The Ultimate M49

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

micaddict

Well-known member
Joined
Jan 3, 2012
Messages
1,805
Location
The Netherlands
There's quite a bit of love for the M49. More than some even prefer it over the famous U47. Yes, application plays an important part, of course.

Anyway, there were several versions of the M49. And the original design is almost seventy years old. (!)
Arguably the biggest change was from -b to -c, when fixed bias became self bias. (I'll ignore the change from M7 to K49 for now.) Singers, like me, tend to prefer the M49b, but there's also a lot going for the M49c.

Unless you're determined to perfectly clone one given version, you could go for a best of both worlds approach. Plus you could implement things that weren't possible or available back then. The latter can be taken too far as is the case IMO with modern Neumanns. The M149 is not fully discrete and it lacks a transformer; the TLM49 lacks the tube to boot. Neither of the two float my boat. And I believe some of the guys here  can come up with a better revised version.

To get the discussion started I'll add some quotes about mixing and/or improving te classic -b and -c versions.

The first is from the late Dennis Stone (kidvybes) who bought Dany Bouchard's M49b prototype (D49 PCB). Brian Fox from Fox Audio Research then made some changes:
...OK, I thought I'd chime in to update the status of the M49b I purchased from Dany a week ago...since the mic had no power supply (and I'm not gifted in the DIY skillset) I had Dany forward the mic to Brian Fox (Foxaudioresearch.ca) who is also in Canada...I've admired Brian's work, and his exceptionally informative website, so I thought this might be a good opportunity to establish a dialog and commission him to modify a PSU to service the M49b...

...Brian has carefully researched the build options (both b and c) and the specific needs in reference to modifying a standard OEM Chinese power supply...he asked me from the get-go if I was locked in to keeping the mic true to Dany's B build specs, or if I would be open to some minor tweaks that might benefit the mic's performance, as per my particular needs...I told him to feel free to suggest options and we would go from there...

...Brian's initial testing focused on the mic's filament bias circuit...noting that in the revised C version, Neumann converted to a cathode bias, he suggested making that modification, to limit the amount of filter capacitance needed in the power supply...he noted the only thing special about the kit supply was that it had enormous amounts for filament power filtering just so the filament bias method would not inject hum into the mic (this was not needed with an indirect heated tube like the 5840)....with my approval, he made the adjustment to generate the bias in cathode as in the C, but it with a fixed voltage of 1.6 like the B...Brian does this by using a red LED in the cathode circuit instead of a resistor, which creates a fixed 1.6 volts with perfect low end response, but with much lower hum and noise than using a filament based bias voltage...

...with that he could re-wire the M49 7pin XLR so that the mic will work with any Chinese supply that is voltage adjusted to 120 volts or less...he modded the OEM chinese supply to produce 116 V volt B+...measurements showed the preamp has a net gain of +1 dB and can produce -10 dBu output before hitting 1% 2nd Harmonic distortion, and distortion is pretty even from 100Hz to 20KHz....

...I hope I have detailed Brian's approach properly (as I'm no "tech" by any measure)...this morning he sent some graphs and a sound file...he offered me the opportunity to further "sculpt" the mic's sonic character, by implementing an EQ cap (270 pF and 470 pF)...but I'm pretty satisfied with what I hear, and would prefer to stay as close to Dany's build, other than the biasing modification...so here is the response graph and sound file...any input or comments are welcomed...


The second is from Dave Thomas (Advanced Audio Microphones) who was recently working on an original M49, but who also produces an M49-inspired mic called the CM49 . I'll start with a Gearslutz member who posted this:
Also, do you know what version M49 you are rebuilding right now? Is it the 49, 49a, 49b or 49c? In my experiences the "b" version sounds best (at least for orchestral film score and vocals), it has the most bottom end while still having that presence bump in the upper mids. The "c" version doesn't have as much bottom end and so it ends up sounding somewhat thin compared to the "b" from my experiences. I know if I'm using a mis-matched pair at a studio on piano or cellos, I can pretty much immediately hear which one is the "b" and which one is the "c" because of the difference in the bottom end. Not sure what the technical differences are between them that creates that difference in sound (maybe a different transformer?). Most studios here in LA have either the 49b or 49c from what I have seen. I haven't noticed any 49 or 49a floating around. If I have used one, I couldn't hear a difference in it and didn't notice it wasn't a "b".

I would be interested to see what your take is on the 49 you are rebuilding!

Cheers Dave and keep up the great work!
And here's Dave's answer:
CM49 design criteria vs M49
Hi Guys, its a little more complicated than that. (re. fixed versus self bias)

True, the M49b uses fixed bias and part of the reason it has more low end is that no cathode bypass capacitor is required which rolls out the low end response.

The M49b with its fixed bias does not require a bypass capacitor and the low frequency response of the tube is "virtually" flat down to DC.

Derek comfirmed this with his observations which agrees with electronic theory.

I have included a screen print of a cathode bypass capacitor calculator for a tube with the same gain as the AC701 and a 25ufd bypass capacitor.

The difference in level between 10hz and 1khz is 10db. With the M49B circuit or our CM49 circuit this line would be dead flat down to DC with a test tone.

Now, in practise the impedance of the capsule and 150M resistor on the input grid of the tube will reduce the low frequency response slightly.

However, this happens in both the M49B and M49C and the difference Derek is hearing relates to the more dramatic cathode bypass capacitor, LF reduction.

The largest capacitor they could find that would fit into the M49c body made back in the day was a 25 mfd which will definitely start to roll out the low end response.

Today, for example in our CM48T with a "traditional" U47 circuit we trade-off the fixed bias for self bias but can easily fit a 220ufd capacitor across the cathode.

With new capacitor technology we can build traditional single stage cathode bias circuits with practically the same LF response as fixed bias circuits.

Self bias also introduces some negative feedback and this will lower the output impedance of the tube circuit which is a good thing.

How, did we address these issues in the CM49:-

We use a brilliant CCDA circuit. This involves using a dual triode tube.

The CCDA class "A" DC coupled circuit is a designers dream.

The first stage uses cathode bias but no bypass capacitor is required because the 2nd half of the tube provides a very high impedance load much higher than the output transformer in a traditional single stage circuit.

The plate of the first stage directly couples to the 2nd stage. We loose about 4db of gain without a bypass capacitor but the 6072 has a few db more gain than the tubes used in a vintage U47 or M49 microphone.

This gives us the advantage of the fixed bias circuit with the reliabilty and lower plate output impedance of the self bias circuit.

As a tube ages, plate current decreases. Fixed bias is calculated for an optimum plate current, as these parameters of the tube change, distortion increases much more dramatically than with self bias.

As, the plate current decreases with age, in a self bias circuit, the bias voltage changes accordingly to minimize distortion.

In the CCDA circuit the final output of the tube is taken from the cathode of the 2nd stage where the output impendance is down at 600 ohms. This is 10 times lower than that of a U47 or M49 circuit.

So, this circuit produces much less low frequency transformer distortion than a single stage circuit because it dampens the transformer 10 times more quickly.

The other advantage of the CCDA is in its name. It a constant current draw amplifier and the two sections act as one closed loop circuit. As the input section goes postive the output section goes negative and the current draw from the power supply remains constant.

The BV18 transformer that we use in our standard CM49 has an 8:1 ratio and produces a smooth response down to 10hz with less distortion than our BV8.

This BV8 with its vintage laminations and lower turns ratio produces more LF transformer distortion increasing the LF content in a musical way.

The BV18 is the same transformer we use in the CM12se and there is not enough room in the CM12se or CM251 body to fit the larger BV8 transformer.

With both our CCDA circuit or the single stage CM48T, CM800T and CM269T (in prototype stage) we use 1G (1000M) grid to ground input resistors which increase the LF response down at 10hz-20hz by nearly 3db.

So the question is, with the current knowledge and materials, but with respect for the classic design, what would be the ultimate M49?
Or how do you guys feel about the approaches as taken in the above quotes?


 
yeah, I read about Brian Fox M49. I'll definatelly try the LED approach in my build. It just sounds like it would deliver the best of both worlds.
There are also some mods to bypass the broadcast filters introduced in the b and c version. Like omitting C6 and R3. They were not there in the IRT version. There is definatelly room for experimentation. 
I'll probably use the LED approach with the IRT circuit.
 
One question: Why creates a red LED in the cathode circuit instead of a resistor fixed 1.6 volts ?
 
Because that's how diodes (normal ones, LEDs or otherwise) work - when forward-biased, they have a (relatively) fixed voltage drop across them.

https://en.wikipedia.org/wiki/Diode#Forward_bias

In a small silicon diode operating at its rated currents, the voltage drop is about 0.6 to 0.7 volts. The value is different for other diode types—Schottky diodes can be rated as low as 0.2 V, germanium diodes 0.25 to 0.3 V, and red or blue light-emitting diodes (LEDs) can have values of 1.4 V and 4.0 V respectively.

kosi said:
One question: Why creates a red LED in the cathode circuit instead of a resistor fixed 1.6 volts ?
 
as far as I experimented higher grid to ground values will give you a clearer and more defined and lower bass. On the other hand it will sound less warm, less tube and less smooth. in a self bias circuit the grid to ground tantalum / lytic is also crucial for the sound.  47uF will give you more bass than 22uF electrolyt will sound faster and more detailed tantals a bit more  soft.
this expirements I did with a 251 circuit but in general they should also be true for the m49.
 
Khron said:
Because that's how diodes (normal ones, LEDs or otherwise) work - when forward-biased, they have a (relatively) fixed voltage drop across them.

https://en.wikipedia.org/wiki/Diode#Forward_bias

Okay ! So he basically uses the 49c schematic and replaces R7 (22k)  with a LED ? What about C9 ?
 
e.oelberg said:
as far as I experimented higher grid to ground values will give you a clearer and more defined and lower bass. On the other hand it will sound less warm, less tube and less smooth. in a self bias circuit the grid to ground tantalum / lytic is also crucial for the sound.  47uF will give you more bass than 22uF electrolyt will sound faster and more detailed tantals a bit more  soft.
this expirements I did with a 251 circuit but in general they should also be true for the m49.

First of all you could just change value of fixed hpf cap - it doesn't change anything else in case of other sound properties, just more low end ;)
 
kosi said:
Okay ! So he basically uses the 49c schematic and replaces R7 (22k)  with a LED ? What about C9 ?

If I can see it right,  C9 is omitted and replaced with a 0,1uF cap in parallel to the LED.
Here is the Schematic from Brian:

https://www.gearslutz.com/board/attachments/so-much-gear-so-little-time/406067d1404706455-adding-go-vocal-mic-m49-microphone-test-report-stone-page-004.jpg

what type of cap would be a good choice in that position? 

Also, Brian suggest to omitt R5 (150M grid to ground resistor) because 5840 doesn't need it. What about 6S6B tube?
And is this 0,1uF cap needed with the 6S6B tube? 
 
Hi,
0,1uf is a bypass cap for the led, makes it (even more) quieter. A film capacitor should do, preferably good quality but no need for a 630v rated monster.

About grid resistor i would say use a 150M with 6s6b... and try without too! But intuitively i would say it is needed.
 
Hmm some Dave infos are, i don't know even how to  specify, especially when i really like this guy ;)
Anyway!

kosi said:
One question: Why creates a red LED in the cathode circuit instead of a resistor fixed 1.6 volts ?

I will ask a little bit more specific - what's benefit (for audio signal) with use of led?
I tried led on cathode only once in my own circuit. It was test (via switch)  led vs. resistor  (both same drop) with led - microphone was sounding worse.
Is there any benefit (except virtual noise and other bla bla)?


 
Use dictionary :D :p

Worse is worse  - i will not describe the subjective feelings differences.
Let say it was worse sounding for me.
It doesn't matter here, for the topic.
 
ln76d said:
Hmm some Dave infos are, i don't know even how to  specify, especially when i really like this guy ;)
Anyway!

I will ask a little bit more specific - what's benefit (for audio signal) with use of led?
I tried led on cathode only once in my own circuit. It was test (via switch)  led vs. resistor  (both same drop) with led - microphone was sounding worse.
Is there any benefit (except virtual noise and other bla bla)?

hmm, I really don't know much about electronics. But when using the LED you can leave the elko cap out. This probably makes a difference, right? As I read, that Elkos are to be avoided where it's possible...
 
Merlin - it depend on which topology you build and both have they own properties.
Only "C" use cap on cathode.
Try Elna Silmic II on cathode it's pretty good and much different than many other electrolytic caps.
Want better result??  Use two in parallel  ;)

 
Khron said:
Without a double-blind test, i'm not buying those results ;) Just saying... :D

First of all - who said(wrote) that i didn't made a blind test?
Maybe just ask first then make your judgement ;)
That's why i don't want to write about my subjective feelings, because it doesn't matter here, and especially for that you wouldn't have to "buy" anything (or any one else).
One thing always surprise me in you answers - you often try to "troll" a little my posts ;P
Why? What for?
 
Nothing personal, it's just that "absolute" declarations along the lines of "this is the (only) right way (because i say so)" kinda tickle me the wrong way ;D

No offense or trolling intended though, i promise :)

ln76d said:
One thing always surprise me in you answers - you often try to "troll" a little my posts ;P
Why? What for?
 
About the led bias, i didn t heard it used the way B. FOX use it in his mod so can t comment on sound but i ve heard this method used with a CCS loaded GC in a preamp stage (Ccs in place of R plate) and it sounded quite nice and for my taste it was an improvement over a standard GC stage using same condition.

However, the ccs GC sounded more 'musical' with a resistor in place of led but less detailed for me...

Anyway, this is a very different topology than the M49 original one and im not sure wether or not it would suit an mic impedance converter or not, neither if it will suit taste or expected sonics of an M49 as this induce less distortion than a conventional GC stage, given some comments about differences between tlm149 and original m49.
 
Back
Top