Greetings, friends. I put a new-to-me rectifier tube in my amp, and it does this:
Tube is a Soviet 5Ц3C, which I believe to be their version of the venerable 5U4G. Amp is fairly low voltage, tube sees maybe 350v on each anode and makes around 130mA. It does this crazy dance at startup (5 times so far) and then behaves normally.
So what gives?
Tube is a Soviet 5Ц3C, which I believe to be their version of the venerable 5U4G. Amp is fairly low voltage, tube sees maybe 350v on each anode and makes around 130mA. It does this crazy dance at startup (5 times so far) and then behaves normally.
So what gives?
POS rectifier tube. Back before 1N4007’s we’re generally available rectifier tubes didn’t usually do that.
With decent rectifier tubes, 1N4007 or any other rectifier diode is not required. It's a bad tube, or maybe the input filter capacitor value is too high.
i use 10ufd/630vdc polypropylene caps right at the rectifier cathodes, then a choke and a bigger ecaps,
you may also put a 47ohm 3 watt resistor between the cathode and the first cap...
this resistor softens up the start up and makes your rectifier live longer...
you are one lucky dude, your rectifier tube did not fail on you..
you may also put a 47ohm 3 watt resistor between the cathode and the first cap...
this resistor softens up the start up and makes your rectifier live longer...
you are one lucky dude, your rectifier tube did not fail on you..
Last edited:
all the more with new production rectifiers like the JJ5ar4, one really needs to do mitigating measures...
How about in parallel with each, but with a padding resistor that's say, double the resistance of each section when hot?
Sort of rectifier tube on crutches, keeping the soft aspect of it intact. Get that bulk cap up some during heating, so it's not coming on into 0V -
Of course I dont know off hand what that resistance is, for the various models. I believe 5AR4 is lower than 5U4...
The voltage drop is lower. The parallel rectifier would do almost all the work, if not all given the voltage drops.
I use New Sensor tubes, and I have zero problems. Just use a good tube in old gear and you'll be fine unless someone increases the filter cap capacitance.
I use New Sensor tubes, and I have zero problems. Just use a good tube in old gear and you'll be fine unless someone increases the filter cap capacitance.
The ‘modern’ way is the 1N4007 does the actual reverse blocking (because you’re more likely to get rectifier tube that’s NFG today than you were 50 years ago), and the tube just provides the “proper” forward characteristic and delayed turn on. It’s a band aid for bad tubes.
Some New Sensor rectifiers do fine. Buy one the next week and it might arc. NOS thats been sitting in a warehouse 50 years getting gassy may or may not be any better. It’s a crap shoot. You either keep plugging them in till one works or use the diodes and live with whatever tube you get.
Some New Sensor rectifiers do fine. Buy one the next week and it might arc. NOS thats been sitting in a warehouse 50 years getting gassy may or may not be any better. It’s a crap shoot. You either keep plugging them in till one works or use the diodes and live with whatever tube you get.
I think the TV repairman next door tried to explain that to me as a kid; "The tube wont fire at that voltage". Looks like I still dont get it. So, what voltage does it take to get some electrons emitted and across the evacuated space?
Oh, looks like it's something...
Just not very many of those electrons. You will always get a few - just because the cathode is hot, even if there is no field. Yes, thermal noise, and it will tend to “rectify” a bit. But I’ll bet your current meter can’t measure it. A noise figure meter would….
Hi wg_ski,
I have yet to have problems with New Sensor tubes. Same for good NOS, but people want way too much $$ for those. You will always get the odd bad tube, but when I was young selling New GE and Westinghouse, we had failures, and more of them. It's the nature of tubes. Having said that, the less good brands had much higher issues, and the cheap ones ... well. Hazards in a box.
Hi jjasniew,
Typical drop for a 5AR4 depending on current was lower than a 5U4, and a mercury rectifier was around 15 VDC (much lower than normal rectifier tubes). Forward drops for a 1N4007 ranges from 0.6 VDC to 1.2 VDC at higher current than you're talking about here. See the issue?
I have yet to have problems with New Sensor tubes. Same for good NOS, but people want way too much $$ for those. You will always get the odd bad tube, but when I was young selling New GE and Westinghouse, we had failures, and more of them. It's the nature of tubes. Having said that, the less good brands had much higher issues, and the cheap ones ... well. Hazards in a box.
Hi jjasniew,
Typical drop for a 5AR4 depending on current was lower than a 5U4, and a mercury rectifier was around 15 VDC (much lower than normal rectifier tubes). Forward drops for a 1N4007 ranges from 0.6 VDC to 1.2 VDC at higher current than you're talking about here. See the issue?
That was the Port Arthur config by Gerald Weber from 2005 : https://www.dalmura.com.au/static/Port Arthur Rectifier.pdf
So, the tube spec gives this "voltage regulation" value of 15V, for a current step of 175mA (175 -> 350); I assume it forward drops 15V with that load change. That implies a "model" of an ideal diode with a ~85 Ohm resistor in series; which I assume would do about the same thing in the same circuit, wrt that current step.
If you put the diode I suggested with, say, 150 Ohm in series with it in parallel with the tube, how would that conduct all the current around the tube part?
What it might do is - when the tube is cold and power is switched on - pull up the filter capacitor voltage some, so by the time the tube cathode starts conducting, that capacitor is at least some way above 0V, before the tube kicks in - with it's lower impedance connection than the diode arrangement.
Granted, 85//150 giving a drop from 85 to ~55 Ohms of "tube rectifier softness" might change the sound of the amp. I'd be surprised if that detriment beats being able to use more tube rectifier units successfully, without arcing inside.
If you put the diode I suggested with, say, 150 Ohm in series with it in parallel with the tube, how would that conduct all the current around the tube part?
What it might do is - when the tube is cold and power is switched on - pull up the filter capacitor voltage some, so by the time the tube cathode starts conducting, that capacitor is at least some way above 0V, before the tube kicks in - with it's lower impedance connection than the diode arrangement.
Granted, 85//150 giving a drop from 85 to ~55 Ohms of "tube rectifier softness" might change the sound of the amp. I'd be surprised if that detriment beats being able to use more tube rectifier units successfully, without arcing inside.
Hi jjasniew,
The biggest problems I see in service are new products where the input capacitance is way too high - that destroys even the best rectifier tubes. The other is in "upgraded" rebuilds where the idiot "tech" has increased the capacitor sizes with the same results.
I'll be honest, I have fewer tube problems today using currently manufactured tubes than I did when I started in the 1970's when we were working off NOS stock. Russian tube manufacturing had improved steadily and the tubes are also far more consistent today. That's if you buy good tubes. I absolutely do not see the point in getting fancy with silicon rectifiers, just do it right and use good parts for pete's sake!
If you're going to use just silicon rectifiers - GREAT! That's called progression. For my new designs I do not use tube rectifiers and haven't for a very long time. Decades. But the rectifier section is not a straight diode deal. This does not change the sound in audio reproduction equipment, it may in a guitar head, but they went to solid state rectifiers in the 1960's.
Anyway, get off the fence. Either use good tubes in a properly designed power supply, or design with solid state diodes. Don't stick solid state diodes in with tubes in an existing design and call it better unless you're going to completely replace the tube and compensate (adjust) the existing circuit. You're going to save 10 to 15 watts of power just by eliminating the rectifier heater (filament) from the equation. The power transformer will thank you!
The biggest problems I see in service are new products where the input capacitance is way too high - that destroys even the best rectifier tubes. The other is in "upgraded" rebuilds where the idiot "tech" has increased the capacitor sizes with the same results.
I'll be honest, I have fewer tube problems today using currently manufactured tubes than I did when I started in the 1970's when we were working off NOS stock. Russian tube manufacturing had improved steadily and the tubes are also far more consistent today. That's if you buy good tubes. I absolutely do not see the point in getting fancy with silicon rectifiers, just do it right and use good parts for pete's sake!
If you're going to use just silicon rectifiers - GREAT! That's called progression. For my new designs I do not use tube rectifiers and haven't for a very long time. Decades. But the rectifier section is not a straight diode deal. This does not change the sound in audio reproduction equipment, it may in a guitar head, but they went to solid state rectifiers in the 1960's.
Anyway, get off the fence. Either use good tubes in a properly designed power supply, or design with solid state diodes. Don't stick solid state diodes in with tubes in an existing design and call it better unless you're going to completely replace the tube and compensate (adjust) the existing circuit. You're going to save 10 to 15 watts of power just by eliminating the rectifier heater (filament) from the equation. The power transformer will thank you!
The only reason to use diodes with the tube rectifier is if an existing piece of equipment calls for it - and gives fireworks with New Sensor replacements. Been there bought the T-shirt. Were the caps too big to start with? They were the ones that came with it - same values at least. New axials under the chassis. I don’t design with tube rectifiers, and if I did I’d be using damper diodes. People seem to be very attached to their vacuum rectifiers. And filter chokes. I tend to use silicon to cure hum, and little capacitors (on the rectifiers) to cure the ”bizzzzzzzzzz”.
The rectifier tube lives a hard life. It must power all of the other tubes in the amp, or whatever device it lives in, AND fill those hungry filter caps while it is waking up from sleep. If the cathode is not uniformly coated there will be hot spots in the thinner areas which heat quickest. All of this startup current flows in a small area exceeding the current density of the cathode causing an arc. The same occurs if the cathode is not concentric with the plate.
I have seen 5U4s and 5Y3s that spark on startup then work fine as the cathode / filament gets hot. They seem to live long after a few sparks. Every 5AR4 that I saw spark became useless after that.
The Tubelab SSE board runs the rectifier pretty hard, near but not at or above the ratings with the DCR found in a correctly sized power transformer. All worked well for the first few years of the SSE board. Somewhere around early 2010 batches of new production tubes from New Sensor, JJ and Shuguang started to exhibit spark out on start up. Sometimes it took two or three tubes before finding one that didn't spark, and those lived long happy lives. I added the diodes to all SSE boards and the problem vanished.
I have seen 5U4s and 5Y3s that spark on startup then work fine as the cathode / filament gets hot. They seem to live long after a few sparks. Every 5AR4 that I saw spark became useless after that.
The Tubelab SSE board runs the rectifier pretty hard, near but not at or above the ratings with the DCR found in a correctly sized power transformer. All worked well for the first few years of the SSE board. Somewhere around early 2010 batches of new production tubes from New Sensor, JJ and Shuguang started to exhibit spark out on start up. Sometimes it took two or three tubes before finding one that didn't spark, and those lived long happy lives. I added the diodes to all SSE boards and the problem vanished.