Planning to build benchtop 500V tube PSU, help needed.

[001]

  and a lot of the parts in that power supply are slightly odd choices and look suspiciously like they were originally specced for other equipment. 

never heard of special " benchtop 500V tube PSU" parts 

[samofab]

There are tubes that are specifically marketed for use in series regulator circuit. e.g. 8068

[David Hess]

There are tubes that are specifically marketed for use in series regulator circuit. e.g. 8068

What makes them specifically suitable for series pass elements?

I seem to recall that these tubes had unusually low anode resistance and would operate at low anode voltages but I do not remember how this was achieved.

[samofab]

What makes them specifically suitable for series pass elements?

I seem to recall that these tubes had unusually low anode resistance and would operate at low anode voltages but I do not remember how this was achieved.

I don't really know and the datasheet is quite sparse (at least for me). I always assumed that intended use is just a marketing term . Perhaps this particular tube could be used in audio amplifier just as well (e.g. for electrostatic speakers).

[Circlotron]

Such a tube might have an unusually low voltage drop, just the ticket for series pass regulator duty, but the construction techniques used to get this low voltage drop might also make the grid volts vs anode current quite nonlinear and therefore not suitable as a normal grounded cathode amplifier. Similar to bipolar transistors - some are optimised for linearity, some are optimised for switching usage.

[duak]

Probably hard to find these days are television horizontal sweep tubes.  These are tubes optimized for switching applications - in a way like power FETs and IGBTs are optimized for low RDS,  high breakdown voltages and minimal drive and switching losses.  Radio amateurs use them in power applications where non-linearity isn't such a big deal.  There are a few articles on the tubes here:  http://www.tetrode.co.uk/sweep.html  I think one of them says that to get low plate resistance, the cathode is larger than one would expect. Scanning the vast wasteland of the 'net, I see that tube audio people that have tried them, especially the single ended Class A open loop folk, don't seem to care for them.  I wonder if the thrash guitarists would?

I would think that as long as the transfer function is monotonic, a good power supply could be designed using a sweep tube for the pass device.

[T3sl4co1l]

Nonlinearity isn't such a big deal, it's mostly the 3/2 power law, typical of any tube.  Frame grid and planar types tend to be worse actually, because of cathode space-charge physics.

The relevant properties are: very high perveance (large cathode, small grid cross-section), close anode spacing (low Rp), and relatively high anode dissipation.  This gives low mu, or relatively low gm, but as the cathode-grid voltage drop is usually used for the anode resistor of the error amp, it really only affects the headroom of the error amp.

Which means if you are okay with a 250-500V supply, you probably won't need a negative supply, but if you need 100-500V say, you'll likely need a -50 to -100V supply, and then you might as well go slightly further and make it a -250V supply and go for 0-500V output range instead.  (And, mind that now you're talking a >800V range on the error amp, and you may want to use a small power tube just to handle that; like, say, 6S4 but if it had higher mu?)

Sweeps are the same way, but with the plate replaced by the screen of course (in terms of grid-anode / grid-screen spacing being similar), the grids aligned (to reduce screen current), and somewhat tighter grids to give higher Gm and mu_g2g1.  Here the peak current is as important as the high perveance.  Some other tweaks to beam tetrode structure and you get a useful saturation voltage under 100V.

Tim

[David Hess]

Tektronix used tubes like the 6080 (sometimes with both triodes directly in parallel) and the 12B4 (sometimes lots in parallel).

The 6080 was apparently intended for series regulators and the 12B4 was often used for them.

[H713]

I would use tubes that are cheap and available long-term. Look at plate dissipation ratings. The 6L6GC might be a good choice- lots of good power supplies have been built around them, and there's a Russian equivalent that is very cheap.

There are advantages of using tetrodes and pentodes instead of triodes for the pass elements, most notably what Pete Millett describes as a "simple, brain-damaged half-assed current limit" that can be implemented by limiting the screen voltage. The other big advantage is that big triodes are very costly in comparison. The 6AS7 was a very common pass tube back in the day, but both new production and NOS tubes are very expensive. Demand has brought down the price of new-production tubes used in guitar amps, and as such 6L6GCs in particular are pretty economical and have a 30W plate dissipation rating. BTW, for those unaware, there are multiple different tubes called the 6L6, and not all of them have the same ratings. Throw an original 6L6 into some circuits designed for the 6L6GC, 5881 or 7581 (all closely related to the original 6L6 in terms of characteristics) and there's a good chance of it red-plating or melting the screen grid (UL operation is very hard on the screen grid- the early tubes weren't designed to take that kind of abuse). Plate dissipation ratings are also lower on the old tubes.

  and a lot of the parts in that power supply are slightly odd choices and look suspiciously like they were originally specced for other equipment. 

never heard of special " benchtop 500V tube PSU" parts 

Of course there are no special parts for a bench 500V power supply, but there are parts that one would expect to use and parts that aren't. There are a lot of parts in that particular power supply that I probably would not have chosen if I were designing a bench power supply, but suspiciously turn up in other pieces of Heathkit gear. In other words, they were originally chosen and purchased for other kits.

[T3sl4co1l]

Also 6BG6 (sweep version), 807 (coke bottle, RF), 1614 (metal, RF), etc.

Mind that new "6L6s" are all over the place, at least in the cheaper brands.  Everyone makes 6L6GCs, because of course they do, but they don't all handle 30W like a 6L6GC is supposed to.  Derate generously (say 30-50% below rated), treat them more like 6L6GBs if that.

Even then, the material quality is poor, so don't expect great lifetime.

That's a nice advantage to industrial and PQ types: better materials, higher ratings, longer life.  A $100 tube that lasts 10khr is a better deal than a $50 tube that lasts less than 5k -- assuming you have the budget to do so, and will be using the equipment frequently.

Or conversely, if not, it may actually be a better deal to use the cheaper part, and then maybe keep spares handy, accepting that when (not if) the originals fail, you'll have the annoyance of changing them out, but you won't be SOL for a week awaiting delivery of new parts.

One last thing.  Mind that poorly constructed tubes are more likely to melt or arc over and fail shorted.  Have some protection method in place (a crowbar?) to deal with this.  It's most likely to happen at the lowest output voltages, where the load will be most vulnerable to a surge to full B+...

Tim

[H713]

Also 6BG6 (sweep version), 807 (coke bottle, RF), 1614 (metal, RF), etc.

Mind that new "6L6s" are all over the place, at least in the cheaper brands.  Everyone makes 6L6GCs, because of course they do, but they don't all handle 30W like a 6L6GC is supposed to.  Derate generously (say 30-50% below rated), treat them more like 6L6GBs if that.

Even then, the material quality is poor, so don't expect great lifetime.

That's a nice advantage to industrial and PQ types: better materials, higher ratings, longer life.  A $100 tube that lasts 10khr is a better deal than a $50 tube that lasts less than 5k -- assuming you have the budget to do so, and will be using the equipment frequently.

Or conversely, if not, it may actually be a better deal to use the cheaper part, and then maybe keep spares handy, accepting that when (not if) the originals fail, you'll have the annoyance of changing them out, but you won't be SOL for a week awaiting delivery of new parts.

One last thing.  Mind that poorly constructed tubes are more likely to melt or arc over and fail shorted.  Have some protection method in place (a crowbar?) to deal with this.  It's most likely to happen at the lowest output voltages, where the load will be most vulnerable to a surge to full B+...

Tim

Just my personal experience, but I've found that the new production 6L6GCs are pretty good. They've certainly gotten better over the last few years. Buy sets that have been burned in, you should be fine. I've pushed some of them pretty hard, and they take it. I'm running Tung Sol (new production) 6L6GC-STR tubes in one of my power amps, and they've been rock solid for over a year (figure five hours a day?), running at about 28W plate dissipation.

Keep in mind what most of these are being used for- Guitar amps. Lots of them push tubes really hard, and in a lot of cases they are sitting just a few inches away from one or two twelve inch speakers that are producing earth-shattering SPLs. And remember- a lot of these amps are being blasted for quite a few hours at a time, and they aren't being babied. Your bench power supply will be sitting on a bench, away from speakers, and you probably won't be running them any harder.

As a contrast, when I built my latest power amp using 1625 tubes (12V heater version of the 807, itself a version of the 6L6), I had two NOS RCA tubes and one NOS National Union tube that shorted after a total of about eight hours of usage running at about 19W plate dissipation.

In addition, a lot of the alternative tubes that could be used have dwindling supplies that I would prefer to see used to keep existing equipment operating. Aside from my personal feelings on this subject, there is nothing more frustrating than having a piece of gear that uses an unobtainium part with a finite lifespan. Just ask owners of U47s and U48s how they feel about the supply of VF14Ms.

Also, I would definitely suggest implementing something to minimize how much power is dissipated in the series pass elements. This could be as simple as a carefully chosen dropping resistors, or as elaborate as a variac controlled by a servo motor. I can attest that the later does work quite well. If you use the Antek toroid I suggested, you have a pair of HV windings. Switching transformer taps is a pretty easy way to handle this issue. That combined with a simple crowbar would go a long way to improving the safety of this power supply, though one must always remember that anything that can source 500V at this level of current can easily kill you.

[001]

Any news?

[geggi1]

If you want at rough and simple 500DCV PSU without any features the resepy is simpel.
- One transformer with secondary voltage of 350V.
- Four diodes capable of at least 500V. Recomend 1N4007 or better.
- Electrolytic capacitors with a rating of at least 500V (go for at least 600V capacitors) with plenty of capacitance

Optional
- One variac to get ajustable voltage.
- Toroide bifilar winded core for noise reduction.

Rectify the 350VAC with a four diode rectifier bridge. Smooth the voltage with plenty of electrolytic capacitors. Add a bifilar wounded inductor to reduce the noise. The variac can be placed in front of the 350V transformer to have some voltage regulagtion.
If avalable get a 230/380 transformer this will be avalable at companies like RS, but you will have to use a variac to ajust the voltage dowin to 500v

[001]

If you want at rough and simple 500DCV PSU without any features the resepy is simpel.
- One transformer with secondary voltage of 350V.
- Four diodes capable of at least 500V. Recomend 1N4007 or better.
- Electrolytic capacitors with a rating of at least 500V (go for at least 600V capacitors) with plenty of capacitance

Optional
- One variac to get ajustable voltage.
- Toroide bifilar winded core for noise reduction.

Rectify the 350VAC with a four diode rectifier bridge. Smooth the voltage with plenty of electrolytic capacitors. Add a bifilar wounded inductor to reduce the noise. The variac can be placed in front of the 350V transformer to have some voltage regulagtion.
If avalable get a 230/380 transformer this will be avalable at companies like RS, but you will have to use a variac to ajust the voltage dowin to 500v

Honey. It is EE forum.    Everybody can build simple Herz bridge here
The story of HV  benchtop tube PSU is always WERY tricky. So any benchtop  PSU requires constant voltage and constant current modes. And big caps at lab PSU output is BAD idea 

[TERRA Operative]

I still got all the parts here for the unit as it is in the first post. I'll have to filter through the thread and see what mods I want to use for my final design...

I haven't done any actual work on it yet, I keep getting distracted with the next shiny thing, but once my current project (A Fluke 332D restoration) is done, I'll get back onto this one for sure.

[001]

I haven't done any actual work on it yet, I keep getting distracted with the next shiny thing, but once my current project (A Fluke 332D restoration) is done, I'll get back onto this one for sure.

Best wishes
Let`s us to know how