This is interesting.
The windings have to 'share' the same core?
I'm thinking about using a toroid PT as a PP output transformer, connecting the two 120vac primaries to form a 'center-tapped' primary. Others have reported success with this.
It does seem similar to the idea that emosms is considering.
Yes in a PP transformer the windings have to share the same core, the opposing dc fields in the two halves of the primary cancel each other leaving only the AC component - this allows for a significantly smaller core.
Several people here have extensive experience with toroid power transformers as OPTs.
perhaps this is a better alternative:
http://www.diyaudio.com/forums/tubes-valves/109335-crowhursts-twin-coupled-amplifier.html
http://www.audioxpress.com/magsdirx/ax/addenda/media/1960crowhurst.pdf
Last edited:
Why not take two E-I transformers, Strip off the lamms and rearrange so they can face each other with the 'E's used in each one, and the 'I''s discarded....
Make up a frame to keep them held together, and you can arrange the 'primaries' to magnetically cancel the 'DC' component in the coil/iron...
I did this many years ago with some SE transformers from TV sets, Worked out pretty good too....
Make up a frame to keep them held together, and you can arrange the 'primaries' to magnetically cancel the 'DC' component in the coil/iron...
I did this many years ago with some SE transformers from TV sets, Worked out pretty good too....
Maybe a kludge of this sort could work:
The center tapped low voltage (2 Volt maybe) DC supply (T3 and bridge rectifier) provides current to the secondaries so as to null out the primary DC tube currents in each OT xfmr. (The audio AC is nulled out at the DC supply insertion point) The speaker has the DC nulled out by connecting to the center tap of T3.
edit: The Rlimit resistors for current control would be better put on the DC sides of the bridge rect. than on the AC sides as shown. Then any mismatch between them would only inject a bit of DC to the speaker (rather than 60 Hz hum as configured).
The center tapped low voltage (2 Volt maybe) DC supply (T3 and bridge rectifier) provides current to the secondaries so as to null out the primary DC tube currents in each OT xfmr. (The audio AC is nulled out at the DC supply insertion point) The speaker has the DC nulled out by connecting to the center tap of T3.
edit: The Rlimit resistors for current control would be better put on the DC sides of the bridge rect. than on the AC sides as shown. Then any mismatch between them would only inject a bit of DC to the speaker (rather than 60 Hz hum as configured).
Attachments
Last edited:
Do you mean to use 2 big expensive transformers instead of one smaller and cheaper?
Yes, why not?
"Connection in series, to get the center tap "
For two "OTs" with no primary center tap on each and no air gap, the only practical way is in post #7, or go Circlotron mode, or use a totem pole tube output.
If the "OTs" do have dual primaries, then it's much easier, since each side of the P-P tubes can go thru a primary on both xfmrs to get DC balance. The Crowhurst Twin Coupled will work then too.
note:
Another improvement on the post #7 diagram: replace the C with two C in series and connect their center point to the CT on T3. That will eliminate any chance of bridge generated glitches from 60 Hz AC commutation.
For two "OTs" with no primary center tap on each and no air gap, the only practical way is in post #7, or go Circlotron mode, or use a totem pole tube output.
If the "OTs" do have dual primaries, then it's much easier, since each side of the P-P tubes can go thru a primary on both xfmrs to get DC balance. The Crowhurst Twin Coupled will work then too.
note:
Another improvement on the post #7 diagram: replace the C with two C in series and connect their center point to the CT on T3. That will eliminate any chance of bridge generated glitches from 60 Hz AC commutation.
Last edited:
It may work fine for a class A output stage, but for class B it will have poor coupling between the tubes. So the crossover between tube conductions in class B will likely have some inductive glitching (from leakage inductance) in the separate xfmrs. Class AB would depend on the amount of overlap in the crossover region. Might be OK.
Last edited:
This is really a genius idea smoking-amp !
To validate it, I make some simulations with two equivalent circuits with my SIMetrix simulator which is so usefull, the results tell me your circuit should work perfectly in real life ...
I first draw a normal push-pull circuit using a 5000 / 8 ohms transformer with a centertap on primary to make a precise comparison ... The two diodes cut-off all the opposite amplitude just like tubes in class "A" :
An externally hosted image should be here but it was not working when we last tested it.
Then I draw a circuit equivalent to your's using two SE transformer instead of one PP transformer with battery representing your DC supply which purpose is annulating part of the magnetic DC saturation :
An externally hosted image should be here but it was not working when we last tested it.
As you can read my schematic, the power is the same and there is absolutelly no DC voltage in the coil of the speaker, there seem to be just a little bit more distortion. But the most important thing to said is the primary impedance of the two SE tranformers must be the fourth of the primary impedance of the PP transformer ...
In my tests, the PP is 5000 to 8 ohms and the two SE are 1250 to 8 ohms ... This is because the ratio of the transformer and the parallel wiring of the two secondary ... R2 and R3 in the second schematic represent the DC resistances of the secondary winding, I have to put them on unless the simulator report an error.
But with those 0,5 ohms resistances and the 1,4 VDC of the battery, the current in the winding is quite high, 2,8A ... So the limiting resistances you use in your circuit are very importants to protect the secondary of the transformers from excessive heating.
I like very much this idea because it can be usefull in some specials kinds of circuits. If two big SE transformers are used, since there is less DC saturation, the output power in deep bass can be improved. And if two small SE are used, they can together replace a bigger PP !
Cheers,
Alain.
re the Kludge Scheme:
Good to see the simulation looks positive!!
If the current in the secondaries is correctly set to null the DC magnetization from the xfmr primaries (equal ampere-turns primary and secondary, so need to know the tube idle DC currents), then no air gap is needed in the xfmrs. So one is using two small xfmrs to make a bigger center tapped one. The LV DC comp. supply could come from a center tapped 6.3 VAC filament winding on the power xfmr maybe. I do have to agree about the over-complexity issue though, since adjusting the tube idle currents (bias adj.) will require adjusting the DC current comps. (using the series Rlimit pots, and they have to balance each other too, to null out DC in the speaker)
Maybe some special applications for this? (Note: the scheme is not useful for SE Stereo since it requires the same, but inverted, AC signal in both xfmrs. Although one might try the idea of a difference channel to sum and sub from two identical channels.)
Having fiddled around with some filament like xfmrs to make an OT though already, for the $100 guitar amp challenge thread, I must say finding an xfmr with two 120 VAC primaries is much more desirable to make an OT. I found some 24 Watt, 50 Hz, 8 VAC xfmrs (for $ 4.50 surplus sourced) that had around 50 Hz to 7 KHz bandpass with an 8 Ohm load. Then by connecting two of them in a Crowhurst Twin Coupled arrangement, with cross coupled caps, the bandwidth increased to 50 Hz to 30 KHz. (note that the primary voltage rating effectively doubles too in this series'd Twin C config. as does also the eff. primary inductance double ) The big HF end increase is due to the leakage inductance of all four primary windings then becoming paralleled and the load on each xfmr is effectively upped to 16 Ohms. Result: a 48 Watt 50 Hz to 30 KHz OT combo for $ 9.00!! A serious breakthru there. (could also be re-rated to 12 Watt 25 Hz to 30 KHz, or 24 Watt 35 Hz to 30 KHz due to the doubled primary L) Unfortunately, the surplus vendor increased the price to $8.95 each after I ordered a bunch. Probably hadn't sold any of them for 20 years and realized the price was still un-inflated. Now he won't sell any more for another 20 years.
Good to see the simulation looks positive!!
If the current in the secondaries is correctly set to null the DC magnetization from the xfmr primaries (equal ampere-turns primary and secondary, so need to know the tube idle DC currents), then no air gap is needed in the xfmrs. So one is using two small xfmrs to make a bigger center tapped one. The LV DC comp. supply could come from a center tapped 6.3 VAC filament winding on the power xfmr maybe. I do have to agree about the over-complexity issue though, since adjusting the tube idle currents (bias adj.) will require adjusting the DC current comps. (using the series Rlimit pots, and they have to balance each other too, to null out DC in the speaker)
Maybe some special applications for this? (Note: the scheme is not useful for SE Stereo since it requires the same, but inverted, AC signal in both xfmrs. Although one might try the idea of a difference channel to sum and sub from two identical channels.)
Having fiddled around with some filament like xfmrs to make an OT though already, for the $100 guitar amp challenge thread, I must say finding an xfmr with two 120 VAC primaries is much more desirable to make an OT. I found some 24 Watt, 50 Hz, 8 VAC xfmrs (for $ 4.50 surplus sourced) that had around 50 Hz to 7 KHz bandpass with an 8 Ohm load. Then by connecting two of them in a Crowhurst Twin Coupled arrangement, with cross coupled caps, the bandwidth increased to 50 Hz to 30 KHz. (note that the primary voltage rating effectively doubles too in this series'd Twin C config. as does also the eff. primary inductance double ) The big HF end increase is due to the leakage inductance of all four primary windings then becoming paralleled and the load on each xfmr is effectively upped to 16 Ohms. Result: a 48 Watt 50 Hz to 30 KHz OT combo for $ 9.00!! A serious breakthru there. (could also be re-rated to 12 Watt 25 Hz to 30 KHz, or 24 Watt 35 Hz to 30 KHz due to the doubled primary L) Unfortunately, the surplus vendor increased the price to $8.95 each after I ordered a bunch. Probably hadn't sold any of them for 20 years and realized the price was still un-inflated. Now he won't sell any more for another 20 years.
Last edited:
Another note about using filament like xfmrs for making OTs.
From experience, I can tell you that you do NOT want to use split bobbin or quad bobbin low-profile (primary on one side and secondary on the other side) power xfmrs to make an OT with any scheme. The leakage inductance is way too high. Another problem comes up in some cheap industrial xfmrs where the dual primaries are near-bifilar wound. The capacitance between the windings is way too excessive. (This is not to say that split bobbin designs shouldn't be used for purpose built OTs, they do work great when properly designed with interleaves on each bobbin section)
So some L/C meter measurements and frequency pass tests are essential to finding a useful candidate xfmr. The toroid power xfmrs are also very likely to have way too much distributed (measure the common mode C) capacitance between windings.
From experience, I can tell you that you do NOT want to use split bobbin or quad bobbin low-profile (primary on one side and secondary on the other side) power xfmrs to make an OT with any scheme. The leakage inductance is way too high. Another problem comes up in some cheap industrial xfmrs where the dual primaries are near-bifilar wound. The capacitance between the windings is way too excessive. (This is not to say that split bobbin designs shouldn't be used for purpose built OTs, they do work great when properly designed with interleaves on each bobbin section)
So some L/C meter measurements and frequency pass tests are essential to finding a useful candidate xfmr. The toroid power xfmrs are also very likely to have way too much distributed (measure the common mode C) capacitance between windings.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.