Troubles when elevating a LT supply

[jesusdecos]

Hello,

For those that don't want to read much in the last paragraph is described the issue.

I am new in this forum and also in valve amplifiers construction. I think the better way to start is to introduce myself: I am Telecommunication Engineer and also a music player and lover. I am in a group of music where I need to use a transducer in order to electrify my violin: it is a Fishman pickup (V-200 Classic Series Professional Violin Pickup - Fishman Transducers, Inc.). The pickup is a good one and in the last years I've been frustrated because I didn't like the sound I was getting. I usually connect directly the pickup into an inexpensive mixer console. I recently discovered that this issue can be overcome by using a preamplifier: the transducer has a very high output impedance (capacitive), which steals bass when not matched. I love the tube sound and I want that on my preamplifier, but all preamplifiers sold for acoustic instruments that I found was solid state based.

So, I decided to make my first tube preamplifier. The requirements are the following: high input impedance, low output impedance, tone control, clean sound and minimum distortion. To fulfill this criteria I considered a design composed of four stages: a buffer input stage to match the piezoelectric impedance, a mu follower stage to get some gain before the next stage, a Baxandall tone control circuit (bass/treble 20 dB boost/cut) and the last stage is a mu follower to get low ouput impedance with minimum distortion and to rise signal to line level. In my calculations, E88CC is a good preamplifier tube to make the mu followers because it has enough voltage gain for my purpose and high transconductance, which is interesting to reduce noise. In the power supply design I decided to include a tube rectifier, specifically an EZ81 for HT. Upper valve cathode is about 220 Volts whilst lower valve cathode is about 2.5 Volts. So, I decided to put upper valves in one bottle, lower valves in another. The maximum voltage between cathode and heater for E88CC is about 50 Volts, so I need to use two separate LT supplies for each bottle, one elevated near 220 Volts. To do this, I followed the design found in Morgan Jones book (you can find a schematic below). Input impedance of a tube based stage cannot be greater than about 1 MegOhm which is not enough (due to maximum grid leak resistor value). So, to overcome this I resorted to solid state, specifically a buffer stage based on an audio operational amplifier. This requires a symmetrical supply, so my aim is to use the non elevated heater supply to feed the operational amplifier as well.

I must admit that I hadn't worked previously with valves and all I know is from reading books/articles like the Morgan Jones ones. I followed his book "Valve amplifiers" to help me in my design. I also used PSpice and PSU2 to simulate the circuits. Well, all my calculations and simulations seems to go very well and I decided to build it.

I found a problem in the power supply that I want to discuss with you that is going to make me mad (you can find a schematic below to follow the description). The elevated supply does not rise to 220 Volts, it is elevated to about 20 Volts instead and doesn't give 6.3 Volts between its terminals as it was supposed to do. On the other hand, the non elevated supply is slightly non symmetrical (+3.5/-2.8V instead of +3.15V/-3.15V). I checked all during some days and everything seems to be ok. Now observe the following: suppose that we disconnect the center tap formed by 47 Ohm resistors in the non elevated LT supply from ground (leaving the resistor leads not connected). In this situation, the elevated supply works perfect as it was supposed to do, but the non elevated supply is now elevated like about 300 Volts. This time suppose that we disconnect the elevated supply from the potential divider instead of the non elevated supply. Now, the non elevated supply is perfectly symmetrical (+3.15V/-3.15V). Any ideas of what can be wrong? Thanks for your time in reading this.

[SY]

First step- disconnect the heater supply from the THINGY. What voltage do you see at the emitter of the MPSA42?

[jesusdecos]

Thanks for the reply, SY. Answering your question, the emitter of MPSA42 is at 249 Volts when I disconnect the heater supply from the THINGY (B+ is at 340 Volts).

Of course, all measures are done with upper valves anodes and heaters disconnected from supplies. This implies that HT voltages are increased roughly in 20 Volts (due to EZ81 internal impedance).

[Hi-Q]

As you are using elevated DC heater supplies, you can dispense with the 47 Ohm resistors on both heater circuits. Ensure both heater circuits are isolated from each other and both are isolated from ground (chassis). I use a simple series resistor chain for my phono amplifier (see attached) though purely for hum rejection rather than reducing cathode to heater differential, it provides approximately 40V. No doubt you could make another resistor chain to provide you with 220V. Something like 100k in the top and 150k in the bottom would give you something like 210V with a 350V B+. You would have to fit a capacitor with a higher voltage rating of course as a 63V one would POP! :-)

[SY]

The two 47R can be helpful- if you have one side of the heaters at AC (or DC) ground, the circuit is susceptible to common mode noise.

OK, so without the heater supply connected, the output of the THINGY is correct. And when you connect it, it's not. Have you checked to make sure there is nothing in the floating heater supply contacting ground? There would appear to be a low resistance path to ground somewhere.

Raw voltage (before the reg) look normal?

[jesusdecos]

Thank you both for your replies. As you pointed out (and I initially thought) it seems that there is something interconnecting both LT supplies at any point and the elevated supply with ground thereby. I exhaustive revised the layout and I haven't found any element connecting both supplies. Even I tried desoldering some components (like relays) to check if there is no good isolating among its terminals. One thing that I noticed is that when disconnecting the LT supply from the THINGY and measuring the resistance between the V+ or V- of the LT supply and ground with a multimeter there is like 0.8 MegaOhms. And the thing is wierd because this resistance seems to be increasing with time; I explain myself: the first time I checked this there was like 0.150 MegaOhms. Yesterday there was like 0.4 MegaOhms and now there is like 0.9 MegaOhms. I observed the same behavior in the emitter of MPSA42 when all is put together: the elevated voltage was initially about 16 Volts, yesterday was 20 Volts and now it is 24 Volts.

I attach a photograph of the supply, dismantled and taken out from chassis. I also attach the PCB layout that I used.

[SY]

That may be a capacitor charging, not really DC leakage.

Just for completeness, put the tubes in and, with the supply connected but the THINGY disconnected, see if there's a lowish DC resistance to ground from the heaters.

[jesusdecos]

But that requires overcoming the maximum cathode to heater voltage (50 Volts) of upper tubes in mu followers. Is it safe to operate like that?

[SY]

I hope that when you measure resistance, the power is off! What we're trying to determine is if there's an inadvertent leakage path through the tubes.

[jesusdecos]

OK, my bad, I understood the first time that you want me to connect the tubes to the +B also. If I am not wrong this time what you tell me to do is just disconnecting the elevated supply from the THINGY, put the signal valves into their respective sockets and check now the resistance between heaters (V+ or V- of the supply) and ground. The resistive value that I am measuring is 0.3 MegaOhms when the bottle with upper tubes is inserted into its socket. Without the tube it is 0.9 MegaOhms as I previously stated. I think that I forgot to say that I use a separate PCB for signal path and all these tests are realised without connecting it to the supply PCB, so the problem is on the supply side.