[Engineer]

Hi, Vacuumlanders.
Further to my earlier post on dumpster diving, I've discovered an
interesting approach to B+ power supplies based on the DC module (only)
contained in switching power supplies (PSU's.)

In North America (120 VAC mains), the DC module in these switchers
generates a high voltage DC from the 120 VDC using a ½ wave voltage
doubler based on a FW bridge diode package. This is then switched at
high frequency into an HF transformer from which rectifiers generate
the 5 and 12 VDC high current outputs (In Europe it seems to be just a
FW rectifier on the nominal 240 VAC input - there's a jumper to do
this on the units I have.) At the high frequency used, circa 100 kHz,
the DC filters can be very small, the rectifier is directly mains fed
and the isolation is achieved by the HF transformer. Regulators are
used to get good DC regulation and there is lots of other circuitry
to do overload protection, etc.

Recently, when parting out an industrial switching PSU unit, I noticed
that the large electrolytic caps were in series. A test with a variac
showed about 330 volts DC across them (165 VDC each) at 120 VDC in. A
"light went on" in my head, and not due to any accidental
electrical connection! I was looking at a potential tube amp. B+
supply module buried in a low voltage PSU! The current rating is very
high (e.g. over an amp at 330 VDC for a 350 watt unit), the FW bridge
rectifier has a several amp rating and the caps are 650 MFD each (200
VDC working.) - all seriously lethal if mishandled. One problem: the
DC module is not isolated. One side of the DC is about 160 VDC above
neutral and the other is 160 VDC plus 160 VAC (that's 0 to 320 volts
peak at 60 Hz in N.A.) above neutral - neither being the sort of
thing you want on your amplifier chassis! Answer: use a 1:1 isolating
transformer to feed the DC module, unfortunately these are not often
free.

To cut a long story short, I made up a B+ power supply using a spare
1:1 isolation transformer (an old 60 VA, 120 VAC toroid from a vintage
computer CRT display I found put out for the trash on our street.) I
connected the positive output to pin 8 of an octal tube base and
plugged this into the rectifier socket of a 20-watt monobloc tube amp -
the negative going to chassis. After heating up the amplifier tubes, I
brought the DC module up slowly on a variac (no smell or fireworks!) -
just a nice 321 VDC B+ (at 120 VAC input) to a pair of 6L6's. Not a
trace of hum and, of course, perfectly normal sound.

Here are the actual numbers
Input voltage: 117 VAC (from variac, my actual supply was 123 VAC)
No load voltage: 348 VDC (my "1:1" isolator is actually 1: 1.06.
Note: 117 x 2 x 1.414 x 1.06 = 350.72, perhaps the mains dipped as I
measured the DC, but close enough!)
Amplifier has 2 x 6L6 plus 2 x 12AU7 plus 1 x 12AX7:
321.4 VDC at 93 mA
Same, but two 6V6's plugged in:
326.1 VDC at 78 mA

Not bad for an unregulated PSU, the Thevenin equivalent resistance is
just over 280 ohms, much better than a CRC, or even CLC, filtered tube
rectifier. Also, in this case, the B+ is designed to operate near the
peak AC voltage already, so there's no 40% surge at switch-on as
there is with a 5U4 tube rectifier.

This set me thinking... PC power supplies use the same principle and
there is a pile of scrapped units out there. In the ones I've
stripped, the DC sections contained 470 or 680 MFD caps of 200 VDC
working (two in series.) But, unlike the above industrial PSU's, the
PC PSU is all on one PCB. If this board could be carefully stripped of
all digital and logic circuits we would be left with the core of a DC
PSU for, say, a 20 watt monobloc at about 300 VDC B+. Just add the
mandatory 1:1 isolation transformer and a 6.3 VAC heater transformer
- might just be cheaper than the classic, vintage 350-0-350, 5 and
6.3 volt power transformer. I've not looked onto this yet as I've
still several of the industrial PSU's (with separate DC modules) to
cannibalize first. I'll have to leave the PC PSU experiments to
others for now.

Of course, doing the above we lose the heater/filament glow from the
classic tube rectifier (I love those fat 5U4's!) So, if we also use
those black metal 6L6's the world gets bit darker all round and
we'll need pilot lights!

Warning and disclaimer: the above describes potentially lethal, high
voltage circuitry. Do NOT try any of the above ideas unless you are
fully qualified to design, build and test such circuits.

Happy New Year!
Cheers,
Roger

[robert casey]

Recently, when parting out an industrial switching PSU unit, I noticed
that the large electrolytic caps were in series. A test with a variac
showed about 330 volts DC across them (165 VDC each) at 120 VDC in. A
"light went on" in my head, and not due to any accidental
electrical connection! I was looking at a potential tube amp. B+
supply module buried in a low voltage PSU! The current rating is very
high (e.g. over an amp at 330 VDC for a 350 watt unit), the FW bridge
rectifier has a several amp rating and the caps are 650 MFD each (200
VDC working.) - all seriously lethal if mishandled. One problem: the
DC module is not isolated. One side of the DC is about 160 VDC above
neutral and the other is 160 VDC plus 160 VAC (that's 0 to 320 volts
peak at 60 Hz in N.A.) above neutral - neither being the sort of
thing you want on your amplifier chassis! Answer: use a 1:1 isolating
transformer to feed the DC module, unfortunately these are not often
free.

To cut a long story short, I made up a B+ power supply using a spare
1:1 isolation transformer (an old 60 VA, 120 VAC toroid from a vintage
computer CRT display I found put out for the trash on our street.)

I've used old PC power supply circuits myself. After removing the
switching circuits, they make a handy circuit board platform to house
the B+ filter circuits in various tube projects I've done. Sometimes,
the newly bare sections of the circuit board can provide enough circuit
board lands that I can use for a preamp or such, when using submini
tubes with long wire leads. Though it's a little hard to see, take a
look at http://home.earthlink.net/~wa2ise/im...yo-cd-tube.jpg over
at the lower right. You'll see a daughter board that used to be a PC
power supply. The small "plate" power transformer feeds the PC power
supply rectifier, and also the tube heater.

Another project I used a switching power supply board from a dead DVD
player. That pretty much uses the same topography as a PC supply. But
here I did a voltage multiplying trick to obtain some high voltage at
low current from a solid state receiver amp for a tube preamp. See
http://home.earthlink.net/~wa2ise/im...defollower.gif
And http://pw2.netcom.com/~wa2ise/radios...html#tubepress for
some more background on this strange looking circuit.

[Patrick Turner]

Engineer wrote:

Hi, Vacuumlanders.
Further to my earlier post on dumpster diving, I've discovered an
interesting approach to B+ power supplies based on the DC module (only)
contained in switching power supplies (PSU's.)

In North America (120 VAC mains), the DC module in these switchers
generates a high voltage DC from the 120 VDC using a ½ wave voltage
doubler based on a FW bridge diode package. This is then switched at
high frequency into an HF transformer from which rectifiers generate
the 5 and 12 VDC high current outputs (In Europe it seems to be just a
FW rectifier on the nominal 240 VAC input - there's a jumper to do
this on the units I have.) At the high frequency used, circa 100 kHz,
the DC filters can be very small, the rectifier is directly mains fed
and the isolation is achieved by the HF transformer. Regulators are
used to get good DC regulation and there is lots of other circuitry
to do overload protection, etc.

Recently, when parting out an industrial switching PSU unit, I noticed
that the large electrolytic caps were in series. A test with a variac
showed about 330 volts DC across them (165 VDC each) at 120 VDC in. A
"light went on" in my head, and not due to any accidental
electrical connection! I was looking at a potential tube amp. B+
supply module buried in a low voltage PSU! The current rating is very
high (e.g. over an amp at 330 VDC for a 350 watt unit), the FW bridge
rectifier has a several amp rating and the caps are 650 MFD each (200
VDC working.) - all seriously lethal if mishandled. One problem: the
DC module is not isolated. One side of the DC is about 160 VDC above
neutral and the other is 160 VDC plus 160 VAC (that's 0 to 320 volts
peak at 60 Hz in N.A.) above neutral - neither being the sort of
thing you want on your amplifier chassis! Answer: use a 1:1 isolating
transformer to feed the DC module, unfortunately these are not often
free.

To cut a long story short, I made up a B+ power supply using a spare
1:1 isolation transformer (an old 60 VA, 120 VAC toroid from a vintage
computer CRT display I found put out for the trash on our street.) I
connected the positive output to pin 8 of an octal tube base and
plugged this into the rectifier socket of a 20-watt monobloc tube amp -
the negative going to chassis. After heating up the amplifier tubes, I
brought the DC module up slowly on a variac (no smell or fireworks!) -
just a nice 321 VDC B+ (at 120 VAC input) to a pair of 6L6's. Not a
trace of hum and, of course, perfectly normal sound.

Here are the actual numbers
Input voltage: 117 VAC (from variac, my actual supply was 123 VAC)
No load voltage: 348 VDC (my "1:1" isolator is actually 1: 1.06.
Note: 117 x 2 x 1.414 x 1.06 = 350.72, perhaps the mains dipped as I
measured the DC, but close enough!)
Amplifier has 2 x 6L6 plus 2 x 12AU7 plus 1 x 12AX7:
321.4 VDC at 93 mA
Same, but two 6V6's plugged in:
326.1 VDC at 78 mA

Not bad for an unregulated PSU, the Thevenin equivalent resistance is
just over 280 ohms, much better than a CRC, or even CLC, filtered tube
rectifier. Also, in this case, the B+ is designed to operate near the
peak AC voltage already, so there's no 40% surge at switch-on as
there is with a 5U4 tube rectifier.

This set me thinking... PC power supplies use the same principle and
there is a pile of scrapped units out there. In the ones I've
stripped, the DC sections contained 470 or 680 MFD caps of 200 VDC
working (two in series.) But, unlike the above industrial PSU's, the
PC PSU is all on one PCB. If this board could be carefully stripped of
all digital and logic circuits we would be left with the core of a DC
PSU for, say, a 20 watt monobloc at about 300 VDC B+. Just add the
mandatory 1:1 isolation transformer and a 6.3 VAC heater transformer
- might just be cheaper than the classic, vintage 350-0-350, 5 and
6.3 volt power transformer. I've not looked onto this yet as I've
still several of the industrial PSU's (with separate DC modules) to
cannibalize first. I'll have to leave the PC PSU experiments to
others for now.

Of course, doing the above we lose the heater/filament glow from the
classic tube rectifier (I love those fat 5U4's!) So, if we also use
those black metal 6L6's the world gets bit darker all round and
we'll need pilot lights!

Warning and disclaimer: the above describes potentially lethal, high
voltage circuitry. Do NOT try any of the above ideas unless you are
fully qualified to design, build and test such circuits.

Happy New Year!
Cheers,
Roger

We have SMPS usage to thank for the easy availablity of electro caps
capable of sustaining the peak value of whatever mains is applied,
hence many C values with 450V ratings common.

The SMPS simply rectifies the mains without any transformer to make
a dc voltage and here in OZ we have 240Vrms, so we get 340 peak volts
with a solid state diode.
Ripple is 50Hz, but its not much if the C value is high enough.
Many 470uF caps are rated for 2A ripple current, which means Vr =
13.6Vrms at 50Hz.
At such a stressed level, the Idc output can be 1.45A, and this means
the rating of the rest of the supply could be 340V x 1.45 = 493VA.
But we wouldn't want to stress the input cap too much,
so having a VA a good bit less is the safer & more reliable approach,
IMHO.
I am not well experienced with SMPS, but I assume the designers do make
some concessions
to safety and reliablility, despite the bean counters in the company.
Noise from SMPS oten sounds like a buzzy sort of hum; 'tis the
wave form at the input cap being picked up
since the ripple voltage of say 50Hz, (or 100Hz perhaps) is modulating
the 100kHz square wave which has a huge range of propergating RF
harmonics,
all crudely amplitude modulated by the HF square wave which is a sort of
RF carrier.



After the input cap there is a chopper circuit which switches the 340V
on and off to make a square wave
at some high F and this wave is fed to a small sized light weight CHEAP
isolation tranny,
and the output recfified to make whatever Vdc wanted, +12V and +5V
usually for a PC and with
power output routinely being 300VA.

If the isolation tranny was 1:1 then you'd get +340V at the rectifier
after the 1:1 secondary.
This is where the isolation and safety is maintained; without the
little tranny, you'd have all sorts of problems with hum and RF.

The circuit before the tranny primary all floats at the neutral
potential, and is said to be
"live", and hence dangerous to touch especially if someone
has reverse connected the active and neutral lines from the wall
and there is no earth wire to the chassis anchoring down the 0V line of
the amp/device.

My guess is that one has to have that 1:1 tranny capable of accepting
fast
square wavees without losses.
Theoretically an air cored tranny may do the trick, but ferrite cores
are the cheaper option because they will offer a useful amount of
permeability at high F, so less turns are needed than having an air
core.
Bean counters in television and PC making companies always want the
cheapest
lightest effective solutions.

So presumuably an existing HF tranny used in a typical SMPS for a PC
could be
re-wound with a secondary of fine wires like the primary
after removing the low voltage secondary.

This does not seem to be an insurmountable problem.
In many existing SMPS, and those from old PCs are abundant and cheap,
the filtering of HF interference is largely addressed already,
or must be otherwise the PC would suffer such interference.

Halco use SMPS, and that brand is a high end state of the art type of SS
design so SMPS should be doable with tubes and the reason why we don't
see SMPS in
say better brands such as ARC and CJ is that the hi-fi cognescenti have
an irrational prejudice against modernity and innovation.

I have never found the time to experiment with such SMPS supplies for
tube amps
and it seems nobody has made a business of providing ready made tube amp
SMPS
with convenient ratings of say +500Vdc at 1Adc. Searches on Google
bring up no such suppliers. Everyone ignores SMPS and states reasons
why they should not be used, but IMHO a successful implementation
of SMPS in tube amps would contribute to the savings in amp costs,
and savings of greenhouse gas prouctions, and ensure that the world's
supplies of energy hungry GOSS maufacture was better used in motors etc
where it is best utilised, and not in inefficient low frequency
ancient power supply designs that need a forklift to move them.

Its easy to justify SMPS with altruisms.

But its got to be profitable.

Nobody at r.a.t has yet tried out a prototype SMPS and fully de-bugged
it so its suitable for any tube amp that could be made
easily by anyone here.


Patrick Turner

[robert casey]

Patrick Turner wrote:


So presumuably an existing HF tranny used in a typical SMPS for a PC
could be
re-wound with a secondary of fine wires like the primary
after removing the low voltage secondary.

I'd keep one low voltage secondary, for running tube heaters. That also
allows one to keep intact the voltage regulation feedback loop (most
supplies look at the +5V output and via an optocoupler or small
transformer adjusts the chopping circuits on the "hot" side of the main
high frequency transformer. Most can be changed to 6.3V. Maybe the 5V
can heat a 5U4 to rectify the high frequency high voltage secondary? :-)
The waveform may not be summectrical enough to make a centertapped
secondary full wave circuit to work right. A TV set damper diode tube
might be a better choice.

This does not seem to be an insurmountable problem.
In many existing SMPS, and those from old PCs are abundant and cheap,
the filtering of HF interference is largely addressed already,
or must be otherwise the PC would suffer such interference.

Halco use SMPS, and that brand is a high end state of the art type of SS
design so SMPS should be doable with tubes and the reason why we don't
see SMPS in
say better brands such as ARC and CJ is that the hi-fi cognescenti have
an irrational prejudice against modernity and innovation.

Old fashioned 60Hz power transformer power supplies are not as hard to
understand vs a SMPS. Also tube circuits usually don't expect tightly
regulated B+, just low ripple. But such is rather heavy and takes up a
lot of room.

I have never found the time to experiment with such SMPS supplies for
tube amps
and it seems nobody has made a business of providing ready made tube amp
SMPS
with convenient ratings of say +500Vdc at 1Adc.

Such a commercially made supply would need to have some way to select
the desired output B+. Probably via a jumper. Did you want 300V, 325V,
350V, 430V, etc?