Just wondering: if I took one of my open frame OPT's and did some surgery to take the UL tapped primary and split it up into 2 isolated primaries, would it be useable in a cathode feedback amp? I realize that since the UL tap is typically at 43%, the resulting "main" primary winding would have decreased resistance and inductance and probably be best suited to a different tube than the OPT was originally designed for. How about the newly split off primary, since it was a 43% tap would it have too much resistance or inductance to be useable for cathode feedback?
If there is some space on the bobbin available, I'd rather try to apply extra CFB windings, as these are typically on the order of 10% of the primary. Splitting a primary almost in the middle might not give usable CFB windings.
It's unclear if additional windings will work out, stabilitywise. All windings should be tightly coupled (and symmetrical also) which would not be achievable with this technique. For experimenting it might suffice...
- Klaus
It's unclear if additional windings will work out, stabilitywise. All windings should be tightly coupled (and symmetrical also) which would not be achievable with this technique. For experimenting it might suffice...
- Klaus
Putting another winding on top in leftover space will give very poor coupling, and CFB needs excellent coupling. The CFB windings should be layered between secondary winding layers.
Some Hammond 1650T xfmrs I have looked at, have the 40% UL windings between the secondary layers, so if you could isolate their connections to the rest of the primary, they might work for CFB. If you have an open frame xfmr, check and see where the windings are located in the layer buildup.
But why bother with all this, this problem was solved many moons ago:
http://www.diyaudio.com/forums/showthread.php?postid=228796#post228796
or try N. Crowhurst's "Twin Coupled Amplifier", which uses two output xfmrs. I have seen it around on the net. Two variants:
Radio Electronics Oct, 1960
Radio Electronics Nov, 1957
Don
Some Hammond 1650T xfmrs I have looked at, have the 40% UL windings between the secondary layers, so if you could isolate their connections to the rest of the primary, they might work for CFB. If you have an open frame xfmr, check and see where the windings are located in the layer buildup.
But why bother with all this, this problem was solved many moons ago:
http://www.diyaudio.com/forums/showthread.php?postid=228796#post228796
or try N. Crowhurst's "Twin Coupled Amplifier", which uses two output xfmrs. I have seen it around on the net. Two variants:
Radio Electronics Oct, 1960
Radio Electronics Nov, 1957
Don
I have done some experimenting with CFB. My first observation is that the UL winding will have too many turns to be useful for CFB. By adding CFB to an output stage you are taking a sample of the output signal and inserting it in series with the cathode. The voltage inserted into the cathode circuit is out of phase with the driving signal, thus increasing the drive requirements of the output stage by an amount equal to the voltage applied to the cathode. Now if you took 43% of the plate voltage (more since the entire plate winding in no longer used) and put it in the cathode circuit, the drive requirements would be huge (hundreds of volts P-P).
I have found that the speaker winding works out to be about the correct amount, depending on the output tube being used.
You did not state if the amp in question is SE or P-P. A P-P amp has the additional requirement of balance between the two primary halves and the two CFB winding halves over the entire audio range. Electrically the 4 ohm tap is the center tap of the 16 ohm speaker winding. You can ground the 4 ohm tap, connect the speaker between the 0 ohm tap (not grounded) and the 8 ohm tap. the 0 ohm and 16 ohm taps are connected to the cathodes. See the schematics for most Audio Research amplifiers for details. I have found that CFB can be used to improve the response (especially bass) of many SE OPT's. It often shows the best improvement on cheap OPT's, like the Hammond 125 series or the Edcor XSE series. CFB will often make the response of a low cost P-P transformer worse because the balance between the two halves varies ofer the audio range. This leads to some weird looking frequency response plots, and it doesn't sound good either.
In either case some experimentation is needed to find out what works for you.
For those people who have forged new ground by experimenting with mains (power) toroids as P-P OPT's, I offer this suggestion. Use the 230 VCT primary for your P-P primary. Use the low voltage secondaries (6 to 18 volts) for the CFB windings, and wind a new secondary over the top of of the existing windings for the speaker output. This will allow for swapping windings around to find the best combination. It works with some transformers, but not all.
I have found that the speaker winding works out to be about the correct amount, depending on the output tube being used.
You did not state if the amp in question is SE or P-P. A P-P amp has the additional requirement of balance between the two primary halves and the two CFB winding halves over the entire audio range. Electrically the 4 ohm tap is the center tap of the 16 ohm speaker winding. You can ground the 4 ohm tap, connect the speaker between the 0 ohm tap (not grounded) and the 8 ohm tap. the 0 ohm and 16 ohm taps are connected to the cathodes. See the schematics for most Audio Research amplifiers for details. I have found that CFB can be used to improve the response (especially bass) of many SE OPT's. It often shows the best improvement on cheap OPT's, like the Hammond 125 series or the Edcor XSE series. CFB will often make the response of a low cost P-P transformer worse because the balance between the two halves varies ofer the audio range. This leads to some weird looking frequency response plots, and it doesn't sound good either.
In either case some experimentation is needed to find out what works for you.
For those people who have forged new ground by experimenting with mains (power) toroids as P-P OPT's, I offer this suggestion. Use the 230 VCT primary for your P-P primary. Use the low voltage secondaries (6 to 18 volts) for the CFB windings, and wind a new secondary over the top of of the existing windings for the speaker output. This will allow for swapping windings around to find the best combination. It works with some transformers, but not all.
Thanks for the replies.
Both the dual supply and CT choke solutions look tougher to execute to me
I don't know, where am I going to find equal resistance PP OPT's?
I was actually thinking about the XSE's, even though I wanted a general answer. How did you implement CFB with the XSE's, they only have one non-tapped secondary?
smoking-amp said:
Both the dual supply and CT choke solutions look tougher to execute to me
smoking-amp said:
I don't know, where am I going to find equal resistance PP OPT's?
tubelab.com said:
I was actually thinking about the XSE's, even though I wanted a general answer. How did you implement CFB with the XSE's, they only have one non-tapped secondary?
Well, for equal resistance, just put a series compensating resistor in the lower R primary side.
For dual supplies, I use two dual bobbin 120V/240V industrial xfmrs, like Signal TGC175-230. Hammond, Magnatec .... they all sell similar ones, all cheapies made in China.
For a center tapped inductor, I found some surplus quad bobbin low profile 120V/120V/35V/35V transformers. I put two units in series to get dual 310V windings.
If you are contemplating actually buying a new output xfmr for this, why not just call up Edcor and tell them you want a CFB version of one of their output xfmrs. I did, model CXPP100-SP1, has a dual 20% CFB winding instead of 40% UL sections, similar price even.
I should point out, however, that for class AB, it is best to use the Elliptron circuit, since it uses both "CFB sections" for each tube. Giving much better close coupling to the secondary and primaries when only one tube is in conduction.
Don
For dual supplies, I use two dual bobbin 120V/240V industrial xfmrs, like Signal TGC175-230. Hammond, Magnatec .... they all sell similar ones, all cheapies made in China.
For a center tapped inductor, I found some surplus quad bobbin low profile 120V/120V/35V/35V transformers. I put two units in series to get dual 310V windings.
If you are contemplating actually buying a new output xfmr for this, why not just call up Edcor and tell them you want a CFB version of one of their output xfmrs. I did, model CXPP100-SP1, has a dual 20% CFB winding instead of 40% UL sections, similar price even.
I should point out, however, that for class AB, it is best to use the Elliptron circuit, since it uses both "CFB sections" for each tube. Giving much better close coupling to the secondary and primaries when only one tube is in conduction.
Don
The simplest solution is to just wire the secondary (with speaker attached) in series with the cathode. Since the tubes cathode current flows through the transformer secondary a few millivolts will be developed across the secondary, which will be applied to the speaker. Most speakers don't mind this, but some high efficiency drivers may not.
There is another solution. If the output stage uses cathode bias, route the cathode bypass capacitor to the speaker hot lead (and one side of the OPT secondary). The speaker common is grounded. Reverse the transformer secondary connections as needed to achieve negative feedback. Try both ways and choose the connection that gives the lowest gain. See the SimpleSE schematic for an example.
http://www.tubelab.com/AssemblyManualSimpleSE/schematic_SSE.htm
The XSE transformer allows only one level of feedback since there is only one tap. It turns out to be 3 to 6 db of feedback depending on the output tubes characteristics and whether UL mode is used.
There is another solution. If the output stage uses cathode bias, route the cathode bypass capacitor to the speaker hot lead (and one side of the OPT secondary). The speaker common is grounded. Reverse the transformer secondary connections as needed to achieve negative feedback. Try both ways and choose the connection that gives the lowest gain. See the SimpleSE schematic for an example.
http://www.tubelab.com/AssemblyManualSimpleSE/schematic_SSE.htm
The XSE transformer allows only one level of feedback since there is only one tap. It turns out to be 3 to 6 db of feedback depending on the output tubes characteristics and whether UL mode is used.
"Does twin coupling work for SE as well?"
Well, you could kinda force it to work using two SE air gapped xfmrs and connecting secondaries in parallel, splitting the primaries between plate and CFB windings, but there's not really much point in doing all that. Or you could use two conventional P-P xfmrs in a twin coupling design with a current source in place of one tube.
If you have a SE air gapped xfmr with a UL tap, then you could ground the UL tap and use the 40% part for CFB. (however, winding capacitance to the steel core may not be optimal if the original B+ end was wound closest to the core) Or as Tubelab has suggested, just use the secondary winding for the CFB part. (similar capacitance considerations apply here too for AC grounds)
Another SE CFB approach would be to use a normal P-P xfmr with its center tap (or a UL tap) grounded and one primary side for the plate winding and the other side (or it's UL tap) for the CFB winding. (floating B+ supply in series with the plate) Then you have to put a current source through part of the primary to ground, to neutralize the magnetization in the core. (current level adjusted to make up for the turns ratio between the tube circuit and CCS circuit)
There are OTHER ways to do local feedback besides CFB which don't require special windings. Search on Partial FDBK or Shunt FDBK or Plate FDBK.
Don
Well, you could kinda force it to work using two SE air gapped xfmrs and connecting secondaries in parallel, splitting the primaries between plate and CFB windings, but there's not really much point in doing all that. Or you could use two conventional P-P xfmrs in a twin coupling design with a current source in place of one tube.
If you have a SE air gapped xfmr with a UL tap, then you could ground the UL tap and use the 40% part for CFB. (however, winding capacitance to the steel core may not be optimal if the original B+ end was wound closest to the core) Or as Tubelab has suggested, just use the secondary winding for the CFB part. (similar capacitance considerations apply here too for AC grounds)
Another SE CFB approach would be to use a normal P-P xfmr with its center tap (or a UL tap) grounded and one primary side for the plate winding and the other side (or it's UL tap) for the CFB winding. (floating B+ supply in series with the plate) Then you have to put a current source through part of the primary to ground, to neutralize the magnetization in the core. (current level adjusted to make up for the turns ratio between the tube circuit and CCS circuit)
There are OTHER ways to do local feedback besides CFB which don't require special windings. Search on Partial FDBK or Shunt FDBK or Plate FDBK.
Don
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.