G7 Capsule Recommendations Please!

[mhelin]

Gyraf himself wrote on the project page that "The spources of inspiration were the Neumann U67 and U47, two legendary microphones that I was lucky enough to have around for reference and comparison when I did this project." Well the capsules back then that were offered and available were ct12 an ck12. So in the end it ended up being a cross of Neumann (circuit and tube) and Akg (capsule and the voltages used which changed everything). If the mic was designed today it might be a little bit different I guess.

 

[ln76d]

That's the thing. These mics have never sounded that clean. Actually kind of 'creamy' especially in the top end. I often have a problem with singers....dirty "esses"

This what you can find on schematic i posted should solve your problems.

Thanks for this info! That's quite a lot of changes to the schematic. Even the output cap is very small. Can you explain the thoughts behind these changes?

Most important change is input topology.
I prefer much more M49 option than C12 option. Also capacitor between both diaphragms 1nF is enough. You could also test lower values from 470pF - 1nF.  For cardioid best option is disconnected back diaphragm (or 1nF cap) but for this you will need to install switch or relay.
EF86 doesn't work great with 1G at the grid - test lower values in original circuit and you should notice difference. 80V (actually 75V or so) will work for most M7 capsules, but dfinately going below 70V you should notice also change not only in sensitivity.
Polarisation voltages also could be better matched. Depending on biasing and drop of the plate voltage, main polarisation voltage is changing from let say 70-80V, where the back diaphragm polarisation voltage is always around 80V (for cardioid setup - fig.8 160V). 
Output cap? I didn't measured it for lundahl transformer - but with other 6.5:1, 0.5uF-3.3uF will give you flat response to 20Hz.
Try 0.5uF, for example paper cap, you should like it. I wouldn't go higher than 1uF at this position. Strongly recommend use 4.7k trimpot, instead 1.6k resistor, to set the bias.
You have two microphones, so i recommend you to try each option in comparison to original circuit. Step by step. You should be able to judge easily what makes changes. I would start from input topology, change polarisation voltage, high z resistors, then biasing and at the end for example output cap.
BTW. Use definatley biasing scheme as it is on my version schematic. Mixed option as on original schematic with extra trimpot was for "test" features. Definataly no applying any feedback which have any specific feature makes it useless. You could even start with replacing 220ohm resistor with jumper and here for the start you should hear the difference.

 

[e.oelberg]

than again Kingston might be right. To me the shape of the head basket looks like a  acoustic lens. It might just cancel or amplify a certain part of the spectrum. Think of standing in the middle of a dome.

 

[Freddy G]

This what you can find on schematic i posted should solve your problems.

Most important change is input topology.
I prefer much more M49 option than C12 option. Also capacitor between both diaphragms 1nF is enough. You could also test lower values from 470pF - 1nF.  For cardioid best option is disconnected back diaphragm (or 1nF cap) but for this you will need to install switch or relay.
EF86 doesn't work great with 1G at the grid - test lower values in original circuit and you should notice difference. 80V (actually 75V or so) will work for most M7 capsules, but dfinately going below 70V you should notice also change not only in sensitivity.
Polarisation voltages also could be better matched. Depending on biasing and drop of the plate voltage, main polarisation voltage is changing from let say 70-80V, where the back diaphragm polarisation voltage is always around 80V (for cardioid setup - fig.8 160V). 
Output cap? I didn't measured it for lundahl transformer - but with other 6.5:1, 0.5uF-3.3uF will give you flat response to 20Hz.
Try 0.5uF, for example paper cap, you should like it. I wouldn't go higher than 1uF at this position. Strongly recommend use 4.7k trimpot, instead 1.6k resistor, to set the bias.
You have two microphones, so i recommend you to try each option in comparison to original circuit. Step by step. You should be able to judge easily what makes changes. I would start from input topology, change polarisation voltage, high z resistors, then biasing and at the end for example output cap.
BTW. Use definatley biasing scheme as it is on my version schematic. Mixed option as on original schematic with extra trimpot was for "test" features. Definataly no applying any feedback which have any specific feature makes it useless. You could even start with replacing 220ohm resistor with jumper and here for the start you should hear the difference.

I will try these mods, thanks for the explanation. Which is the mod that drops the polarization voltage?

than again Kingston might be right. To me the shape of the head basket looks like a  acoustic lens. It might just cancel or amplify a certain part of the spectrum. Think of standing in the middle of a dome.

I may also fabricate a body with a more traditional basket to see what that does.

 

[ln76d]

Shape of headbasket looks good - there's question how mesh affect response. Try definately without internal layer.
Anyway best option is just try microphone without headbasket and with it - this would give you most correct answer.
For the polarisation voltage mod, best would be if you could measure plate voltage after 10K resistor.
Anyway if you want to try, then change first (not connected to ground) 470k resistor to 680k  and inside PSU first (not connected to ground) 100k resistor to 150k.  This should give you around 64V for cardioid. If you want exactly same voltage, then you can make other mod - disconnect separate heater ground from the cable, connect this ground inside to circuit ground inside microphone and reuse free wire to make extra polarisation node which you need to connect between two 100k resistors (or 150k/100k if you want go lower with voltage), then you can remove 470k/470k divider inside microphone and just left one 470k resistor connected to pcb only from one side, from the other attach new wire node. Then you have exactly the same both polarsiation voltages for the cardioid. With this extra connection you can also replace first 100k resistor in PSU with 200k potentiometer and easily set most optimal for your needs polarisation voltage.

 

[dmp]

I have a G7 using a Dale capsule, same as yours. It's a nice mic but I think you'll be happy to have some other options. I would recommend a U47 style circuit, since the G7 is actually closer to a C12, IMO, due to the tube biasing.
I'll offer my $0.02 on the options out there. This is based on about 10 other tube mics that I have built.

- Capsule makes a huge difference. If you want one all around fantastic mic, you will need one of the best capsules, imo.  That said, I have C12 & U47 style tube mics that work great on certain instruments with cheap Chinese capsules. My go-to mics for drum overheads have Chinese capsules.

- The tube biasing is the most significant circuit ingredient to a mic's sound. The G7 is cathode biased like a C12, which puts a capacitor on the cathode resistor. Switch to a U47/U67 fixed bias and you here a distinct difference.

- in my experience, component differences are difficult to hear - you'll have trouble in blind tests to notice them. Some people love to tinker with it though and that's fine, but if you want a really noticeable change, you'll need to build a different style mic and try some different capsules.

 

[ln76d]

- The tube biasing is the most significant circuit ingredient to a mic's sound. The G7 is cathode biased like a C12, which puts a capacitor on the cathode resistor. Switch to a U47/U67 fixed bias and you here a distinct difference.

Fully agree except that C12 use different kind of biasing. G7kind of biasing you can find in ELA M250/M251 and M49C.
U47 have different bias than U67. U47 have grounded grid resistor, U67 with "cold" grounded cathode and negative ca. -1.8V applied to first grid. Some guys here will tell you that biasing have no impact on tube microphone work
Anyway for G7 type of biasing realy important is to find most optimal setup. So definately worth to use potentiometer and set it as best as possible

- in my experience, component differences are difficult to hear - you'll have trouble in blind tests to notice them. Some people love to tinker with it though and that's fine, but if you want a really noticeable change, you'll need to build a different style mic and try some different capsules.

Definately more important is component value than its type and here definately it's more audible

 

[dmp]

Fully agree except that C12 use different kind of biasing. G7kind of biasing you can find in ELA M250/M251 and M49C.

Yes, good clarification. But if you look at the C12 schematic, the grounded cathode actually goes back to the PSU where it is revealed to be a floating ground, which passes through a resistor (with a capacitor) to the true ground. So it is pretty similar to the more obvious cathode bias circuits (251, 49b). In my experience the mic sound reflects this.

U47 stiffens the cathode bias by putting the heater current through a smaller resistor without a cap. U67 fully grounds the cathode and makes a negative voltage in the PSU for biasing. Big difference in sound.

Definately more important is component value than its type and here definately it's more audible

Yes, agreed, I was talking about swapping components of same value. Start switching component values and you can really change how a mic sounds. You can make it stop making any sound at all, haha

 

[Freddy G]

Fascinating discussion both of you! Thank you.

I'm reading your comments and while studying the circuits you mention.

 

[Kingston]

Shape of headbasket looks good - there's question how mesh affect response. Try definately without internal layer.

Funny that. I would have said try with one more layer - or completely replace them with different thicknesses. Think of it as renovating the acoustic treatment of a room. It looks a little cramped in there.  I'd use more baffles.

 

[Freddy G]

I originally built the mic with a fine inner brass mesh layer.  After I installed Dale's capsule (originally I just had a cheap Chinese capsule in there) I removed the fine mesh from one of the mics and after A/B-ing both mics I decided I liked the fine mesh removed. It was a pretty subtle difference though.

 

[ln76d]

Yes, good clarification. But if you look at the C12 schematic, the grounded cathode actually goes back to the PSU where it is revealed to be a floating ground, which passes through a resistor (with a capacitor) to the true ground. So it is pretty similar to the more obvious cathode bias circuits (251, 49b). In my experience the mic sound reflects this.

Not exactly Dan, if you look on schematic. cathode is fully grounded at pin6 - main ground. There goes -1.1V from PSU via 30M and 250M resistor direct to grid so this design is much more similar to U67

Fascinating discussion both of you! Thank you.

I'm reading your comments and while studying the circuits you mention.

The whole fun with tube mikes is full circuit optimisation for the tube, capsule and output transformer. Microphone body and headbasket is another thing
Really worth to try many things especially that you have two microphones, so the comparison is pretty simple. You can learn a little bit from this

  I'd use more baffles.

And this usual ends with boosted sibilants (of course depends on mesh density). Even Neumann changed from three layers to two layers in most of current products. This shape of headbasket should be more transparent than for example M250/M251 headbasket.  Of course positioning of the capsule inside can also change a little. Best option is just test microphone with headbasket and without. THis clearly shows is it affect sibilants.

 

[dmp]

Not exactly Dan, if you look on schematic. cathode is fully grounded at pin6 - main ground. There goes -1.1V from PSU via 30M and 250M resistor direct to grid so this design is much more similar to U67

I just looked at the C12 schematic again and it is a little different than I described. Cathode is at the pin6 ground, however I called this a 'virtual' ground since it still is upstream of true ground with R8, which sets a slightly negative bias voltage from the return current.  Variations in return current however will 'wiggle' the bias, so are stabilized with C6. Power supply needs to be considered a part of a tube mic - plays a significant role in the behavior of the mic.  I still think the C12 fits more closely with cathode bias mics than U47 / U67 types but I could be wrong.. Anyway, you make good clarifications.

It will be tough to try different biasing on the G7 without building a new psu that can do the U47 method (requires more heater filtering) or U67 (negative bias voltage)

 

[ln76d]

I just looked at the C12 schematic again and it is a little different than I described. Cathode is at the pin6 ground, however I called this a 'virtual' ground since it still is upstream of true ground with R8, which sets a slightly negative bias voltage from the return current.  Variations in return current however will 'wiggle' the bias, so are stabilized with C6. Power supply needs to be considered a part of a tube mic - plays a significant role in the behavior of the mic.  I still think the C12 fits more closely with cathode bias mics than U47 / U67 types but I could be wrong.. Anyway, you make good clarifications.

It will be tough to try different biasing on the G7 without building a new psu that can do the U47 method (requires more heater filtering) or U67 (negative bias voltage)

U47 bias type could be done same way as Oliver did in his ef80 circuit. G7 PSU is definately good option to make negative heater voltage, worse is make proper connection on the mic psu but still possible. To make negative 6.3 there's a need to reverse rectifier connection, reverse electrolytic caps connection and change LM317 to LM337 and that's it
Still G7 biasing is pretty ok but definately i would replace 200ohm resistor with jumper and increase the value at least to 2.7k.

 

[e.oelberg]

there is one thing in the g7 circuit I really didn't like and it's the bypass cap for the bias 47u cap.

 

[Freddy G]

there is one thing in the g7 circuit I really didn't like and it's the bypass cap for the bias 47u cap.

Do you suggest removing it entirely?

 

[gyraf]

Do you suggest removing it entirely?

It's redundant - background story is that I got a big batch of REALLY low-grade electrolytics around that time, and only later realized that the parallel-polyester really shouldn't be necessary...

Jakob E.

 

[dmp]

U47 bias type could be done same way as Oliver did in his ef80 circuit.

Yes, this is what I was referring to since I used that method in the EF800 u47 style mic I built. The heater needs more filtering however than the G7 has.  If the G7 PSU were used unmodified the mic would be noisy.

 

[Freddy G]

Here's what you can do with the circuit.
100M as example, it can be 150M, 200M, 250M, 300M etc. I wouldn't go higher than 470M.
Also i recommend E80F tube if you will be able to fit it.
Dale M7 is good capsule i would definately keep it here.
Also for it i would reccomend to lower polarisation voltage if it's 80V like in original circuit.
Adjust bias on cathode doesn't hurt also

Hi In76d,

I got around to testing some mods. The headbasket test proved to be very subtle. So I don't think that's a problem.
The output cap got switched out to 1uf (I know you recommended 0.5uf, but 1uf was all I had at hand) and it definitely reduced the low end. Then I removed the 220r resistor and replaced with a jumper. Then installed a 4k7 trimpot and tested various bias settings.
Everything was going along well until I modified the section near the capsule. The two 1 gig resistors swapped for 100Meg, the removal of the 1n cap between the one of the 100M resistors and the center of the capsule. The swap of 1n, 10n capacitors and all the rest as shown in red on the mod scheme. Now I get very low gain and bad sound. I've checked and double checked and I can't see any error I made. Not sure what to do now!

By the way, here are some A-B test files if you care to listen.

Unmodded G7:  https://app.box.com/s/9zf237xy2h3ibpz3jp132aw2l32e9sho
1uf output cap:  https://app.box.com/s/s2ypnihcav8wll8prdrf7w4rra53uq9q
220r jumpered:  https://app.box.com/s/riq2gjex996mj3hi9ak0e3421ij72w1y
2.7k bias :  https://app.box.com/s/t59bd3itg1gp00vbcb84yamv8hrymd5f
2k bias:  https://app.box.com/s/felx41e8rkp77f63x3n2i082ya5ok0ou
4.7k bias:  https://app.box.com/s/cjcp29lq732di57wxk1glv65n8y8rq39
700r bias: https://app.box.com/s/jnlrafbuqgdhi6c5b71fgdv62ag92f5t
input mods: https://app.box.com/s/okbg1zy3711ux5dzikxcqux53h1n4h4n


And just for reference, this is the other unmodded G7 from a recording session today. It's on the ukelele and also the vocal. No EQ on the vocal, just compression and reverb :  https://app.box.com/s/wl2aso1a5kmc3rxrxxodlslpqkavcjfn

 

[Kingston]

And just for reference, this is the other unmodded G7 from a recording session today. It's on the ukelele and also the vocal. No EQ on the vocal, just compression and reverb :  https://app.box.com/s/wl2aso1a5kmc3rxrxxodlslpqkavcjfn

You should have started with a sample like this and we could have told you to stop and simply enjoy your superior mic.  It needs absolutely zero adjustment or further tweaking.