Neumann TLM circuits

[Gus]

Has anyone traced the TLM103 M147, M149 or have schematics?

The reason I ask is I have a TLM103. It does not sound bad however it has a edge to my ears. when I opened it up it LOOKS like it uses ceramic SMD caps in the signal path and the output electros seem to be on the small side, I don't know what brand they are because I have not peeled the sticker neumann used to show it is a circuit diagram 2.

I was going to trace the circuit : however I can't see any markings on the chip caps and some of the transistor don't seem to have a marking at all. I have measures the caps in my rode nt1000 they were easy to remove the tlm103 has a coating that looks to hard to remove.

The bottom of the board looks to have a standard 74c14 dc to dc converter circuit.

Why am I asking? I think the design of the TLM103 is good but maybe some of the parts(the caps) are not as good as they should be. When I replaced the ceramic caps in the signal path of my rode nt1000 with polypros and a polystyene and used two 100uf HFs I had saved to replace the output caps it sounded alot better less edgy IMO.

I will trace the tlm103 eventually but I thought I would ask first.

I do understand the the use of a low z drive circuit being used with two macthed resistors as the type of circuit used. Neumann's infopool has a good PDF in german on transformerless microhones and a few circuits are shown

I guess a the big problem tracing the circuit is the chip cap values I don't know how to remove them without breaking them they look like they are glued down with a comformal coating..

 

[soundguy]

wow, if there is a doable way to improve the sound of a 103, Id be all over that. Every time I set that mic up, I always feel like it could be %60 better, such a disappointment.

dave

 

[soundguy]

I suppose this translates as:

"We've worked very hard to make our new microphone designs as uninteresting sounding and bland as possible, and under the direction of sennheiser, we need to keep this top secret from all the other companies who want to make sucky characterless mics. If you want a cool sounding mic, I reccomend searching out available neumann schematics from older neumann models which actually sound cool, boo hoo for you though, you'll have to use a dreaded transformer..."

Ooh, I sound bitter.

dave

 

[JPrisus]

Bitter but definitely right on target. That's one company that won't get a dime from me unless they make a complete 180, what with all of other options out there. Soundelux makes better Neumanns than neumann lol.

 

[dale116dot7]

There are about two caps in the signal path - the one from the capsule to the preamp, and from the preamp out to the mic cable. The other caps do not matter at all - their frequencies of interest are way too high, or they are for DC filtering, where ceramic caps do just fine.

Caution - soldering on a ceramic PC board can be very difficult if you aren't set up for it, and if you don't have the experience to do it.

Anyways, the schematic for the preamp part only is here:

http://www3.telus.net/~dulan/tlm103sch.pdf

 

[dale116dot7]

Oh yes, this was meant to operate in a simulator so V3 is where the capsule would be, the other DC voltages come from the charge pump circuit located on the back of the board.

This is configured as a pseudo-op-amp circuit. The output comes from the output bipolars, the bottom JFET is set up for source-fed feedback, and the top JFET is used to help keep the JFET linear. It's similar to what AKG has done for years. This preamp should be quite linear, and very transparent. Except maybe the caps.

 

[volki]

dale, interesting.
similarities with akg's later designs indeed come to mind. IIRC, they used a 3rd FET on the source leg of the middle FET, instead of a BJT (Q2).

so Q6 (middle FET) and Q2 obviously form a kind of compound cascode, or something? but where do all those other BJT's come in there? that's also what i never quite understood about the newer c414-circuit, either... :sad:

 

[volki]

dale, btw, did you reverse-engineer this? i thought all the tlm designs feature IC gain blocks, no discrete parts except maybe those 2 FET's...? like e.g. TLM 50, TLM 170...? but this one looks pretty discrete to me...=)

 

[dale116dot7]

Yea, all discrete except for the charge pump that I didn't do yet. The operation isn't as obvious as you'd think, but it is pretty straightforward.

Q2 and Q3 make up a push-pull output stage, D2 and Q4 bias them. Q1 is a voltage amplifier, Q5 is used to linearize and raise the input impedance of Q6 by forcing the drain of Q6 to follow the source - not quite a cascode, and Q6 is basically configured as a common-source amplifier with 100% feedback applied to the source - so the whole mess acts as a unity-gain buffer.

 

[PRR]

> a kind of compound cascode, or something? but where do all those other BJT's come in there?

The drawing layout style is misleading. Probably for simulator convenience. I don't swear this is accurate or complete, but it makes more sense to me if re-drawn this way.

First stage an FET (with cascode).

Second stage a PNP BJT, with NPN current-source load.

Third stage an ordinary NPN-PNP complementary output pair.

Unity-gain DC and AC feedback to the input stage.

 

[volki]

PRR, that makes sense indeed. i think i should redraw those u47 fet and u89 circuits as well, they might get a lil' clearer then...?

dale, so i figure the tlm103 is mostly smd, then...?! or not, even :?:

regarding the negative feedback (0db gain)... that should also increase the input headroom for the FET stage, right? so it won't distort until spl gets really high (plus, there's an additional pad which lowers the polarizing voltage, IIRC). although high NFB greatly reduces low order harmonic distortion, it also means a) increase of high order harmonics and b) very abrupt clipping instead of soft onset of saturation.
however, a great part of the soudn that some of the "TLM haters" :green: dislike seems to be due to the capsule freq response...

 

[Gus]

I have mine in pieces(TLM103) and am finding caps that I can make fit. I will take pictures of the work. The edge I hear sounds like the use of a ceramic with a good amount of voltage difference across it, this stresses the piezo materal and it also has a time constant. FWIW the Sencore lc102 has a good section in the manual about how it tests around ceramic cap reactions to voltage.

I have not tried a paper and oil cap at spot yet but I do like polystyrenes.

What I don't understand it if you look at older Neumann microphones like the km84 you will find polystyrenes at that spot. Polystyrenes need good soldering skill as not to over heat them I guess that does not work with SMD.

Tim H. maybe you can give your take on the sound change.

One of the simple schematic in the neumann PDF microphones is of the fet47.

I finished another ss microphone yesterday.

 

[volki]

gus, dumb question: Sencore lc102 - what's that all about? is there a url?

about those ceramic's, where exactly are they located? are those the dc blocking caps before the FET gate and after the output stage?? (yikes)

 

[PRR]

> so it won't distort until spl gets really high ... although high NFB greatly reduces low order harmonic distortion, it also means a) increase of high order harmonics and b) very abrupt clipping instead of soft onset of saturation.

Well, yeah, but clipping will be up around 3 Volts RMS. Your mike preamp will overload first. And as you note, there is a way to drop capsule output.

I have not analyzed this, but a design like that should have TONS of feedback, enough to put all distortion products down in the noise at any sane level. Compensation does not look too heavy-handed.

You might check the idle current in the output stage. DC voltage across R7 should be more than 30mV.

The output cap is none too generous. It is probably fine, but 100uFd might be different.

If you are hearing something funny, it may be the capsule. Or even your mike amp not liking this low-low 20 ohm source.

 

[volki]

[quote author="PRR"]
Well, yeah, but clipping will be up around 3 Volts RMS. Your mike preamp will overload first. [/quote]
definitely. distortion at high spl is too often attributed to the mic itself, although, especially with more recent high output mic's, it does occur in the pre.

You might check the idle current in the output stage. DC voltage across R7 should be more than 30mV.

thanks for the hint, but i'm not talking about a particular unit (since i don't own one myself :green: ), but in general, rather.

... your mike amp not liking this low-low 20 ohm source.

i think this will be mostly true with xformer-coupled pre's, since the complex behaviour of the associated inductances/capacitances might get shifted to freq's where they're not supposed to be.

- something i just thought of: what about optimum source impedance, anyway? the TLM's actually boast super low noise, but wouldn't that feature be lost with a pre that is actually designed for an OSI of 10 times higher? so if presented with a source impedance of only 20 R, i assume noise will increase?

 

[dale116dot7]

I drew it quickly and didn't make it that easy to follow, sorry. Yea, your photo is a lot more clear. I don't know about some of the cap values, I just guessed by their colour. Most of the resistors are printed resistive ink with laser trimming.

If you're getting up to 3 volts of output from the mic, you're well into distortion from other places like the diaphragm smashing against the backplate and that sort of thing.

I can't remember if I made some mistakes in transcribing the circuit, I may have a cap value off here or there - maybe the output cap is bigger, I just don't remember.

-Dale

 

[volki]

[quote author="dale116dot7"]If you're getting up to 3 volts of output from the mic, you're well into distortion from other places like the diaphragm smashing against the backplate and that sort of thing.[/quote]

:green:

a very respected mic tech i know just told me recently that the collapsing of a membrane to the backplate practically only occurs if the polarizing voltage is made too high, and that it's supposed to be very hard (if not impossible) to make the diaphragm touch the backplate by applying high SPL. but maybe he was only talking about certain types of capsules...? from an example in the Neumann microphone book, it says that at 94db SPL, a certain type of their capsules, with diaphragm spacing of about 30 micro-meters, only exhibits diaphragm movement in the magnitude of nano-meters...! how much might that be for 150 SPL? i figure that with the complex mechanical laws, it's not a simple logarithmic relationship...?!

of course, the closer the diaphragm excurses to the backplate, the more you do get electrostatic attraction to the very backplate itself...

the onset of membrane distortion is gradual, anyway. but i was always wondering what max. voltage SD or LD capsules can deliver, even when distorting, and regardless of overload in any successive stage?

 

[Gus]

The two output caps on mine are marked 47uf 50V. It looks like the output resistors for the balanced drive might be laser trimmed to take into acount the finite output R of the amp section. one side is driven by the output of the amp and the other side goes to ground via a 47uf cap and a resistor both resistors need to be matched.

AKG, mackie,neumann,rode and ? use this type of balanced drive.

 

[PRR]

> so if presented with a source impedance of only 20 R, i assume noise will increase?

A) No.

B) The noise level in this type capacitor mike is usually set by the FET at the capsule (and by acoustic resistance), not the preamp following the mike.

> diaphragm smashing against the backplate and that sort of thing.

On these high-output mikes, 12mV at 94dB SPL, 3V output is "only" 142dB SPL.

Mikes are usually not harmed by acoustic pressures well above 150dB SPL.

> with diaphragm spacing of about 30 micro-meters, only exhibits diaphragm movement in the magnitude of nano-meters...! how much might that be for 150 SPL? i figure that with the complex mechanical laws, it's not a simple logarithmic relationship...?!

To a good approximation, it "has to be" a linear relationship, sound pressure to displacement. So if it were 1nM at 94dB SPL, it would be 631nM (0.6uM) at 150dB SPL. If it weren't linear, it would distort.

In these mikes, 150dB SPL is about 7V RMS output. Most mikes will clip in the internal amplifier (unless you use the capsule pad). Many modern mikes use an internal 9V to 15V supply for the amplifer, and unity output coupling, so 3V RMS is about all they can do. (and with phantom supply, mike loaded in a 200 ohm test load, this is about all the power you can extract from Phantom.) Some non-phantom tube mikes use 100V-200V supplies with a 7:1 output transformer, which comes to about the same thing after tube losses.

> the closer the diaphragm excurses to the backplate, the more you do get electrostatic attraction to the very backplate itself...

In fact, yes it does distort. There is a second harmonic that rises with level. At 150dB SPL, 0.6uM in a 30uM space, the 2nd harmonic is measurable but not objectionable. It gets gross by around 180dB SPL. At this acoustic level, deafness is instant. Output voltage for mikes like these computes to 300V RMS, so you will have clipping.

 

[volki]

To a good approximation, it "has to be" a linear relationship, sound pressure to displacement. So if it were 1nM at 94dB SPL, it would be 631nM (0.6uM) at 150dB SPL. If it weren't linear, it would distort.

of course the relationship has to be linear until you reach very high SPL. what i mean is that a) with increasing displacement, compliance will start to decrease at some point, which is where distortion starts. now this effect is linearized by damping of the air cushion between diaphragm and the backplate with all it's holes, but at very high spl, i figure it will occur in interaction with the electrostatic attraction i mentioned last time.

another thing not to be neglected would be the fact that the diaphragm displacement is not distributed equally across it's area - at the edges, it will be low, and high at the center, and in between, it doesn't develop a linear shap, so the more displaced regions don't exactly make up for the less displaced ones. so if you're really picky, this means that, although the displacement - SPL relationship is linear, the spacing of diaphragm and backplate, and thus the resultant change in capacity, only approximates to linearity to a certain degree. i figure this effect occurs less with true omni capsules, which have a closer packplate spacing (and higher diaphragm tension)...? - there is an iteresting publication by neumann on the subject, they do measure different amounts of difference-freq distortion with omni's and cardioids. i believe it's on their website info-pool.

In these mikes, 150dB SPL is about 7V RMS output. Most mikes will clip in the internal amplifier (unless you use the capsule pad)

.
of course all this discussed distortion will mostly be masked by the head amp's behaviour. when you use the pad, it's only when you close-mic very loud sources anyway. so, since you wouldn't even be listening to these loud sources in the nearfield with your ears (unless you want them gone), you can't even tell exactly how it actually sounds like - so you can't compare that to the sound of the source recorded thru the mic, which means that it's not relevant if the timbre might change a little due to some distortion in the capsule. does that make sense...?