Benjamin,
I generally want to not go crazy because i'd like the respect of any engineers that might look at my designs down the road and not get the "This guy didn't have a clue" response.
I have some of those large Hg rectifiers and matching thyratrons.
If you want a no-RFI problem rectifier that will drop into the 866A socket with identical filament, voltage and current ratings use a 3B28 Helium filled rectifier tube. No pretty, UV intensive, potentially eye damaging glow however.
I generally want to not go crazy because i'd like the respect of any engineers that might look at my designs down the road and not get the "This guy didn't have a clue" response.
I have some of those large Hg rectifiers and matching thyratrons.
If you want a no-RFI problem rectifier that will drop into the 866A socket with identical filament, voltage and current ratings use a 3B28 Helium filled rectifier tube. No pretty, UV intensive, potentially eye damaging glow however.
rcavictim said:
Oh if anyone asks me that I'll just say that I'm building for future experimentation with higher operating points.
I personally wouldn't mind 2.5kV on an 813 for audio purposes...
3B28 can be another option, Cheers!
The 6CJ3 I think is my best option right now, apart from a hybrid SS/Tube rectified setup, I think the power supply will be an on-going problem.
I'm assuming the 6CJ3 will be choke input? There isn't very much data available on it..
Okay here is a good design:
Sowter M057, 780vCT 400mA (These are DC values)
6CJ3 Full-wave CT Rectification.
C1 16uF, Parallel 8uF Motorstart 933vDC simple capacitor filter.
Now, C2 comes off this choke:
Hammond 193Q at 10H, 1000v DC Max, 500mA DC Max to filter to ground.
C2 33.5uF combination of 8uF caps and 17.5uF caps in ser/par for a min voltage of 933vDC.
C3 is a simple Resistor filter with the shown combination up-to 33.5uF so I can nudge the voltage in the right direction with nothing but a resistor change.
Inbetween here a
The load is represented as 900v 200mA, so 1.8k ohm.
Now, Sowter say that they have provided the full-load voltage rating of the transformer, this means that the voltage will increase as current is pulled back.
If I am going to get a 400mA transformer and only draw 200mA from it, surely the voltage will increase.
I will simply decrease the voltage as necessary via the decoupling resistor and a bleeder resistor.
decoupling resistor = 50 ohm for starters.
bleeder = 50k ohm, 25 watt
Sowter M057, 780vCT 400mA (These are DC values)
6CJ3 Full-wave CT Rectification.
C1 16uF, Parallel 8uF Motorstart 933vDC simple capacitor filter.
Now, C2 comes off this choke:
Hammond 193Q at 10H, 1000v DC Max, 500mA DC Max to filter to ground.
C2 33.5uF combination of 8uF caps and 17.5uF caps in ser/par for a min voltage of 933vDC.
C3 is a simple Resistor filter with the shown combination up-to 33.5uF so I can nudge the voltage in the right direction with nothing but a resistor change.
Inbetween here a
The load is represented as 900v 200mA, so 1.8k ohm.
Now, Sowter say that they have provided the full-load voltage rating of the transformer, this means that the voltage will increase as current is pulled back.
If I am going to get a 400mA transformer and only draw 200mA from it, surely the voltage will increase.
I will simply decrease the voltage as necessary via the decoupling resistor and a bleeder resistor.
decoupling resistor = 50 ohm for starters.
bleeder = 50k ohm, 25 watt
Mercury Rectifiers
575A and 872 are quite easy to find out there and are cheap to acquire,
but, not cheap to operate. 872's want 10V 5A to light, so to use a pair
of these in an amp means dissipating 100W just for the filaments and
then buying a 15A 10VCT filament transformer which will be neither
cheap nor small! Not worth the hassle... If you had to rectify off
a 4800VCT tranny - maybe these tubes are you only good option...
As far as RFI, I've used 866JR, 866A and 3B28 and the sound that
comes out of my amps is fresh, lively and warm - not a hint of RFI.
I have 4kV 10pF ceramic disk between anode and CT to shunt HF.
Jack Elliano told me that their attempts to use the 813 were difficult
due to oscillation. That "this tube really want to go RF" and they tried
all kinds of tricks to tame it. It might be that MV in conjunction with the
813 is asking for trouble, whereas with the 845 - things are more tame.
-- Jim
575A and 872 are quite easy to find out there and are cheap to acquire,
but, not cheap to operate. 872's want 10V 5A to light, so to use a pair
of these in an amp means dissipating 100W just for the filaments and
then buying a 15A 10VCT filament transformer which will be neither
cheap nor small! Not worth the hassle... If you had to rectify off
a 4800VCT tranny - maybe these tubes are you only good option...
As far as RFI, I've used 866JR, 866A and 3B28 and the sound that
comes out of my amps is fresh, lively and warm - not a hint of RFI.
I have 4kV 10pF ceramic disk between anode and CT to shunt HF.
Jack Elliano told me that their attempts to use the 813 were difficult
due to oscillation. That "this tube really want to go RF" and they tried
all kinds of tricks to tame it. It might be that MV in conjunction with the
813 is asking for trouble, whereas with the 845 - things are more tame.
-- Jim
/
Er, 660VCT with a FWCT rectifier will do about 900V. Think peak voltage (cap only charges on peaks). Average is about 0.9 times so choke input (chokes average voltage and try to average current) is a bit lower, should you be so inclined.
Er, no? You set the resistance right? Or are you using bad diodes?
A little higher, but what do I know, I like good diodes...tube diodes can swing 300V easily...(a 5Y3 can go from 420V open circuit to 300V at 100mA from a 600VCT supply.)
Okie...
Hmm, interesting! Then so far all transformers I've used are overrated, none have melted...
Hammonds must be rated for very large C then, as my 269AX is still going strong at 65mA DC into 50uF (FWB to get 320V instead of the 100mA rated FWCT). Mind exactly that - transformers may be rated in AC or DC terms.
Well screw you too.
Hey, everything I calculate manually works perfectly*, I just never had a reliable reference for average vs. RMS and assumed it was around sqrt(2), which assumes giving up voltage for current.
*Well okay, Hept'AU7 only passes the three second test, and I have asbestos fingers. It's class A right now, so has a bit more current draw than usual, which does the 10 second test. But I don't know what's up with the scrap PT anyway... it looks 150W but has awfully thin wires... probably 25Hz or something...IOW, not my fault
Tim
Layberinthius said:So a 660-0-660 at 400mA would be after full wave = 660vDC ?????
Confusing!
Er, 660VCT with a FWCT rectifier will do about 900V. Think peak voltage (cap only charges on peaks). Average is about 0.9 times so choke input (chokes average voltage and try to average current) is a bit lower, should you be so inclined.
Er, no? You set the resistance right? Or are you using bad diodes?
A little higher, but what do I know, I like good diodes...tube diodes can swing 300V easily...(a 5Y3 can go from 420V open circuit to 300V at 100mA from a 600VCT supply.)
Knarf said:
Okie...
Hmm, interesting! Then so far all transformers I've used are overrated, none have melted...
So for a load current of 100mA, the transformer should be 'overrated' between 2.6 and 5.6 times in its AC current rating in order for it not to be run outside specs, meaning 260mA and up to 560mA.The reason, of course, is that the peak current drawn when charging the capacitor is much, much higher than the average load current, so even though the pulses are comparatively shorter as compared to running DC through the trafo loss 'resistor', the losses are proportional with I^2 and they catch up with us.
Hammonds must be rated for very large C then, as my 269AX is still going strong at 65mA DC into 50uF (FWB to get 320V instead of the 100mA rated FWCT). Mind exactly that - transformers may be rated in AC or DC terms.
Well screw you too.
Hey, everything I calculate manually works perfectly*, I just never had a reliable reference for average vs. RMS and assumed it was around sqrt(2), which assumes giving up voltage for current.*Well okay, Hept'AU7 only passes the three second test, and I have asbestos fingers. It's class A right now, so has a bit more current draw than usual, which does the 10 second test. But I don't know what's up with the scrap PT anyway... it looks 150W but has awfully thin wires... probably 25Hz or something...IOW, not my fault
Tim
adalin said:
Yes but not quite. Take a stock transformer, plug the primary into the AC 120 volt line, no secondary load and watch it overheat and melt down in several hours or less, what would be normal daily listening time for an amplifier.
Trouble is they don't use enough turns on the primary for the amount of iron which is also skimpy. Take a clamp-on amp meter and monitor the unloaded transformer primary current while raising the primary voltage to the transformer with a variac. Above about 90 VAC it starts rising as the transformer saturates. By 120 volts you have three or four amps at 120 volts flowing. This makes a lot of core and winding heat. If you use an external bucking transformer to reduce the oven xfmer to 60 volts primary then you will have a useable 1000 volt output xfmer. Two can be used to provide 1000-0-1000. This takes three transformers though.
Unfortunately it is not a simple thing to wire two transformer's primaries in series to place 60 volts across each transformer because the load on each occurs on opposite half cycles and the transformer not loaded at the time just swings like a current limiting reactor cutting back the current to the loaded transformer. There is a way around this. You hammer out the magnetic shunts to create some core window and wind an equal coil of heavy wire, say #12 or #14 AWG around each core, as many turns as you can fit, hopefully at least 20. Then you connect these two coils, one in each transformer together. If you get the polarity correct this does the same effect as having a common magnetic core.
Now you have a useable 1000-0-1000 plate xfmer good for 500 mA or so for a big tube amp. The transformers will still get hot though because of eddy curents where the laminations are welded together, but with half the normal magnetic flux in the core it won't be too serious.
I like to use scavenged microwave oven xfmers as custom filament transformers. I saw off the HV secondary and knock out the shunts. In the core space I wind extra primary turns to take care of the core saturation issue, then in the space left I wind my filament windings. You get about 1/2 to 3/4 volt per turn, so you don't need much wire to get 5, 6.3 or even 10 volts. Ten amp windings are easy with #14 magnet wire.
Graph
Mains trafo is Sowter M045, (Enclosed) (Estimated price of: £189.54)
660vCT 400mA
6.3vAC (no-ct) 8A
45vAC (no-ct) 100mA
8.5 kg (18.7 lbs)
Choke 1 can be Sowter CB16, (Estimated price of: £62.81)
10H
300mA
55 ohm
2.6 kg (5.7 lbs)
Choke 2 can be Sowter CA08, (Estimated price of: £48.07)
5 H
300mA
40 ohm
1.2 kg (2.6lbs)
Prices are UK Pounds and off their website, I have no affilliation with advertising for E A Sowter Ltd, reference was purely for advertising that this is a cheaper alternative to hacked Microwave transformers =)
Mains trafo is Sowter M045, (Enclosed) (Estimated price of: £189.54)
660vCT 400mA
6.3vAC (no-ct) 8A
45vAC (no-ct) 100mA
8.5 kg (18.7 lbs)
Choke 1 can be Sowter CB16, (Estimated price of: £62.81)
10H
300mA
55 ohm
2.6 kg (5.7 lbs)
Choke 2 can be Sowter CA08, (Estimated price of: £48.07)
5 H
300mA
40 ohm
1.2 kg (2.6lbs)
Prices are UK Pounds and off their website, I have no affilliation with advertising for E A Sowter Ltd, reference was purely for advertising that this is a cheaper alternative to hacked Microwave transformers =)
Attachments
Labyrinth and your 900V supply
Heya. I'd be concerned about the 900 mA ripple current pulses.
Although the tranny might be able to source 500mA or so, there might
not be sufficient magnetic flux remaining. If that is the case, the transformer will saturate every 2*Power Line Frequency periods and
the transformer will buzz, heat up and lots of leakage flux will circulate.
Only the tranny mfg can tell you if it can take the 900mA pulses.
If you look at the specs for the Hammond 700 series, you see that
the CCS (continuous current service) versus ICAS (intemittent current)
differs only by 30%. In your model, you are counting on a 100%
spread between CCS and ICAS numbers, this might be overly optimistic.
A further study of the Hammond 700 series specs reveals something
more interesting... Every transformer that can provide 300mA or
more, regardless of voltages, weighs a minimum of 12 lbs. Further
400 and 500mA transformers, regardless of voltage, seem to weigh
in between 17-25 lbs.
The transformer that you are looking at ought to weight at least this...
If it doesnt, you might not have the right transformer.
-- Jim
Heya. I'd be concerned about the 900 mA ripple current pulses.
Although the tranny might be able to source 500mA or so, there might
not be sufficient magnetic flux remaining. If that is the case, the transformer will saturate every 2*Power Line Frequency periods and
the transformer will buzz, heat up and lots of leakage flux will circulate.
Only the tranny mfg can tell you if it can take the 900mA pulses.
If you look at the specs for the Hammond 700 series, you see that
the CCS (continuous current service) versus ICAS (intemittent current)
differs only by 30%. In your model, you are counting on a 100%
spread between CCS and ICAS numbers, this might be overly optimistic.
A further study of the Hammond 700 series specs reveals something
more interesting... Every transformer that can provide 300mA or
more, regardless of voltages, weighs a minimum of 12 lbs. Further
400 and 500mA transformers, regardless of voltage, seem to weigh
in between 17-25 lbs.
The transformer that you are looking at ought to weight at least this...
If it doesnt, you might not have the right transformer.
-- Jim
Well, it might be easier if you could relax the requirement of 550mA CCS.
At that power level, you honestly should be talking to someone with
RF Linear Amplifier design experience and you should be planning for
the size and expense associated with the parts needed to get the
job done. Peter W. Dahl & CO (www.pwdahl.com) has transformers for
sale for this purpose. It is going to be big (150 lbs or more), heavy and
the shipping costs for the parts might cost as much as a NOS tubes.
I'd not use Electrolytics caps for high voltage, particularly as an input
capacitor. The high ripple pulse currents cause them to heat up. The
failure modes of Electrolytics are unreasonable at 1000V. Paper in oil
caps have softer failure modes. They are self-healing for modest internal faults. I would say that a paper (MP) or film (MKP) is a must for an
input capacitor!
As a reference case, Morgan Jones talkes about a HV PP amp in his
book Valve Amplifiers 2nd Edn. Its called the "Beast". Its a PP 845 rig
that uses 900V at 200mA per channel. In the case study, you will
also see that he needs to use a 350mA rated choke even though he is
drawing 200mA CCS (if you read the article and follow the analysis, you
will see why this overrated choke is utterly necessary). A 10H 350mA
rated choke is going to weigh in around 10 lbs. A choke rated at 1000mA
to account for 500mA CCS might be a 40-50 lb device and chances are
it was designed for 5000V service for a transmitter somewhere which means its going to be big and very ugly to look at.... Your listening
room might look more like a substation yard than a living space.
Reclosers and OCBs might be needed for your mute/operate switch
and you might want to invest in a fiberglass pole and rubber gloves
for operating the switches...
I think a 900V 200mA supply might be reasonable to talk about...
-- Jim
At that power level, you honestly should be talking to someone with
RF Linear Amplifier design experience and you should be planning for
the size and expense associated with the parts needed to get the
job done. Peter W. Dahl & CO (www.pwdahl.com) has transformers for
sale for this purpose. It is going to be big (150 lbs or more), heavy and
the shipping costs for the parts might cost as much as a NOS tubes.
I'd not use Electrolytics caps for high voltage, particularly as an input
capacitor. The high ripple pulse currents cause them to heat up. The
failure modes of Electrolytics are unreasonable at 1000V. Paper in oil
caps have softer failure modes. They are self-healing for modest internal faults. I would say that a paper (MP) or film (MKP) is a must for an
input capacitor!
As a reference case, Morgan Jones talkes about a HV PP amp in his
book Valve Amplifiers 2nd Edn. Its called the "Beast". Its a PP 845 rig
that uses 900V at 200mA per channel. In the case study, you will
also see that he needs to use a 350mA rated choke even though he is
drawing 200mA CCS (if you read the article and follow the analysis, you
will see why this overrated choke is utterly necessary). A 10H 350mA
rated choke is going to weigh in around 10 lbs. A choke rated at 1000mA
to account for 500mA CCS might be a 40-50 lb device and chances are
it was designed for 5000V service for a transmitter somewhere which means its going to be big and very ugly to look at.... Your listening
room might look more like a substation yard than a living space.
Reclosers and OCBs might be needed for your mute/operate switch
and you might want to invest in a fiberglass pole and rubber gloves
for operating the switches...
I think a 900V 200mA supply might be reasonable to talk about...
-- Jim
Alright I'm apparently two pages behind so I'm just going to reply to what I see...
1. You DO NOT ever need two chokes. 10H as-is is too much in my book... I'd use 470uF after mind you. (Note I wind my own chokes, if I were buying, I'd go with whatever I can get... and it just so happens 10H is an extremely common value. Or if it's what you have on hand, all the more filtering, eh.) At 800V this isn't as easy but that's why God allowed electrolytics to be connected in series, with the added advantage of bleeder resistors to boot. If you have oil caps on hand and just want to use them, whatever. But you don't get as much stiffness... your LF response funeral...
2. MOT's aren't evil, but as-is they are very close to saturation and have low inductance (read: high idle current). Remember, reading a current monitoring resistor doesn't tell you anything, you have to scope it. If it is indeed drawing large currents at the peaks, then it is saturating. If sinusoidal, it isn't saturating, no way no how. Mmmmm'k?
3. Erm.....primaries in series? That puts 60V per primary and lets you ground the cores for a CT without rewiring them, which is dubious at best as the secondaries aren't designed to handle voltages between the core and the beginning of the secondary winding.
4. 1N4007 x 12
Tim
1. You DO NOT ever need two chokes. 10H as-is is too much in my book... I'd use 470uF after mind you. (Note I wind my own chokes, if I were buying, I'd go with whatever I can get... and it just so happens 10H is an extremely common value. Or if it's what you have on hand, all the more filtering, eh.) At 800V this isn't as easy but that's why God allowed electrolytics to be connected in series, with the added advantage of bleeder resistors to boot. If you have oil caps on hand and just want to use them, whatever. But you don't get as much stiffness... your LF response funeral...
2. MOT's aren't evil, but as-is they are very close to saturation and have low inductance (read: high idle current). Remember, reading a current monitoring resistor doesn't tell you anything, you have to scope it. If it is indeed drawing large currents at the peaks, then it is saturating. If sinusoidal, it isn't saturating, no way no how. Mmmmm'k?
3. Erm.....primaries in series?
That puts 60V per primary and lets you ground the cores for a CT without rewiring them, which is dubious at best as the secondaries aren't designed to handle voltages between the core and the beginning of the secondary winding.4. 1N4007 x 12
Tim
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.