KX-100/X-100B -- Fitting A New Output Transformer

dcgillespie

Fisher SA-100 Clone
Subscriber
Anyone who has lurked around the AK Fisher Forum for very long has probably come across threads by Fisher Folks who either bought a KX-100 or X-100B with one or both OPTs bad, or had an otherwise good one die on them while in service. This usually prompts calls for help either in fitting or trying to locate a suitable aftermarket replace piece, or in obtaining a good original piece from a parts donor unit somebody has on their shelf. It's a story that's repeated itself seemingly a number of times here on AK. And so, my own brush with this issue has now come to life.

Some years ago, I had restored a KX-100 for a client, where in that process, I had also modified the unit with many of the modifications/recommendations I've posted on this forum for Fisher equipment. For the purposes of this thread, the only modification that is relevant from that effort was the installation of a separate DC Heater/Bias supply that allowed the unit to have its output stage converted to fixed bias operation, with the end result that output tube quiescent current could then be reduced to sane levels that would benefit the life of the tubes and all three transformers, as well as providing significant performance improvements as well. Just recently then, he has sent this unit back to me - still in perfect working order -- asking only for me to install the EFB™ modification into it as well.

That process begins with running a series of tests, to determine the optimum parameters to set the EFB Screen Grid and Control Grid Regulators at for optimum performance of the unit. One of those tests involves driving both channels to the onset of clipping at the same time, to gauge power supply performance under that condition. And so the test began, where above about 15 watts of power, the left channel scope display started to scatter and display added trash to the sine wave that was starting to partially break up as well. I backed the drive down, checked all connections, turned up the drive again, where upon it did the same thing. More testing showed that the drive signal to the output tubes was clean on both sides of the push-pull connection, so the next hope was that it was the tubes or their sockets causing the issue.

I swapped the power tubes between the channels, and gave it another whirl. This time, the left channel rose right up to the onset of clipping without incident, which was a relief. The trouble is, so did the right channel as well. That seemed to eliminate the tubes, leaving (hopefully) just the sockets. After a good cleaning, the tubes were reinstalled in their original locations, and the test run again. This time, the left channel again rose cleanly right up to the onset of clipping, so at that point, the previous scattered display was (cautiously) written off to a dirty tube socket, which I informed my client of as the only thing the unit apparently suffered from over the time since I last had the unit on my bench.

The rest of the testing was then finished uneventfully, parameters determined, and the EFB circuits installed. At that point, more testing then began to determine if the new circuits were performing as intended, and the expected increase in performance was achieved. Unfortunately, the same hash/trash appeared at the same power level, in the same channel, once again. So as before, the power tubes were swapped. But unlike before, the same hash/trash still appeared at the same power level with now cleaned (and well tensioned) sockets, in the same channel, using different tubes. So now the drive, the tubes, and the sockets were all positively eliminated. Uh oh.......

What I was witnessing was an original KX-100 OPT going through the process of passing from this world to the next. Below 15 watts of power, there were no apparent problems. Clearly, my client had never pushed the amplifier beyond this power level, as he had no complaints as to its operation at all -- in fact, he loved it. He just wanted EFB installed. Whenever the hash/trash appeared, I would always back off on the drive, not wanting to hasten the transformer's demise, but at this point, there was no doubt -- it was only a matter of time before the transformer would die completely. Of course, when that happens, it often does damage to the tubes involved, as well as potentially damage to other components as well, so I made the recommendation that the transformer(s) get replaced. Hardly what my client was expecting from a request to simply install EFB into a perfectly operating amplifier, and yet, unfortunately, far too common an event for owners of X-100B and KX-100 amplifiers. I have to believe that had the biasing system still been stock, with the output tubes running at much higher quiescent current levels, at or beyond their dissipation limits, and everything getting cooked in the amplifier at much higher temperatures, that this transformer would have been toast much sooner.

In the discussions that followed with my client, I expressed my concern with using the common go-to transformer manufacturers/re-winders, as near direct experience with the go-to re-winder when rewinding OPTs yielded basically incompetent results, and otherwise only scant claims of great work performed (but with no published performance/evidence), or, with with the go-to new manufacturer, direct experience (and observed performance published by others) has shown that performance is well off the mark of the high quality high fidelity pieces of yesteryear -- including those of the KX-100/X-100B as well. Alternately, finding a good original replacement is always possible given enough time, but of course then the same fate is likely inevitable to occur at some point.

With the recent apparent shuttering of Transcendar, I was always disappointed that I never got a chance to try out some of their products. They certainly look to be quality devices (but then so do other's products), and have built up their followers who praise them (as have others) -- but again, I am not aware of any published performance data as a result of using their products in vintage designs. However, based on their description as to how their transformers are built, it offered hope. And, there have been sputtering signs of life at Transcendar lately, if only to (possibly) sell off some of their remaining inventory. With that thought, I decided to check on line where at auction, as luck would have it, Transcendar was offering for sale a few direct replacement OPTs specifically for the KX-100/X-100B (isn't that telling). I immediately informed my client, where upon the decision was made to purchase two of them. In spite of an indicated March delivery, the two transformers arrived in one short week's worth of time.

With that, new tests were then run to determine just how good the new transformers were, how well they mimicked the original pieces, and how well they could perform in the KX-100/X-100B design.

I'll get into all of that next time.

Dave
 
Is the failure likely inside the windings or could it be one of the connections of an external wire to the winding? If the latter, it might be worth opening the transformer and inspecting those connections.
 
I took a quick peak inside, Fred. Nothing visually apparent. At this point, the transformer consistently passes power levels of at least 10 watts cleanly before the shenanigans begin -- which I don't think would be consistent with a loose connection, which I know has occurred in some other transformers. This thing gives all the indications of internally arcing over when high AC voltages are superimposed on top of the B+ voltage. The transformer also still shows good correlation to the good one with respect to the DCR of the windings, and infinite resistance between the windings, and to the case. It would seem then that what ever is happening is in the beginning stages, so that nothing appears until the transformer's insulation is stressed with the application of a high level signal. Of course, once breakdown begins, it grows over time like a deadly disease. As well, I did not want to tear into the transformer until I know of my clients intentions for it!

Dave
 
COMPARISON

I. Freestanding Tests:

DC Resistance and Reflected Impedance

A. Primary DCR (red/blue-yellow / red/blue) followed by any relevant notes:
a. Transcendar: 142Ω / 150Ω Very close DCR between the two primary halves. Both transformers measured identically.
b. Original/Good: 145Ω / 165Ω Typical DCR spread seen between the primary halves in vintage transformers of this size.
c. Original/Bad: 142Ω / 157Ω Closer DCR spread than the original good unit, but hardly enough difference to indicate anything bad brewing on the inside.

B. Secondary DCR (Common to 4Ω, 8Ω, and 16Ω) followed by any relevant notes:
a Transcendar: 0.29Ω / 0.49Ω / 0.74Ω Notably lower DCR than produced by typical vintage transformers of this size. Both transformers measured identically.
b. Original/Good: 0.32Ω / 0.66Ω / 1.00Ω Very typical DCR produced by typical vintage transformers of this size.
c. Original/Bad: 0.31Ω / 0.67Ω / 1.01Ω Insignificant difference to the original good transformer.
Readings based on meter lead resistance adjusted for 0.00Ω.

Comment: The DCR readings of the Transcendar transformers are certainly quite close to the original, but notable for the Transcendars is the close DCR for each half of the primary winding, and lower DCR for the secondary winding, providing for greater transformer efficiency and lower distortion.

C. Reflected Impedance (Common to 4Ω, 8Ω, and 16Ω, measured at 60 Hz, unloaded) followed by any relevant notes:
a. Transcendar: 6719Ω / 6530Ω / 6530Ω The close agreement between the 8Ω and 16Ω reflected impedance is extraordinary. Both transformers measured the same.
b. Original/Good: 6556Ω / 5940Ω / 6747Ω The difference shown at the 8Ω tap is typical of vintage transformers. Can be higher or lower, but invariably one way or the other.
c. Original/Bad: Identical to the original good transformer.

Comment: Transcendar specifies their transformer as having a 6400Ω primary impedance. The results achieved by measurement are close enough, with the small difference shown no doubt due to a difference in measurement method. No specification is known for the Fisher transformers, but the reflected impedance (based on full secondary winding) is quite close.


II. Installed Tests:

These are best shown by a series of scope shots with attending comments. As I customarily do, only one channel is modified first, so that unless specifically indicated, the upper (Channel A) trace is the Transcendar transformer, while the lower (Channel B) trace is the good Fisher transformer. All traces are generated with the generator directly driving the power amplifier section.

Below: 10 kHz square waves. Hardly bad installed into the original design without modification. However, the Transcendar transformer showed a rising response (never good for stability) peaking at +2 db @ 50 kHz, while response for the Fisher transformer was down 0.5 db @ 50 kHz. The Fisher transformer showed perfect manners during HF stability testing, while the Transcendar transformer -- still remaining stable -- had far less margin. Note the exaggerated peak, and reduced rise time.
SAM_2646.JPG

Below: 10 kHz square waves. After modification, HF transient response for the channel with the Transcendar transformer now closely mimics that with the Fisher transformer. Response of the Transcendar channel is now down .5 db at 50 kHz as well, with a smooth roll off and no peaking at higher frequencies. All succeeding scope shots of the Transcendar channel include the modified response network.
SAM_2651.JPG

Below: 10 kHz square waves. To show just how closely the frequency, transient, and rise time of the Transcendar channel now matches the performance of of that with the original Fisher transformer, the two wave forms have been increased in amplitude, and placed on top of each other. Agreement is extremely close.
SAM_2655.JPG

Below: 10 kHz square waves. Unloaded, the Transcendar channel shows far less change in transient response, looking much like it did when loaded before the response was modified for it. Both channels are extremely stable.
SAM_2652.JPG

Below: 10 kHz square waves: With a .1 uF cap only load applied (the most reactive value for both transformers), the Transcendar transformer is exposed as having an additional (higher) resonant peak that the Fisher transformer is void of, but for all intents and purposes, the channel still displays the same excellent HF stability that the Fisher transformer channel does.
SAM_2654.JPG

Below: 10 kHz square waves. Satisfied with the first Transcendar's supersonic performance, the second Transcendar transformer was installed, with the exact same values in the modified HF response network. It is notable that the two transformers produce such identical performance with the same response tailoring networks.
SAM_2659.JPG

Below: Once again, the two images are superimposed on top of each other, showing the matched supersonic response of the two channels now with both sporting the Transcendar transformers. It is also well worth noting that the waveform shown -- other than in amplitude -- presents identically at each output tap.
SAM_2658.JPG

You may wonder why all the attention to a range of frequencies that we can't hear. In a feedback amplifier, it is vitally important, as anyone who has worked with the design of NFB amplifiers knows. How the amplifier responds to various loading and transient signal conditions in the supersonic region can have a profound impact on the frequencies we can in fact hear. Attention to this region then ensures that nothing going on up there will impact the audio frequency range, which of course is the primary frequency range of interest. When changing output transformers for those built without the original winding specifications, this is the region where the most trouble can start. With proper attention, any potential trouble can be eliminated, and with capable transformers, performance be made to closely match that of the original pieces.


Below: Back down in the audio frequency realm, with the Transcendar transformer still shown in the upper trace and the Fisher in the lower, full power 1 kHz sine waves are indistinguishable, with clipping commencing at the same power level. These wave forms represent a power output of 33 watts RMS each (with both channels obviously driven at the same time), which is significantly more than the stock amplifier is capable of, thanks to the added DC Heater/Bias Supply, and now the addition of EFB. Distortion in the Transcendar channel is .38%, while in the Fisher channel, THD = .48%. By contrast, the stock amplifier can only produce 20 watts per channel with both channels driven, producing over 1% THD at that power level.
SAM_2660.JPG

Below: Of significant note is power output at 25 Hz -- and here again at 33 watts RMS! Because my scope is not a digital storage scope, the trace is hard to see due to the timing of the camera shutter versus the scope's time setting. Looking closely however, you can see that both channels are clearly producing elevated distortion due to the transformers starting to saturate, but you can also see that the Transcendar is hanging in there every step of the way with the Fisher transformer, if not even staying a step ahead. Waveform deformity is virtually entirely due to the transformers themselves, as at this point, the amplifiers are just below the onset of clipping.
SAM_2656.JPG

Below: Where the rubber really meets the road: You always hear everybody comment on how big or how nice the iron looks. But for the vast majority of the audio spectrum (above ~1 kHz), the iron does nothing at all, and can even detract from performance. At frequencies above 1 kHz, its all about the interleaving and how well the coil form is wound. It's at these frequencies (as well as at supersonic frequencies) where a real high fidelity transformer can leave hifi "looking" transformers in the dust. The waveforms shown here are (again) produced at 33 watts RMS, but this time, at 20 kHz. Note the lack of distress that imitation transformers so often show at this frequency and power level. Again, the Transcendar transformer is doing just fine against the original Fisher device. At this point, THD is still well below 1% for the modified amplifier in either channel, with both channels producing about .80% THD.
SAM_2657.JPG


So the Transcendar TT-433-OT, with slight circuit adjustment, proves to be the real thing, displaying all the performance that the original Fisher transformers did. For this project, though unfortunate in terms of need, they were a life saver. As available, I whole heartedly recommend these transformers to those needing a replacement for their KX-100/X-100B amplifiers, where upon they can expect every bit of the original performance that the amplifier is capable of.

It is to Transcendar's credit that these transformers perform as well as they do, placing them (in my book) on basically the same level as the superb Dynaclone transformers that have been made available in recent years as well. For all the performance they pack, their price of $80 each would seem to be quite reasonable as well.

I've run out of room for any more pics in this post, so in my next post, I'll finish up with a pic of the new transformers installed, and a peak under the hood.

Dave
 
Last edited:
As promised, topside and bottom side pics of the finished amplifier.

Below: Not quite the flat black appearance of the original Fisher transformers, but looking fine enough, and performing every bit as good as the originals.
SAM_2661.JPG

Below: Underside, with the lead dress of the new transformers mimicking that of the original transformers to maintain the appearance of the original build.
SAM_2662.JPG

Dave
 
KX-200 owner here...interested in this thread (I know the 100 is probably quite different than the 200).

So your o-scope methodology is to scope a 10khz signal (or any frequency) while connected to the secondary outputs (L and R) of the output transformer and observe the curves, ideally looking for L and R to be identical? Into which KX-100 input do you play the signal? Are there adjustments one could make in the circuitry to get the signals to be closer to each other? I guess it depends on what is being measured.
 
Thorn - A 10 kHz square wave is used, because to faithfully display a 10 kHz square wave, the circuit must be able to pass up to 10X and down to 1/10X the fundamental frequency of the square wave, without any significant frequency discrimination within that band. Some really good audio transformers have a relatively flat response up to 100 kHz, but can also have resonant frequencies below and above that frequency as well. Since any distortion of frequency response shows up as a deformity of the waveform, a 10 kHz square wave then makes for the perfect analytical tool for checking supersonic response in a vacuum tube power amplifier. If you use a higher frequency, you end up inherently distorting the wave since you are asking the transformer to pass frequencies that it cannot without attenuation or phase shift. Any lower, and you might not catch the resonant frequency of the transformer. For the purposes of this exercise, a 10 kHz square waveform is the standard go-to test frequency.

The supersonic display does not need to be identical between the channels -- although ideally it will be. The display is primarily used to ensure that complete stability is achieved under all possible loading conditions -- this because an 8Ω speaker is not 8Ω at all frequencies. At some frequencies, it can look like nearly a dead short, while at others, it can appear as an open circuit. Therefore, the amplifier must be able to remain stable under all of these conceivable conditions. If it becomes unstable at any time, the amplifier can then consume power as it becomes an oscillator (usually ultrasonic) that then does in fact affect what you hear. It can also do great damage to the tweeter drivers in your speakers. Therefore, the stability of a NFB amplifier is hugely important. Quality NFB amplifiers are always designed around the characteristics of the output transformer. The circuit and transformer are married in many ways so that there is always the full power capability of the amplifier available at any frequency and loading condition, so that you don't hear the amplifier, you hear the music.

In the case of this project, the OPT was being changed out for one that was claimed by the manufacturer to be a clone of the original piece. But to my knowledge, none had ever been installed and tested to know that for a fact. The effort here then was not only to fit the new transformers for optimum performance, but also to assess just how close they represented the original piece.

The fitting of a new output transformer to an established circuit is the subject of a great many articles and books from back in the day. Throughout the 50s, many of the major manufacturers (including Fisher) got the problem of properly stabilizing NFB amplifiers wrong. It really wasn't until David Hafler came along and turned the industry on its ear that it started getting addressed properly. Just take a look at my thread on the Fisher 50A amplifier and you'll see how Fisher struggled mightily to produce good performance and stability in that amplifier both at the same time!

Dave
 
So I guess time will tell how well your new KX-100 replacement transformer matches up to the "days of yore." Looking forward to it. Thorne

PS - Come to think of it, I'm quite curious about Negative Feedback (NFB) loops. I have it on my Fender guitar amp. I installed a miniature, inline "piano-style" on/off switch so I can have it both ways. I added a bias pot so I can play with bias (so far in the 70's percent of max is what I like). From what I've read NFB is about reducing noise in the audio signal. I'd like to understand how the properties of the resistors and capacitors operate in an NFB. If you don't mind a few words... Thanks.
 
Last edited:
NFB is a cure-all for many things. It reduces gain, but in return, it reduces distortion, flattens and extends frequency response, reduces output impedance, and stabilizes performance from one unit to the next. For high fidelity amplifiers, it is just the ticket to neutralize the effects of individual component variations, and produce a flat response presentation. It is often not used in guitar amps, but when it is, it is used sparingly. Leo applied about 5-6 db to his designs (2X reduction) that both acts to protect the output transformer if the amplifier ever gets driven without a load, and acts to control the damping of the speaker.

The basic idea of NFB is to take a portion of the output signal, and apply it in opposite phase back into an earlier stage, so that it acts to cancel out anything in the output that was not present in the original input signal to begin with.
_____________________________________________________________________________________________________________________________________________________________________

I meant to add a disclaimer earlier, that in my endorsement of these particular Transcendar transformers, I know no one, nor have ever talked with anyone at Transcendar. I've never purchased any of their products for either mine, or any of my clients pieces until this project, with these transformers being purchased by way of auction, and therefore generating no direct contact with Transcendar. Therefore, I have no stake in whether you do or do not decide to use their transformers.

Dave
 
The other edge to NFB is that it is more likely to make an amplifier oscillate or be generally ill-mannered if it is not implemented properly. A lot of that has to do with the quality of the output transformer, but circuit design choices and layout also play into it.
 
Dave;

Could you give us some details on what you did to optimize the Transcendar transformers?

Thanks in advance.

Joe
 
Thanks Joe -- I meant to include that info earlier! In the KX-100 and X-100B, the HF response of the power amplifier section is controlled by two components. Using the schematic of the X-100B to identify them, they consist of C23 and C29 in Channel A, and C24 and C30 in Channel B. The adjustments made that resulted in the performance data shown then are as follows:

1. C29 and C30 should be changed from their original value of 680 pF, to 750 pF.

2. C23 and C24 (originally 100 pF) should be removed and each replaced with a step network consisting of a 68 pF cap and a 33K resistor connected in series. The networks are connected in place where C23 and C24 once resided. It makes no difference which way the networks are installed, but it is easiest to install them with the cap end connected to the tube, and the resistor end connected to the T-strip ground terminal.

And that's it!

Dave
 
These capacitor changes were made to the new transformer to bring it up to spec with the transformer that was replaced? What was the process for determining the desired capacitance, voltage and resistance? I don't need to know the math, but a brief description will suffice, if you don't mind. Thanks, Thorne
 
Dave,

The KX-100 is the kit version of the X-100, correct? So the original transformer issue applies to those as well? (Assuming same model configuration).
My X-100 (serial 30xxx, 5AR4, 7189s) is still sitting awaiting restoration, so I'm paying close attention to this thread!

*As a side note, is there a thread with instructions convert the X-100 output to fixed bias? I can't seem to find one.

Thanks!
Keith
 
Hi Keith,

The KX-100 is the kit version of the X-100-B only (I have both). There are other versions of the X-100 which are different enough that I would not attempt this transformer replacement without additional study. The KX-100 and X-100-B have 7868 output tubes which are different than the 7189 output tubes you mentioned. That alone would give me pause in adapting what is being discussed here. Other experts might not be so timid, but that is me.

In terms of changing the bias, there are a number of threads about using the "Enhanced Fixed Bias" (EFB tm) modification on some of the Fisher models. An evaluation of exactly your amp model would need to be done to see how to adapt it to your unit. The EFB is trademarked by none other than Dave himself! I would suspect another thread would be in order since Dave seems to have targeted the transformer topic in this thread.

I'm going to go silent now and continue to enjoy the show..... :lurk:
 
Bacon -- I know that the 7868 based KX-100 and the X-100B are basically the same units, but I don't know about earlier iterations of either unit. At one point, I believe they were both a 7189 based unit, but then later converted to the 7868. The serial number of the KX-100 I have here is 12525B.

I never posted any instructions for converting these units to fixed bias, but have always directed folks to my thread some years ago on the X-101C. That project was a major one which included a conversion to fixed bias, with a schematic posted for that work as well. While not exactly the same, the effort would be nearly an identical project relative to that portion of the project.

Thorn -- The subject of applying NFB around a vacuum tube amplifier which includes the OPT (i.e., a global FB loop) is complex, involving a number of issues that all have to be understood and coordinated through intelligent compromise into a stable, high performance result. There is certainly a basic formula to establish a given amount of FB for a given design, but then having it remain stable into complex loads after application is what committed so much of the topic to print back in the day. Because of the complexity, there were plenty of commercial mis-steps in implementing NFB correctly back in the day. It's a topic I have studied continuously now for over 40 years so unfortunately, there are no cliff notes or 1,2,3 process I can lay out in a post or even a thread to easily describe the process. As well, because of the mis-steps I've seen, much of the process I use in stabilizing NFB amplifiers I have devised on my own -- which started with defining what good stability actually looks like to begin with. I know what the stability looks like for many pieces of commercial equipment that were offered as being of a stable design -- and the problems they caused. So I started by learning from that, and then created my own approach from definition, through to result. I know that's not the answer you're looking for, but if the subject interests you, I would encourage you to do a search on the topic to access the significant amount of material that exists on it. One author in particular to search for would be Norman Chowhurst, who has written extensively on the topic.

Dave
 
Ahh, thanks guys! These Fisher X-series models are myriad! I'll check out that X-101C thread.

Love all of Dave's thorough and insightful "investigation" threads. Class is in session!

-keith
 
I've seen the Fisher X-101-C thread. Very good work! I'm researching that as the basis for an EFB project for my X-100-B. The circuits between the two models are very close.
 
Back
Top Bottom