethal level. Beyond economic repair.

One 2.2kohm in series with the input would have saved it easily.

Dumb design, seen over and over, in many products.

** Not very convinced about that.

Rane is accompany that rose from the ashes of Phase Linear - IOW ex factory staff. They publish lots of audio app notes that look wrong to me. Opinion presented as fact, mostly.

Their own stuff is far from impressive.

That said, the thing with RFI is something HAS to detect the RF signal.

Usually an input device junction or even triode tube will do.

So designers usually isolate them from the source of RF with a series resistor.

The world of audio is filled with silly myths, most of which are near impossible to kill off.

That twisted pair balanced mic lines have far better mag field hum rejection than a co-axial cable is one.

Go try it, co-ax wins.

For non phantom powered mics, it should be the norm.

..... Phil

Large enough Transzorbs across the power rails?

In the music scene, it seems that all connections should be made thunderstorm protected.

--

-TV

The preamp sports a 1nV input noise level. A 2.2k resistor has about 6nV of Johnson noise. A 47-ohm value would have about 0.9nV, but fail to provide protection you're seeking (THAT suggests 10 ohms in their 1510 datasheet). It's often suggested when a protection diode is added, connected to a supply rail, that a big zener to ground be placed on the supply, rated a few volts higher. Or a simple BJT network could shunt the higher current to ground. A elegant way to limit input current, is to use two depletion-mode MOSFETs, back-to-back. See AoE III, Figure 5.80, page 361. There we show two LND150, and an R4 = 1k resistor to set current-limit to 1mA.

Hopefully this scheme can be used without adding too much noise, but 1k isn't much better than 2.2k. With no resistor, the maximum current is equal to Idss, or 3mA for an LND150. But LND150 has Rds(on) = 650 ohms, and 2x = 1.3k, far too high. Another favorite of ours is the BSS126, but it has Rds(on) = 320 ohms. Awwkk, we want both low Idss and low Rds(on).

The BSP135 has Rds(on) = 30 ohms, excellent, but its Idss = 20mA value is a minimum, no maximum is given. However, it has a very low Vgs, so a small R4 = 27 ohms could limit the maximum current to 25 to 50mA, and total resistance would be 87 ohms, adding 1.2 nV. Or use 56 ohms. The withstand voltage is +/-600V, so one could safely hook it up to the AC line. :-)

--
 Thanks, 
    - Win

Low Rds-on, low Idss is all we want. Sounds like some sort of Phemt. But phemts aren't very appealing as input protection devices!

DN2530 is a nice little fet, 12 ohms at Vgs=0, mostly off at -2 or so.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Winfield Hill Rides Again:

--------------------------- >

** Hello Win, are you there ?

The subject changed from the posted mic-pre to real world items a few paras ago.

My comment was about a line level input on a Mackie desk.

Adding a series 2.2k would have no detrimental effect.

..... Phil

OK, but the issue of protecting sensitive preamp inputs, from abuse, is still a valid and non-trivial issue worth thinking about, wouldn't you agree? In the example from our book, the op-amp has a low 10uV offset, but a fairly- high 8nV/rt-Hz noise level, so a simple circuit and a non-critical choice of parts and resistor values was OK. That's not the case with a 1nV mic input stage. I came up with a candidate part, and John suggested a possibly better one, this is something we're enjoying discussing.

--
 Thanks, 
    - Win

As a student of your book allow me to reiterate that a 1K ohm resistor exhibits 4nV of Johnson noise. You multiply that by the square root of the desired resistance to derive its noise level. Your above commentary leads me to believe that MOSFETs can be treated as simple resistors in regards to Johnson noise.

Thank you, 73,

--
Don Kuenz KB7RPU 
There was a young lady named Bright Whose speed was far faster than light; 
She set out one day In a relative way And returned on the previous night.

Rds(on), with zero DC current flowing, yes, I assume so.

--
 Thanks, 
    - Win

Everyone is supposed to know a "Mackie 16 ch desk" doesn't have any low-level mic inputs? Back in my broadcast days, all mixers had dynamic mic inputs. At age 17, I spent a summer designing and building a mixer for our local FM station, which they used for five years until they could purchase a new one, having blown their budget on a new GATES transmitter and antenna. My mixer had many dynamic mic inputs.

--
 Thanks, 
    - Win

From the days that stations bought equipment, expecting 25 years of reliabl e service. Gate and RCA built some real workhorses before Harris bought the m. The first station I engineered at had one AM and two VHF High Band Gates TV transmitters. The board in the radio studio was an early Gates, and we had an original Gates Spotmaster cart machine. Everything except the cart m achine was vacuum tube. Everything was transformer coupled, but I had a pro blem with ground loops because of a bad RF grounding system. The solution w as to cut all the shields from the ground terminals at one end.

The last TV station I worked at had a now 67 year old RCA UHF transmitter. It cleared final test at RCA on my date of birth! :)

---------------------------------------------

** Wins asks, having snipped his reading error out of sight. ** So does the Mackie - 16 of 'em.

Good low noise pres too and 48V phantom.

Guitar amps use 1/4 inch jacks, same as for LINE inputs on desks.

** Using tubes and input matching transformers ? Or some of 'dem new fangled Silicon BJTs ?

.... Phil