I'm planning out a rather hi-fi SE guitar power amp with dual KT120. I want to build it capable of 60 watts but bias it class "A" for about 25 watts output.
If I use a really big 75-watt output transformer, they usually lose some of the highest treble. Smaller ones lose some of the lowest bass. Wouldn't usually matter, but this time I want to try broader bandwidth, hi-fi type transformers.
So I'd like one output transformer for each tube.
The question is, can I connect the secondary windings from the two transformers (in either series or parallel) so that they only need one cable to one speaker cabinet? I can imagine possible problems with back-emf etc.
If I use a really big 75-watt output transformer, they usually lose some of the highest treble. Smaller ones lose some of the lowest bass. Wouldn't usually matter, but this time I want to try broader bandwidth, hi-fi type transformers.
So I'd like one output transformer for each tube.
The question is, can I connect the secondary windings from the two transformers (in either series or parallel) so that they only need one cable to one speaker cabinet? I can imagine possible problems with back-emf etc.
Last edited:
Yes and I think the easiest way would be parallel. Remember if you use an 8 ohm load use the 16 ohm taps on the transformers.
I have paralleled both channels on my Tubelab SSE's for guitar amp use several times. It works good and nothing bad happens. As stated the load impedance gets halved. In my case I stick a 4 ohm cabinet on the paralleled 8 ohm channels.
I have been using the big Edcor CXSE's with KT88's in pentode mode. I get 15 watts per channel and 30 watts from the paralleled pair. Playing a 4 string bass guitar through this works great as does guitar and keyboards.
I have been using the big Edcor CXSE's with KT88's in pentode mode. I get 15 watts per channel and 30 watts from the paralleled pair. Playing a 4 string bass guitar through this works great as does guitar and keyboards.
Tubelab, I remember you posting long ago about a huge Hammond SE OT that lost some extreme treble, probably due to its size. Some of those problems are due to things like small degrees of magnetization of the big core, and exaggeration of the hysteresis loop. But I wonder whether some of the problem would diminish with a lesser turns ratio, especially with the high voltages you work with. Your big SE amps are not portable anyway. So of course my idea is inefficient in many ways, but...if you used your thousand-volt huge triodes or beam tubes or sweep tubes or enormous transmitter tubes or whatever...and had a big SE transformer made with a magnetic gap but wound with only maybe a 1.5:1 winding ratio, that would block the DC but perhaps have better bandwidth than a Hammond 1642SE. Then you could take the output from that transformer's secondary (which is now at a normal level) and run it into the primary (ingoring the center tap) of one of those fabulous efficient high-power push/pull Plitron toroidial output transformers you have, without any problems with DC saturation. Though going thru two transformers, and despite twice the cost, nearly twice the weight and space...I wonder how it would perform.
It might be a dumb idea, because I don't really know enough about how/why an oversize transformer has high-frequency rolloff.
It might be a dumb idea, because I don't really know enough about how/why an oversize transformer has high-frequency rolloff.
If I knew how to make my own capacitors for cheap I might try that myself. But caps that big would be EXPENSIVE if you want full range.
Last edited:
The OPT in question was the Hammond 1628SE. That piece of iron had design issues from the start that made it rather useless as a full range OPT if the secondaries were configured for normal 8 ohm operation. Hammond created a newer version the 1628SEA which is what I ordered, but AES shipped the old versions. AES offered to exchange them if I paid to ship them back. The shipping cost was $54, so complained. Their second offer was to sell me a pair of the new ones at their cost and split the shipping, which I wound up doing. I eventually sold the old versions at a loss after I couldn't make them work right in any circuit without a ton of feedback.
All transformers are imperfect devices. There are several engineering factors that determine the quality of an OPT, but the there are 4 main factors that we can look at. These are simplified explanations since these factors all interact, and can not really be looked at independently without some serious math.
First is core size. It must be large enough to support all the magnetic energy needed to transfer all the audio from the primary to the secondary. The lower frequencies tend to transfer more energy through the iron core than the highs, so the core must be large enough to support the lowest frequencies at the highest expected power level without saturation.
Second is the primary inductance. This appears to the tube as an inductance in parallel with an ideal (theoretically perfect) transformer. An inductance acts as a resistance whose value decreases as the audio frequency decreases. Thus, if the inductance isn't large enough some of the low frequency energy will be shunted past the transformer, and the load that the tube sees will appear lower than it is at higher frequencies, causing distortion at low frequencies even if the core is large enough to avoid saturation. This can be measured by connecting an inductance meter to the primary with the secondary not connected to anything.
Third is leakage inductance. If the transformer is perfect, all of the energy applied to the primary will be transferred to the secondary equally at all frequencies. The impedance ratio would be the square of the turns ratio at all frequencies. The primary inductance would appear to be zero if the secondary was shorted, but this is not the case. Some of the energy applied to the primary never makes it to the secondary for many reasons. One of these reasons is called leakage inductance. If you measure the inductance of the primary with the secondary shorted you will get a number that we call the leakage inductance. This appears to the tube to be an inductance in series with the primary.
Fourth in distributed interwinding capacitance. Each turn of wire in the coil has some capacitance between it and all the other metal in the transformer. This all adds up to be an amount of capacitance that appears to the tube as a capacitor in parallel with the transformers primary.
So we have a inductor in series with the primary, and some capacitance to ground. This creates a series resonant circuit, which acts to short the transformer primary at the resonant frequency. Good winding techniques can coerce this resonance well above the audio band, but this imposes an upper limit on the size of the transformer. A 5 pound transformer with no interleaving whatsoever may have the resonant notch at 25 KHz, while an excellent quality transformer of the same size may have the notch at 75 KHz. A poorly designed 12 pound OPT like the 1628SE has the notch at 16 KHz.
Ah. I knew some of that, but not the hierarchy of priority of impact.
The 1642 or 1628? The 1642 seems simply amazing at 28 pounds, but must be subject to all those limitations of big transformers (but not saturation). So, do you have any idea what's structurally different about the SE versus SEA in their model designations? How they interleave? Size of the gap? I'd noticed that Digi-Key has great prices and listed a 1642SE and 1642SEA, but the Hammond catalog only mentions the 1642SE. It also doesn't list a 1628SE, only a 1628SEA.
His parafeed comment was interesting, and I just always thought of that as capacitor-coupling the transformer, but learning a bit I see it's also common to use a choke to provide the DC to the plate with parafeed. Which explains why I've seen 'Parafeed" output transformers, which I assume are various ways to incorporate the choke and the output transformer in one? I'll see if there's a Wikipedia page or better ref, as I imagine many complications possible.
So the parafeed delivers the B supply to the plate thru a choke. That seems like a good thing, not just for keeping the supply ripple and sag and modulation out of the output and giving the tube a stiff well-regulated clean supply, but also keeping the tube from modulating the supply rail. Are such plate chokes ever used on traditional transformer-coupled amps, much like local caps at each tube might sometimes be used? It sounds worthwhile if you're hanging a lot of different tube sections off of one power supply.
I've heard of using chokes on the fixed-bias supply too, and that also seems like a good idea.
It really seems to me a fully damped audio-band tank circuit at each plate and another of smaller components at the grid would really clean things up in any amp...tube, transistor, SE, PP...but especially in the SE because supply ripple does not cancel. Or alternatively it might be ways to make a cheaper smaller less than optimal supply to the plate that still improves on the basic resistor.
The common design filters better as it drops voltage, and the most sensitive first preamp stage is decently filtered, which is great because noise injected there is amplified so much. But in real use, sometimes you turn down the master volume and the preamp noise goes away, leaving you with the power amp noise predominant.
Ideas like these, or using big-bottle preamp tubes, are driving me to try another scratch-build instead of modifying old amps, where you run out of space.
The 1642 or 1628? The 1642 seems simply amazing at 28 pounds, but must be subject to all those limitations of big transformers (but not saturation). So, do you have any idea what's structurally different about the SE versus SEA in their model designations? How they interleave? Size of the gap? I'd noticed that Digi-Key has great prices and listed a 1642SE and 1642SEA, but the Hammond catalog only mentions the 1642SE. It also doesn't list a 1628SE, only a 1628SEA.
His parafeed comment was interesting, and I just always thought of that as capacitor-coupling the transformer, but learning a bit I see it's also common to use a choke to provide the DC to the plate with parafeed. Which explains why I've seen 'Parafeed" output transformers, which I assume are various ways to incorporate the choke and the output transformer in one? I'll see if there's a Wikipedia page or better ref, as I imagine many complications possible.
So the parafeed delivers the B supply to the plate thru a choke. That seems like a good thing, not just for keeping the supply ripple and sag and modulation out of the output and giving the tube a stiff well-regulated clean supply, but also keeping the tube from modulating the supply rail. Are such plate chokes ever used on traditional transformer-coupled amps, much like local caps at each tube might sometimes be used? It sounds worthwhile if you're hanging a lot of different tube sections off of one power supply.
I've heard of using chokes on the fixed-bias supply too, and that also seems like a good idea.
It really seems to me a fully damped audio-band tank circuit at each plate and another of smaller components at the grid would really clean things up in any amp...tube, transistor, SE, PP...but especially in the SE because supply ripple does not cancel. Or alternatively it might be ways to make a cheaper smaller less than optimal supply to the plate that still improves on the basic resistor.
The common design filters better as it drops voltage, and the most sensitive first preamp stage is decently filtered, which is great because noise injected there is amplified so much. But in real use, sometimes you turn down the master volume and the preamp noise goes away, leaving you with the power amp noise predominant.
Ideas like these, or using big-bottle preamp tubes, are driving me to try another scratch-build instead of modifying old amps, where you run out of space.
It seems to me that the proper reactive network at the plate supply would also be more efficient than a resistor burning off the power.
Or, a damping resistor in that individual plate supply network there might handle enough of the power to get away with smaller choke and cap, at least in voltage and current handling if not inductance and capacitance.
From an opposed viewpoint, when I see a lot of paralleled output tubes, why are there separate plate supply resistors and separate bias resistors for each? Sometimes transistors do that, other times not (transistors more literally just paralleled). Are the tubes just more mismatched? Or is it in case of a tube failure? I guess several small resistors don't cost much more than 1 big one...so there's no reason not to use individual ones...
Or, a damping resistor in that individual plate supply network there might handle enough of the power to get away with smaller choke and cap, at least in voltage and current handling if not inductance and capacitance.
From an opposed viewpoint, when I see a lot of paralleled output tubes, why are there separate plate supply resistors and separate bias resistors for each? Sometimes transistors do that, other times not (transistors more literally just paralleled). Are the tubes just more mismatched? Or is it in case of a tube failure? I guess several small resistors don't cost much more than 1 big one...so there's no reason not to use individual ones...
The older SE models had two different secondary windings distributed throughout the primary. You had to interconnect them differently to get each different secondary impedance making a Marshall style speaker impedance switch difficult. The SEA models had a single tapped winding like just about every other transformer made. The SEA models were supposed to have improved specs.
I got a pair of the 1628SEA's and found them better than the 1628SE, but still not 20Hz to 20KHz +/- 1db as their specs indicate. I tested the transformers in a SSE type amplifier using a single KT88 tube with no feedback. This is a real world application. I still have them in a box somewhere. I will try them in some sort of feedback design in the future. The usual high frequency notch would not be as deep if the transformer was driven from a lower impedance source such as a tube with negative feedback applied.
I have no experience with the 1642. I asked about it on this forum somewhere when I was looking for a transformer to use with the 833A tube. I seems that nobody had tested them for frequency response in a real amplifier.
All of my observations were based on the design of a high fidelity tube amp for listening to music, and possible resale. This means response to 20KHz. A guitar amp doesn't need to go that high. All of the anomalies I reported in these OPT's involve the frequencies from 15KHz up. Any of these OPT's will work up to 15KHz which is far beyond the reach of the most harmonically rich guitar + effects system that I could think of.
A parafeed amp eliminates the DC through the OPT which allows the use of a much smaller core for a given power level, but it substitutes two more problems for the one it solves. Assuming a perfect parafeed capacitor, the choke and the OPT primary are effectively in parallel. Inductors in parallel work like resistors in parallel, the inductance value drops according to the same formula, so you need more inductance in each coil. Second, capacitors are not perfect. You, don't however need huge capacitance values since you are working at 5000 ohms, not 8. I have seen good results with a 30 uF motor run cap from an air conditioner motor. It was probably overkill, but I had it.
Parafeed with a resistor wastes some DC, but the resistance is effectively in parallel with the OPT so it eats up to half of your AUDIO power too! You can feed the plate of the output tube with a CCS circuit. It is a very high impedance so it will not cost you any audio power, but it will require a much higher B+ voltage. This is not a big deal in a low powered amp, but may be a showstopper in anything bigger than 10 watts.
I don't know what you have in mind for your final design, but I once owned a rather novel guitar amp. Mine was a Guild Ultraflex, but I have seen the same amp branded Magnatone, and Panaramic. It had two separate push pull power amps of about 30 watts each. One amp had a small OPT and drove a 8 inch speaker and the other amp had a big OPT and drove a 12 or 15 inch speaker. The "Ultraflex" knob was the "balance" control between the two amps. It made the best blues tones that I ever heard.
Thank you for your time and knowledge and experience.
I'm not sure what I'm going to build either, but my parts box sure is getting interesting. I seem to have a lot of chokes of various sizes, including one really monstrous one. A KT120 that looks lonely LOL.
I'm not sure what I'm going to build either, but my parts box sure is getting interesting. I seem to have a lot of chokes of various sizes, including one really monstrous one. A KT120 that looks lonely LOL.
Member
Joined 2009
Paid Member
Excellent post. I thought I knew all this, but your explanation actually clarified some fuzzy areas for me.
- Status
- Not open for further replies.