|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.223
In Reply to: RE: So... posted by Nickel Core on October 26, 2014 at 07:04:30
"Remove C4, C5 and D6?"
I should probably back up and ask whether you're using self bias or a negative bias supply. I assumed the former, and that the schematic you posted with the regulator grounded was just "boilerplate," but maybe that's incorrect.
In either case, it's not clear to me how the PS output caps will alleviate the issue of the DC (and therefore AC) gradient across the filament. The impedance of the filament in this mode is a little over four ohms (E/I). At 20 Hz, a 100uF cap exhibits about 80 ohms. That's not nearly low enough to smooth the AC induced across the filament. The caps are more effective at 1K and up, but perhaps modulation of the regulator by the low frequency energy creates intermod or other audible effects at those higher frequencies.
If you really want to use DC for this, I think the "voltage controlled current source" Alex recommended is the only option. I'm not familiar with the Coleman regulator, but it may also be appropriate.
Of course, this all changes if you're elevating the entire filament above ground with a bias resistor.
--------------------------
Buy Chinese. Bury freedom.
Follow Ups:
I use fixed bias.
I don't "really" want to use DC, but AC for the 10Y drivers of my amp doesn't look like an option from a hum point of view.
I will like in the Ronan Regulator or HF options again.
Thanks.
NC
If you are using a 10Y as driver, standard AC is out of the question because of hum, which is going to be further amplified by the output stage.
At 7.5V and 1.25A this tubes is rather easy to do. Two tubes, one oer channel, will consume about 18.75W, thus you can use a 50W electronic transformer. After modifying the input diode bridge (added cap across and added NTC in series with the input to protect the input bridge from current surge - you should get slightly above 16.5V output. Now you can unsolder the secondary and split it approximately in half: assuming 9 turns secondary, you should get 1.83V per turn. Now you can use the same wire and turns as 4 for one tube and 4 for the other. The result should be 7.33V per tube, which I guess is just perfect. Carefully extend the secondary wires for each split, and fix the output transformer with some plastic string or similar. You are ready to go, it's as simple as that.
No hum, I guess you are not going to hear anything but music if the cap is approximately 300uF. Dead silent with no signal. As for the sound, cleaner than normal AC and more lively and detailed and fast than the best DC: also less noise/buzz/hum than DC.
******
http://rh-amps.blogspot.com/
Just to expand on what Alex is describing... You will need to be sure the electronic transformer (power oscillator) has a toroidal transformer at its output. The modules I've purchased in the past used miniature conventional transformers, much more difficult to modify (I wouldn't try). Searching eBay recently, I did find a few sellers that post pics of the inside of their modules, and several show a toroidal design in that position. Also, just to be clear, Alex is telling you to add a CT to the output transformer in order to power the filament. I assume the CT will be grounded, just as in the case of a 60Hz filament transformer.
Alex, I'm pretty sure you said recently that you've found some of the electronic transformers that are already designed to oscillate at 50 or 60 kHz. All the modules I've purchased ran at roughly 20 kHz, so I had to modify them. Can you point us to a specific seller for this? In my case at least, it would need to have a 120V primary, not 220V.
--------------------------
Buy Chinese. Bury freedom.
TK, I have to correct one thing you got wrong because my explanation was not good enough: there is no CT but 2 separate 4 turn secondaries. The idea is to unsolder the output transformer from the board and split the secondary (assuming it has 9 turns) in half.
By splitting the 9 turns secondary you are going to "loose" one turn to the split, thus 2 halves with 4 turns each. Now extend those secondaries by soldering and isolating the joining points, so that you have longer wire to extend out of the original box: you should use stranded wire, not thick solid. Now solder the primary back to the board, leaving the secondaries "free" and protruding out of the box. You should fix somehow the transformer to the board, to make it a fixed assembly - I use plastic strips.
Thus each of the two secondary windings carrying approximately 7.5V HF AC is left floating and should be connected to the same pins where you would connect a normal AC secondary: a humdinger arrangement of fixed matched resistors is advisable, and bypass to ground is from the midpoint only: do not bypass both ends of the filament like you should do in a standard AC arrangement because to HF AC the cap is a dead short to ground!
The situation nowadays has improved as even the cheapest electronic transformers operate at 30kHz (my DMM cannot measure above 20kHz and they are sometimes declared as 30kHz, which is good enough for me).
Higher quality electronic transformers (branded) operate at 50-60kHz. Philips is one example. Most Philips and similar electronic transformers incorporate additional protection devices - no-load, shorted wires, overload) which are not vital but come in handy.
As for the output transformer, you need to check before ordering (look for pictures and even try to count the number of secondary turns)! There is little customization possible with conventional ferrite output transformer types, and lots of possibility with toroid cores.
Last but not least, check the specs for minimum load. And, obviously you must buy for your mains: 110/120 or 230/240V.
******
http://rh-amps.blogspot.com/
Some years ago on this forum, Kurt Strain proposed the use of a transformer designed for fluorescent lighting that operates at very high frequencies, I think even higher than 60kHz, to run AC filaments. Am I remembering correctly? Does this make sense?
Of course, the HF AC heaters supply can be used as well with fixed bias: only one side of the filament will be connected to ground (grounded), or even better in my view, the midpoint of the humdinger resistors arrangement will be connected to ground (instead of being connected to whichever cathode resistor or CCS might be used - which in turn is grounded on the other end).
******
http://rh-amps.blogspot.com/
I'm still not sure I understand the need to split the secondary. My 211s use fixed bias, and I simply apply the single secondary across the hum pot circuit. The center of that circuit is grounded, but as you say, neither end is bypassed. Thanks for the info on the newer supplies. I'll buy a few and see how they compare.
--------------------------
Buy Chinese. Bury freedom.
"I'm still not sure I understand the need to split the secondary."
The original secondary is probably 9 turns, for an output of 12V.
After installing the cap across the rectifiers bridge, the output is going to rise probably to 16.68V (this depends on various factors, but can be assumed as a rule of thumb to simplify the calculation).
Now 16.68V are too much for 10Y, aren't they? Instead of burning voltage across resistors (more unnecessary heat), and adding a second unit for the other tube, we can do a simple calculation: 7.5V + 7.5V = 15V.
This means that we already have the voltage and the space for two windings, the only thing we need to do is split the original winding into two (now separate) windings. While doing that, we are going to loose 1 turn, which turns out to be beneficial, since 16.68V/9 turns = 1.853V per turn.
This leaves us with 2 separate 4 turns windings, each of them will yield 4x 1.853V = 7.413V (which is an almost perfect approximation of 7.5V).
Not only have we avoided to burn voltage across resistors, but we are using the same original wire, which was good for 50W (i.e. 4.16A) so it must be more than good for our purpose (1.25A).
You could probably use the same 50W transformer for one 211, which requires 10V 3.75A -- in this case you would not split (i.e. cut in half) the 9 turns winding, but rather unwind the superfluous turns. Assuming the same unit and the same voltages, with 6 turns you are most probably going to get 11.12V - which is slightly too much for 10V, and thus would leave you with two alternatives:
1) Burn the excess 1.12V with resistors (0.3 ohm would suffice) which is rather easy to do, or
2) Count the number of turns on the primary and calculate how many do you have to remove in order to get as close to 10V as possible (usually 2 or 3 for Europe, probably 1 or 2 for US) by setting an adequate primary to secondary ratio.
Removing a couple of turns from the primary will not endanger the transformer. Take into account that we plan on using a 50W unit for 37.5W, so there is enough room to play and retain all the reliability.
NOTE:
In practice, most electronic transformers delivery slightly less than 12V, usually between 11.4 and 11.8 depending on the load - the higher the load, the less the output voltage. Since we are planning to use those units at less than rated load, their original output will most probably be 11.6-11.8V and from there should a precise calculation start. But assuming 12V instead leaves us in the worst case with slightly lower voltage, which is not going to endanger the tube: it is not so low to be considered as starvation, and when it comes to thoriated thungsten filaments, most are aware that 9.6 or 9.7V are a better option than 10.2V ...
******
http://rh-amps.blogspot.com/
How well do these electronic transformers do with regards to the cold resistance of the filament ? I'd love to find a practical way of lighting up VT127 (5V 10A) . PC power supplies sulk and choke input choke smoothed supplies are too large to be practical .
Al
When adequately "sized", they do absolutely perfect. The filament lights up effortlessly and reliably.
I have initially used higher quality 105VA units for each 813 (10V 5A). Higher quality units already include overload protection (they simply power down when overheated or overtaxed), short-circuit protection (they power down until the short-circuit cause is solved and removed), no-load condition (they simply do not turn on, although buzz can be heard from the inside of the units).
The second step was to use a 160VA unit for both tubes, and in the end I added another secondary to power the driver and CF tube (6.3V 1A approx.). The 160VA unit is a cheap but well made Chinese product and it powers everything up without a fuss, is rather cold in operation even after several hours, and has a very diminutive size.
You can see several illustrations on my blog - the small black box is the modified transformer with 3 secondaries... compare that to the size of (even) SMPS power supplies, let alone adequately sized power transformers, diode bridges, (someone mentioned chokes?!)... not to mention the heat dissipated by the regulators if you were to use them.
With all that, the most important feature is noise - I am unable to hear any. It's as "noiseless" as if the tubes were indirectly heated. Sounds almost too good to be true - but true it is.
******
http://rh-amps.blogspot.com/
'Sounds almost too good to be true - but true it is.'
A potential 'iron-law' breaker . I like that :) Any suggestions on a unit that could do that 5V 10A filament ?
Al
5V 10A is 50VA after all, so any unit capable of 50W should be able to do it: consider it a light bulb! In practice, I would suggest a 105VA unit for each or a 160VA unit for both 5V 10A filaments.
10A means thicker wire, but since you must use stranded or Litz, this translates to "bifilar wound" secondaries used in parallel.
For 813s, I ended up making 4 turns of 1.5mm2 stranded isolated wire for 10V 5A. For 5V 10A I would do just 2 turns and parallel the two windings for double current. On the primary side nothing changes, it's 50VA.
BTW, the VT127 seems to be 4V 1.25A, and indirectly heated at that? What is the direct heated tube that you are going to use with 5V 10A filament?
******
http://rh-amps.blogspot.com/
"5V 10A is 50VA after all, so any unit capable of 50W should be able to do it: consider it a light bulb!"
Amperage at the output transformer must also be considered. There's no way to know whether the windings of the 50W unit (~ 4A) can handle the increased current, or how the user will reduce the voltage. Rewinding the transformer is best, of course, but if a resistive drop is used, the 10A current draw from the smaller unit will overload it. That's why I recommended a 160W supply (13.3A) for this purpose. It's roughly twice the cost, but still very inexpensive. Incidentally, as a result of recent discussions here, I opened one of the 60W supplies that I had purchased last year as a spare. The output transformer is conventional, not toroidal, much more difficult to rewind.
--------------------------
Buy Chinese. Bury freedom.
If we consider the option to use original secondary wire, we need to provide a 160W unit like you're suggesting. If the amp is stereo, we can use it for both tubes, as 2 secondaries, as explained for 10Y and the 50W unit.
While cheap, it is not at all necessary to use a 160W unit to power a 50W load. If the secondary is removed and a new secondary with adequate wire is wound instead, a 105W unit is literally twice as powerful as needed. Besides the fact that there is no need to burn voltage across resistors (low values of discrepancy can be adjusted by unwinding a few turns from the primary), you would have to burn a lot of W to overload the unit! At 10A, .5 ohms is already 50W, so why throw away all that power and have to deal with huge dissipation? That's what got me to get in the HF AC story in the first place: unnecessary losses in power consumption and heat dissipation - so why would I get back to that once I took a different road?
Once you make it possible for 2x 50W of heaters and a few more W for the drivers to be effortlessly produced by a rather cold (40 degrees C after several hours of operation) and small box... you just cannot get back to burning excess voltage in heat.
As for 50W, imagine 2 "normal" mains transformers. One is 10V 5A, the other 5V 10A. They would be identical size, wouldn't they? The difference would be in the secondary wire thickness and turns number... but if we assume the transformer secondary is bifilar, 2 secondaries of 5V 5A each, than you can do both options with the same transformer. In series 10V 5A, parallel 5V 10A. It's literally as simple as that. The same is valid for the HF AC output transformer, only the size is very much different, as well as the type of wire used (both are frequency dependant).
******
http://rh-amps.blogspot.com/
A 10A filament will need a 160W transformer. You'll need to remove turns from the secondary of the output transformer to reduce it to 5V. You might also want to install a couple low value resistors to "trim" the final value (that's what I do in my 211 SETs).
I searched eBay for "160W electronic transformer" while writing this and came up with quite a few hits for 120V and 220V versions. Unfortunately, none of the listings show the inner workings, so there's no way to be certain if the output transformers are toroidal. On the other hand, none of them appear to be potted, so it's a simple matter to pop one apart and look. At worst, you'll lose $15 for your effort.
Just so you know, where Alex and I primarily differ in all this is the operating frequency. Alex is using newer transformers that operate at 30kHz or so out of the box. The transformers I modified years ago are running at 65kHz. Lynn Olson cautioned me at one time regarding the possible side effects of using such a low frequency (65kHz), and although they sound fine to me, I do wish they were operating at 100kHz or more. In fact, if memory serves, Davie Berning's amps use a minimum frequency of 250kHz. I suspect the upper frequency limit of these little transformers is determined by the capacitance of the active devices. Someday I'll revisit this, just to see if they can be made to do better.
--------------------------
Buy Chinese. Bury freedom.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: