I'm not using triodes at all. Am using six 6550 pentode power output tubes, standard equipment in many guitar and stereo amp units...see impedance measurement thread for details....nothing of intrest happens until page 2, then I shift from impedance related to circlotron related. Not sure what "3 or 4 PSs" means.....do you mean "3 or 4 power supplies" ? If so, am using 4 supplies. The standard amount for circlotron topology, 2 B+, 2 BIAS.....Have been thinking about using a series string filament supply as filament transformer getting hot. Plus keeping down weight and saving space would be good. A rectified 120V AC would also allow filtering in the preamp/driver section giving DC to those filaments, while outputs would be on seperate string using unfiltered pulsating DC, as current requirements would be too high to try to filter.
Thanks for your intrest..midmoe
Thanks for your intrest..midmoe
I used tubes that I have lying around the house, plus I wanted to use commonly available, inexpensive tubes, and not too many of them. I didn't want to have to enlist the help of a team of volunteers to move it, or require an air conditioner to cool it....and I'm not even sure what a "frame grid" tube is....please explain. Perhaps if these frame grid tubes can be used in my setup, which is a maximum of 3X3 output tubes in push-pull, then I will persue the matter. I was skeptical that the wierd looking circlotron diagrams would actually work. To my pleasant suprise, they do and I'm sitting here right now listening to Colin James and the big band 2. Swing music, jump blues, big band stuff...the new amp sounds so much better than the mac 275, I'm thinking about selling it to pay for my trip to europe...please see the "impedance measurement" thread of mine for details...I will switch over to this thread for further updates since my main concern is no longer related to the impedance of the circlotron output, but a investigation of the properties of the output topology...I'm using high impedance speakers, so the impedance issue is fading at least for me. I've rewired my bose 901s in series (64 ohms), put both systems in series (128 ohms) and now they are a close enough impedance match to wire directly to the cathodes, no matching xfmr. required. I estimate the amp's impedance output to be around 200 ohms using crude assessment techniques. Am working daily on getting impedance down, and still have a few tricks to try. The subs are pulled off the plates, using a 120 ohm 3 watt resistor in series to increase dc resistance....suprisingly high output from this, thinking I'd still have to use a xfmr for the subs. l'll be looking for another set of bose 901s to rewire, and any high end 18" woofers with a 16 ohm or higher impedance. Anyone know of 18"ers with a higher than 8 ohm impedance ? Thanks, midmoe.
Last edited:
diytube.com • View topic - OTL amplifier
You can think......cheep amplifier....
I used 6AS7G double triod- about 6-12$.
+U fase splitter must be 2 and more time than +U output tubes for driving them, that have bias -/80 - 100/ v.
You can think......cheep amplifier....
I used 6AS7G double triod- about 6-12$.
+U fase splitter must be 2 and more time than +U output tubes for driving them, that have bias -/80 - 100/ v.
Last edited:
I'm not sure quite what you mean by this. Seems like they would then be at B+ voltage? And as far as the audio path is concerned, I wouldn't think it should matter whether you hook them to the plates or the cathodes, since the power supplies are (in principle, at least) more or less dead shorts to audio frequencies. So wouldn't it be better to hook them up to the cathodes, just like the other speakers?
It would be interesting to see your schematic, certainly.
Chris
there seems to be a power-sharing thing going on between the plates and the cathodes...yes the sub is a +170, no big deal on that for me as am still in prototype phase. Identical information seems to be expressed at both cathode and plate rails...and if I present a high freq. load to the plates, it reduces highs expressed at the cathodes...leaving me to presume that if you were to remove low freq energy from the plates (with sub and associated cross.) then it would be reduced in the cathodes in direct proportion to removal from plates....hence the natural crossover effect. At least that seems to what's happening. Will do further tests tommorow about this. I'm getting the bias contols re-set today as I added a 100K between grid and plate rails for more neg. feedback and it threw my bias off beyond the control's range to compensate. Several diagrams show a 100K resistor, so I use a 2.2 meg in there now and it made quite a difference in the stage's behavior in a positive way, and within the bias conrol setting range. See entry in page two of "impedance measurement" thread for details. All posts related to the circlotron should have been made here, sorry for any confusion. Still a newbie about certain computer tasks, such as uploading pics. My friend Russel is coming up tommorow to help me out, and a functional schematic should be available in the evening. I will post new schematics as improvemets are made, but the basics shouldn't change. Thank you for your intrest. midmoe
Yes, that's what is to be expected, I think. Which means it should make no significant difference whether you take the signal for the plates or the cathodes.
I'm not arguing against spicing things up a little by floating the subs at B+ voltage, just pointing out that it probably has no significantly different effect from hooking them instead to the cathodes.
Chris
I don't think, with the circlotron topology, that that makes a difference. The cathode of one tube is connected, through a power supply, to the anode of the other. Since the power supply (should, at least) approximate to zero impedance at audio frequencies, then it should not affect the output impedance whether the output it is taken from the cathodes or the anodes.
Chris
yes , your'e probably right about it not making any difference....only going by ear so far on this one, will be hopefully uploadiing schematic tommorow. You can begin to see what's what by looking at the schmatic of the Grommes 260A. That is closely the pre/phase inverter I use, driving an added 12BH7 to get enough voltage swing to output stage grids....then the rest is easy, follow commonly posted diagrams, but don't tie the 1Ks to ground that are shown across the cathode rails. A drastic improvement in stability and distortion reduction is achieved thru tying them to the cathodes of the opposite driver stages (extra BH7). I ended up using a .47 to block the +5 present on the cathodes from causing trouble later, but didn't seem to make any difference, anyway, seems a good idea at the time. On the screen supply. don't hook them to opposite plate rails with resistors directly, I use a simple screen rail system that allows some mixing between all screens. This seems to have stabilized bias balance and reduced spurious responses on heavy bass notes. See posts on "impedance measurement" for a better discription although, it's admittedly sketchy, wil be easily understood with schematic. With later mods, perhaps a more direct hookup could be done, but it's good enough for now, so am working on bias issues. A rather mundane issue is the heating of the filament xfmr...have to have a mufin fan on it if left on more than a couple hours...am toying with the idea of tying the outputs filaments in series string, and the preamp/driver fils. in a seperate series string....and using a half wave diode ckt. to half the volts, needing less extra volts to get rid of. And added advantage is that the pre/driver tubes have a low enough current so as to filter the pulsating dc down to a fairly smooth negative dc for added hum and noise reductions. That will be in the companion model (for stereo), unless problems are found. I am trying to minimize space, weight, and energy consumption issues. Getting rig of the 80 watt filament xfmr. would cut down weight and reduce heating inside, although I am only too aware of the "qiuck and dirty" madman muntz quality of series string filaments, if it works, and is reliable, so be it. I used to work on those cheap B&W tvs with series strings fils when I was a kid and didn't like them. Perhaps a sweeter technical idea would be a small switching power supply on board to accomplish filament heating, but with added complexity. Will be looking for a simple SCR based design, using a minimum of components and no ICs to efficiently chop off sinewave peaks to achieve results. A humerous idea was to mount 4 more 6550s on top and just use them as votage droppers in the fil. ckt..!..don't hook them up in the output stage, just use them in a circut used in the '40s called "ballast tubes" in early ac/dc cheap radios..they also use "reisitance line cords" to get rid of extra volts ! This would give the amp a built-in spare set of output tubes, ready to go...haha just kidding...of course. But the additional tubes hooked up in series might elimenate any extra filament supply needs. there is no reason yet seen why a person couldn't hook them up fully into the output stage...the B+ transformer runs cool with 6 tubes and is rated at 175 watts...giving additional drop in output impedance....but my original goal was to achieve a functional medium power amp, using just enough output tubes to run my modified boses/electovoice subs to a good volume which I have achieved using just 6 outputs. The only significant mod. to grommes is to unhook the 100K to the plate of the 12AU7 from +250 and hook it to the top plate of the 12BH7 phase inverter....as gain was too high. This introduced neg feedback in the stage so as to have to re-jigger associated plate resistors as now the grounded grid side has more gain than the other. Will have to go back and attempt to equalize gains in bot sides, but it's ok for now. I am using Triad transformers purchased from Allied, output tubes are JJ brand (slovakian) purchased thru Antique Electronic Supply, Tempe Az. Most all the rest is cobbled from several junked organ amp chassis or parts currently in stock in my store....midmoe
rewired the bias supply to allow a lower value of resistor from plates to cathodes. Changes to the bias supplies has allowed me to lower the plate to grid resistance from 2.2 meg down to 820K, with no discernable sonic difference noticed. The next tweaks will require setting up a distortion analyzer, which I need to get, along with equalizing gains in the phase splitter section. I'm wondering if my filament xfmr. heating is partly due to the transformer itself. As I recall, it's a universal type, and so could be used on 50 or 60 hz. Does this mean that it's tuned to 55hz and so is a little off tuning wise? Perhaps a xfmr. specifically for the USA 60 hz would lessen the problem. A small,quiet muffin fan I guess is acceptable.I would certainly like to keep the conventional filament setup....midmoe
I'm not sure what voltage you might be wanting for your series-connected heater chain, but another way of dropping some voltage, without any significant power dissipation, is to put a rectifier diode in series with the chain. This will effectively drop the voltage by a factor of 1/sqrt2 (since the power will be halved). I am planning to do that in a heater chain for an amplifier I am dreaming up at the moment.
Chris
yes, a diode would be necessary in the series string idea in order to get the 10 6550s to do the job. If the volts is roughly cut in half with the diode, then you have approx 60 vdc to use on 10X6.3=63 volts.....probably won't do that though. I will use transformer for the 6550 in the next unit, and series string the preamp/drivers...also needing same voltage requirements, but much less current. This would take some of the work off the fil. xfrm. and give oppurtunity to clean up the series string on the 12AU7, 12BH7s giving dc negative to all the more hum sensitive tubes....haven't done the math yet on how much available energy is available after the diode, but capping the neg should skew the math a little. Wouldn't that give a higher voltage? I believe the 1.414 rule would come into play, but this is half wave, not full wave, so would have to experiment a little, should be doable. Am going to town now to get 8"X14" aluminum panels and 1/2" oak for stereo companion. Should have that ready within 2 wks. Mono is ok for now, but I've noticed some added harshness in certain passages when preamp set to mono that dissapears when switched to left of right...seems with the mixing of L&R there is some occaisional subtraction multiplexing effect causing a loss of integrity, sounds like a smearing of wavefronts from two different mics....usually only listen in mono position when playing 78s as then the noise cancelling property works to advantage. midmoe
No, the voltage is effectively cut to 1/sqrt2 of the original rms voltage by the diode.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.