What I mean is: why do we need the C-shape at all - just to thread the wire through the centre of the toroid?
So my idea was just use a wheel as you suggested with the wire around the rim and going through the centre of the toroid which is pulled away from the wheel by a spring for tensioning.
So just the same as the patent but the toroid is not in the any C-slot, it is pulled away from the wheel which contains the wire on it's rim and it stretches to the toroid through it's centre!
So my idea was just use a wheel as you suggested with the wire around the rim and going through the centre of the toroid which is pulled away from the wheel by a spring for tensioning.
So just the same as the patent but the toroid is not in the any C-slot, it is pulled away from the wheel which contains the wire on it's rim and it stretches to the toroid through it's centre!
You start with the toroid pulled away from the bobbin (wheel) and thread the wire by hand through the toroid & around the wheel rim as per Don's description.
Basically, everything is done the same as per Don's description except the toroid is held away from the bobbin by a tensioning spring.
The original C-shaped sausage belt is conceptually the same as a circle!
Basically, everything is done the same as per Don's description except the toroid is held away from the bobbin by a tensioning spring.
The original C-shaped sausage belt is conceptually the same as a circle!
Hmmmm.... I think I see what jkeny is trying to do, but as kenpeter said: what? The pre-wound wire in either scheme is a big circle passing thru the core. I don't see how you can move it away from the core dynamically for tensioning. You could move the C drive away, but the wire bundle thru the core would then have to dynamically deform every rotation. Not too good.
Let me mention a Japanese patent. They wanted a scheme to wind tiny tape head cores. Obviously it had to be shuttleless. Instead of winding the wire on something thru the core, they just spread the wire out in a long trough with air flow thru it, like the way the old 9 track computer tape columns used for tensioning the tape. The wire end was then guided around thru the core opening by air jets then back into the trough. One turn at a time getting wound on the core. I don't think this would allow accurate positioning of the wires on a core without some very fancy mechanical contraption though.
I think the Potthoff scheme is reasonably workable. Maybe could use velvet against the teflon C side for the wire accumulator, for smooth tension control.
Another concern I have for the Potthoff scheme is how the wire gets back into the accumulator slot after the spillover point passes thru the core cutout section in the C. It has to get back up on the rounded Teflon C edge again. This seems tricky to me. Likely requires some tapered section of the rounded C edge to lift it back up. But the wire could just as easily decide to get snagged in the cog belt to Teflon C edge clearance. (ie, between the belt and the Teflon edge)
Don
Let me mention a Japanese patent. They wanted a scheme to wind tiny tape head cores. Obviously it had to be shuttleless. Instead of winding the wire on something thru the core, they just spread the wire out in a long trough with air flow thru it, like the way the old 9 track computer tape columns used for tensioning the tape. The wire end was then guided around thru the core opening by air jets then back into the trough. One turn at a time getting wound on the core. I don't think this would allow accurate positioning of the wires on a core without some very fancy mechanical contraption though.
I think the Potthoff scheme is reasonably workable. Maybe could use velvet against the teflon C side for the wire accumulator, for smooth tension control.
Another concern I have for the Potthoff scheme is how the wire gets back into the accumulator slot after the spillover point passes thru the core cutout section in the C. It has to get back up on the rounded Teflon C edge again. This seems tricky to me. Likely requires some tapered section of the rounded C edge to lift it back up. But the wire could just as easily decide to get snagged in the cog belt to Teflon C edge clearance. (ie, between the belt and the Teflon edge)
Don
The Holy Smoking Toroid Winder
Just for laughs, here's an idea:
We start with a very long U shaped pipe with a narrow slot cut in the inside side along the length (ie, inside side of the U) Air is blown into the pipe from one end. Very small primary wire is then fed into the air inlet end of the U tube and blown down and around the U until it emerges from the other end. Air jets then blow this wire end thru the toroid and back into the inlet end of the U tube.
As the wire sails around thru the toroid it gets shorter in the U tube and pulls across thru the slot in the U (ie, takes a shortcut short of the U bottom end eventually short cutting the whole U and pulling up tight on the toroid). Since each turn wound on the toroid will then shorten the remaining wire, we move the core gradually down the U tube with each turn so that it will blow the wire down around the U and back to where the toroid is. So the wire now enters and leaves the U tube thru the slots near the toroid and this point progressively moves down toward the U end as the wire is used up. Each turn still shortcuts across the U thru the slots as it pulls up tight. (shortcutting starts at the U bottom end each time)
The tube cross section would need to be a circle with a V extension on the side protruding out to the slot, so that the wire can easily find its way to the slot as it pulls tight. Well, actually, only the end of the U needs this special shape.
Of course, with 1000s of turns, I would hesitate to calculate the length of custom pipe needed. And the amount of air flow required.
Don
Just for laughs, here's an idea:
We start with a very long U shaped pipe with a narrow slot cut in the inside side along the length (ie, inside side of the U) Air is blown into the pipe from one end. Very small primary wire is then fed into the air inlet end of the U tube and blown down and around the U until it emerges from the other end. Air jets then blow this wire end thru the toroid and back into the inlet end of the U tube.
As the wire sails around thru the toroid it gets shorter in the U tube and pulls across thru the slot in the U (ie, takes a shortcut short of the U bottom end eventually short cutting the whole U and pulling up tight on the toroid). Since each turn wound on the toroid will then shorten the remaining wire, we move the core gradually down the U tube with each turn so that it will blow the wire down around the U and back to where the toroid is. So the wire now enters and leaves the U tube thru the slots near the toroid and this point progressively moves down toward the U end as the wire is used up. Each turn still shortcuts across the U thru the slots as it pulls up tight. (shortcutting starts at the U bottom end each time)
The tube cross section would need to be a circle with a V extension on the side protruding out to the slot, so that the wire can easily find its way to the slot as it pulls tight. Well, actually, only the end of the U needs this special shape.
Of course, with 1000s of turns, I would hesitate to calculate the length of custom pipe needed. And the amount of air flow required.
Don
This idea hasn't died - its percolating quietly in the background. I've decided for the time being on Edcor output transformers, as they are cheapest and I like the looks. If anyone has had experience with the Edcor 5k, 100W push-pull transformers, please speak up.I haven't bought any output iron yet, and won't for a while.
Private communications have given me some options to try to boost HF bandwidth. These will be held in reserve until I see how the simplest version of this amp performs. The screens of the output tubes will be driven with source followers, so there will be plenty of current available for the screen and for Mr. Miller.
I've been collecting 35LR6s to see who is best in terms of husky plate construction. Mum's the word on my favorites until I get my share....
Private communications have given me some options to try to boost HF bandwidth. These will be held in reserve until I see how the simplest version of this amp performs. The screens of the output tubes will be driven with source followers, so there will be plenty of current available for the screen and for Mr. Miller.
I've been collecting 35LR6s to see who is best in terms of husky plate construction. Mum's the word on my favorites until I get my share....
I have no experience with either of these exact items, but I offer this observation. I have been collecting 6, 26 and 36LW6's for years. They are a similar large sweep tube. They are found with several different plate constructions from big to immense. The wimpiest versions can dissipate 50 watts without glow, and the big guys smile at 90 watts. I have cranked these up to the 150 watts per pair level in both conventional G1 drive and screen drive. I find that these tubes work best with a much lower load impedance than 5K ohms. Most of my experiments were done using a 6600 ohm OPT with the 8 ohm load on the 16 ohm tap for a 3.3 K load. I have also used a 1250 ohm BIG 400 watt Plitron toroid OPT with good results. SY uses a 1250 ohm load in his 6LF6 screen drive amp. Both of these tubes are a bit bigger than the 6LR6.
I got 8 6LR6's off of Ebay for cheap a while back. Two are obviously toast (noted in the auction) so these will get "tested". The good ones will see some P-P amp testing as time permits.
wrenchone said:
I own a pair. Recall measuring primary inductance at 20H per plate! Far too many Henries for my LCR meter plate to plate...
I'm guessing thats 80H P-P whilst both plates are in ClassA?
But only 20H for one plate in Class B. Or is that not the math
is done???
I forget now what I had measured for secondaries...
I can check that for you this weekend.
They are pretty and blue. But white Edcor logo is on a clear
peel-off sticker that tends to have ugly air bubbles under it,
so you will most likely wanna tear that sticker right off.
The mounting tabs on the bell housing don't quite seem as
substantial as a transformer of this mass deserves... Maybe
the bellhousing was stamped from too thin a sheet?
Edcor leads and insulation are substantial, much better than
the crummy leads on my Hammond choke of similar size. But
Hammond's ugly black bell seems to have the stronger tabs.
A typical LCR meter uses a small test signal (like 1V or so) that does not get the magnetic field up enough to see large signal inductance. M6 (or most xfmr steels) have low Mu at tiny field strength, then improve considerably up around 1/2 Bmax. The usual inductance figures given for xfmrs are measured with large excitation (like from a Variac). But the initial Mu measured by the L meter is an important parameter for small signal performance. 70% Nickel alloys do much better at the low excitation level than the M6.
Don
Don
If I wanted to do a "full George" and beat the living snot out of the tubes, I could use a lower impedance transformer than 5k P-P. However, I was keeping in mind the relatively high plate resistance of most screen driven sweep tubes and catering to that with a reasonable value of transformer impedance so as not to be so terribly dependent on feedback for low output impedance - correct me if this line of thought is faulty. A clean and honest 100W RMS would be more than enough for me - I'm 55, but I still seem to have relatively intact ears...
The 35LR6s were a very nice chance buy during the feeding frenzy at AES a while back - I couldn't resist hefty NOS sweep tubes at $5 each. AES has since recovered their senses, and the 35LR6s are now over 3X that price, though still a bargain compared to a lot of the big sweep tubes.
So far, I am seeing two types of 35LR6s - the GE type with a big plate with two simple side fins for heat dissipation, and the Sylvania type that has an equally big plate and a complex fin arrangement intended to spread out the radiating area some more. If I get around to it, I'll post a picture of the two types side by side. I can't recall if I also have an RCA version for comparison. The alt brand stuff I have so far seems to be relabeled GE or Sylvania.
The 35LR6s were a very nice chance buy during the feeding frenzy at AES a while back - I couldn't resist hefty NOS sweep tubes at $5 each. AES has since recovered their senses, and the 35LR6s are now over 3X that price, though still a bargain compared to a lot of the big sweep tubes.
So far, I am seeing two types of 35LR6s - the GE type with a big plate with two simple side fins for heat dissipation, and the Sylvania type that has an equally big plate and a complex fin arrangement intended to spread out the radiating area some more. If I get around to it, I'll post a picture of the two types side by side. I can't recall if I also have an RCA version for comparison. The alt brand stuff I have so far seems to be relabeled GE or Sylvania.
Well, I was planning to try a pair of matched JFETs cascoded by an 8CG7 as a differential gain stage/splitter, with another 8CG7 as voltage amp after that if necessary. If you look at the schematic for the old Berning sweep tube amp, they split up the gain stages in a similar fashion, though without the sand. I'll use source followers to drive the screens on the finals. I'm trying the JFET cascode first driving a pair of Russian 7189 clones as part of rebuilding an old Lafayette integrated amp to test the concept and see how much gain/drive voltage I get. Slow and easy does it, especially as I seem to be getting sidetracked into a gainclone project for a short while (curiosity killed the cat).
Some more back of the envelope stuff regarding the gain stages - in another project using PN4393 JFETS running at 2.5ma under a 12AU7, 250V B+, I'm getting a gain of about 125X. Taking this gain value as an attainable example, and assuming 80V drive will get me to the stated goal of 400ma peak plate current, and tracking through an output transformer with a 12.5:1 turns ratio on each side (5k), I get an open loop gain of 62.5. I would settle for a closed loop gain of 30-35X or so, so I will need some more open loop gain to lower the output impedance via feedback. I think I can get some more gain out of the input stage using the 8CG7 on top of the FETs by upping the drain current (more transconductance). I may look at using J310s or 2SK170s to pump up the gain. The higher available plate voltage will also allow me to use higher value load resistors. This may allow me to get all the gain I need (along with the phase splitter) in the first cascoded differential stage. If not, I have the option of using a modest common cathode stage like the Berning amp to eke out some more gain. I'd like to avoid this if I can.
One question - will the 8CG7s put up with 500V on their plates while they're warming up, or will I need to drop the B+ feeding them?
One question - will the 8CG7s put up with 500V on their plates while they're warming up, or will I need to drop the B+ feeding them?
Take a look at 6GF5, 700 V DC rating, 9 Watt, $1. Doesn't have the high Gm of the Fets though, so you will need to cascode still. Just run them with a fixed screen voltage. Could run local feedbacks to the g1's (Ummm... that's only working if there is a triode like device underneath though).
Actually, I was thinking of some other equally humble options. The 8CM7 and 8CS7 dissimilar double triodes both can withstand 500V on their plates, and triode 2 of either tube will allow sufficient voltage across the bottom JFETs at 5ma or so if used as a simple cascode with grid pinned to the JFET source. This leaves triode 1, which can be left idle or used for a second gain stage if necessary.
I would have to knock the plate supply down a bit from the main B+ to stay within the ratings, but I could tap the main output winding or just use a modest regulator, assuring a really clean supply for the input stage. I'd like to stay with the JFET front end, as it is then really easy to match devices for a differential stage, and you get a lot of gain with just one stage. Besides, I haven't seen anyone else trying it, and I like eccentric solutions...
I would have to knock the plate supply down a bit from the main B+ to stay within the ratings, but I could tap the main output winding or just use a modest regulator, assuring a really clean supply for the input stage. I'd like to stay with the JFET front end, as it is then really easy to match devices for a differential stage, and you get a lot of gain with just one stage. Besides, I haven't seen anyone else trying it, and I like eccentric solutions...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.