Fisher X-1000 integrated - Welcome

[audiodon]

What's the current draw of a standard 12AU7 at 12V?
150ma.
What's the current draw of a 5814?
175ma.
What's the current draw of a 12BH7?
300ma
What's the current draw of a 6CG7?
600ma but @6V.

In tube rolling in this particular unit, I have experienced rolled tubes that degrade performance due to excessive current draw draining the filaments of a series connected heater tube. So, generally, except for that one 5814 subbed for the original Telefunken 12AU7, which I labeled and kept, I believe I stuck with specified tube types in all positions.

 

[dcgillespie]

Documentation

This post will mark the beginning of a series of posts documenting the measured performance achieved between the stock (Ch A) design of the X-1000, and that of the completely redesigned Ch B installation. It should be noted however that no attempt was made to alter the design of the tone control circuits in channel B -- this because normal operation of the new design bypasses them anyway (making their absolute accuracy basically a moot point), and because with their circuits are primarily included within two PECs, making it not particularly practical to go to such lengths for a mode of operation that Don has indicated will rarely be used by him.

This particular post will document the HF response performance of the control unit sections. For both channels, all input signal were presented to the Aux 1 inputs. For Channel A, the output was taken at the grid of the EF86 pentode (with the tube removed), which represents the input of the power amp section. For Channel B, the output was taken from the rear panel preamp output jack, with the jumper to the Channel B power amp input removed. In all cases, the volume control was at max, with the tone controls set to their center tap position. This position represents not only the electrical center of the physical controls, but also of the tone control circuits, and coincides well with their marked center point of rotation on the front panel.

Pics with explanations include:

1. This is a 2 kHz square wave as presented to the Channel A power amplifier from the stock control section design of the X-1000. Since an accurate display of a square waves requires a flat response to at least 10X the fundamental frequency, a 2 kHz square wave can then be used to judge frequency response to 20 kHz. The actual response of the Channel A control section in this amplifier is -.5 db @ 10 kHz, and -1.5 db @ 20 kHz.

2. This is a 10 kHz square wave through the same Channel A control section. It shows the acute loss of response for frequencies above 20 kHz. It also shows why 10 kHz square wave testing is not particularly useful in preamplifiers that include tone controls, as this is where much of the high frequency response is lost in preamplifier design that includes such controls, as will be seen in a moment.

3. Here is a 2 kHz square wave from Channel B as presented at the rear panel Preamp Output jack. It is included for comparison to the image in pic #1 to show that with the tone controls turned ON in Channel B, not only are the two channels very similar in response, but also, since Channel B does not include all the HF filter circuitry of Channel A, it shows that virtually all of the HF response is lost in Channel A because of the tone control section. As a side note, this also indicates then that the original HF filter design does a credible job of limiting response droop when turned to the off position (at least to 20 kHz anyway), and that the two tone control sections are very well matched in HF characteristics.

4. Here, a 2 kHz square wave is passing through channel B with the tone control switch in its normal "off" position. Measured response of this channel is +/-0.0 db to 20 kHz, as the resulting square wave display shows.

5. Here, the rubber really meets the road. This is a 10 kHz square wave passing through channel B, with the tone controls off. While I have not measured the response of the active line amplifiers in channel A (which I will do before ripping it out), this at least give some indication of just how much HF response is lost through the tone control stage. In fact, while response is dead flat to 20 kHz as indicated in pic #4, the extended response of the new Channel B line amplifier is flat to 70 kHz, +0/-.5 db.

Comment: Fisher specifications for the X-1000 indicate that overall response is within .5 db to 20 kHz. Previous tests have already indicated that the power amplifier sections alone are already down 1 db at 20 kHz, so adding an addition 1.5 db of loss at 20 kHz from the stock control section indicates that the actual 20 kHz response from Aux In to speaker output is more on the order of 2.5 db down in Channel A. I will also confirm this as well before beginning work on that channel.

The next post will detail the low end response of the two control units.

Dave

Post Note: The response of the basic line amplifiers in Channel A at 70kHz was found to be very similar to that of the new line amplifier in Channel B, being down .75 db at that frequency. However, getting to that frequency was another story as the response TO 70 kHz displayed a rising characteristic by over 2 db, before finally falling to -.75 db at 70 kHz. This was directly traced to the effects of the high filter circuit when in the off mode, however, judging from the displays in pic #1 and #3, it did little to help the overall response, with pic #3 (Ch B through the tone controls) actually being the better displays of the two. By comparison, the new Channel B line amplifier displays a very smooth gradual drop off above 50 kHz, reaching -.4 db at 70 kHz.

The response of the complete stock Channel A circuitry from Aux 1 input to speaker output was in fact found to be down -2.55 db at 20 kHz. I will detail the response of the complete Channel B circuitry after the documentation for the power amp improvements in Channel B has been presented.

 

[dcgillespie]

Documentation -- Con't.

This post continues the documentation comparing the response of the control section portions of the stock (Ch A) and modified (Ch B) X-1000 amplifier. Test conditions were the same as outlined in my previous post.

On the low end of the spectrum, all is as would be expected, with one particular point of interest: The built in, non-switchable low frequency filter. It's effects can clearly be shown.

Again, Pics and explanations include:

1. Here is a 200 Hz square wave through the control section of Channel A. Because accurate display of a square wave requires response to at least 1/10 that of the fundamental, a 200 Hz square wave is then useful as an indication in response down to 20 Hz. In this pic, the presence of the non-switchable LF filter is clearly shown by the notably slanted wave tops. Because of R/C coupling used throughout vacuum tube amplifier design, some slant is inevitable at this frequency, but this pic shows the severity of slant produced in the stock design. In fact, the LF response of the stock (Ch A) control section is -.5 db at 50 Hz, and -2.75 db at 20 Hz.

2. Here is the same frequency square wave passing through the Channel B control section, with the tone controls ON. Since pic #1 (obviously) had the tone controls on, the difference between these two pics shows the direct effect of the non-switchable LF filter built into Channel A. With the tone controls ON, the response of Channel B is down only .25 db at 20 Hz.

3. Finally, here is the same 200 Hz square wave passing through Channel B with the tone controls OFF. Note that response has improved, but only slightly. While the tone controls have a big impact on the high frequency response of the control section, note that they have only very little effect on the low frequency response. The lion's share of response drop off in Channel A is clearly due to the non-switchable LF filter built into that channel in the stock design.

From the standpoint of complete response then, the response of the modified Channel B control section with the tone control switch OFF is +0/-.1 db from 20 Hz to 20 kHz, and +0/-.5 db to 70 kHz. With the tone controls ON, the low end response would still fall within this specification, while the high end will be rolled off ~ 1 db at 20 kHz, due entirely to the effects of the tone control stage.

The response of the stock Channel A control section is -2.75 db at 20 Hz, to -1.5 db at 20 kHz. Since 1 kHz is the 0 db reference point frequency wise, this clearly illustrates the predominance of the Fisher mid-range sound, with a falling response on either side of the reference.

A bit of a break, and then documentation of the power amps will begin, with distortion performance bringing up the rear. Then, channel A will have its modifications installed.

Dave

 

[sony6060]

Appreciate all the time you take to share your designs.

 

[audiodon]

Dave,
Are you telling us that, for a square wave output to be a duplicate of the input, an accurate square wave output requires accurate response from 1/10 of the fundamental frequency to 10 times the fundamental frequency +- amplitude gain or loss?

 

[dcgillespie]

Absolutely Don. That's because relative to a sine wave, a square wave is a sine wave signal containing a predominate fundamental frequency, along with an infinite number of odd harmonics of decreasing amplitude relative to the fundamental. Therefore, relative to a harmonically pure sine wave, a square wave needs extreme bandwidth (relative to the fundamental) for proper display. If response of the circuit that the square wave is passing through is limited, then the waveform becomes less than ideal, because the higher harmonics of the waveform are not being properly represented in the display. This causes the waveform to take on the rounded shapes you've seen presented.

This is why square waves are such a wonderful tool when examining frequency response issues. Circuits that will accurately pass square waves whose predominant harmonics lie within the band of interest will almost certainly treat all the frequencies within that band without coloration. As a given circuit distorts a square wave presentation, then that circuit must be altering the signal in some non-linear way, relative to the overall amplification the circuit seeks to provide.

Dave

 

[vinylisfinal]

i know!

 

[dcgillespie]

Total Overall Response

This final post relating to response performance will detail the difference between the two channels, from Aux 1 input, to speaker output.

Performance for the individual control and power amp sections of both the original (Channel A) and modified (Channel B) can be found in previous posts to this one.

Two pics are provided:

1. A 10 kHz square wave passing through the complete (stock) Channel A, as produced from a circuit response within the audio band measured at -2.55 db @ 20 kHz.

2. The same 10 kHz square wave passing through the complete Channel B (tone controls OFF), as produced from a circuit response within the audio band as measured at -.2 db @ 20 kHz.

Those following along might recognize that the response and square wave presentation of the complete channel B is virtually the same as that of the modified Channel B power amplifier alone -- which is true. That's because with the tone controls removed from the signal path, the power amplifier then dominate the response curve characteristic, since the response of the new channel B control circuits is much greater than that of the Channel B power amplifier, even with its improved performance.

Since the distortion performance of both the stock and modified controls sections is basically unmeasurable, I will now begin installing the modified control circuits of Channel B into Channel A. This will allow all of the significant work from that development process to be finalized.

Overall distortion in the X-1000 then is almost entirely a product of the power amplifier sections. With the Channel A control circuits finished, distortion of both the stock Channel A and modified Channel B power amplifiers can then easily be measured through the new Power Amplifier Input jacks which would then be available in both channels.

Dave

 

[audiodon]

Well . . . the evidence is in.
That picture #2 is very convincing.

 

[dcgillespie]

Mulling While Working.......

The first part of the channel A installation I completed was the Power Amplifier Input Jack. With this complete, I can now easily measure the distortion characteristics of the power amplifiers, driving those sections directly through their new direct input jacks. Since I knew I had some time consuming work ahead installing the new control circuits in Channel A, I thought I'd like to know what was ahead when I got to those distortion measurements, so I could use the time now to mull over whatever those results might be.

Fisher rated the X-1000 has having .5% THD at 1000 Hz at rated power output. Steady state RMS based power is not given, but the 55 watts IHFM rated power that is given is basically equal to 50 watts RMS, so that is the power level that the distortion rating is taken to represent. And of course, this is based on only one channel being driven.

Each channel alone can in fact deliver 50 watts RMS at 1000 Hz in the X-1000, although with the CL-70 current limiter in circuit, power drops to about 45 watts for each channel alone, and closer to 40 watts RMS with both channels driven -- this because of the voltage drop produced across this device even when fully heated, and all as powered from a 121 vac AC line.

With the higher power supply impedance created by the CL-70, distortion at 50 watts RMS power output averaged .70% in each channel at 1000 Hz when driven individually. Without the limiter in place, distortion dropped down to just over .5 % on average in each channel, under the same test conditions. This was with the bias and balance controls set to factory settings, and the phase inverter control set for minimum THD in both channels.

I have come to realize that this level of distortion performance for the stock design is typical of other pieces of Fisher equipment I have tested -- certainly at least of the stock SA-100 power amplifier and 400 receiver I tested in earlier threads. Of course with both channels driven, the distortion problem only gets worse, with both channels developing at least 1% distortion at 1 kHz at maximum power output, with the X-1000 reacting in virtually identical fashion. At frequency extremes, the distortion picture only gets worse.

For the level of performance that the X-1000 implies, and certainly that which the new control section now represents, this is rather disappointing performance. A few quick tests revealed that the problem lies squarely in the output stages, with EL34 tubes not particularly happy operating as pentodes under the operating conditions provided. But, they do develop 50 watts of power in those conditions, and we all know how important the power race was back at that time. Actually, its to the EL34's credit that they are producing the performance levels they are in the stock design. As specified by Fisher, they only dissipate about 16 watts each under quiescent conditions, which makes it pretty tough to produce 50 low distortion watts in typical designs of the day.

At any rate, I'm glad I made the quick checks, as now I have some time to mull those results over -- which I will do as I continue work on installing the new Channel A control circuits.

Regarding that effort, the chassis has now been stripped of all the original control section circuits: In the pic provided, if you see it there, it's because it is or will be part of the new design. Channel B is all in place of course, with the cleaned out areas ready for installation of the new circuits for Channel A.

This amplifier has come such a long way from when it was first received, not only with the basic repairs made, but with greatly improved frequency response, elimination of basically all noise, and features made much more appropriate for today's audio use. So, options will be considered for this last area of performance opportunity, while work on Channel A's new control section continues.

Dave

 

[audiodon]

Such a big amp. Perhaps it needs higher current thermisters like CL-40s, which are current rated for 6 amps, but then the series resistance drops to 4 ohms. CL-60s will carry 5 amps with a series resistance of 10 ohms, instead of the CL-70s 16 ohms@4 amps.
Everything's a trade-off, right?

After your work, thermisters might not be needed as voltage dropping tools at all.

Will you just jumper them in series for power tests?

This likely wouldn't be a problem unless you're pushing >35 watts, right?

 

[dcgillespie]

Current limiters are best rated according to the fuse size a given unit requires. For the X-1000 then, while the single 4 amp CL-70 installed "works", it is on the small side, with a 5 amp CL-60 device being ideal.

When line voltage is high, these devices can serve a purpose, although I personally prefer a zero impedance current limiter circuit consisting of a resistor and suitable relay. With this approach, when the limiter has done its job, it removes itself from the circuit completely to eliminate any effects of its presence. Generally high line voltage then is best addressed with an external bucking transformer -- again, to minimize the increase in impedance that any voltage altering device will add to the AC line supply.

If the line voltage is high in your area, lowering it will not be so important from a B+ level in the X-1000, but it will be from the standpoint of the heaters. Operating them at a notably higher level than necessary needlessly shortens tube life in general.

I will run power output levels without the limiter installed. Generally, if the unit is rated for 50 watts, that's what it should be capable of delivering, since that is part of the performance it offers itself to provide as a benefit over lower powered models.

Dave

 

[dcgillespie]

Book I -- Epilog

All of the original plans and goals for Don's X-1000 have now been completed, with all expectations met. I have been able to do some extensive listening tests this weekend -- which for the first time, represented both channels of the new design. Up to this point, Channel A always represented the original design while the new design was being developed in Channel B. Having both channels updated to the new design then is truly a perfect example where the sum is greater than that of the individual parts.

I think there will be little doubt regarding the wonderfully enhanced qualities this very large integrated amplifier now represents. Beyond all the improved performance data the lab can now demonstrate, the very real and practical improvements available now make this amplifier a real joy to have and use. These boil down to:

1. By eliminating the original high gain multiple tube line stages with all their various built in switchable and non-switchable filters, and replacing them with a high performance single tube line stage design with the proper amount of gain to perform JUST the basic line amp function, the noise contributed by the amplification process has been cut by 70% over that created by the original design. The sonic improvement from the ruler flat response of this new line amplifier is audible, but more importantly for this point, it removes that emotional itch of always wondering what all the complexity of the original circuitry was doing to the sound........

2. The combination of the improved power supply filtering and grounding systems in conjunction with the noise reduction from the new line stages, has left this amplifier utterly silent with no signal present on my 101 db efficient speakers. Whereas before, hum and noise was readily apparent from my listening position in between tracks or when changing source material, now there is nothing -- nothing until you press your ear right up against the speaker board where the faintest noise is present, which is always such a satisfying feeling from an amplifier with so much power on tap. It's very much like having a high performance Vette idle like a Corolla in your listening room -- you have no idea it's even there.... until a signal is applied.

3. The Tone Control On/Off switch very clearly delineates the impact on the sound that these controls introduce. With the controls in the normal OUT position (switch down) the presentation is very full across the audible spectrum, detailed and precise, but hardly edgy or bright. It is simply accurate. With the tone controls IN (switch up) and set electrically flat, the lowest octave drops out, and there is a loss of dynamic energy. The volume level (gain) is unchanged between the IN and OUT positions, but the OUT position is very much more preferable for most material.

4. Having the Common speaker terminals both represent ground (as opposed to the 4 Ohm tap in the original design) has no impact on the sound, but is certainly nice to have with common ground speaker systems, switching systems, and in the lab during testing. The old system facilitated a powered center channel speaker (i.e., eliminating the need for a separate center channel amplifier), but is a feature no one will miss today, and often in fact, becomes a nuisance. Front panel variable level control of the center channel output is still available for Don's Sub Woofer addiction, although I hope he'll give the extended response of the modified X-1000 a chance before going down that route with this unit. The extended LF response coupled with the high power output available and excellent OPTs may just make the need for such add-ons unnecessary with this unit.

5. The new Preamp Output/ Power Amp Input jacks are very nice to have on such a high quality unit. The internal power amps can now be driven from any quality external preamp capable of delivering a minimum of 2.0 volts output from the sources it uses, or conversely, the internal preamps can now be used drive any external power amps requiring even obscene drive levels (they can provide 50+ volts of undistorted output with the tone controls OUT), but of course, the sensitivity level is based on providing a 2.0 volt output. Additionally, because the output impedance of the new line stage design is very low (~500 ohms with the tone controls IN or OUT), they can also be used to drive SS designs, or extended length cable runs with virtually no impact on the sound quality.

6. Similarly, the Record Output signal now represents a very low output impedance as well, with the outputs fully isolated from the signal path within the amplifier: A dead short on the recording output jacks produces no change in the signal being produced at the speakers. The importance of this is that now these jacks can properly drive any type of modern SS external equipment -- recording or otherwise, to enhance capability.

7. Finally, the Aux 1/Tape Monitor bleed circuits have been removed, so that these inputs can properly function as two islolated inputs. This then allows for the Tape Monitor input to fully support any signal recording or processing needs, while leaving the Aux 1 input available for its own independent source connection. This work, in conjunction with the new Recording Output circuits, was also implemented in such a manner as to eliminate all the various phase problems that existed between these inputs and outputs in the original design.

8. And really finally, no more having to source rare and expensive quality 7247 tubes.

The list clearly shows that the vast majority of the improvements were all aimed at the line amp/control section and power supply section of the X-1000, with the only power amplifier modifications being to improve the response of those sections, and converting the output transformer/NFB connections to facilitate the Common output terminals operating at ground level. As mentioned earlier however, the power amplifier sections have shown their own weaknesses regarding distortion performance, so Book II might take a crack at that. While that will be discussed in the next post, the most workable potential solution would appear to be including a modified form of EFB, similar to that which I developed for my 400 in the thread regarding that effort. However, that will take more tests to determine before going forward with that project.

For now, final pics to close out Book I include:

1. The 10 kHz square wave response of Channel A is now materially no different than

2. The 10 kHz square wave response of Channel B (both with the tone controls OUT).

Most preamplifiers would be very hard pressed to produce anywhere near this kind of flat, extended response. Such response guarantees accurate transmission of HF transient information, without making the circuits needlessly sensitive to RFI problems.

3. The new Record Output buffers located in the old V2 position.

4. The new Line stage/control section output buffers located at the old V6 position.

5. The new Channel A line stage amplifier, located at the old V3.

Later this week, I will do some evaluations to see if EFB can help the output stage operation, and then the following week, I'll be out of town. For now however, the tunes are sounding better than ever from this Fisher!

Dave

 

[audiodon]

A pro cleaned up this pre.

Dave,
What vendor made those blue caps?

 

[dcgillespie]

I'll try to check Don, but I seem to remember they are German made. I got a number of them in bulk at a Ham Fest some years ago, and have used them in many applications. They measure quite well, and I've never been able to destroy one yet......

Dave

 

[sony6060]

Thanks for sharing. One super amplifier upgrade with all the re-engineering involved and the scope does not lie. :thmbsp:

 

[dannyr]

Dave, great work. How's the 1000 so far stack up sonicly to your "pimped" 400?

Dan

 

[dcgillespie]

Hi Dan -- With the exception of power, very very similar.

Given any specific engineering goal, there are always may ways to achieve the desired outcome. Take the generation of electric power. You can use hydro, gas, coal, or nuclear for energy, and direct drive, steam turbine, or otherwise to drive the generators -- but given a specific design goal, all of these various energy and mechanical approaches can be used in one form or another to achieve the same exact outcome in terms of power developed.

The outcome I strive for in audio is accuracy. Whether it's triodes, pentodes, or UL output stages, Williamson, Mullard, or classic console type front end/driver circuits, they too can all be used in a variety of forms to achieve the common outcome of accuracy.

With accuracy the common goal then, as all of the various circuits I work with start to achieve this goal, then the individual signatures that triode or UL output stages, pentode or triode input stages, or Williamson or Mullard driver designs contribute, all start to disappear. And so it is between the 400, and the X-1000. The 400 uses an all triode Williamson "type" front end/driver design in the power amp section, while the X-1000 uses the same approach, but with a pentode input design to get the job done. The X-1000 has twice the power output capability of the 400, but within the power output capabilities of both, they sound very similar -- as well they should. After all, accuracy can't be defined one way for one piece of equipment, and another way for a different piece. Accuracy is accuracy, and as it becomes achieved by multiple pieces, the more the sonic differences between those pieces disappear.

The X-1000 has been a greater challenge for sure, but I expect no less in terms of accuracy from either of these fine instruments.

Dave

 

[dannyr]

Dave, thanks for the reply. Great work. I wasn't sure that the 400 would stand up to the 1000's iron etc so I figured that I'd ask. I'm even more impressed.

Dan

 

[dcgillespie]

OK. With the exception of finishing up a couple of small details (the rear screws mounting the plastic feet to the bottom cover are dangerously close to an output tube socket and AC power wiring when this cover is installed, and the premium output tube coupling caps ultimately need to be secured as well), I will close out this chapter of the X-1000 adventure. It's been a long road, but one I feel has paid off with many handsome returns in terms of feature enhancements and listening room performance.

Over the weekend, I was again able to get some good listening time in, with my impression identical to that which I had before I left town a week ago: The response is seamless across the audio bandwidth, with excellent presentation of fine detail, timbre, and human vocal qualities. Bass elements are presented with authority, in a tight, controlled fashion, without boom or false emphasis. The amount of power available is seemingly limitless, which is certainly made all the more apparent by my high efficiency Cornwalls, while with no signal present, the unit is dead silent. The resulting stereo image and soundstage presented is very defined, with individual instruments and voices easily identified and located within the sonic presentation.

While this description accurately describes my impression of the X-1000's current sonic performance, it also must be recognized that even during the most demanding passages of these listening tests, power levels only peaked at about 10 watts per channel due to the efficiencies of my speaker system. At such levels, the (basically) stock power amplifiers of the X-1000 are coasting along with ease. But the X-1000 is a high power integrated amplifier, presumably offered to address the power needs of less efficient speaker systems or distribution systems. At the elevated power levels these systems require, tests indicate that the fine performance now produced at lower power levels becomes compromised with elevated distortion levels, as power is increased to the upper ranges of this unit's capabilities.

As a side note, this of course is the reason that many prefer an amplifier with significant reserve power capability, so that performance delivered at much lower typical listening levels has no chance of becoming tainted from any distortion products. Still, how nice would it be if the excellent performance delivered at normal listening levels, could be available through out the entire power range that this unit represents?

Previous tests have already shown that the elevated distortion levels are a product of the output stage, as opposed that of the power amplifier driver circuitry. Therefore, efforts to address this final element of the X-1000's performance will be aimed squarely at the output stage, to see what potential for improvement can be had through the use of EFB, in conjunction with the basic output tube/transformer elements originally offered. That effort will be much simpler of course than all the work that has preceded with this project. But because EFB is its own design entity, I will start a new thread documenting that effort. I'll be back with that final effort shortly!

Dave