I built this unit for myself so I put into it the features that I wanted.
So there it is.
Tony
- Operates from a 6.3A transformer (this complicated things)
- 1.6A CCS with a 400mA standby
- The Standby signal is opto-isolated so I can float the whole supply
- Voltage controlled Soft-start for cold heaters filaments
- Scalable architecture
So there it is.
Tony
Hi dtproff,
You have me wondering. Are you running a bunch of heaters in parallel with a constant current source by chance? This raises a number of concerns if that is the case.
I haven't followed the entire schematic (since it's much easier to ask you). I'm going to guess you keep the heaters partially on while in the off mode. Since you do have a soft-start built in, is the partial on state necessary? In these days of high efficiency everything, I'm surprised at the 400 mA delivered in the off state. There isn't much of a plus keeping the heaters warmed up unless you are dealing with mercury vapour rectifiers or thyratrons. Those heaters would be running a little hotter then anyway.
Floating the heaters is a method I use often to reduce heater coupled noise. Are you going negative with that, or positive? Looks impressive, and I am curious now.
-Chris
You have me wondering. Are you running a bunch of heaters in parallel with a constant current source by chance? This raises a number of concerns if that is the case.
I haven't followed the entire schematic (since it's much easier to ask you). I'm going to guess you keep the heaters partially on while in the off mode. Since you do have a soft-start built in, is the partial on state necessary? In these days of high efficiency everything, I'm surprised at the 400 mA delivered in the off state. There isn't much of a plus keeping the heaters warmed up unless you are dealing with mercury vapour rectifiers or thyratrons. Those heaters would be running a little hotter then anyway.
Floating the heaters is a method I use often to reduce heater coupled noise. Are you going negative with that, or positive? Looks impressive, and I am curious now.
-Chris
You do ensure that the HT supply is removed when the heaters are on 'standby', don't you? Otherwise you are setting up the valves for cathode damage.
By the way, what is the point? Most valves work best with a constant voltage supply; that is what they were designed for. However, a set of valves run in parallel from a CCS means that each one individually sees an approximation to a voltage supply so I suppose they will be happy.
By the way, what is the point? Most valves work best with a constant voltage supply; that is what they were designed for. However, a set of valves run in parallel from a CCS means that each one individually sees an approximation to a voltage supply so I suppose they will be happy.
Chris;
No... this is for a single 1.6A heater coil.
If I were doing a CCS <1A I would use a standard TL431/transistor circuit with a diode/cap/resistor for the soft-start function and use a MOSFET to parallel a resistor on the current sense resistor to get the standby I needed. However because it was 1.6A and a TL431 needs 1.24V the I2R losses would have meant that I was dissipating more power in the enclosure than I wanted.
The soft-start function is for the heaters and not the valve. I have an analog compandor that looks at the audio port (rigged as a dB meter) if I am quiet for too long (I forgot to turn the unit off when I wasn't using it) then it flips me into standby mode. By the same token, when it wakes things up (mutes the input till everything stabilizes out), I delay the HT line until the heaters hit temperature and then turn the AC mains on at the zero crossing. I do the same for disconnecting from the mains in that I wait for the zero crossing to disconnect the mains.
DF96
The HT mains is disabled when it goes to standby.
This is strictly about noise reduction, valve life, and power dissipation in the box.
Not totally sure I agree with you that constant voltage is the bests solution. Think of the filament on a standard bulb. It usually pops because of inrush current. The coil is cold and at a low impedance so if you happen to turn on with a 90 degree phase angle on the AC mains then the inrush current can be quite substantial (remember the voltage regulator is not in regulation yet because the power supply is still starting up). Even ignoring the inrush current at various phase angles the 5ms time it takes for the AC mains to reach it's peak voltage doesn't provide a lot of time for the heater coil to get to the correct temperature and resistance. So for a period of time you put more current through than you would probably want.
Overall, does this make a huge difference? Maybe not. As Chris said just because you can doesn't mean you should. I did this because this is what I wanted. I offer my thought process so you can judge for yourself.
Tony
No... this is for a single 1.6A heater coil.
If I were doing a CCS <1A I would use a standard TL431/transistor circuit with a diode/cap/resistor for the soft-start function and use a MOSFET to parallel a resistor on the current sense resistor to get the standby I needed. However because it was 1.6A and a TL431 needs 1.24V the I2R losses would have meant that I was dissipating more power in the enclosure than I wanted.
The soft-start function is for the heaters and not the valve. I have an analog compandor that looks at the audio port (rigged as a dB meter) if I am quiet for too long (I forgot to turn the unit off when I wasn't using it) then it flips me into standby mode. By the same token, when it wakes things up (mutes the input till everything stabilizes out), I delay the HT line until the heaters hit temperature and then turn the AC mains on at the zero crossing. I do the same for disconnecting from the mains in that I wait for the zero crossing to disconnect the mains.
DF96
The HT mains is disabled when it goes to standby.
This is strictly about noise reduction, valve life, and power dissipation in the box.
Not totally sure I agree with you that constant voltage is the bests solution. Think of the filament on a standard bulb. It usually pops because of inrush current. The coil is cold and at a low impedance so if you happen to turn on with a 90 degree phase angle on the AC mains then the inrush current can be quite substantial (remember the voltage regulator is not in regulation yet because the power supply is still starting up). Even ignoring the inrush current at various phase angles the 5ms time it takes for the AC mains to reach it's peak voltage doesn't provide a lot of time for the heater coil to get to the correct temperature and resistance. So for a period of time you put more current through than you would probably want.
Overall, does this make a huge difference? Maybe not. As Chris said just because you can doesn't mean you should. I did this because this is what I wanted. I offer my thought process so you can judge for yourself.
Tony
Yes... As I said, it is an overly complicated circuit. However, it does accomplish exactly what I wanted using the 6.3Vac power transformer.
I would be interested in hearing how others would handle the same task for the same features using a standard 6.3Vac transformer and little power dissipation in the box. I did a switcher version running as a CCS source. I got it working the way I wanted but ultimately went this way since I didn't want a switching PSU in the box.
Tony
I would be interested in hearing how others would handle the same task for the same features using a standard 6.3Vac transformer and little power dissipation in the box. I did a switcher version running as a CCS source. I got it working the way I wanted but ultimately went this way since I didn't want a switching PSU in the box.
Tony
Yes but at 1.6A and a 6V drop would be 9.6W. I would call that pretty exciting actually. 10W power dissipation and another 10W for the heater means 20W total. Yuck!
Tony
A few months ago I tried operating ECC88s with series heaters, dc, higher voltage, better smoothing in the power supply and all that. Did I get a shock! (No, not electrically from the 48V odd). Some lit up like a christmas tree while others were dull. Dashing for the turn-off switch, I went back to data: 300mA acording to some, 365mA (!) according to others. So I measured the heater current for some 8 valves at piously 6.30V on the heater. The currents varied from 290mA to 420mA! - some of the same make. I was forced to shelve my efficient series design and go to a rather hot 6,3V output supply.
I have some 25 x 6L6GCs. I dread doing the same measurements on those - not that I would use series heaters there. But wth is going on here with quality control? Or has it been like that all the time (for me that will be over 50 years)? One never measured heater current; the ubiquitious 6V winding did service and that was that. I am also using a dc heater supply for ECF82 input tubes in a power amplifier, and found the same thing there, though not nearly as severe. Still, again I was forced to stick to a 6,3V (hot) heater supply. (All the above are regulated in a pious attempt to cope with the varying mains supply, load-wise and from town to town, in RSA.)
I would like to know whether others had any experiences in this direction. I cannot imagine this is typical of the heater world; I do not have enough examples of other types to conduct a similar experiment.
Oh, sorry Dtproff - perhaps somewhat of a thread-jack, but I thought since the matter of heater supply is discussed here - and this matter would be important in heater supply design, series operation being more efficient.
I have some 25 x 6L6GCs. I dread doing the same measurements on those - not that I would use series heaters there. But wth is going on here with quality control? Or has it been like that all the time (for me that will be over 50 years)? One never measured heater current; the ubiquitious 6V winding did service and that was that. I am also using a dc heater supply for ECF82 input tubes in a power amplifier, and found the same thing there, though not nearly as severe. Still, again I was forced to stick to a 6,3V (hot) heater supply. (All the above are regulated in a pious attempt to cope with the varying mains supply, load-wise and from town to town, in RSA.)
I would like to know whether others had any experiences in this direction. I cannot imagine this is typical of the heater world; I do not have enough examples of other types to conduct a similar experiment.
Oh, sorry Dtproff - perhaps somewhat of a thread-jack, but I thought since the matter of heater supply is discussed here - and this matter would be important in heater supply design, series operation being more efficient.
Tom;
This was designed for a KT88 so I got what I wanted. You can easily change the scale on the amplifier for 650mA or whatever the requirement is (alternatively change the dividers on the TL431). You no longer care what the filament voltage is, you care about what the required current is. This is now a voltage controlled current source.
John;
Your experience is precisely why I went CCS for the heater supply (as well as noise immunity). This way I don't have to wonder what the heater voltages are supposed to be. I simply select the required current and go. Also... sometimes the mains in the house is a little low and sometimes a little high. That is about a 10% variance also so you are probably looking at 20% total variance for line and load.
For the 6L6 that would be 6.3V/0.9 for 7 ohms hot. Your problem is that the heater wires don't start out hot. It takes them a while to reach thermal equilibrium. The heater coils don't thermally heat at exactly the same rate. until they hit the same temperature they won't have the same resistance. Since the voltage is fixed at 48V some of the resistors had a larger drop across them than the 6.3V, 900mA, 7 ohm hot resistance. For a very short time right at startup they are almost a dead short. Hence the differential heating and the different observed coils temps. Had you been using a CCS supply this would not have been an issue since you controlled the current through the string instead of the voltage. I still recommend a soft-start on startup on the CCS supply to help control the cold-heater coil issue.
Morgan Jones book discusses this in detail and is a really good reference on the subject.
Hope that helped.
Tony
This was designed for a KT88 so I got what I wanted. You can easily change the scale on the amplifier for 650mA or whatever the requirement is (alternatively change the dividers on the TL431). You no longer care what the filament voltage is, you care about what the required current is. This is now a voltage controlled current source.
John;
Your experience is precisely why I went CCS for the heater supply (as well as noise immunity). This way I don't have to wonder what the heater voltages are supposed to be. I simply select the required current and go. Also... sometimes the mains in the house is a little low and sometimes a little high. That is about a 10% variance also so you are probably looking at 20% total variance for line and load.
For the 6L6 that would be 6.3V/0.9 for 7 ohms hot. Your problem is that the heater wires don't start out hot. It takes them a while to reach thermal equilibrium. The heater coils don't thermally heat at exactly the same rate. until they hit the same temperature they won't have the same resistance. Since the voltage is fixed at 48V some of the resistors had a larger drop across them than the 6.3V, 900mA, 7 ohm hot resistance. For a very short time right at startup they are almost a dead short. Hence the differential heating and the different observed coils temps. Had you been using a CCS supply this would not have been an issue since you controlled the current through the string instead of the voltage. I still recommend a soft-start on startup on the CCS supply to help control the cold-heater coil issue.
Morgan Jones book discusses this in detail and is a really good reference on the subject.
Hope that helped.
Tony
ECC88 and E88CC have different heater currents, although both are 6.3V. It is quite common for the special quality version of a valve to have a higher heater power. These days it can be unpredictable what a modern 'ECC88' is actually meant to be. The primary issue is not quality control but naming: an E88CC is not just a reliable version of ECC88 but a slightly different valve. The distinction between 'replacement', 'substitute' and 'equivalent' needs to be emphasised more often.Johan Potgieter said:
6.3V valves which were commonly used in European TV sets (such as EF80, EF183) are OK for 300mA operation too. For others you have to check carefully.
But as said, I did not get equal brightness from the cathodes. Some heaters dropped over 7V across them with this arrangement, which would have shortened their lives. In that sense constant current did not work there for me. Anyway.
DF96 said:
I am aware of this, but none of those in my test were E88CCs. Even if there was a misnaming (the possibility of which I am only too aware of!), one should have had two sets of currents. But in the measurements I did the variation was all over the show between the limits mentioned - 290mA - 420mA - not only 300mA and 365mA.
But off-topic; I just thought others might have had the same experience.
I see... Sorry for the misunderstanding. I don't usually series heaters so I hadn't noticed the issue. Thanks again for the heads up.
Tony
Hi dtproff,
Morgan's books are excellent in scope, plus an easy read. A great source of information.
What I would do in your case is use a bridge rectifier with some series resistance to cut down the peak current spikes. Then I would have used a National "simple switcher" that will regulate step up or step down (it's automatic). The LM2588 flyback regulator. It will provide soft start and other advantages. You can turn the unit off with a logic condition. Nice and easy, plus a high frequency (easier to filter and well above your hearing) in order to avoid hum from a normal supply. It's very efficient too.
-Chris
Morgan's books are excellent in scope, plus an easy read. A great source of information.
What I would do in your case is use a bridge rectifier with some series resistance to cut down the peak current spikes. Then I would have used a National "simple switcher" that will regulate step up or step down (it's automatic). The LM2588 flyback regulator. It will provide soft start and other advantages. You can turn the unit off with a logic condition. Nice and easy, plus a high frequency (easier to filter and well above your hearing) in order to avoid hum from a normal supply. It's very efficient too.
-Chris
Sick minds think alike!
Actually I used a TI part for the switcher. It worked fine. However, the purist in me wanted to know what it would look like as a linear design. I always wonder if the HF noise will mix in somewhere. Usually it won't because I normally choose topologies with high PSRR.
Alternatively I have another 10 Flyback configured as a CC/CV (like a sell phone charger) that has a slow softstart.
Tony
I see what you are getting at. Believe I have thrown that scrap of paper away after having seen that I cannot operate in series.
Perhaps I could re-measure plus quite a few more tubes thisaway and report back here. I have not been concerned with this before either. I must admit that in the past I had no trouble when operating 6,3V/12,6V tubes (ECC-types et al) in series. Thanks for the Mullard-Sylvania information - I did not notice that.
Hi Tony,
The thing with high frequency switching is that you can box it all in - literally. Bring the output through a feed-through capacitor and hang an inductor on each side. You can also do something similar with the sense lead that will correct all voltage drops along the way to the load. Just watch your bode plot and watch for any ringing or instability. The control loop is analog after all.
Ripple. I forgot what the switching frequency is on this one, 260 KHz? So that frequency ripple is well beyond any information and hearing. Filtering it is extremely easy and with a stable frequency you can make a notch filter. Many of these will also lock to another oscillator if you want to hammer the switching frequency down. If you have to run a pair, just run the other 180° out of phase. That should greatly reduce any supply side leak before you even begin filtering.
I don't believe there will be any interaction between the heater supply and the rest of the world, unless the execution stinks. It doesn't matter what brand you use either, all quality chips will work fine here as long as you do your part. I firmly believe this will not be a challenge to you.
The idea of constant current is attractive as an inherent soft-start. The problems arise when you begin looking at all the various similar tubes that might end up in that socket. Never mind production tolerances, often different tubes may be marked as another tube type. In a voltage controlled heater power supply, not much in the way of trouble should result as long as there is reasonable current headroom in the design. This can get dicey when there are many parallel tubes, or in a series string as discussed. Tubes were designed to be operated in parallel except those that are designed for series string use. These will also be labelled as having a "controlled warm up time". Of course we are talking about the 150mA, 300mA and 600mA tube types. Most developed for television use or transformer-less table radios. Don't do this though (operate directly off the line). It's not only dangerous, but it is also not legal to the best of my knowledge. Darwin would approve, and that is not a positive endorsement by any means!
Look at how much sensitive test equipment is designed with switching power supplies. Not a real problem.
-Chris
The thing with high frequency switching is that you can box it all in - literally. Bring the output through a feed-through capacitor and hang an inductor on each side. You can also do something similar with the sense lead that will correct all voltage drops along the way to the load. Just watch your bode plot and watch for any ringing or instability. The control loop is analog after all.
Ripple. I forgot what the switching frequency is on this one, 260 KHz? So that frequency ripple is well beyond any information and hearing. Filtering it is extremely easy and with a stable frequency you can make a notch filter. Many of these will also lock to another oscillator if you want to hammer the switching frequency down. If you have to run a pair, just run the other 180° out of phase. That should greatly reduce any supply side leak before you even begin filtering.
I don't believe there will be any interaction between the heater supply and the rest of the world, unless the execution stinks. It doesn't matter what brand you use either, all quality chips will work fine here as long as you do your part. I firmly believe this will not be a challenge to you.
The idea of constant current is attractive as an inherent soft-start. The problems arise when you begin looking at all the various similar tubes that might end up in that socket. Never mind production tolerances, often different tubes may be marked as another tube type. In a voltage controlled heater power supply, not much in the way of trouble should result as long as there is reasonable current headroom in the design. This can get dicey when there are many parallel tubes, or in a series string as discussed. Tubes were designed to be operated in parallel except those that are designed for series string use. These will also be labelled as having a "controlled warm up time". Of course we are talking about the 150mA, 300mA and 600mA tube types. Most developed for television use or transformer-less table radios. Don't do this though (operate directly off the line). It's not only dangerous, but it is also not legal to the best of my knowledge. Darwin would approve, and that is not a positive endorsement by any means!
Look at how much sensitive test equipment is designed with switching power supplies. Not a real problem.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.