I read this yesterday on diyAudio:
For those who do [not] like [a particular directly heated rectifier tube - in this case a 5Y3], only because it does not have delayed B+ . . .
Then add an additional 6.3V filament winding (a rectifier dedicated winding, not for the other tube filaments), and . . .
Then connect a 0.65 Ohm 5 Watt resistor in series from one end of the 6.3V to the [tube's] filament, and the other end of the 6.3V to the other end of [its] filament.
That makes a Slow Start B+.
Is this true and as straightforward as it seems?
I have a 5U4G (actually a Svetlana 5C3S / 5Ц3С) that I am proposing to use (I like the coke bottle shape and its voltage drop suits, and I have a few) in an RH84 amplifier, using 6P14P-EVs. I may have a spare 6.3V secondary that I could use this way. Is it advantageous for me to do so?
Simon
For those who do [not] like [a particular directly heated rectifier tube - in this case a 5Y3], only because it does not have delayed B+ . . .
Then add an additional 6.3V filament winding (a rectifier dedicated winding, not for the other tube filaments), and . . .
Then connect a 0.65 Ohm 5 Watt resistor in series from one end of the 6.3V to the [tube's] filament, and the other end of the 6.3V to the other end of [its] filament.
That makes a Slow Start B+.
Is this true and as straightforward as it seems?
I have a 5U4G (actually a Svetlana 5C3S / 5Ц3С) that I am proposing to use (I like the coke bottle shape and its voltage drop suits, and I have a few) in an RH84 amplifier, using 6P14P-EVs. I may have a spare 6.3V secondary that I could use this way. Is it advantageous for me to do so?
Simon
You might get away with this for a very low current application.
But even one 6P14P needs 40-50mA+, and this will cause arcovers in directly heated rectifiers, if the 6P14Ps are already warmed up. (if they are starting cold at the same time, the B+ will rise immediately, even with this circuit).
Once a DH rectifier has arced over, it will repeat this trick in the same location at every startup, each time adding gas to the vacuum. Outcome: dead rectifier
Damper diodes may not be so pretty, but they provide a very smooth and slow rise time, and are much more reliable than the older big bottle types.
But even one 6P14P needs 40-50mA+, and this will cause arcovers in directly heated rectifiers, if the 6P14Ps are already warmed up. (if they are starting cold at the same time, the B+ will rise immediately, even with this circuit).
Once a DH rectifier has arced over, it will repeat this trick in the same location at every startup, each time adding gas to the vacuum. Outcome: dead rectifier
Damper diodes may not be so pretty, but they provide a very smooth and slow rise time, and are much more reliable than the older big bottle types.
Typically a relay contact is needed to short out any added series resistance in the rectifier heater, in order to obtain nominal voltage for the rectifier heater once enough time has elapsed for the remaining valves in the amp to start conducting. That can be done in a few ways, such as with a cheap ebay 555 timer with rely pcb. The way I do it now for a 5U4 is to use a particular NTC resistor (5D-15) and a common 12Vdc relay (Carlo Gavazi MZPA002 44 05).
Given this is adding complexity, imho there should be a good reason for doing it, such as your filter caps will get stressed because you are using a choke-input filter. If your reason is to reduce some concept of surge, or shock, or to improve valve lifetime, then I think they are typically invalid and unsupported reasons for common valve amp applications.
Given this is adding complexity, imho there should be a good reason for doing it, such as your filter caps will get stressed because you are using a choke-input filter. If your reason is to reduce some concept of surge, or shock, or to improve valve lifetime, then I think they are typically invalid and unsupported reasons for common valve amp applications.
The idea came from post #2 in a thread about substitutes for a 5Y3 tube. The poster proposed running a 5 volt DH rectifier, the 5Y3 from an ISOLATED 6.3 volt winding with a suitable resistor in series. The DH rectifier has a much lower heater resistance when cold that it does when warmed up. Most of the heater supply voltage will appear across the resistor at power on, but the voltage will appear across the tube as its heater resistance rises with temperature. This provides a slower start than the typical DH rectifier, but it's not as slow as a big damper tube with a large fat cathode. There is no need to short the resistor with a relay since it is sized to put 5 volts across a hot rectifier tube.
The 6.3 volt winding used by this method will be elevated to the full B+ voltage so it must not be shared with other tubes, or their H/K breakdown ratings will be severely violated leading to lots of burnt parts.
Virtually zero 6.3 volt windings actually produce exactly 6.3 volts today due to higher line voltages and a bunch of other variables, so this resistor value will likely need to be determined empirically. Math says 0.65 ohms for the 5Y3 and 0.43 ohms for a 5U4. Since it will see a big surge at turn on, the resistor should be rated for 5 watts or more for a 5Y3 and 10 watts for a 5U4.
The 6.3 volt winding used by this method will be elevated to the full B+ voltage so it must not be shared with other tubes, or their H/K breakdown ratings will be severely violated leading to lots of burnt parts.
Virtually zero 6.3 volt windings actually produce exactly 6.3 volts today due to higher line voltages and a bunch of other variables, so this resistor value will likely need to be determined empirically. Math says 0.65 ohms for the 5Y3 and 0.43 ohms for a 5U4. Since it will see a big surge at turn on, the resistor should be rated for 5 watts or more for a 5Y3 and 10 watts for a 5U4.
Voltage drop across the slow start device may need to be taken into consideration. The damper diode will have a lower drop than a 5Y3 or 5U4 would have.You might get away with this for a very low current application.
But even one 6P14P needs 40-50mA+, and this will cause arcovers in directly heated rectifiers, if the 6P14Ps are already warmed up. (if they are starting cold at the same time, the B+ will rise immediately, even with this circuit).
Once a DH rectifier has arced over, it will repeat this trick in the same location at every startup, each time adding gas to the vacuum. Outcome: dead rectifier
Damper diodes may not be so pretty, but they provide a very smooth and slow rise time, and are much more reliable than the older big bottle types.
For all of my amplifiers, I use solid stage B+ rectifiers.
Some will say, OH NO! (both for sound, and for tube life; . . . so they say).
I have not had reduced output tube life, even though the B+ comes up quickly (for the various output tube circuits I use; they are either individual self biased, or PP tubes sharing a single self bias resistor with Very well matched tubes).
A couple of decades ago, this also worked for DHT output tubes that had battery bias.
Your mileage may vary.
What I did with 300B tubes will cause some to cringe:
I used brute force DC supplies for the DHT 300B filaments.
6.3VAC filament winding, Schottky bridge, 22,000uF first cap, 2 Ohm series resistor (adjust for 5V on the 300B filament when it finally warms up), and 22,000uF second cap across the 300B filament. 1mV to 2mV ripple, good enough for single ended output stage, unless you use headphones instead of speakers.
Your mileage may vary.
I try and make my designs reliable, including living through brief power mains interruptions (Hot Starts).
In addition to the topologies I use, running tubes lower than rated dissipation; . . . I also design additional protection.
Example: I have a series string of fast-blow and slow-blow fuses in my power transformer's primary circuit.
One amplifier uses a 1.25A fast-blow in series with a 600mA slow-blow.
Others vary from that, according to the inrush current which is higher than the warmed up Run-Current.
One amplifier would run for 1 to 3 months with a 0.5A slow blow, then it would finally fatigue and open, so I changed the slow blow to 0.6A (600mA),
Still running with that fuse.
When using '5V' rectifier tubes, they do not always exactly meet their rated current draws of 2 Amps, 3 Amps, etc.
Series resistors have to be adjusted accordingly so that when the filament is warm it is 5V.
Always measure and test your designs for proper current, voltage, dissipation, etc.
My experience with power-up arcing of an indirect heated 5AR4 was found to be the 100uF capacitor that immediately followed the 5AR4 cathode.
Do Not do that!
Just my opinions, and my experience.
Worked good with the tubes and parts and circuits I designed.
Your mileage may vary
Some will say, OH NO! (both for sound, and for tube life; . . . so they say).
I have not had reduced output tube life, even though the B+ comes up quickly (for the various output tube circuits I use; they are either individual self biased, or PP tubes sharing a single self bias resistor with Very well matched tubes).
A couple of decades ago, this also worked for DHT output tubes that had battery bias.
Your mileage may vary.
What I did with 300B tubes will cause some to cringe:
I used brute force DC supplies for the DHT 300B filaments.
6.3VAC filament winding, Schottky bridge, 22,000uF first cap, 2 Ohm series resistor (adjust for 5V on the 300B filament when it finally warms up), and 22,000uF second cap across the 300B filament. 1mV to 2mV ripple, good enough for single ended output stage, unless you use headphones instead of speakers.
Your mileage may vary.
I try and make my designs reliable, including living through brief power mains interruptions (Hot Starts).
In addition to the topologies I use, running tubes lower than rated dissipation; . . . I also design additional protection.
Example: I have a series string of fast-blow and slow-blow fuses in my power transformer's primary circuit.
One amplifier uses a 1.25A fast-blow in series with a 600mA slow-blow.
Others vary from that, according to the inrush current which is higher than the warmed up Run-Current.
One amplifier would run for 1 to 3 months with a 0.5A slow blow, then it would finally fatigue and open, so I changed the slow blow to 0.6A (600mA),
Still running with that fuse.
When using '5V' rectifier tubes, they do not always exactly meet their rated current draws of 2 Amps, 3 Amps, etc.
Series resistors have to be adjusted accordingly so that when the filament is warm it is 5V.
Always measure and test your designs for proper current, voltage, dissipation, etc.
My experience with power-up arcing of an indirect heated 5AR4 was found to be the 100uF capacitor that immediately followed the 5AR4 cathode.
Do Not do that!
Just my opinions, and my experience.
Worked good with the tubes and parts and circuits I designed.
Your mileage may vary
Last edited:
"1.25A fast-blow in series with a 600mA"
A range from .6 to 1.25A where it will blow if it sees that load for long enough but still handle short surges into that range. And if anything serious happens and hits over 1.25A it will cut the connection fast to have the best chance at saving components.
If you just have the 0.6A slow blow it could take too long to blow under a bad failure.
A range from .6 to 1.25A where it will blow if it sees that load for long enough but still handle short surges into that range. And if anything serious happens and hits over 1.25A it will cut the connection fast to have the best chance at saving components.
If you just have the 0.6A slow blow it could take too long to blow under a bad failure.
You guys would be using UL rated fuses (as they are the default standard parts used in USA), where the difference between slow and fast blow ratings are related to what happens when the fuse experiences 2x the rated current. At the 2x current level, the fast blow must have opened within 5 seconds, and the slow blow must hang on for at least 5 seconds. So at 2.5A, for the example 1.25A fast + 0.6A slow, both fuses should have blown within 5 seconds, but imho there is no guarantee that either fuse would blow soon after 5 secs at 1.25A.
Yes there is a characteristic I-t curve that provides some insight as to what happens at other conditions, but characteristics vary with the manufacturer and model, and actual fusing times are a statistical pin the tail on the donkey around any I-t characteristic you may be looking at.
Yes there is a characteristic I-t curve that provides some insight as to what happens at other conditions, but characteristics vary with the manufacturer and model, and actual fusing times are a statistical pin the tail on the donkey around any I-t characteristic you may be looking at.
1. The difference in actual use of certain fuses is much more complex than just the < 5 second rule and/or the > 5 second rule.
That is one measurement at one condition.
As an example (not the actual result) The inrush current might be 3 Amps for 0.3 second, and still not blow the 1.25A fast blow fuse.
2. I once shut down a production line of a major spectrum analyzer line. My boss was gone that day, so I went to the quality manager.
The problem was a line fuse in the power supply.
It had a very specific rating, but I no longer remember the Amp rating or the fast blow or slow blow rating.
I always inspected the fuse during test and calibration of the power supply. Sometimes they put in an incorrectly rated fuse type; I was doubtful of the 'new and improved' fuse of the correct rating.
The original particular rated fuse was in the form of a zig zag wire like the hills / valleys of a washboard.
The new fuse, with the exact same rating was in the form of an aluminum foil that was cut like a zig zag lightning bolt.
Same rating, but a much different response in actual use. After 2 or 3 cold starts of the power supply,
the new (and less expensively made) fuse opened.
At risk were two things, not meeting government required delivery dates, and meeting the delivery with DOA spectrum analyzers.
Fortunately, they got me the correctly responding 'washboard' fuses within 2 hours; the deliveries were not delayed.
Stick your neck out, shut down the production line, and save the company some money.
4. When I have a new tube amp power supply design, I often measure the inrush current. Not easy to do, you have to power-on and power down a number of times in order to get a measurement of the current when the switch is closed at the crest of the sine wave (I have no phase adjustable turn on tester).
Sometimes, I do not catch the worst case power-on transient.
Example: I might measure a 1.5 amp inrush, and use a 1 Amp fuse that holds because of the short transient inrush. But later, at the power-up with the switch turned on the actual voltage crest, the fuse opens. So I next try a 1.25A fuse, and see if it holds each time. Etc.
Take it in small steps, that way you will not use an overrated fuse.
As I stated before, one amplifier that I had used a 0.5A slow blow, that lasted for 1 to 3 months of power-ons; but finally it would open.
So I changed to a 600mA (0.6A) slow blow, and it has held to this day.
That is one measurement at one condition.
As an example (not the actual result) The inrush current might be 3 Amps for 0.3 second, and still not blow the 1.25A fast blow fuse.
2. I once shut down a production line of a major spectrum analyzer line. My boss was gone that day, so I went to the quality manager.
The problem was a line fuse in the power supply.
It had a very specific rating, but I no longer remember the Amp rating or the fast blow or slow blow rating.
I always inspected the fuse during test and calibration of the power supply. Sometimes they put in an incorrectly rated fuse type; I was doubtful of the 'new and improved' fuse of the correct rating.
The original particular rated fuse was in the form of a zig zag wire like the hills / valleys of a washboard.
The new fuse, with the exact same rating was in the form of an aluminum foil that was cut like a zig zag lightning bolt.
Same rating, but a much different response in actual use. After 2 or 3 cold starts of the power supply,
the new (and less expensively made) fuse opened.
At risk were two things, not meeting government required delivery dates, and meeting the delivery with DOA spectrum analyzers.
Fortunately, they got me the correctly responding 'washboard' fuses within 2 hours; the deliveries were not delayed.
Stick your neck out, shut down the production line, and save the company some money.
4. When I have a new tube amp power supply design, I often measure the inrush current. Not easy to do, you have to power-on and power down a number of times in order to get a measurement of the current when the switch is closed at the crest of the sine wave (I have no phase adjustable turn on tester).
Sometimes, I do not catch the worst case power-on transient.
Example: I might measure a 1.5 amp inrush, and use a 1 Amp fuse that holds because of the short transient inrush. But later, at the power-up with the switch turned on the actual voltage crest, the fuse opens. So I next try a 1.25A fuse, and see if it holds each time. Etc.
Take it in small steps, that way you will not use an overrated fuse.
As I stated before, one amplifier that I had used a 0.5A slow blow, that lasted for 1 to 3 months of power-ons; but finally it would open.
So I changed to a 600mA (0.6A) slow blow, and it has held to this day.
Last edited:
Agreed, that is the deficiency of the UL rated fuses typically sold in USA. The IEC rated fuse typically available in Europe and many other countries is much more defined for surge/in-rush, so not as much of a hit/miss effort, and much easier to design for using tools like PSUD2. Of course it is now relatively easy to buy any type of fuse, and anywhere, so hit and miss can likely be avoided, as can the need for 2 fuses in series.
There is a maximum Inrush current at Cold power-up.
There is a maximum Inrush current at Hot-Start power-up.
There is a relatively constant current for a warmed up operating single ended amplifier.
There is a maximum current for a warmed up push pull amplifier that is operated at the severe clipping level (like a stereo amplifier with the volume control improperly set; or just a guitar amplifier when the guitar strings are practically snapped to get the max volume).
None of the above have the same current versus time integrals.
A fuse generally will blow open at a certain current x time integral, Right?
How much trouble is it to use a slow blow and a fast blow in series?
There are many more complex protection schemes than that, some of them have so many parts that the reliability goes down (proportionate to the number of connections, and the number of components).
Sometimes deficiency is in the head of the designer; careful heads versus careless heads.
Just my opinions.
There is a maximum Inrush current at Hot-Start power-up.
There is a relatively constant current for a warmed up operating single ended amplifier.
There is a maximum current for a warmed up push pull amplifier that is operated at the severe clipping level (like a stereo amplifier with the volume control improperly set; or just a guitar amplifier when the guitar strings are practically snapped to get the max volume).
None of the above have the same current versus time integrals.
A fuse generally will blow open at a certain current x time integral, Right?
How much trouble is it to use a slow blow and a fast blow in series?
There are many more complex protection schemes than that, some of them have so many parts that the reliability goes down (proportionate to the number of connections, and the number of components).
Sometimes deficiency is in the head of the designer; careful heads versus careless heads.
Just my opinions.
Last edited:
That's fine by me. I'm just offering an alternate path that uses one fuse to adequately parse all those operating conditions.
The design process I use is discussed in: https://dalmura.com.au/static/Valve amp fusing.pdf
The design process I use is discussed in: https://dalmura.com.au/static/Valve amp fusing.pdf
There is nothing new under the sun, really. The key issue is faults and overloads are completely different phenomenon. One does not want to protect from a fault in the same manner one wants to protect from an overload. UL listed / FM approved transformers have been doing this (two series fuses) for years.
Fast protection is desired to protect against faults, be it transformer faults or secondary shorts. Coupled with this behavior of a short circuit, the fuse requires suitable interrupting rating, which is the maximum available current from the circuit. This protection is optimized with a current-limiting fuse. These fuses are designed to arc as they blow, which adds impedance to the device and serves to assist the current-limiting behavior. They actually benefit from high current to open quickly. The physical design of the element is such that these fuses are poor at protecting from overloads, and they do not like to be run anywhere near their 'rated' ampacity.
Time-delayed protection is desired to protect against overloads, while still being able to accommodate inrush/startup behaviors. This protection will be fine running near its rated ampacity, and has a melting style element that is responsive to an I^2 characteristic. It is effectively a heat-responding element, with different types of curves available to allow the fuse to coordinate with up and down stream protection. Due to the melting behavior (as opposed to the arcing and expulsion behaviors) it is not optimized by design to provide high amp interrupting and/or current limiting.
Fuses are getting better at trying to provide both, called dual element fuses, but they are not perfect. One still has to compromise. I would not normally select a Class J fuse (very fast, highly protective) for a circuit heavy in motor load that starts across the line. You pick a Class R for that, which is optimized for overloads and heating characteristic.
Semiconductor fuses are designed for extremely fast operation (fault protection) and are poor at handling overloads. Since those circuits can normally limit continuous current through their controllers, they inhibit overloads by nature, and do not require the overload protection via the fuse.
Fast protection is desired to protect against faults, be it transformer faults or secondary shorts. Coupled with this behavior of a short circuit, the fuse requires suitable interrupting rating, which is the maximum available current from the circuit. This protection is optimized with a current-limiting fuse. These fuses are designed to arc as they blow, which adds impedance to the device and serves to assist the current-limiting behavior. They actually benefit from high current to open quickly. The physical design of the element is such that these fuses are poor at protecting from overloads, and they do not like to be run anywhere near their 'rated' ampacity.
Time-delayed protection is desired to protect against overloads, while still being able to accommodate inrush/startup behaviors. This protection will be fine running near its rated ampacity, and has a melting style element that is responsive to an I^2 characteristic. It is effectively a heat-responding element, with different types of curves available to allow the fuse to coordinate with up and down stream protection. Due to the melting behavior (as opposed to the arcing and expulsion behaviors) it is not optimized by design to provide high amp interrupting and/or current limiting.
Fuses are getting better at trying to provide both, called dual element fuses, but they are not perfect. One still has to compromise. I would not normally select a Class J fuse (very fast, highly protective) for a circuit heavy in motor load that starts across the line. You pick a Class R for that, which is optimized for overloads and heating characteristic.
Semiconductor fuses are designed for extremely fast operation (fault protection) and are poor at handling overloads. Since those circuits can normally limit continuous current through their controllers, they inhibit overloads by nature, and do not require the overload protection via the fuse.
Yes, as some of you have noted, and also in the linked Australian article:
( I)squared x time integral
I mis-spoke when I said: I x time integral
( I)squared x time integral
I mis-spoke when I said: I x time integral
I'm grateful for all the discussion and contributions here.
I decided to test 6A3sUMMER's suggestion as quoted in post #1, of obtaining a slow start B+ by running a 5V rectifier on a 6.3V supply with a series resistor. The only DHT rectifiers I have are 5Ц3С so I set one of these up so it would receive 5V after stabilizing - actually about 4.5V, to give it a better chance of being "slow".
The upshot was that it didn't really work for me with these rectifiers. It possibly slowed start-up by maybe a second or so but I think this is not long enough to allow indirectly heated tubes warm up and begin operating.
See the attached video demonstration, below. The four meters (from L to R) show the filament voltage directly out of the PT (6.2V), the voltage across the rectifier filament after the series resistor (4.5V), the B+ voltage (368V) and the filament current (2.9A). I ran my variac at 230V rather than my usual PS testing voltage of 240V for this demo.
Other DHT rectifiers may behave differently?
Simon
I decided to test 6A3sUMMER's suggestion as quoted in post #1, of obtaining a slow start B+ by running a 5V rectifier on a 6.3V supply with a series resistor. The only DHT rectifiers I have are 5Ц3С so I set one of these up so it would receive 5V after stabilizing - actually about 4.5V, to give it a better chance of being "slow".
The upshot was that it didn't really work for me with these rectifiers. It possibly slowed start-up by maybe a second or so but I think this is not long enough to allow indirectly heated tubes warm up and begin operating.
See the attached video demonstration, below. The four meters (from L to R) show the filament voltage directly out of the PT (6.2V), the voltage across the rectifier filament after the series resistor (4.5V), the B+ voltage (368V) and the filament current (2.9A). I ran my variac at 230V rather than my usual PS testing voltage of 240V for this demo.
Other DHT rectifiers may behave differently?
Simon
- Home
- Amplifiers
- Tubes / Valves
- Easy Slow Start with Directly Heated Rectifier?