Quote:
Originally Posted by patrick-turner
On Friday, 18 January 2013 11:56:25 UTC+11, John L Stewart wrote:
In this one the 6N7s need to be driven into grid current to get anykind of useable output current. But the stabilization factor is porportional to the pass tube mu (35) & amplifier gain. The gain is reduced a little by the 12B4 CF. Looking back in from the output terminals the source impedance is proportional to the amplifier gain & gm of the 6N7s, so 3x whatever the 6N7s are doing at that load. Notice it can be swiched for either C or L filter input. So at lower voltages, higher currents are possible. Stability depends mostly on the gas tube & output sampling resistors. Altogether a passable PS for ordinary work. Why did I use 6N7s? We had a pile of them & $$$ were short. Overtime I also used some 6A3s & 6B4Gs. Sure would not do that now! Cheers to all, John +-------------------------------------------------------------------+ |Filename: 6N7 Regulated Power Supply (PS) 10W 5 75 40 Contrast.jpg | |Download: http://www.audiobanter.com/attachment.php?attachmentid=325| +-------------------------------------------------------------------+ -- John L Stewart
About 12 months ago I revised a PSU bench top supply for testing KT120 tubes. I found that if you could regulate both EG2 AND Ea at +500Vdc, then 2 x KT120 could produce about 140W AB1, and KT88 about 135W, so all KT120, 90, 88 and 6550can do this although the most unhappy tube is the 6550 which has a lower Eg2 rating. The happiest tube is the KT120 which has the highest Pda rating of 60W, so they cope if RLa-a drifts low for any reason.
If for any reason the Ea and or Eg2 sags under increased DC current then you just can't get the power I just mentioned. So it pays to remember that what data sheets say is possible, may be impossible in the amp made by the average Joe Blogz, that guy who never regulates his tube amp power rails. It isn't necessary if the amp has huge PSU caps and the duty cycle of music waves means average PO 50W when maybe some peaks in the music are just beginning to clip.
The circuit of my regulated +500Vdc has adjustable 50Vdc B+ steps down to 250Vdc by means of a string of zener diodes fed by resistors from the regulated output voltage. The B+ produced by HT supply PT winding and caps and diodes is adjustable by including variations from cap input to choke input plus series R to prevent pass elements overheating because of Ea x Ia product exceeding Pda ratings. Pass elements are two 6AS7 tubes with all 4 triodes paralleled which allows 250mA dc without grid current.
Driver for 6AS7 is 6BX6 gain element with its own floating 300Vdc supply referenced to the cathode so that the anode RL isn't 1M and anode current is not feeble, hence 6BX6 Gm is high and so gain is high so Rout of final product is low, despite Iout change from 20mA to 200mAdc. This floating supply comes from a transformer I had handy from junk box and performs the task of keeping any hum from main PSU from entering 6AS7 grids via the 6BX6 anode resistance which is in a divider with the high pentode Ra.
Anyway, my circuit well out performs the circuit referred to by our dear John Stewart, which follows classical 1955 thinking. Such classical thinking was in fact status quo generic thinking which could be argued was lowest common denominator but my expectations are always higher than what was good enough for 1955. I like testing junk over a wider range of Ea and Ia and with lower noise and lower Rout, so I don't regret the 2 weeks I spent sweating over my PSU which I first made in 1993.
I might add that back then I initially tried to use BU208A transistors for pass elements but after managing to fuse a few despite having quick acting over-current protection, I changed to using tube pass elements. I never figured why the BU208 failed so easily despite working below their quite high and impressive voltage ratings. I suspect I didn't have a diode across the choke to short out back EMF allowing collector voltage to soar for an instant thus causing bjt to fail. John's circuit would benefit from such a thing.. If the load current is cut off sharply, choke voltage tries to soar positively, like the anode voltage of a tube in SET amp with SE OPT - unless you have a nice big fat cap to slow it all down. But, I'd had a +400Vdc regulator with BU208 supplying 2 x 6550 in my general purpose workshop amp for years now and that has not ever failed, and B+ is much better filtered and stabilised than by any huge choke and caps could do. In fact hum at output 0.1mV and the poweramp can be used as preamp if wanted with gain = 15, 20Hz to 65kHz, at 15Vrms output, Rout = 0.5 ohms, THD 0.1%, all with this result aided by the regulated PSU with BU208. The BU108 is also not a bad bjt. But once above 500Vdc, watch out, because then things get iffy.
Patrick Turner.
|
We didn’t have tubes such as the KT120 or KT 90 back in those days. But we did have 807s (lots of them), 813s & 304TH. Many of those found their way into my designs. I also used KT88s & 6550s when it made sense. By using error amplifiers of two or more stages I managed to avoid filter chokes completely.
By the time I joined HP in 1965 most regulated PS were SS. To get around the potential failure problems, pre-regulators using phase controlled SCRs was common. I’ve not built a tube based regulator since then. And for high voltages used in photomultipliers there are some nice hybrid designs.
All my higher powered designs had separate +ve & -ve rails for the error amplifiers. I also did several large supplies where the current needed to be regulated. Voltage & current regulators are not much different. But if you see a 4-quadrant regulator it may seem familiar. Just another name for a power Op-Amp. They can Sink or Source current, something needed with active loads.
Best common tube reference is the 85A2. But for ordinary applications all the others, 0A2, 0B2, Etc work quite well. I found HV Zeners tend to drift as they warm up.
A single pentode error amp will also drift, all caused by the heater temperature changes as the line moves around. But again, OK for ordinary work. A diff amp for the 1st stage is a good cure for that. But adds to the complexity.
One oddity we see is in the example given in the RCA tube manual is their placement of the equalization resistors in the plates of the 6AS7. They should be in the cathodes where they will have (1 + mu) times the effectiveness. Perhaps designed by someone barely out of school.
The transistor failure in the regulated PS Patrick tried was most likely caused by something called, ’Second Breakdown’. I will leave it to the curious to look that up. It is well known that a transistor is a better fuse than a real fuse. I noticed that early on!
My personal regulated PS is something I built more than 55 years ago. It still uses the 25 Hz Hammond iron I had previously used in a PA based on an 815 running Class AB2. It made a lot of music at many parties. I junked the amp & built my regulator. It has a catcher diode so that while running in the choke input mode, the turnoff transient does not cause problems.
Best way to switch a tube regulator on/off is on the AC secondary of the HT transformer. Like always, easier to switch AC than the DC you are making. Much easier on the switch. But never had a switch fail, whether on AC or DC.
These days N-Channel MOS-Fets make a lot of sense. If a lot of power, then a preregulator to keep the lid on things. All somewhat more complicated but a lot more efficient & stable.
Monday morning quarterbacks are everywhere. We did what we could with the devices available to us at the time. For me it was a break to work in the Physics lab at U of T. By interning there I got a PEng out of it. And a lot of common sense too!
Having designed & built something like 40 regulated PS & a lot of other stuff, My thoughts, anyway. John