In a desperate attempt to get less microphonics from my 1H4 types I wanted to try the 1G4 (another nice and cheap DHT) instead of the somewhat bigger 1H4. The 1G4 has the same base and pin-outs which make adjusting the existing preamp a piece of cake. The 1G4 has a GT envelope which is generally less microphonic due to the smaller glass area that can start to vibrate plus it is sturdier. The 1G4 runs at about half the voltage that a 1H4 does, this means I had to drop quite a few volts in my power supply. I would recommend that anyone building the 1G4 version take a lower transformer voltage. Less voltage drop (less resistance) opens the sound up a little more than this way. I do get a very good rejection this way though, measured ripple is in the 2mV range. The gain of both 1H4 and 1G4 is around 6 times, this should be enough for a line-level preamp.
The 1G4 gets my vote out of these two tubes. It sounds less aggressive in the mid/high region and a bit warmer (more mellow) than the 1H4 as well.
One warning though: be very careful when setting the filament voltages. The filaments are very very thin and can't handle an overload like the more sturdy oxide cathodes of say the 12, 26, 45 etc... 2V instead of 1.4V can kill the tube... PLING!!!!
I am giving the circuit a second try, what else to do during the Christmas holidays :) I have changed the circuits a bit to improve the supply and signal circuit. I hope to remove the microphonics a bit as well. The filament is now supplied by a 6V lead/acid battery and voltage regulator. The 1G4 pulls about 50mA of current, the battery is 4000mAh so it should last for quite a while. I will report the results after I finish the last modifications. How about ten 9V NiMH batteries for a B supply next time? ;))))))
Wiring is completed, everything works like expected (voltages, currents) except gain is quite low; about 4 times instead of 6.5x. Ah well, must be the bypass cap, will try that later today. I used 6V batteries and a voltage regulator to get the filament voltage right. I know, not the best way... Using a current regulator is a pain here since the filament only pulls 50mA, a small turn on the pot (no matter what value I use) gives a big difference in current to the filament. Not very accurate, and with the tiny filament wire of the 1G4 I don't dare take the risk of blowing another.
I changed the entire power supply from what I had last time. No more big MKP's just to get rid of all the crap in one blow. I now use an oil-filled MP at the front, followed by a 5uF paper-in-oil and then a 50uF Micro electrolytic (yeah yeah: they're really good!!!!) as the last stage. Calculations gave a 10mV ripple with these last two, maybe I will use something smaller later on if I decide to keep the preamp the way it is right now.
Sound is as I remember from last time: very smooth and musical, although a bit on the fat side. Microphonics are a lot less as well. I can clap my hands and hear the filament vibrate for a second or two, but no singing around (feedback) like last time. I think I will listen to it for a few days and then give my final verdict. The 1G4 is a lovely tube, underrated by many people because of it's size.
I updated the schematics as to what they were in the final stages.
Signal section, RC loaded, not much else to say.
B+ supplies for both channels.
The switch on the batteries connects them to either the filament of the 1G4 or the charger circuit.
finished version, now to find some knobs for the input and volume controls
internal wiring, old version
new version, everything put in place with silicone.
Measurement were taken using a Philips Signal Generator and a very nice Fluke digital "handheld" scope with tons of neat tricks in it
Sqaure wave at 20Hz, 40Hz, 100Hz and 1kHz. Pretty okay until below 40Hz.
Square wave at 50kHz, 100kHz, 200kHz and 400kHz. Pretty good up to 150-200kHz.
Frequency range between 6Hz and 656kHz at -3dB, not bad!
Input signal and output signal with volume pot at full. Input signal was chosen so that the output did not show any sign of distorting or clipping. As you can see the gain of the circuit is around 6 times.