Greetings everyone,
So, I seem to be at a loss, I have 4 Edcor 3.3KΩ/100W transformers. When I measure the resistance between the red and blue leads and then the red and brown leads, there seems to be a difference of 10-15 ohms. Is this by design or is this manufacturing tolerances?
Thanks,
Ray
So, I seem to be at a loss, I have 4 Edcor 3.3KΩ/100W transformers. When I measure the resistance between the red and blue leads and then the red and brown leads, there seems to be a difference of 10-15 ohms. Is this by design or is this manufacturing tolerances?
Thanks,
Ray
What you're measuring is not necessarily an inaccuracy, but has to do with the wire length as the coil build-up becomes larger. The outer layers require more length per turn so the resistance increases. A high quality transformer will have progressively slightly larger wire size to compensate for this as the build-up expands. If you use low voltage AC on the secondary to measure each half of the primary, you'll probably find a balanced CT.
There are other (and better) ways to ensure accurate CT's on push-pull transformers.
Measuring the primary through a voltage source on the secondary does not tell the whole story, as in this case there is no DC current flowing in the primary.
Inaccurate CT's will cause extra odd order harmonic distortion.
High quality OTs use a split bobbin with equal resistance windups for each primary side. The cheaper ones (most) just use a single bobbin, so one side of the primary is wound at larger diameter than the other. So the resistances come out unequal unless different wire sizes are used to compensate. Good amp designs will then use an equalizing resistor in series with the lower R winding to equalize the sides. This equalization helps to reduce distortion (gain symmetry) but will increase losses and output Z a little.
Another factor to check for is unequal leakage inductance on the two primary sides. This causes some distortion at high frequencies. Typically the inner UL 40% winding sections are between secondaries and have the best coupling. The 60 % sections have a large difference in leakage L, since one is wound at the beginning of the bobbin and the other is wound on the outside.
This is one reason I like the "Elliptron" configuration, since each tube uses both of the good 40% sections, with the CFB taken off of one 40% section for each tube. Using the 40% taps for differential local feedbacks in the amp works out well too, since they are both always driven windings (each tube drives both). Whereas, the typical local plate "Schade" feedback scheme uses feedback points that have a large difference in leakage L and only one of them is driven at any time (in class AB). So the other feedback is really not local since it has to pass thru the xfmr.
Another factor to check for is unequal leakage inductance on the two primary sides. This causes some distortion at high frequencies. Typically the inner UL 40% winding sections are between secondaries and have the best coupling. The 60 % sections have a large difference in leakage L, since one is wound at the beginning of the bobbin and the other is wound on the outside.
This is one reason I like the "Elliptron" configuration, since each tube uses both of the good 40% sections, with the CFB taken off of one 40% section for each tube. Using the 40% taps for differential local feedbacks in the amp works out well too, since they are both always driven windings (each tube drives both). Whereas, the typical local plate "Schade" feedback scheme uses feedback points that have a large difference in leakage L and only one of them is driven at any time (in class AB). So the other feedback is really not local since it has to pass thru the xfmr.
Last edited:
OK, I have seen the introduction of some series R to balance unequal primary DC resistance either side of the CT reduce distortion in small signal stuff (even this was debatable) but introducing a series resistance in a power application ie. an output stage is daft.
If the CT is truely not a CT then send it back.
I would love to see some measurements on a 100W output tranny with unequal primary DC RESISTANCE that has been balanced with some series R that show less distortion in this configuration.
Anyway, I have no experience of Edcor trannys and have only read good things about them. If the OP is not happy with his 3K3 trannys I am sure I can make use of them, send them this way for proper disposal
Cheers Matt.
If the CT is truely not a CT then send it back.
I would love to see some measurements on a 100W output tranny with unequal primary DC RESISTANCE that has been balanced with some series R that show less distortion in this configuration.
Anyway, I have no experience of Edcor trannys and have only read good things about them. If the OP is not happy with his 3K3 trannys I am sure I can make use of them, send them this way for proper disposal
Cheers Matt.
The conventional Elliptron configuration is just a Circlotron with less cathode winding %. A 40% UL tap gives (.4/1.4 = .286 or 28.6 % CFB). No question it works, but requires two inconvenient floating power supplies. I use some Lambda power factor modules to get 360 V regulated from some split bobbin isolation xfmrs (low common mode capacitance). A big advantage is that Elliptron mode essentially doubles (1.4 squared) the primary impedance seen by the tubes. Some of those 10 K Ohm P-P xfmrs out there (for GU50 say) don't have any bandwidth, but a 5K P-P Elliptron'd (to 10K) OT is quite practical.
Things get more tricky when trying to get by with just one power supply in the Elliptron. I have just acquired some high bandwidth balancing inductors to try the cathode DC feed scheme. An issue comes up here concerning DC resistance of the inductor feeds, since they set up a time constant with the two floating power capacitors, which affects the tube biasing with class AB and signal evelope bursts. One can try to brute force solve that by using a low DC resistance balancing inductor to feed the two floating power caps, but this gets expensive to also keep the magnetizing current low (inductance high).
I'm exploring a servo circuit from a small negative supply to regulate the cap voltages and hence the tube biasing. So this is still under development. Although there is an alternate approach of using drivers with their plate loads bootstrapped from the output tube cathodes to effectively remove the bias shifts due to the floating power cap voltage variations. I have no doubt the single supply scheme can be made to work, its just a question of finding the most effective and cost effective approach.
----------
The balancing resistor for the OTs are typically only 15 Ohms or so as Rsumperl mentioned. Not a problem in kiloOhm level outputs. People put that much resistance in the cathode circuits just to measure idle current. Certainly depends on the standards one wants to achieve as to distortion too. The usual couple percent dist. amps won't notice any improvement from balancing R. 15 Ohms imbalance for 1000 Ohms per side (a 4K P-P OT) is a 1.5% gain error, that makes real distortion. One can try to null that out with an AC gain balance pot in the driver stage, but that only works for a fixed load impedance. Real speakers have varying Zload with frequency.
Things get more tricky when trying to get by with just one power supply in the Elliptron. I have just acquired some high bandwidth balancing inductors to try the cathode DC feed scheme. An issue comes up here concerning DC resistance of the inductor feeds, since they set up a time constant with the two floating power capacitors, which affects the tube biasing with class AB and signal evelope bursts. One can try to brute force solve that by using a low DC resistance balancing inductor to feed the two floating power caps, but this gets expensive to also keep the magnetizing current low (inductance high).
I'm exploring a servo circuit from a small negative supply to regulate the cap voltages and hence the tube biasing. So this is still under development. Although there is an alternate approach of using drivers with their plate loads bootstrapped from the output tube cathodes to effectively remove the bias shifts due to the floating power cap voltage variations. I have no doubt the single supply scheme can be made to work, its just a question of finding the most effective and cost effective approach.
----------
The balancing resistor for the OTs are typically only 15 Ohms or so as Rsumperl mentioned. Not a problem in kiloOhm level outputs. People put that much resistance in the cathode circuits just to measure idle current. Certainly depends on the standards one wants to achieve as to distortion too. The usual couple percent dist. amps won't notice any improvement from balancing R. 15 Ohms imbalance for 1000 Ohms per side (a 4K P-P OT) is a 1.5% gain error, that makes real distortion. One can try to null that out with an AC gain balance pot in the driver stage, but that only works for a fixed load impedance. Real speakers have varying Zload with frequency.
Last edited:
Split bobbins or two identical coils on a c-core are techniques to ensure accurate CT's.
But even with the single bobbin and the same wire size it is not that difficult to get an accurate CT.
How about this one to tease your brains: 1+4=2+3, or 1+4+5+8=2+3+6+7.
Happy Eastern
He put them up for sale here last year, and I offered to buy, but nothing ever happened. I doubt they are going across the pond for free......
I can't speak for the interleaving in an Edcor since I have never taken one apart. I have some cheap P-P OPT's that have no interleaving at all. They are wound with one half-primary, then the entire secondary, then the other half-primary. The outer half primary is wound whth thicker wire but the DC resistances are not equal. They however sound much better than they should.
The unequal DC resistances can cause a tiny bit of imbalance, but there are other factors that are much more dominant to deal with in OPT design.
Thanks for the reply Smoking Amp, sorry for the delay in my reply.
Perhaps you should start a new thread on your Eliptron design.
I have no need for the single supply version of your design and it also seems to be one hell of a balancing act, as are most ppp designs.
I have an ex BT 48V 60A switching supply with PF correction that I have modified the secondary side to produce alsorts of voltages x2 only as a bench supply but very, very useful.
Anyway the inter supply capacitance for two 500V secondarys is only a few pF. I didnt originally plan this, just figured two variable 500V floating 1.5A supplys would be usefull. Plus it has any number of easily wired heater supplys and two whatever use bias supplys.
Could you start a thread on your Eliptron design? Obviously for a stand alone design the PS is a real problem as with any circlotron style design. Still for me and I guess many others it would be vey interesting to see your developments.
Thanks Matt.
Perhaps you should start a new thread on your Eliptron design.
I have no need for the single supply version of your design and it also seems to be one hell of a balancing act, as are most ppp designs.
I have an ex BT 48V 60A switching supply with PF correction that I have modified the secondary side to produce alsorts of voltages x2
only as a bench supply but very, very useful.Anyway the inter supply capacitance for two 500V secondarys is only a few pF. I didnt originally plan this, just figured two variable 500V floating 1.5A supplys would be usefull. Plus it has any number of easily wired heater supplys and two whatever use bias supplys.
Could you start a thread on your Eliptron design? Obviously for a stand alone design the PS is a real problem as with any circlotron style design. Still for me and I guess many others it would be vey interesting to see your developments.
Thanks Matt.
There was an old Elliptron thread, but it's a bit out of date now. Not really much new here with respect to a standard Circlotron design. Just lower CFB %. This makes the driver stage easier to design at least. Same issue comes up for both as to screen grid supplies. Either floating screen supplies from some split bobbin xfmrs (or switcher windings), or bootstrapped/zenered screen supplies from the opposite plates (like McIntosh), or a fixed (to ground) common screen supply for both outputs, which gives UL mode effectively. (the 28% UL resulting for a Mu 4 Sweep tube looks like 56% UL for a typical Mu 8 audio tube, so is a bit excessive) One can use bootstrapped loads on the driver stage (like McIntosh, but mainly for pentode drivers) to reduce the CFB% and lower drive stage requirements. Doesn't lower the UL % though.
One issue I would see with the switcher supplies is the capacitive coupling from the switcher primary to at least one of the secondaries. Usually switchers have some small HF caps to ground for the secondary (and primary) to short out the common mode coupling. If your secondaries have good thick insulation (which an oversized switcher would have room for) then maybe this will not be a problem.
Using PFC modules for the floating supplies (off of split bobbin isolation xfmrs) like I am using has an issue with 60 Hz ripple still, due to the primary side voltage dropouts at 60 Hz. I'm experimenting currently with a 60 Hz resonant LC connection for the PFC inputs (leakage L of the split bobbin xfmr, and an elec. cap after the input rectifier for the PFC) to eliminate the ripple.
I'm planning to use this Elliptron stage as the output stage for a variant of Pete Millett's "Engineer's Ampl". But one can sub in the Elliptron stage for most any design.
One issue I would see with the switcher supplies is the capacitive coupling from the switcher primary to at least one of the secondaries. Usually switchers have some small HF caps to ground for the secondary (and primary) to short out the common mode coupling. If your secondaries have good thick insulation (which an oversized switcher would have room for) then maybe this will not be a problem.
Using PFC modules for the floating supplies (off of split bobbin isolation xfmrs) like I am using has an issue with 60 Hz ripple still, due to the primary side voltage dropouts at 60 Hz. I'm experimenting currently with a 60 Hz resonant LC connection for the PFC inputs (leakage L of the split bobbin xfmr, and an elec. cap after the input rectifier for the PFC) to eliminate the ripple.
I'm planning to use this Elliptron stage as the output stage for a variant of Pete Millett's "Engineer's Ampl". But one can sub in the Elliptron stage for most any design.
Last edited:
nice kuya tonyi have used 1+4 = 2+3, resistance is within 1 to 3 ohms of each other...
Happy Easter to ALL.....
it took me a while before realizing it was haflers mind
nice math
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.