Improving the Fisher X-101C

[dcgillespie]

Hi Fred -- Absolutely correct. The LM-4562 has unbelievable specifications as compared to vacuum tube equipment -- but then so does the TL-082 for that matter. Ultimately, since the circuit operates the op-amp in a non-inverting, unity gain mode (100% NFB), the performance of either device in the buffer circuit is so far superior to the circuits of the X-101C, that either component will do the job admirably. Still, for those who want the very best, the LM-4562 is the superior device. Thanks for providing the comparison information!

Dave

 

[eduarsan1]

If we are talking about op-amp performance the opa2134 is better in harmonic distortion ; dannyr did this in their 400 improvements

The opa2134 has 0.00008% distortion.

More info about the opa2134 here

But I don´t wanna create a storm in a cup of water so Fred betewen the opa-2134 and the lm-4562 what do you think would be the best option??


And Dave I´m pending for your feedback about the drawings to start when possible!!.

Regards

Luis

 

[sony6060]

OPA2134 is my standby when upgrading SS products. Like a K40Y-9 capacitor it never seems to disappoint. Not to harsh, not too clinical.

 

[dcgillespie]

Output Stage Acrobatics

I mentioned earlier that installing PO/PAI jacks allows for each section of an integrated amplifier to be optimized individually. Accordingly, when these were installed in the X-101C, it allowed the power amplifier (PA) sections to then be laid bare warts and all, so that their stand alone performance could be plainly seen. When this was done, previous testing indicated that their transient response could use some work.

To be fair, my post #17 pic 1 offered a depiction of power amplifier square wave/transient response for the stock PA design, with text indicating excellent stability, but being over compensated to produce a rather poor response of -.75 db at 20 kHz. However, this was with the test signal injected at the top of the volume control, with that control turned to maximum. However, as will be shown, when the PA and preamp sections were properly split, the volume and balance controls stayed with the preamp section, as they should. When this was done, the stability/transient response of the power amplifier sections became compromised. In reality, they always were. But Fisher used a band-aid to cover the wart up.

In the stock design, the balance control alone (whose wiper represents the output of the preamp section) adds nearly 100k of resistance to the output impedance of the preamp section in each channel. It is clear that Fisher was using this very high output impedance of the preamp section to allow Miller in the AF amplifier stage of the power amplifier section to roll off the response of the PA section. When this resistance was removed and the power amplifiers were then driven from a low impedance test source into the 10K input resistors of the stock PA design, the wart was then exposed. This must be properly dealt with, as any good preamp design represents a low output impedance. Therefore, the PAs of the X-101C must be modified to have inherently good stability/transient response on their own, without depending on the impedance of the source driving them, be it any new external preamp that might be used, or the internal preamps of the X-101C. This is particularly true now the the new buffer stage has been installed between the balance control wipers and the PO jacks: The output of the internal preamps is also quite low now, so that they can be used to properly drive any external power amplifier if desired. Since the output impedance of the internal preamps is now very low, the power amplifiers in the X-101C must be modified to have stand alone stability.

If anything, this shows why casually separating the preamp and power amp sections of an integrated amplifier design to operate separately as stand alone units can be fraught with problems. Fisher was hardly alone in using a high preamp section source resistance to help tame response in the power amp section. After all, whoever dreamed that some goof would want to come along and try to separate them!

Achieving stand alone stability in the X-101C's power amplifier section was a bit of a chore. As originally designed, the NFB is (effectively) taken from the 4 ohm tap. I say effectively, because the original design has the 4 ohm tap grounded, with the Com lead supplying the NFB. When the revised output connections were installed, the Com lead was then grounded, with the actual 4 ohm tap then supplying the NFB, with the OPT primary leads swapped to maintain correct phase. In any event, in both cases, the NFB was being supplied by the 4 ohm section of the secondary winding. It was necessary to use just half of the secondary winding to generate the NFB signal, so that the powered center channel output feature of this amplifier could be implemented. But using the 4 ohm portion of the secondary winding to provide NFB also ultimately created the stability/transient response wart Fisher had to cover up by the elevated source impedance at the output of the preamp section.

Tests showed that using the 4 ohm winding to provide NFB produced a rather wicked double peak in the frequency response of the stock stand alone PAs. Using the full secondary helped a little, but still did not produce the desired result. Finally, a dual tap NFB network was developed that produced very good results, allowing the internal PAs to operate stand alone without regard to the impedance of the source driving them. A new Transient Phase Shift Network was also employed to achieve the correct overall response.

This effort with the PA section, along with the installation of the new buffer stage for the output of the preamp section, will now allow both sections of the X-101C to operate very well independently of each other. And, as with any good design, the response of the unit (as separate elements or strapped together), will no longer vary with the setting of the volume or balance controls.

Pics include:

1. Because of the revised output connections, both channels of the scope can be used to present both channels of the power amp section at once. Here, the L (top) and R (bottom) channels are shown with a 10 kHz sq wave driving the 10K input resistors of the stock power amp section, with the outputs loaded with an 8 ohm resistive load on the 8 ohm tap. A comparison with pic 1 in post 17 will show the deterioration in stability and transient response that the stock PAs display operating in stand alone fashion.

2. Here, the left channel is still stock, while the right channel clearly shows the improvements that the new NFB network and HF stability measures produce at 10 kHz. As with the stock channel, HF response is -1db at 50 kHz, but unlike the stock design, the response contains no double peaks, with response tapering off smoothly above 50 kHz. This is an absolute necessity in achieving good stability in a NFB amplifier. The slight hump in the middle of both wave presentations is a pretty stubborn resonance in the design of these OPTs.

3. With no load at all, the improvement in stability and damping can clearly be seen. Note how the right channel is well damped after one cycle of the initial rise, while the stock channel continues to ring across the entire wave top. The modified amplifier continues to operate with the stock level of NFB.

4. Alligator clip leads are your friend with this work!

The new networks will be permanently installed, and then the final effort will get underway to installed the tone control bypass feature.

Dave

 

[audiodon]

I guess two wrongs were used to make a right . . . with mixed results.
It'll be interesting to listen to this amp when it comes back.

 

[fred soop]

Never used the OPA2134 and it would take a while to compare the test conditions for the various specs. Douglas Self ranks these (TL072 rather than TL082) along with LM833 (I have a pile of those), LM4562, OPA2134, and others.

For noise, the TL072 is the worst at 18 (en(nV/SQRT(Hz))) (if you can follow that), OPA2134 at 8, LM833 at 4.5, LM4562 at 2.7. But in a table where these are tested as phono preamps, the OPA2134 beats the LM4562 by 2 dB. He ends up preferring a 5532 (dual) or 5534 (single) for best noise performance in the phono preamp.

Slew rate: LM833 slowest at 7 V/μS, TL072 at 13, OPA2134 and LM4562 at 20.

Bottom line: None of us will be able to hear any difference in any of the above.

Dave, when you are done with the modifications, can you post a partial schematic showing the changes? I'm also assuming that the 100 ohm resistors on the buffer card are output series resistors so the buffer can handle a shorted load.

 

[dcgillespie]

I will post all of the changes made when finished.

The 100 ohm resistors on the output of the op-amps are primarily for stability. They do certainly add a degree of protection. But due to the extremely low output impedance of the unity gain configuration, placing any capacitance directly on the output of the op-amps is a sure fire invitation for instability. The resistors then act to isolate the output of the op-amps from any capacitive element in whatever is connected to the PO jacks, without unduly raising the output impedance in the process.

Dave

 

[dcgillespie]

Down The Home Stretch

The last two days brought final tweaking to the new NFB networks, and chasing one last (known) ghost out of the ratrig.

Installation of the new dual tap NFB networks was a project all unto itself because of the old nemesis: No room. No room and tight space always up the ante when trying to make for a neat installation. It did require making some "air" connections, but due to the way they are made, they are as secure as any other connection under the chassis -- but all of which does beg a point that needs to be made.

The total scope of this project from a stock unit to its current condition is not for beginners. This unit had already been recapped in both the power supply and audio circuits, and had most resistors replaced in the power amplifier section as well. Therefore, it was hardly in a mode of needing basic refurbishment work done. Still, at this point in the project, there is basically no circuit in this unit that has not been reworked, modified, or rebuilt in some way, other than the phono preamp circuits, which only had their heater circuits rewired. This is hardly to scare anyone off who wants to incorporate these changes into their unit. It can be done, and done neatly, but represents a significant amount of work.

On the plus side however, the transformation of this unit is nothing short of stunning. It simply must be heard to be appreciated now, with it no longer being the economy entry level into Fisher integrated amplifiers. From a power delivery standpoint, this puppy packs a wallop now that it never had before (even when the stock unit was fully repaired), and a clarity that is unmistakable in presence. The new dual NFB loops accomplish what I have heard so many times before when the shortcomings of such networks are addressed: All of the detail is there, but presented with a smoothness that is natural, rather than generated by a rolled off response. Detail without edginess is what produces clarity, and in that department the re-engineering excels. The significantly reduced distortion numbers also contribute to these results as well, while the EFB controlled output stages are now allowing the output tubes to run considerably cooler than they did in the original design (each dissipating < 14 watts now) -- a design that produced considerably less power output, and considerably more distortion. So while the effort to duplicate these modifications is significant, the results are ever so much more so. Don of course will be the final judge, so his comments will be interesting from a comparative standpoint.

Beyond that, in putting the unit through its paces in the listening room today, the loudness circuit was the only casualty of the experience, with the left channel intermittently increasing in volume, and then just as suddenly falling back in line -- but only when the loudness switch was turned on. This was rather quickly traced to an intermittent .022 uF cap on the loudness switch for the left channel. As a result, these caps were replaced in both channels, producing normal, dependable operation.

So, all that is left is to remove the permanent filter circuits and install the tone control bypass switch, which I hope to make significant headway on tomorrow. Waiting in the wings is the second phase of the Bogen MO-200A project, whose first phase preceded this one. I hope to have this unit finished up by this weekend, so the last phase of the Bogen project can get started the first of next week.

Pics include:

1. The new dual tap NFB networks are embedded into the area where the original feedback networks were, but require off T-strip connections to accommodate them. If you look closely, you can see the two air connections per channel to make the installation. Still however, all connections are tight and secure, with no significant movement of the off strip connections possible. The new step networks required to achieve the final result are also visible for each channel.

2. A 10 kHz square wave driving the new power amplifiers directly through their Power Amplifier Input jacks, Left channel on top, Right on the bottom. A new step network in the plate circuit of the AF amplifier stage was tweaked for best performance in each channel. This required the R component of these networks to be 12K in the Right channel, and 8.2K in the Left. This is hardly a significant shift in component value, with such differences not being uncommon when best matching performance is sought. The different values required are due to slight differences in wiring layout and response of the output transformers.

Final performance for the new power amplifiers has the response of each channel +0/-0.1 db to 20 kHz (essentially flat), and +0/-1 db at 50 kHz. The amplifiers are extremely well behaved, remaining absolutely stable into any type of load. This represents a significant improvement in response performance, where the original design was -.75 db at 20 kHz (worse at normal volume settings), and displayed significant ringing into non-resistive loads.

3. The same signal is applied as in #2 above, but here, the traces for each channel are relocated on top of each other to show just how identical the frequency, amplitude, and transient response is for the two power amplifiers now.

4. New caps installed at the loudness switch nailed down the erratic Left channel performance when the switch was turned on.

So its on to the line stage and tone control bypass feature. Back soon with the results of that effort.

Dave

 

[audiodon]

I must say that my anticipation is building.
Dave is right in that this is a substantial rebuild.

In the "is it worth it?" department, I'll leave you with some words of wisdom from the contractor who put an addition on our house, repoured the basement cement floor and gutted and rehabbed the whole house.
From Dave the Builder, "It would have been cheaper to demolish the house and build a new one."

Is cheaper better?

I gave someone a walking tour through the house just two weeks ago and his comment was, "You'd never know that was an addition. The whole house looks like it was always this way."

And so it is with a complete rebuild like this Fisher X-101-C.
You can do what the Fisher does for less money and labor, but you will not necessarily better the whole package in looks and performance.
The original cosmetics have a strong allure.
The X-101-C is a contender for the best of the mid-century modern designs ever created in a standalone unit for audio applications department.
I'd venture that, among integrated amps, there are few rivals in the design elegance, quality user interface, ease of use, and good looks categories.

As a reminder for those who haven't followed the links, here's the X-101-C's initial state:
http://www.audiokarma.org/forums/showpost.php?p=6371635&postcount=2

 

[dcgillespie]

Tone Control Bypass and Filter Removal

Engineering is often a game of dominoes, wherein you have to do this because of that, which means you have to do something else because of this, which means you have to...... You get the idea. In the X-101C, the original "that" of this conversation begins with the output stage: The biasing and loading conditions employed in this stage required substantially more NFB to be used in the power amplifier section (the "this") to achieve similar performance levels, as the otherwise almost identical fixed bias version of this power amplifier section as used in so many of the Fisher stereo receivers.

Continuing on, this expedient then required more gain from the line/tone amplifier stages to achieve a reasonable input sensitivity, in keeping with other Fisher products. The extra gain was found by eliminating the negative voltage feedback typically placed around the tone amplifier stage (which immediately precedes the power amplifier section), as used in virtually all other examples of this overall design. Even at that, the X-101C still has a somewhat greater input signal requirement (less sensitivity) than its many brother, sister, and cousin designs.

This last expedient then is what (in part) caused so much of the rolled off response as documented in post 17, and leads to the challenge of this final modification: Remove the fixed filter circuits, flatten and extend the response of the line and tone stage amplifiers, and do it without the aid of any negative voltage feedback for the tone amplifier stage. To punctuate just how poor the response of this stage is, review pic #2 of post 17, where in the test signal is applied directly to the tone stack and following tone amplifier stage, via the tape monitor input. At this point, response is down nearly 3 db at 20 kHz in the stock design (and even more at reduced volume control settings).

You might recall that Fisher's approach to this problem was to make the preceding line amplifier stage have a rising response to help compensate for the reduced response of the tone amplifier stage. This works after a fashion, but also makes the signal provided at the Recording Output jack have a rising response characteristic versus frequency, since that jack is tapped into the circuit immediately after the line amplifier stage. It also makes for poor performance from any source played through the tape monitor input, since those signals don't pass through the line amplifier stage. Finally, due to the tolerances in the various components used to shape the rising response of the line stage, it can also lead to a dis-similar response between the two channels, as pics #3 and #4 of post 17 show. With the volume control at more normal operating levels, pic #5 then is what you received from this approach.

Adding a tone control bypass feature then represented an opportunity to address the concerns noted above (in addition to adding a simple switch), to make the performance of the preamplifier/control section come up to speed with the improved response of the new power amplifier section. In doing this, advantage was taken of the negative current feedback generated across the unbypassed cathode resistors of the line and tone amplifier stages, along with the usual bypass cap placed across the input resistor of the line amplifier stage. By installing appropriately small value bypass capacitors across these resistors, the response at the Recording Output jack is now very flat to 20 kHz (+0/-.1 db), and through the complete preamplifier/control section, now down only .2 db at 20 kHz with the tone controls disengaged. In fact, overall response from the line level inputs, through the complete preamplifier/control section and power amplifier section to speaker outputs is now +0/- .25 db at 20 kHz at any volume control setting. This is a major improvement over that of the original design, where response at normal volume settings approached -2.5 db at 20 kHz in this same scenario.

With the tone controls engaged, the response of the complete preamplifier/control section is now only somewhat better than it was before, as the response is now primarily a product of the actual tone controls. Even though the response of the tone amplifier stage is now much improved, the rising response of the line amplifier stage before it has been eliminated, thus canceling out a significant portion of the response improvements made to the tone amplifier stage. Therefore, the response is almost entirely driven by the tone controls themselves, with the complete preamplifier/control section producing an overall response of +0/-1.5 db @ 20 kHz at any volume setting with the tone controls engaged.

An opportunity was also taken to help flatten the response of the tone control networks, within the bandpass of their flat setting characteristic. This helped to achieve the the overall response mentioned in the previous paragraph. It also produced a slight inequity in volume level (about 1 db) as the new tone control engage/disengage switch is activated, with a the gain advantage being in the disengaged position. While this could have been leveled out more closely, it would have required more components to achieve non-production component values in an already space challenged chassis. The slight difference in volume level between the two settings from using standard bypass position component values cannot hide the obvious improvements made by removing the tone controls from the signal path.

In all then, the improvements made to the preamplifier/control section are as significant as those made to the power amplifier section, both from a measured and sonic presentation standpoint. The only downside to any of this (if loss of of the original Hi Filter switch feature is in fact a loss), is that shorting plugs now need to be installed in the Tape Head input jacks IF the Tape Head EQ position is selected, and no Tape Head source is plugged in. In that scenario (and only that scenario), due to the HF characteristics of the Tape Head EQ curve, the amplifier will now oscillate with an unterminated input, due to the extended HF response all of the circuits ahead of it now achieve. This is one of the limiting factors in trying to pack so much performance into so little space. The current improvements represent the realistic limits of what the existing physical layout can support. Additional shielding could be installed, but then that can create its own problems as well. The concern is completely eliminated with the use of shorting plugs, or by simply leaving the EQ switch in the Phono position, which is what I'm sure Don will likely do.

Since the likelihood of ever using the Tape Head EQ setting is beyond remote, the problem could also likely be eradicated completely by moving the discrete phono EQ networks installed previously over to the phono preamp tubes, and permanently install them there. This would eliminate the wiring to the EQ switch (and the Tape Head EQ option), but still allow the Tape Head input jacks to serve as inputs for a second TT. Don?

Pics include:

1. The modified line/tone amp stages. Even though some new components are in place, the reworked circuits presents a less cluttered look.

2. A 10 kHz sq wave passing though the complete control section with the tone controls engaged. Circuit response is down 1.5 db at 20 kHz.

3. A 10 kHz sq wave passing through the complete control section with the tone controls disengaged. Circuit response is down just 0.2 db at 20 kHz.

4. Here is an extremely tough waveform to reproduce accurately. This is a 2 kHz sq wave, first passing through a reverse RIAA network, driving the phono preamp inputs, with the tone controls disengaged, and the scope connected to the Preamp Output jack. The quality of this waveform shows the extreme accuracy of the phono, line, and tone stages now. And who says the Fisher phono preamp circuit isn't accurate? These traces were produced using 5751 tubes in the phono preamp positions. For best LF RIAA accuracy, you should always use 12AX7 tubes in these positions.

5. Full bottom view of the modified amplifier with all modifications complete.

With the new preamp Output Jacks, it is so nice now to be able to just yank the output tubes, and work on the preamp/control section without the unit producing any heat to speak of. Thanks to the new heater supply, the output tubes are no longer needed to heat the small signal tubes, so the unit can be operated permanently that way as a quality preamp/control center if desired.

I should also point out that unlike the Fisher receivers, the line/tone amp stages are both contained within the same tube for each channel in the X-101C. However, because no loop NFB is used for the tone amplifier stage, channel balance can be affected by the match between these two tubes, although the effect of any routine mismatch is very slight.

Time to clean up the lab, hook up the ratrig for some extended listening tests, and post the final Summary. What a ride this one has been!

Dave

 

[audiodon]

Dave,
It seems to me like shorting plugs are the way to go.

I don't believe I've ever engaged the tape head circuit on any Fisher that's ever come into the house.

On the phono pecs, there are three leads. Lead two is to the tape head input as lead three is the phono input. Snipping lead two would lead to a permanent open when you want that circuit shorted, correct?

Probably no need to move the PEC. Shorting plugs will do.

 

[ferninando]

It mite be extreemly helpful to see a schematic with all the modifications ID'd somehow so that we could see the whole picture.
It may be too much to ask so if it's not possible we'll have to do our best to follow the text..
thanks for documenting your work as you've done.

 

[dcgillespie]

With your discrete RIAA networks installed, in the phono position, the PEC is doing nothing. But because of the way the discrete network is installed, the switch is still capable of selecting between your discrete RIAA network, or the internal Tape Head EQ network of the PEC.

If the Tape Head EQ will never be used, the suggestion is to completely remove the PECs and wiring to the switch, and move the discrete RIAA phono networks over to the preamp circuits themselves. This would eliminate the EQ circuits of the preamp tubes from being wired directly to a switch that is in close proximity to a switch that controls the signal path for the output of the line amplifier stage, which very likely will eliminate the problem -- or at the very least would represent good build practice. It wouldn't take but a few minutes to make the change, so I'd be happy to do it if you like.

However, the shorting plug option works just as well, with no effort on my part to sway you one way or the other, but simply to present the best options available.

Dave

 

[dcgillespie]

Ferninando -- I plan on posting the revised schematics. I'm just trying to finish up the physical work on this unit today, so I can move it off the bench and into the listening room, and get set up on the next project. I hope to be able to get the schematics posted later this week.

Thanks for the interest!

Dave

 

[fred soop]

.....

With the tone controls engaged, the response of the complete preamplifier/control section is now only somewhat better than it was before, as the response is now primarily a product of the actual tone controls. Even though the response of the tone amplifier stage is now much improved, the rising response of the line amplifier stage before it has been eliminated, thus canceling out a significant portion of the response improvements made to the tone amplifier stage. Therefore, the response is almost entirely driven by the tone controls themselves, with the complete preamplifier/control section producing an overall response of +0/-1.5 db @ 20 kHz at any volume setting with the tone controls engaged.
.....
Since the likelihood of ever using the Tape Head EQ setting is beyond remote, the problem could also likely be eradicated completely by moving the discrete phono EQ networks installed previously over to the phono preamp tubes, and permanently install them there. This would eliminate the wiring to the EQ switch (and the Tape Head EQ option), but still allow the Tape Head input jacks to serve as inputs for a second TT. Don?

Dave

The -1.5 dB @ 20 kHz is going to be trumped by the fact that the tone control pots themselves are inaccurate, even more so if they are log taper. McIntosh solved that on one of their preamps by using fixed resistors and rotary switches for the tone controls.

Removal of TAPE HEAD and moving EQ close to the photo stage is a good improvement. On my 600-T this removed 2 round trips to the selector switch of very low level signals. In my case, the TH input is now high level. If it was used for a second TT, then one round trip to the selector with the low level signal would still be required.

 

[audiodon]

I vote for removing the PECs and moving my aftermarket phono PEC replacements.

 

[dcgillespie]

The on center accuracy of the tone control networks is pretty good as passive designs go (pic #2) -- I've certainly seen much worse for sure for the frequency of the test signal used. As you say however, all bets are off regarding the tracking of the controls when moving from the on center position.

Dave

 

[tube-a-lou]

I'm amazed at the work you do with the Fisher amps, my 101 is suffering from a
Humming transformer which I'll have to fix some how, but your upgrades are nice.

Tube

 

[ferninando]

Dave, I meant no hurry on schems. Just somday is good enuf. I know the work invokved in all this.
At your liesure.

 

[dcgillespie]

Phono EQ Followup

After the last discussion, I went in and removed the PECs and associated wiring, and relocated the discrete RIAA EQ networks over to the phono preamp tube area. As originally thought, this completely solved the feedback issue, so no need for shorting plugs or a source of any kind to be plugged into any of the low level input jacks to maintain stability at any volume setting or tone control setting if activated.

A few followup pics of the resulting work:

1. A close up of the phono preamp area behind the EQ switch where the PECs once resided. The switch wiring has all been removed, and the new discrete RIAA networks relocated under/within existing preamp tube wiring. Talk about spring cleaning! The area almost looks bare now.

2. A new full shot of the underside with this last detail attended to.

3. The ratrig is now fully buttoned up, and has moved over into the listening room, for some serious run and listening time. But I gotta tell you, it's no ratrig anymore.......

A final post will summarize the project, but other than that and drawing up the schematics, the book on this case is all but closed.

Dave