My U47 408A (D47) Build

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
A very nice and informative post!
Circuit design comment:
The high-ish heater voltage of the 408A is utilized with two tubes in series, to match the heater requirements of a VF14 replacement.
This configuration would be physically different anyway, so why bother with two tubes, when a larger dropping resistor could work? Does it make the noise 3dB lower? Copying a crisis era circuit with their contemporary materials limitations, for some sonic magic may not be such a good idea.
The cable used has several paralleled conductors, which could be broken out to provide a separate heater power, of any suitable voltage, for any suitable tube. Why not?
The 1.5K heater dropping resistor carries all the whatever noise at 42.4mA from the HV PSU, and this gets divided down with a 27R to be ~39dB lower signal (or some other number with resistor inductance included) which is injected into the input signal. No bypassing is done here. Why not? The PSU box has plenty of space.
The heater current would essentially swamp out any current variation of the tube(s) and negate any negative feedback from this resistor, which would not be much anyway with a 100K load of the plates. (-73dB). The current from the heaters and the tube(s) would be ~42.4mA (a little less than the 50mA spec for the tube), 2.7W into the 1500 Ohm resistor. Is the location of this resistor not better inside the power supply, or is the extra heating useful for some other purpose?
The comments above would indicate 44.3V (?) on the plates, which would translate into a combined plate current of 0.467mA, well outside the 7 - 9 mA range for "typical" datasheet RF operation of the tube as a pentode. I cannot find any grid/plate curves for this tube in a triode configuration.
Clearly the tube(s) are operated outside their design intention, 100Meg on the grids is only 100 times the max recommended. No grid stopper for RF reduction either. The shielding may be sufficient for that.
The measurement of the voltage divider includes the load of the DMM, typically 10 Meg, so it would be lower than the calculated 63V.
The choice of reed-relay is for a HV type, but the grids carry a very low voltage, unless the capsule breaks down and shorts to the grids with 2 micro joules of charge from the 10nF cap. Not sure how likely this would be.

Some ideas, maybe wrong:
  1. The output transformer could be in series with the 100K plate, the low current should not saturate the core(?) Then the coupling cap is not needed.
  2. Use one tube.
  3. Separate the heater supply with its own wire, DC regulate it to a suitable value for the tube used. The ground return could be used if the regulator noise is low enough. Allows for easy experimentation of heater voltages.
  4. The tube cathode bias could be an IR-LED, seen in tube phono amps, here selected for its forward voltage at the relevant current. Not sure if the noise would be lower. Several types could replace the 29R resistor with 45 mA heater current. I'd be curious to know.
  5. Bypass the 29R or IR-LED with a very low ESR cap, like OSCON or similar. 5 milli Ohm ESR is parallel with 29R or an IR-LED could not hurt. Could be 75dB better.
  6. Battery bias. Batteries have low noise, should last a long time in series with 100Meg. then the cathode(s) could be grounded.
  7. High value resistor have higher noise. Metal foil and wire wound types have better tempco and lower noise, could be physically larger and are not accountant friendly.

my $0.02
 
I love this mic exactly as it is. I have no interest or desire to change any part of the circuit from my previous build, though I do want to make another one.

The one thing I might try differently in the 2nd build is to put the heater resistor in the PSU box instead of the mic. That's not a circuit change, just a location change. I'm not sure why it's located in the mic, I haven't heard an excuse for that yet. Inside the PSU box, we can go even higher wattage/better heatsink size and expect lower heat.

After I build it that way, I'll use my thermal camera to see what the temperatures look like. If it performs as expected, I may modify the original to match.

Thanks,
Josh
 
It’s in the mic because the anode circuit requires the full 105 volts, and the mic was designed to run from a single voltage supply.

A lot of variants have been designed and built that run off a dual voltage supply, but this isn’t one of them. If you do that you don’t even need the dropper, but you do need to utilise an extra conductor in the cable and another pin in the connector.

Why don’t you build the EF version? Then you can compare them 😎
 
It’s in the mic because the anode circuit requires the full 105 volts, and the mic was designed to run from a single voltage supply.

A lot of variants have been designed and built that run off a dual voltage supply, but this isn’t one of them. If you do that you don’t even need the dropper, but you do need to utilise an extra conductor in the cable and another pin in the connector.

Why don’t you build the EF version? Then you can compare them 😎
I did build the EF, and I did compare them:
Isolate left and right speakers to isolate mic builds.

I like both mics, but I prefer the 408A. That EF800 version is in a friend's studio, and he loves it. The EF800 has a Beesneez capsule, and the 408A has a Heiserman capsule, both M7s.

Thanks,
Josh
 
Last edited:
Well, if you look at the schematic you see there's a 30K and the 1500R coming in from the mic cable for B+ and the heaters respectively. If we place the 1500R in the PSU case, and adjust the 30K to 28.7K, then it would be electronically about the same. You need to measure the voltage in the mic anyway when adjusting the B+. So voltage loss over the cable is irrelevant if you've already compensated for it. I could be wrong, but it's worth testing anyway, unless an EE has a reason why the idea sucks.

It doesn’t work because the 1500R is a voltage dropper that uses the heater current to drop the heater voltage to ~40v

If you connect your plate resistor to that node, you no longer have the required voltage at the plate.

So the answer to the problem of the dropper resistor is the separate heater supply, which if you like can still be used to provide fixed bias to the cathode.

Great video, btw.
 
Last edited:
It doesn’t work because the 1500R is a voltage dropper that uses the heater current to drop the heater voltage to ~40v

If you connect your plate resistor to that node, you no longer have the required voltage at the plate.

So the answer to the problem of the dropper resistor is the separate heater supply, which if you like can still be used to provide fixed bias to the cathode.
Ah, okay, that makes sense. Thanks for the explanation. I guess that saves me some time. 😉
 
Of course remoting the heater dropping resistor would require a separate heater wire, kinda selfevident. Maybe a problem in 1947, but as these cables have 7 conductors, 3 for ground, where one can be repurposed.
Putting that ~ 3W heater inside the mike seems redundant, and additionally reduces the choice of tubes.
In 1947 it would have been difficult to make a low noise DC filament supply, which also explains the choice of tubes, not magic.
As low voltage DC regulation difficulty is no longer an issue, it is hard to now defend that design choice.
A variable DC heating supply would be super easy to do, accomodating various tubes and voltages.
 
You're both right, and swapping the heater resistor into the PSU box, and switching the mic cable wires (Gotham 7 wire cable) is easy enough, especially considering 2 wires are currently not in use. However, I love the 408A tubes, and the mic I already made is awesome, so I don't feel like making any changes to it. And since I'm making a 2nd one, I don't want to have any significant difference between them.

OTOH, I was smart enough to connect all 7 wires in the cable, so it's not THAT hard to make the same cables work for both mics, even if I do change the 2nd one up a bit. The real problem is that if I make the 2nd one the modified way, then I'll absolutely want to mod the 1st one. Y'all know how it is. 🙄🤣
 
If you want a matched pair, they’d better have the same capsule!
Yeah, they'll both have Heiserman capsules. Actually, I'm tempted to make another EF800 version with the Heiserman capsule, and then compare again. Then make a 2nd matching mic of the one I like best.
 
That's an interesting idea. Almost all the resistors I use are CMF, and there's not nearly enough heat generated in the mic for those to drift. I would bet that if I moved the resistor and adjusted the wiring to accommodate, that nobody would be able to tell the difference, especially if I match voltages.
 
Triodes are more linear with a constant current plate load, has anyone tried on this mike?
Not sure how much noise an LND150 would add.
A larger signal on a resistor loaded triode plate would have more deviation from the ideal. I have not seen any triode plate curves in typical VF14 mike operation.
Linearty of the circuit could be measured on transformer output with substituting the capsule with a capacitively coupled to a signal generator.

If two tubes are used for heating circuit convenience, why not use the second tube a follower?
 
Did you use pin 3 and 6 and send it to the mic cable?
Thank you
No, I ve used only one pin. The Heater of one Tube draws 0.05A -> 20V. The Filaments of the two used Tubes are connected in serial - so you will have 40V at 0.05A -> no high currents, no problems with Heat. Cassical cheap AWG26 7Pin Miccables like from Thomann, Music Store with AWG26 can draw without any Problems 0.2A,
The Voltage lost over these Cable also has no real impact, we talking about 0.2Ohm/m and less - that means about <0.1V on a 5m Cable.
Regards
 
Last edited:
No, I ve used only one pin. The Heater of one Tube draws 0.05A -> 20V. The Filaments of the two used Tubes are connected in serial - so you will have 40V at 0.05A -> no high currents, no problems with Heat. Cassical cheap AWG26 7Pin Miccables like from Thomann, Music Store with AWG26 can draw without any Problems 0.2A,
The Voltage lost over these Cable also has no real impact, we talking about 0.2Ohm/m and less - that means about <0.1V on a 5m Cable.
Regards
Thank you :)
 
Can someone tell me which resistors in the U47 circuit I should experiment with using carbon film? Which ones alter the sound the most?
 
Josh, thanks for posting and thank you Danny for putting this project together. And thanks to anyone else involved I have overlooked!

Octopart has suggested this replacement part for the 60M RNX02560M0FKLB
resistor: 62M VR68000006205JAC00

https://octopart.com/search?q=RNX02560M0FKLB&currency=USD&specs=0

Link may or may not display what I see, so I'm attaching a screenshot.

Can I get away with this substitution? What would the ramifications be to the rest of the circuit?

Dimensions are just a little shorter. Lmax ~9.09mm vs 7.5mm

Thanks!
Peter
 

Attachments

  • Screen Shot 2024-02-20 at 2.23.14 PM.png
    Screen Shot 2024-02-20 at 2.23.14 PM.png
    150.7 KB

Latest posts

Back
Top