DC supply for Vacuum tube filaments (1/2)
Vložit
- čas přidán 6. 05. 2022
- #159 In this video I start working on a seemingly simple application - turning an AC voltage into a DC one. However the trick is that, since this is a relatively low voltage application, you need to be really careful with the exact component choice, otherwise, it won't work.
Used datasheets:
www.ti.com/lit/ds/snvs774q/sn...
www.st.com/resource/en/datash...
www.analog.com/media/en/techn...
www.st.com/resource/en/datash...
www.st.com/resource/en/datash...
Special Thanks to all my supporters on Patreon! Especially Ralf B., Paul Pr. Richard, Jonathan Alvarado and Tosta!
If you liked this video be sure to check out my other videos and you can also subscribe to be up to date with all the new ones!
If you want to support the creation of more and better videos please consider checking out: / feszelectronics - Věda a technologie
"The voltage at the transistor will never be higher than 18V" - ominous foreshadowing
thanks, much appreciate your well thought out videos, simulations and practical circuits!
You're a legend. Great video!
Now I know the reason for why my Ltspice simulation is too slow.
You should make a video about snubber circuit design in future.
Usually when a simulation is running to slow, or just does not want to start - there is a conversion error; I try adding the ".options cshunt=1e-15" statement. - this adds 1e-15F (0.01pF) to all circuit nodes and usually helps. The exact value you use can be different of course 1e-13 to 1e-16
Thank you for simulating this, and in the process showing how easy it is to add a soft start to a traditional linear regulator, with just one or two components. I bet this would make LED lighting "look" better if it did not instantly illuminate at 100% (like a strobe light) but instead ramped up in intensity over a second or two. Bravo!
(I also loved the active rectifier explanation too.)
Excellent! This is exactly what I was looking for. Thanks a lot!
Can't wait for part 2!
Indeed....what could possibly go wrong!
your videos are really very detailed an easily understandable, please do a video on Class E power amplifier.
c'mon Fesz... we are all waiting for the second part
Its coming out this week :D
Explanation of this massive set of simulation parameters would be great as well.
So... A filament is a dead short when it is cold and it rises in impedance as it heats up, so your start up current will be much higher than you are seeing here. You can just measure the filament on a tube to see what the cold impedance is to get an idea of what the peak inrush current might be. Another thing to keep in mind when doing this is the power factor and diode losses. This is rarely considered when retrofitting DC filaments into an existing gear but it can be significant. Also in the past I have used LT LDR's with tubes and they are (at least the one I used) very fast meaning the one I was using had no tolerance for poor layout and it oscillated. This is the only time I have ever seen a LDO become unstable (but as I said this was wired with... wires _ I recall it had 5 pins and I needed to add bypass caps to get it to behave). I would suspect that the cap in the feedback network you added might be a problem. I was kind of hoping you would be designing an offline 6.3 volt supply with PFC, but I enjoy all the videos you upload - Thanks!
dead short > much lower resistance than when at working temperature.
impedance > RESISTANCE, agree?
@@paulcohen1555 I would say that the load is almost 100 percent resistive hot or cold. I would think that reactive components are minimal.
You ALWAYS need bypass/decoupling caps right_next_to_3 terminal regulators though.
nice video. teach me how to calculate the voltage drop in the plate resistance
LOL, "What could possibly go wrong" 😲
Remember Murphy's LAWS!
Imagine if it literally explodes the next episode
Unforeseen consequences time! Have your crowbar ready just in case.
Failures on Cathode is common as well.
My guess is that the inductor added creates high voltages and pops the transistors or regulator.
Why don't you use linear current regulator? Two resistors, a small signal transistor and a TO-247 power transistor or MOSFET. Can handle down to 0.8V voltage drop.
What is the power factor of this DC supply and how does that effect the required transformer current rating?
very elegant, of course you could just do a full wave Doubler prior to standard linear regulator. it would require a second filter electrolytic capacitor
Why do you need a voltage doubler ?
@@cat-ie6yp The whole point of the video was that a standard bridge rectifier would be inefficient to meet the Dropout voltage of a standard linear regulator
@@davidkclaytonThe doubler will be impractical, as it will require a very high capacitance. The several filaments together may take amperes of current that will drain almost any capacitor much faster than the next charge will come. Doublers work reasonably well for low-current circuits.
Given the regulator involves a series transistor, I do find myself wondering if one could make an all-transistor-based active bridge rectifier that, rather than driving the transistors to saturation, uses a feedback circuit to also regulate the output voltage in some way, thereby reducing the total voltage drop... eh, or perhaps I'm just re-inventing the switching power supply?
I think the closes thing to that, currently in use, would be the active PFC circuit; since the big issue with a rectifying bridge+ the rest of the circuit is still the poor power factor. Anyway; I guess you could adapt an active diode bridge to also work as a voltage regulator - not just to compare the input to output but also to a built in reference.
or just use an active rectifier, and use mosfet isolation transformers as a convenient way of controlling them like active elements.
@@manitoba-op4jx Sounds extra extra complicated, but a sound point.
I've seen the most common solution, no regulator and a 1-2 ohm power resistor in between two filter caps, depending on your filament load
Yes, I have use this many times because of low headroom for front end of guitar amps with 2-3 12a_7 tubes I usually end up with about half an ohm split between top and bottom (1 Ohm total) with the bottom being reverenced to either ground or some small positive voltage. It is fun to see the arc of videos leading up to this one though!
R1 - C6 isn't that a Zobel network - just what you would use on Low frequency amplifiers exit ?
What could possible go wrong? In my experience, quite a lot!!!
I'd rather ditch the 6.3V heater primary constraint (at least when designing a new amp from scratch rather than modifying an existing one or using old iron as a base for a new amp) and make a classic voltage regulator for noise-sensitive tubes, i.e. mic, RIAA or guitar preamp stages.
Tubes.
I also share my opinion, from decades (25) years of experience, I have made tube amplifiers specifically. is as follows, filaments with AC current sound better, more realistic, more analog, and looser than filaments with DC circuits.
main voltage has +-10% variation, you will not get 6,3 DC when the main is at its -10%
That is true.. I guess it would be possible to set the DC supply voltage from the beginning to be ~5.7V (6.3*0.9) to account for this case; or any value that can be obtained under any conditions between 5.7 and 6.3
Why not use 4 mosfets in the bridge?
Also your startup ramp will hit the regulator with huge power dissipation for too long, ..It will likely melt! :(
Looking forward to you next video :)
The 4 mosfet bridge would have needed some more complex circuits to work correctly, and most regulators do have some built in over temperature and over current feature - its highly unlikely they will self destruct during the startup, but that remains to be seen :D
I am a pretty old fart, but I recall from the 'old days' that running tube filaments on DC will shorten their life compared to AC
Has something to due with filament material ?
Differential thermal emission.
If one side of the filament is grounded then the other is 6.3V or whatever negative and electron current (although low) will flow from one side of the filament to the other. Also some ion migration kinda like electroplating.
If you let the filament float that won't save you either because you'll still have one side of the filament -6.3V relative to the other. This also messes with grid voltage and causes uneven heating of the tube because the edge of the grid closest to the most positive part of the heater will fry. Grids are usually rated for dumping, say, 50 mA into the entire grid, but only 25 mA into just one little edge will melt it and short it and "how could running DC filament cause my amp tube to arc over when running way under ratings?"
There's a classic technical paper from 70 years ago from some guy with a B in his name where you take a tube with like 5 heater wires in series and due to drop they're all at different voltages relative to the grid causing some kind of massive linearity problem, you could tell in the linearity graph which PA tube was powered by AC and which powered at DC. Perfect conductors don't exist in the real world and if you have the gear to measure enough dB of SNR it'll show up, even if it may not matter in practical applications.
If you think running DC will prevent hum, surely running ultrasonic AC would prevent hum even betterer. So yeah, decades before 2000's era switching power supplies there were guys running tube heaters at 20 KHz. You'd never do this in production, but typical ham radio stuff you go to a swapfest and come home with a 10 watt audio xformer and ...
As usual with engineering, just because you can prove to two decimal places that running DC will turn a 10K hour life tube into a 9.9K hour life tube, that doesn't mean anyone outside the engineering dept will care. Also most people don't run right up against the limits of grid or plate dissipation nor do they run 1 dB below linearity limits, so "most of the time its OK enough to run DC" Another typical engineering thing is you could buy DC rated heaters which just had larger heavier more expensive wires, but if 99% of non-automotive tubes ran on AC they're going to ship with the cheaper thinner wire.
(edited to add, I forgot about thoriated filaments, thorium is expensive so if they thoriate it they'll apply it evenly which is fine for AC heating but a DC heated filament will emit more from one side or the other so it's thorium layer will "wear out" faster than the other side of the filament. Again, typical engineering stuff where just because you can prove the tube will only meet specs for 9.9K hours after you pay for 10K hours doesn't mean anyone will care, but its a real measurable effect...)
@@VinceMulhollon Ah, excelent point! I pretty much wasnt thinkng of directly heated triodes when watching the video. Has anyone thought of flipping the dc polarity occasionally to "Even out" the wear on the filament? I wonder if that helps?