Ad Widget

Collapse

Announcement

Collapse
No announcement yet.

Voltage doubler heater supply design problem

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Voltage doubler heater supply design problem

    Over Christmas I've been experimenting with a low-voltage reverb circuit - a scrapbox special.

    My PSU consists of a 12v-0v-12v transformer (24v) driving a standard full-wave voltage doubler. This gives me a B+ of 76v, which I'm happy with. It's on the high side due to UK/European voltage inconsistencies. The tube heaters are powered off 12vac (actually 14v) - one leg of the PT to the centre tap. That also works fairly well apart from excessive heater hum. If I power the heaters (with one side grounded) from an external transformer there's no hum.

    So, the problem I have looks like the ground reference to the heaters, which are raised 38v with respect to the chassis and is difficult to resolve because of the lack of common ground path in the supply.

    The simple answer would be to use a half-wave doubler with common grounding, but I'm wondering if anyone else has any ideas? I want to keep the simplicity of the existing transformer and avoid having a separate heater transformer.

  • #2
    Originally posted by Mick Bailey View Post
    I want to keep the simplicity of the existing transformer and avoid having a separate heater transformer.
    Better you have a separate heater transformer, really. Your experiment with one showed you what an improvement it makes.

    I've run across a similar problem in Ampeg SVP preamps, not the first run but after "redesign" to cut corners, they were too clever by half. They use a bridge not V-doubler but no matter, diode switching noise from the DC heater supply shows up in the audio, very obvious and totally not acceptable. The only improvement I could propose with a buzzy SVP was - separate transformer - for which there is no room in the chassis.
    This isn't the future I signed up for.

    Comment


    • #3
      Interesting you should mention diode switching noise. I hadn't noticed this before but when the reverb mix is over to the wet side it shows up. I guess that's inevitable with a doubler circuit. Looks like my original idea of a dual-purposed supply is flawed.

      I've just reconfigured the PSU as a common-ground half-wave supply and increased the capacitance. Ripple now is about the same as it was before at under 3mv. One side of the heaters is grounded but it still hums too much. I was hoping for the same hum-free result as with the separate heater feed but I guess there's more to this than I'd originally figured.

      Comment


      • #4
        If you can post a schematic of the powering and heater circuit that may help, as there are a few rectifier circuits that may use the same 'standard' name.

        Comment


        • #5
          Here's the doubler schematic. Heater run from CT to one side of the secondary;

          Click image for larger version

Name:	voldoub.gif
Views:	1
Size:	2.1 KB
ID:	836427

          Comment


          • #6
            Any chance those are actually two 0-12V windings? You might be able to make that work.

            Or hey, how about using an external wall wart to power the heaters?

            Comment


            • #7
              Ok, so I assume all your reverb circuitry is referenced to a 0V node which is the negative end of the output of the doubler. Hence, one end of heater is effectively sitting at a 'somewhat quiet' DC level (half the doubler DC output). The output supply mid-point will have some ripple on it, but given you aren't running a power amp off the supply then the ripple voltage at that point should be negligible.

              Does your reverb circuit have a low signal preamp valve stage (eg. reverb tank output make-up stage), and is that valve's heater a 6V or 12V heater? If you were using a 12AX7 for example, then the heater used by the low signal level triode should be the heater with one end going to that mid-point, and not going to the PT CT. That is likely to be your lowest hum configuration (which you may have already wired).

              The 'hum' you refer to may be not just 60Hz hum, but also 120Hz plus higher frequency rectification hash. You would likely need a spectrum analyser function to determine. If you are mainly hearing higher frequency rectifier noise, and not the 60Hz fundamental, then there are ways to dampen that higher frequency hum.

              Some valves are also less influenced by heater AC. Even valves such as 12AX7 can vary between samples, due to different heater-cathode resistance. Plus you need to ensure your wiring is not causing heater AC to easily couple to grid - that can be checked by using a temporary separate isolated 12V heater supply and a humdinger pot with wiper connected to 0V.

              Comment


              • #8
                Originally posted by trobbins View Post
                Ok, so I assume all your reverb circuitry is referenced to a 0V node which is the negative end of the output of the doubler.
                That's right.

                The mid-point of the doubler is 38v so I can't use that to power the heater directly. The hum is a mixture of rectification hash and 50 hz (UK). I don't get any hum with the heaters powered off a separate 12vac supply so it's related to the doubler configuration itself. What I'm finding difficult to understand is that the heaters reference to 38v so I wasn't expecting the level of hum. Reversing the heater leads and changing the connection point to the other side of the secondary makes no difference either.

                I've just run through a selection of valves and this seems to be the key. I have two NOS Westinghouse 12AX7 and they almost kill the hum. There's still some diode noise and a little more hum than I would like, but a big improvement.

                Comment


                • #9
                  Hmmm. 50Hz would typically enter the signal chain through either heater-cathode impedance (capacitive and resistive parts), or as a voltage coupling via stray capacitance to the grid.

                  If you have used a separate 12VAC and had not changed the local heater wiring near the valve base and grid wiring, then that should have given the same level of 50Hz hum as what you are getting now. What is changing is that the heater voltage now has a distorted 50Hz waveform (due to rectifying currents in the same winding), plus the whole heater waveform is varying relative to 0V due to the ripple voltage at the mid-point junction compared to the 0V reference at the other end of that bottom filter cap.

                  I'd suggest that the main contributor is the distorted 12V 50Hz waveform, due to the rectifier current pulses passing through that winding. Those current pulses can be lowered in amplitude, and broadened in width by lowering the doubler capacitance, and using post-filtering of the 75VDC rail.

                  You may also be getting rectifier hash, which could be alleviated by a better choice of diode or adding an RC snubber across the 24V winding (some people use a 'bell-ringer' circuit to determine better RC values). Also, shunt filtering the 12VAC heater portion of the PT winding with say a 100nF capacitor may lower higher frequency hash.

                  Comment


                  • #10
                    Originally posted by trobbins View Post
                    Hmmm. 50Hz would typically enter the signal chain through either heater-cathode impedance (capacitive and resistive parts), or as a voltage coupling via stray capacitance to the grid.

                    If you have used a separate 12VAC and had not changed the local heater wiring near the valve base and grid wiring, then that should have given the same level of 50Hz hum as what you are getting now. What is changing is that the heater voltage now has a distorted 50Hz waveform (due to rectifying currents in the same winding), plus the whole heater waveform is varying relative to 0V due to the ripple voltage at the mid-point junction compared to the 0V reference at the other end of that bottom filter cap.

                    I'd suggest that the main contributor is the distorted 12V 50Hz waveform, due to the rectifier current pulses passing through that winding. Those current pulses can be lowered in amplitude, and broadened in width by lowering the doubler capacitance, and using post-filtering of the 75VDC rail.

                    You may also be getting rectifier hash, which could be alleviated by a better choice of diode or adding an RC snubber across the 24V winding (some people use a 'bell-ringer' circuit to determine better RC values). Also, shunt filtering the 12VAC heater portion of the PT winding with say a 100nF capacitor may lower higher frequency hash.
                    Thanks for expanding on what I was trying to get across in #2. There's no free lunch. A separate filament transformer is called for to solve this problem. Get out the drill and warm up the iron Mick. (How Ampeg released their "improved" SVP with this problem built-in is beyond me and bit of a disappointment.)
                    This isn't the future I signed up for.

                    Comment


                    • #11
                      Well, the separate transformer seems to be the real answer. Tried everything I know with filters, snubbers, cap values, etc. Damn. I was hoping I could crack this. Looks like a gutted wall-wart to the rescue.

                      My old woodwork teacher used to say "You want simplicity? I'll give you simplicity - a chair with one leg. Does it work? No. Sometimes simple isn't the best".

                      Comment


                      • #12
                        If the hum is from diode current pulses in the B+ supply getting on the heater supply something like this could work as long as it doesn't draw too much current for a half wave heater supply.

                        Click image for larger version

Name:	Dual Supply.PNG
Views:	1
Size:	17.3 KB
ID:	836456

                        Comment


                        • #13
                          That looks promising. I'll give it a go tomorrow.

                          Comment


                          • #14
                            It may be a bit mind-bending, but you have the opportunity to use a double doubler to generate effectively twice the B+ voltage. The additional doubler would generate a '-76V' rail, which would then become your new 0V rail. I note that you said you were fine with 76V B+, so it is just an option.

                            The circuit is just the existing doubler, but in a reverse polarity format - I can only quickly find the link below to illustrate the circuit which I used.
                            http://dalmura.com.au/projects/McPhe...20modified.pdf

                            The heater would be elevated to 76V in that case.

                            Comment


                            • #15
                              I tried Dave H's suggestion of a v-reg and that worked well. In the process though I tried the whole setup (rather than just the heaters) running off an external transformer and that worked, too. This turned my attention to the transformer; I replaced the original transformer with the external toroidal I was using and it also fixed the problem. So two fixes when using a 1/2 wave rectified setup and common grounding. I can't find a fault with the EI transformer - maybe the toroidal is better at internally damping the switching noise.

                              The McPherson amp is one I've not seen before. Probably one I'll never see in real-life. That sure is a complex way to generate a bias voltage.

                              Comment

                              Working...
                              X