U47 with EF14 tube Build

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
The voltages seem to very slowly be dropping down overall, so something weird is happening.
Nothing gets overly hot, after 5-10 mins reg is about 25C.

I’ll have another look now and see if I can see what’s going on.
 
Nothing gets overly hot, after 5-10 mins reg is about 25C.

That right there should be suspicious. If you're looking for (say) 6.3v at the output, and dropping about 1v across each of those 3.8R series resistors, that would work out to about 8.3v right at the output of the LM317.

15 - 8.3 = 6.7v (voltage drop between regulator input & output)
6.7v x 0.266 = 1.8 Watts (that the LM317 needs to burn off)

Without an infinitely efficient heatsink, that LM317 will NOT be simmering at room temperature.

Could the regulator be shot? Leaky / busted diodes? Wonky solder joints?
 
Last edited:
Without an infinitely efficient heatsink, that LM317 will NOT be simmering at room temperature.

Could the regulator be shot? Leaky / busted diodes? Wonky solder joints?
This is also a possibility that, for example, the LM317 is already in overload protection mode due to too much current being drawn by a low-resistance load (almost short, Diodes, defect cable?) on the secondary side??
 
No secondary fuses for the transformer? :devilish:
View attachment 116604
Edit: I can't see a primary fuse either, have you installed one? I would fuse the transformer on both sides, I have seen a transformer burn down during a live show! Don't save money in the wrong places...
Ah I remember the AKG sSolidtube psu's always burned their mains transformer, MOV's not saved them(however it was undersized)
 
LM317 should have 1.25V (typ.) between Pin 1 and 2. The schematic above with a pot allows for 12.7V output, not great for 6.3V tubes. Why even have a pot there? It is an accident waiting to happen.
Using the above R2 of 220, R3 needs to be 900 Ohms, no pot needed for 6.3V. If you want to reduce the 6.3V put a 500 Ohm pot in series with a 410 ohm.
Using the voltage-doubler may not be needed for a 6.3V output IF an LM350 is used.
https://www.digikey.com/en/products/detail/texas-instruments/LM350T-NOPB/8901Dropout should be around 1.5V for 500mA, so a minimum 8.8V in minimum (peak). Schottky rectifier diodes would help. Excess heat does not help anything.

If one is concerned about tube longevity, the 3-pin regulator can be connected as a current regulator. Cold current inrush current is then not an issue, and extra wire/connector resistance is not important. Adding the last filament filter cap inside the mike makes the lead resistance a part of an RC filter. Noise performance is no different in the constant current mode.

A 6DJ8 heater is rated 6.3V, 365mA, =17.26 Ohm. Cold 2.8 Ohm = 2.25A or over 6 times higher...
 

Attachments

  • current_heater_supply_.png
    current_heater_supply_.png
    43.2 KB
Last edited:
LM317 should have 1.25V (typ.) between Pin 1 and 2. The schematic above with a pot allows for 12.7V output, not great for 6.3V tubes. Why even have a pot there? It is an accident waiting to happen.
Its a universal tube PSU solution for different tubes in the range from 4V up to 12V with a 6.3VAC winding or more with other input voltages. Not all tubes work with 6.3VDC, this board is intentionally designed for flexibility, hence the potentiometer, which by the way works very well in the long run. (you don't have to fit it if you don't need this flexibility, you've probably seen the parallel resistor)
Excess heat does not help anything.
There is no excessive heat if you do it right, I have several units in continuous operation with it, works perfectly, you can easily touch everything, hand-warm maximum.
Using the voltage-doubler may not be needed for a 6.3V output IF an LM350 is used.
https://www.digikey.com/en/products/detail/texas-instruments/LM350T-NOPB/8901 Dropout should be around 1.5V for 500mA, so a minimum 8.8V in minimum (peak). Schottky rectifier diodes would help. Excess heat does not help anything.
I have also tried this, but I found it not really convincing. In addition to the lack of flexibility, you are pretty much at the limit where the regulator still regulates cleanly, depending on the grid fluctuations. You can do it, but I decided otherwise.
My solution works without any problems, is very flexible for various applications, not just microphones, and also harmonises well with many recycling transformers from old devices.

There are always other good solutions, but I'm very happy with this one.
 
Its a universal tube PSU solution for different tubes in the range from 4V up to 12V with a 6.3VAC winding or more with other input voltages. Not all tubes work with 6.3VDC, this board is intentionally designed for flexibility, hence the potentiometer, which by the way works very well in the long run. (you don't have to fit it if you don't need this flexibility, you've probably seen the parallel resistor)

There is no excessive heat if you do it right, I have several units in continuous operation with it, works perfectly, you can easily touch everything, hand-warm maximum.

I have also tried this, but I found it not really convincing. In addition to the lack of flexibility, you are pretty much at the limit where the regulator still regulates cleanly, depending on the grid fluctuations. You can do it, but I decided otherwise.
My solution works without any problems, is very flexible for various applications, not just microphones, and also harmonises well with many recycling transformers from old devices.

There are always other good solutions, but I'm very happy with this one.
If you are stuck with a low voltage transformer, a doubler could work, otherwise check on a scope what the minimum voltage is under load. More capacitance on rectifier will help. There are better LDO's than the LM350, but it is rated for 3A and pin-compatible with 317.

Hard grounding the rectifier diodes may not be optimal, add some resistance to ground. Only having a diode between ground and a transformer winding adds rectifier currents into ground, which in low power/low noise supplies is not desirable.

To get even lower dropout, the 5-pin thru-hole LT1963AET chip could work, very low noise, around 0.4V dropout for 0.5A, max 1.5A, among many choices, including higher current ones.

My choice would be for fixed resistors and jumpers, rather than a pot. Tube heater voltages do not need much change once set, but one may be included in series with fixed resistors, for an adjustment range that will not destroy the tube. Tube specs are very closely related to the heating value, tubes degrade very gradually until emission becomes unusable for the circuit expectations.
Cheers

 
So far it all works as expected, will continue to monitor and see how it goes.
Yes, keep us posted. I would be interested to know how high your noise floor is in this combination with your U47 (EF800) microphone?
You have already built several versions, is the current combination noticeable louder or quiter in this respect? Is the ripple filtering of B+ sufficient?

As it is, it corresponds approximately to an U67 original PSU. In my application (Sela T12 inspired mic with 5654) I have an additional RC filter in the mic body, I don't know if that was really necessary but it is quite enough for serious work.
 
Has anyone tried the 6DJ8 dual triode for U47 type mikes? Very low noise and much more linear than VF14. The two sections could be un in parallel, for lower noise, or in series as a cascode, or use the second section as a follower.
More voltage would make the tube operate in a more linear region, 200V or more would be better.
Has anyone tried using an active plate load, like the LND150, DN250 etc?
A resistance loaded triode will always be current starved with a large +peak input, and have less headroom for the -peak, reducing the voltage swing compared to constant current plate load. Noise is of course an issue with "infinite" resistance, but the plate resistance should dominate. Heat generation is of course an issue, but running the tube with more current would help, maybe upwards to 5mA?
My preamp runs 6DJ with 10mA constant, ~110V on the plate and 180V from power the supply.
Headroom is great with well over 100V peak-to-peak capability, which would be less with a transformer.
Gain is the same as the mu of the tube, here 33, or 30dB. Enough for a mike?
There are higher gain tubes available, if needed.
 
Not included in these PSU circuits are any delay effects for the HV, usually used to avoid, "cathode stripping". How real this problem is I have no data for, but no smoke without fire, so better include it, as good tubes are a pity to waste. The old days this was easy with an indirectly heated rectifier tubes, now pretty expensive. Any crap TV triode hooked up with the grid to the cathode, may work as a substitute, or a 30 second timed relay.
A gradually increasing voltage can also be done with a LM317 buffered with a HV mosfet, but requires more parts. Any voltage ramp can be accomplished this way, including 120dB of PSU noise rejection.
For the low current mike tube a series LND150 feeding an electrolytic could be enough.
Bleeder resistors I would not put on the input of the regulator. It does not need to bleed faster than the time it takes to get the screws off the cover.
 
Has anyone tried the 6DJ8 dual triode for U47 type mikes? Very low noise and much more linear than VF14. The two sections could be un in parallel, for lower noise, or in series as a cascode, or use the second section as a follower.
More voltage would make the tube operate in a more linear region, 200V or more would be better.
Has anyone tried using an active plate load, like the LND150, DN250 etc?
A resistance loaded triode will always be current starved with a large +peak input, and have less headroom for the -peak, reducing the voltage swing compared to constant current plate load. Noise is of course an issue with "infinite" resistance, but the plate resistance should dominate. Heat generation is of course an issue, but running the tube with more current would help, maybe upwards to 5mA?
My preamp runs 6DJ with 10mA constant, ~110V on the plate and 180V from power the supply.
Headroom is great with well over 100V peak-to-peak capability, which would be less with a transformer.
Gain is the same as the mu of the tube, here 33, or 30dB. Enough for a mike?
There are higher gain tubes available, if needed.
https://www.foxaudioresearch.ca/The47.htm
 
Nice, thanks!
The 6DJ8 bests most triodes, in linearity, transconductance low plate resistance, great with a CCS , if a mu of ~33 is enough.
 
Back
Top