Hello,


I am putting together a push-pull power amplifier based on the 813 power
tube. It will have about 1.5 to 1.7kV on the plates and put out at least
260 watts into a 9k load.

I am not new to high voltage tube circuits, but this is the first time I
will be working with potentials this high (above 1kV). I am aware of all of
the usual safety precautions, but I was hoping to get some specific tips for
putting together a design like this from some people who work with these
sort of voltages, hence the posting here.

For the plate supply, I have a Hammond power transformer laying around here
rated at 1000-0-1000 volts @ 200mA. From the ap notes on the 813, I am know
that the tubes will idle at about 50 mA and peak at about 305 mA. Would
this Hammond be ok for this type of use? I dont *think* I would be drawing
300 mils anywhere near continuously, but you never know. This amp will be
part of a bass guitar amplifier for live performance use.

I want to use solid state rectifiers and in studying some older ARRL
handbooks, they tell you to use equalizing resistors and small caps across
series-strung diodes to get up to the peak inverse voltage you need.
However I have seen a newer part in Mouser, made by Rectron, rated at 800mA
and 8kV PIV. Would something like this be all I would need (four in a
bridge configuration)? I would think it would be easier than making all of
those strings up from individual diodes. Or is there a better way?

I want to use a choke-input supply and I happen to have a large old filter
reactor made by Chicago Transformer rated at 20H @ 250mA, with a test
voltage of 6kV.

As for filter caps, I guess the cheapest way to go is to string together
series electrolytics with balancing resistors.

Also I have seen a lot of supplies use soft-start relays in the primary -
would you think this would be necessary in this application? I am looking
to get about 1.6kV out of this supply.

So basically you have a 2000vac secondary feeding a bridge of high voltage
rectifier diodes, then a 20H choke, and then a string of 'lytics. How much
capacitance would be sufficient to elliminate the hum in the plates? I know
that with push-pull and beam power tubes you can probably stang higher
ripple in the plates but is there a good ballpark figure?

Also since the 813 is directly-heated, would it be ok to have one 10v, 10A
filament transformer to light them both or would it be best to have two
separate ones to allow for hum balance -pots or for some other reason?

I also will have to make a screen supply for 750 volts at 45mA max. Would
this have to be a regulated supply?

I had planned on ordering some Belden 18awg high voltage test lead wire
rated to 10kV for all of the plate supply wiring. Would this be ok, or is
there a better wire for this application?

By the way I am having Heyboer custom wind an output transformer for this
thing. It was fairly expensive (around $200) so I want to protect it as
best as possible - fuses, etc - any ideas there?

Basically I'm looking for any ideas that might not be at first obvious, in
terms of providing safety to me, the operator, or the most exensive
components -

Any and all ideas/tips/etc. most welcome

Dave

1) The power supply transformer has a centre tap ,hence is meant to be a
1000 V -200 mA ,hence 200Watts supply device. When you would use it as a
2000V-AC device the primary winding and or the 'iron' will highly
likely be inadequate to produce 400W of power.
So the transformer will NOT be capable of supplying 1600* 0.305 equals
490 Watts. You will need a transformer which can supply at least 2.5
times the power you get from your available transformer .

2)Depending on the transformer voltage available say 1300 V the max
rectified DC voltage (due to the capacitor(s) will be 1.4* 1300 equals
1820 V ,hence the PIV of a bridge diode string would have to be at least
twice as high ,with a safety factor ,say 4000 if not 5000 V
It depends of the type of diode you use but if it is one with a PIV of
1000V you'll need 4 to 5 diodes in a bridge arm string.
Modern avalanche diodes are known not needing parallel capacitors or
resistors . The Recton diodes are highly likely a molded set of lower
PIV diodes . You could find out how many diodes are in series by
applying a low forward bias dc current and measure the voltage across
the module At approx 0.7 V across each diode you then can determine the
number of diodes in the module. You obviously can also use a string of
1N4007 1 Ampere diodes each having a PIV of 1000 V

3)The choke designed for 250 mA might be OK ,although is somewhat
marginal .Its resistance will cause a drop of the DC voltage hence in
order to get the required final DC supply voltage at max load the
transformer voltage has to be adequate . It also means that you have to
determine how much 'voltage sagging' is permissable.

4) If you use a string of electrolytic filter capacitors you need a
sufficient number to suit the rectifier bridge output voltage multiplied
by 1.4 and should apply a decent safety factor. Assuming you use
identical caps ,the overall capacitance is of course the value of 1
divided by the number of caps .
You have to pay much attention to the equalising resistors taking into
account both their voltage and power rating .
A well know RF Power Amplifier designer ,AG6K , recommends MOF =Metal
Oxide Film
resistors , although often one sees high power wire wound resistors on a
ceramic core being used.

5) You also need to decide in which operating Class the valves will have
to work in view of the acceptable distortion level of the amplifier.
Class A would be best but a waste of energy. In the intended push pull
configuration Class B would be OK.

6) The screen voltage of both valves will have to be stabilised

7) You also have to ensure that the valves are being cooled ,by means of
one or more small fans. A good method is to have an airtight
under-chassis pressurised by a small centrifugal fan and the valve
sockets fitted just below the chassis (on stand-offs)such that the air
flows along the valve bases and up. It might also be useful to apply
anode connectors with cooling fins.

8) If amplifier enclosure has a hatch giving access to High Voltage
carrying components it is essential that when the hatch is opened a
microswitch opens ,cutting all external power if left on. Normally of
course power should be cut (power supply plug pulled) and then ,only
after a few minutes, the hatch should be opened, because of retention of
charge by the filter capacitors for some time (depending on the value of
the voltage equalising resistors).

These are just a few of many aspects to consider when building a valve
power amplifier.

Frank GM0CSZ / KN6WH
=================================================
I am putting together a push-pull power amplifier based on the 813 power
tube. It will have about 1.5 to 1.7kV on the plates and put out at least
260 watts into a 9k load.

I am not new to high voltage tube circuits, but this is the first time I
will be working with potentials this high (above 1kV). I am aware of all of
the usual safety precautions, but I was hoping to get some specific tips for
putting together a design like this from some people who work with these
sort of voltages, hence the posting here.

For the plate supply, I have a Hammond power transformer laying around here
rated at 1000-0-1000 volts @ 200mA. From the ap notes on the 813, I am know
that the tubes will idle at about 50 mA and peak at about 305 mA. Would
this Hammond be ok for this type of use? I dont *think* I would be drawing
300 mils anywhere near continuously, but you never know. This amp will be
part of a bass guitar amplifier for live performance use.

I want to use solid state rectifiers and in studying some older ARRL
handbooks, they tell you to use equalizing resistors and small caps across
series-strung diodes to get up to the peak inverse voltage you need.
However I have seen a newer part in Mouser, made by Rectron, rated at 800mA
and 8kV PIV. Would something like this be all I would need (four in a
bridge configuration)? I would think it would be easier than making all of
those strings up from individual diodes. Or is there a better way?

I want to use a choke-input supply and I happen to have a large old filter
reactor made by Chicago Transformer rated at 20H @ 250mA, with a test
voltage of 6kV.

As for filter caps, I guess the cheapest way to go is to string together
series electrolytics with balancing resistors.

Also I have seen a lot of supplies use soft-start relays in the primary -
would you think this would be necessary in this application? I am looking
to get about 1.6kV out of this supply.

So basically you have a 2000vac secondary feeding a bridge of high voltage
rectifier diodes, then a 20H choke, and then a string of 'lytics. How much
capacitance would be sufficient to elliminate the hum in the plates? I know
that with push-pull and beam power tubes you can probably stang higher
ripple in the plates but is there a good ballpark figure?

Also since the 813 is directly-heated, would it be ok to have one 10v, 10A
filament transformer to light them both or would it be best to have two
separate ones to allow for hum balance -pots or for some other reason?

I also will have to make a screen supply for 750 volts at 45mA max. Would
this have to be a regulated supply?

I had planned on ordering some Belden 18awg high voltage test lead wire
rated to 10kV for all of the plate supply wiring. Would this be ok, or is
there a better wire for this application?

By the way I am having Heyboer custom wind an output transformer for this
thing. It was fairly expensive (around $200) so I want to protect it as
best as possible - fuses, etc - any ideas there?

Basically I'm looking for any ideas that might not be at first obvious, in
terms of providing safety to me, the operator, or the most exensive
components -

Any and all ideas/tips/etc. most welcome

Dave

On Fri, 07 Jul 2006 18:04:10 GMT, "dave" wrote:

Hello,


I am putting together a push-pull power amplifier based on the 813 power
tube. It will have about 1.5 to 1.7kV on the plates and put out at least
260 watts into a 9k load.

I am not new to high voltage tube circuits, but this is the first time I
will be working with potentials this high (above 1kV). I am aware of all of
the usual safety precautions, but I was hoping to get some specific tips for
putting together a design like this from some people who work with these
sort of voltages, hence the posting here.

For the plate supply, I have a Hammond power transformer laying around here
rated at 1000-0-1000 volts @ 200mA. From the ap notes on the 813, I am know
that the tubes will idle at about 50 mA and peak at about 305 mA. Would
this Hammond be ok for this type of use? I dont *think* I would be drawing
300 mils anywhere near continuously, but you never know. This amp will be
part of a bass guitar amplifier for live performance use.

If the XFmer is 1000-0-1000 with rectifiers (Full wave configuration)
you can count on around 1550V at 200ma or ~310W DC power.
You can load that higher if the peak to average is in the 300W range.
If you use a full wave bridge you can get twice the voltage and run
the same 200ma at ~2800-3000V for 600W.

The nasty with a 3000V supply is everything has to stand at least
10KV (allow for transients at least 2x or 3x the supply). Also 813s
may arc over at this high a voltage. (use 4-400s!).

The caveat is for sound amps and music overbuilding is a good thing.
Serious overbuilding is sometime required for commercial (stage)
use. Nothing worse than an amp smoking when the perfromer was.
Most I knew back when used multiple smaller amps rather than one
BIG one all driven from one head.


I want to use solid state rectifiers and in studying some older ARRL
handbooks, they tell you to use equalizing resistors and small caps across
series-strung diodes to get up to the peak inverse voltage you need.
However I have seen a newer part in Mouser, made by Rectron, rated at 800mA
and 8kV PIV. Would something like this be all I would need (four in a
bridge configuration)? I would think it would be easier than making all of
those strings up from individual diodes. Or is there a better way?

Either way works. follow the recommeded add xxx percent over rating to
prevent the smoke from escaping.

I want to use a choke-input supply and I happen to have a large old filter
reactor made by Chicago Transformer rated at 20H @ 250mA, with a test
voltage of 6kV.

Ok to 250ma and the power xfmer is 200ma so it's a winner to those
limits. You can go past that but then heating and peak to average
considerations apply (and shorter life is possible).

As for filter caps, I guess the cheapest way to go is to string together
series electrolytics with balancing resistors.

It is the way to go. Lots of them! Don't forget all those resistors
are eating up that limited 200ma, they are needed but it emphasizes
the limited transformer your using.

Also I have seen a lot of supplies use soft-start relays in the primary -
would you think this would be necessary in this application? I am looking
to get about 1.6kV out of this supply.

Yes, saves popping mains fuses from power on surges and also is
easier on the HV components.

FYI: the filliments on the 813s must be up to temp before DC plate or
screen power is applied.

So basically you have a 2000vac secondary feeding a bridge of high voltage
rectifier diodes, then a 20H choke, and then a string of 'lytics. How much
capacitance would be sufficient to elliminate the hum in the plates? I know
that with push-pull and beam power tubes you can probably stang higher
ripple in the plates but is there a good ballpark figure?

Enough to keep the ripple at full load less than 1%. Start around
60-100uf.

See caveat for ~3000V on plates!

Pushpull will help balance out ripple.

Also since the 813 is directly-heated, would it be ok to have one 10v, 10A
filament transformer to light them both or would it be best to have two
separate ones to allow for hum balance -pots or for some other reason?

One of adaquate current rating with a 50 ohm pot with the wiper to
ground for balance will do. That 50ohm pot is a high power resistor.
The alternate is a 10V center tapped or two 5V transformers.

I also will have to make a screen supply for 750 volts at 45mA max. Would
this have to be a regulated supply?

Absolutely. Likely you will need another transformer, rectifier and
filter for that voltage as you only have a 200ma transformer for the
plates so why waste some of it on the screens. Also the screen
supply must not be applied before the plate voltage.

Not needed if you use 811 or other power triode.

I had planned on ordering some Belden 18awg high voltage test lead wire
rated to 10kV for all of the plate supply wiring. Would this be ok, or is
there a better wire for this application?

Consider that the environment is not only HV but also HOT. Make sure
the wire (and all your parts) can stand heat too.

By the way I am having Heyboer custom wind an output transformer for this
thing. It was fairly expensive (around $200) so I want to protect it as
best as possible - fuses, etc - any ideas there?

Fuses should abound. Minimally, mains input, and DC to plate
transformer.

Basically I'm looking for any ideas that might not be at first obvious, in
terms of providing safety to me, the operator, or the most exensive
components -

ENCLOSE everything with HV on it. Finger proof it. and provide
interlocks. Consider cooling as well.

Any and all ideas/tips/etc. most welcome

Reminder 2 813s with only filliments lit are pumping out some 100W
of heat! (10V at 5A).

The tubes should be biased between class AB1 and B for minimum
distortion. that means at idle you likely standing 20-50ma per tube
(another 50-150W of heat!).

You also need a -V (0-100V) negative to bias the tubes.
Another small transformer as trivial current is needed
mostly for bias pot and bleeders.

Mechanical construction is important as with all that iron
(transformers) it's going to be heavy. I mean wheels on this
beast are a must.

You also need a driver and that will have it's own power needs
though only a few watts need be developed. The low level amps
need to be kept far from the power transformers or it will HUM.


Allison



Dave

wrote:
On Fri, 07 Jul 2006 18:04:10 GMT, "dave" wrote:

Hello,


I am putting together a push-pull power amplifier based on the 813 power
tube. It will have about 1.5 to 1.7kV on the plates and put out at least
260 watts into a 9k load.

I am not new to high voltage tube circuits, but this is the first time I
will be working with potentials this high (above 1kV). I am aware of all of
the usual safety precautions, but I was hoping to get some specific tips for
putting together a design like this from some people who work with these
sort of voltages, hence the posting here.

For the plate supply, I have a Hammond power transformer laying around here
rated at 1000-0-1000 volts @ 200mA. From the ap notes on the 813, I am know
that the tubes will idle at about 50 mA and peak at about 305 mA. Would
this Hammond be ok for this type of use? I dont *think* I would be drawing
300 mils anywhere near continuously, but you never know. This amp will be
part of a bass guitar amplifier for live performance use.

If the XFmer is 1000-0-1000 with rectifiers (Full wave configuration)
you can count on around 1550V at 200ma or ~310W DC power.
You can load that higher if the peak to average is in the 300W range.
If you use a full wave bridge you can get twice the voltage and run
the same 200ma at ~2800-3000V for 600W.

The nasty with a 3000V supply is everything has to stand at least
10KV (allow for transients at least 2x or 3x the supply). Also 813s
may arc over at this high a voltage. (use 4-400s!).

The caveat is for sound amps and music overbuilding is a good thing.
Serious overbuilding is sometime required for commercial (stage)
use. Nothing worse than an amp smoking when the perfromer was.
Most I knew back when used multiple smaller amps rather than one
BIG one all driven from one head.


I want to use solid state rectifiers and in studying some older ARRL
handbooks, they tell you to use equalizing resistors and small caps across
series-strung diodes to get up to the peak inverse voltage you need.
However I have seen a newer part in Mouser, made by Rectron, rated at 800mA
and 8kV PIV. Would something like this be all I would need (four in a
bridge configuration)? I would think it would be easier than making all of
those strings up from individual diodes. Or is there a better way?

Either way works. follow the recommeded add xxx percent over rating to
prevent the smoke from escaping.

I want to use a choke-input supply and I happen to have a large old filter
reactor made by Chicago Transformer rated at 20H @ 250mA, with a test
voltage of 6kV.

Ok to 250ma and the power xfmer is 200ma so it's a winner to those
limits. You can go past that but then heating and peak to average
considerations apply (and shorter life is possible).

As for filter caps, I guess the cheapest way to go is to string together
series electrolytics with balancing resistors.

It is the way to go. Lots of them! Don't forget all those resistors
are eating up that limited 200ma, they are needed but it emphasizes
the limited transformer your using.

Also I have seen a lot of supplies use soft-start relays in the primary -
would you think this would be necessary in this application? I am looking
to get about 1.6kV out of this supply.

Yes, saves popping mains fuses from power on surges and also is
easier on the HV components.

FYI: the filliments on the 813s must be up to temp before DC plate or
screen power is applied.

So basically you have a 2000vac secondary feeding a bridge of high voltage
rectifier diodes, then a 20H choke, and then a string of 'lytics. How much
capacitance would be sufficient to elliminate the hum in the plates? I know
that with push-pull and beam power tubes you can probably stang higher
ripple in the plates but is there a good ballpark figure?

Enough to keep the ripple at full load less than 1%. Start around
60-100uf.

See caveat for ~3000V on plates!

Pushpull will help balance out ripple.

Also since the 813 is directly-heated, would it be ok to have one 10v, 10A
filament transformer to light them both or would it be best to have two
separate ones to allow for hum balance -pots or for some other reason?

One of adaquate current rating with a 50 ohm pot with the wiper to
ground for balance will do. That 50ohm pot is a high power resistor.
The alternate is a 10V center tapped or two 5V transformers.

I also will have to make a screen supply for 750 volts at 45mA max. Would
this have to be a regulated supply?

Absolutely. Likely you will need another transformer, rectifier and
filter for that voltage as you only have a 200ma transformer for the
plates so why waste some of it on the screens. Also the screen
supply must not be applied before the plate voltage.

Not needed if you use 811 or other power triode.

I had planned on ordering some Belden 18awg high voltage test lead wire
rated to 10kV for all of the plate supply wiring. Would this be ok, or is
there a better wire for this application?

Consider that the environment is not only HV but also HOT. Make sure
the wire (and all your parts) can stand heat too.

By the way I am having Heyboer custom wind an output transformer for this
thing. It was fairly expensive (around $200) so I want to protect it as
best as possible - fuses, etc - any ideas there?

Fuses should abound. Minimally, mains input, and DC to plate
transformer.

Basically I'm looking for any ideas that might not be at first obvious, in
terms of providing safety to me, the operator, or the most exensive
components -

ENCLOSE everything with HV on it. Finger proof it. and provide
interlocks. Consider cooling as well.

Any and all ideas/tips/etc. most welcome

Reminder 2 813s with only filliments lit are pumping out some 100W
of heat! (10V at 5A).

The tubes should be biased between class AB1 and B for minimum
distortion. that means at idle you likely standing 20-50ma per tube
(another 50-150W of heat!).

You also need a -V (0-100V) negative to bias the tubes.
Another small transformer as trivial current is needed
mostly for bias pot and bleeders.

Mechanical construction is important as with all that iron
(transformers) it's going to be heavy. I mean wheels on this
beast are a must.

You also need a driver and that will have it's own power needs
though only a few watts need be developed. The low level amps
need to be kept far from the power transformers or it will HUM.


Allison



Dave




Have you got the audio output transformer for this beastie? - if not,
get a price before you start - you might reconsider the project.

It is going to be physicaly big, very heavy, and fragile - 813
filaments dont like being knocked around in the back of a truck (and
you will need a truck or SUV to move it) so if its intended to go on
the road, it might be more trouble than its worth.

But, so saying that, ----- nice lunatic fringe project - worth doing,
just for the hell of it, and ignore practical realities......

Andrew VK3BFA.

Sorry - didnt read closely enough - you have the output transformer....

Andrew VK3BFA.

Thanks to all who responded - some very good points to ponder...

I think I should spring for a different plate transformer, since I don't
want to be running this one too marginally. Also, I suppose I could do away
with the choke and still get under 1% ripple. That would definitely lighten
up the weight. The Hammond # 720 is rated 625-0-625 V at 300mA CCS and
500mA ICAS, so I guess that would be adequate.

I am using a free program called PSU Designer II to calculate the power
supply values and outputs. With the Hammond #720 and a 40uF/47ohm/40uF
C-R-C filter after the diode bridge, I should get about 1760VDC at 50mA and
about 1660VDC at 300mA.

I was also thinking of using three toggle switches across the front panel
for filaments, plates, and screens, arranged so as to reflect the order in
which to apply the power. I suppose some sort of relay or other device
could be used to make sure the screens never got power before or in place of
the plates. It is a good idea about the interlocks on the chassis too - I
know nobody else will be using this amp besides me but still always a good
idea to design in the safety. Also it won't be toured around as I play in
basement bands most of the time so it probably won't see too much road use.
It's more of a crazy way to generate 300 watts rather than go with the
ordinary 'three pairs of KT88 's' route.

thanks again sincerely for all the input!

Dave

I was also thinking of using three toggle switches across the front panel
for filaments, plates, and screens, arranged so as to reflect the order in
which to apply the power. I suppose some sort of relay or other device
could be used to make sure the screens never got power before or in place of
the plates. It is a good idea about the interlocks on the chassis too - I
know nobody else will be using this amp besides me but still always a good
idea to design in the safety.
=====================================
Instead of having 3 separate switches ,go for a single manual switch , a
sequential timer or separate timers and have 3 front panel lights for
filaments ,anodes and screen voltages. With 3 separate switches you
might operate these in the wrong sequence. With the 3 lights to come on
sequentially you quickly can detect any possible sequential switching
problems.

Frank GM0CSZ / KN6WH