• The move to the new server is done. There are some software and database maintenance updates in process. This has us passing the hat around to help out. We appreciate any donations. Seriously, even a dollar helps. The payment page may be found here - https://www.audiokarma.org/support.html

Rare Beauty: Fairchild's 275 Amplifier

Thanks Gordon. The measured undistorted power response of the original design was 20 Hz to 20 kHz within 1 db of 60 watts, being down not quite 1 db at at 20 kHz. But that was with the updated Fairchild circuit installed that was so overly compensated in the supersonic region, which in my experience can often actually reduce power output at 20 kHz due to the excessive phase shift created from the use of such heavy handed HF compensation. While hardly bad as is, it shows the inherent quality of the Fairchild OPT, and will no doubt be better in the modified design due to the HF transient corrections made -- but this has yet to be determined, which of course it will be and published.

As for 1 watt frequency response, it is now truly ruler flat from 20 Hz to 20 kHz. On the low end, it is flat to 5 Hz, being down 6 db at 2 Hz (this is unchanged from the original design). On the high end, the difference is quite notable compared to the original design, now being down less than -.5 db at 40 kz, and down just under 1 db at 60 kHz. Once the amplifiers are done being restored, I'll give a full report on their documented as finished performance.

Dave
 
All of the wiring/rewiring on the underside of the component tag board is finished now -- jumpers rerouted as necessary, unused jumpers removed, leads to/from the tag board reconnected to alternate posts as necessary, and unused terminals cleared of excessive solder. This means that for the first time in likely a long, long time, the tag board can be, and now has been securely mounted back down in place, as all remaining work now resides on the component side of the board, at the can caps, and at the 6AB4 socket:
SAM_3001.JPG

One thing I would warn others about who restore these amplifiers: While there have been no mishaps with the amplifiers here, the gray paint has only marginal (imo) adherence to the chassis, as I (regularly enough) see paint chips on the underside of the chassis. I don't believe there is a primer coat under the gray paint, and if the paint was baked, there certainly isn't the strength to it that I am accustomed to with other pieces employing a painted chassis. Again, not a call out on Fairchild, as the coating certainly appears to be adequate for normal use of the amplifiers. But during extensive work like a restoration, the paint might be challenged. Something to be aware of........

Of course truth be told, the help behind the scene has been enormous, so if anybody wonders who the real brain trust has been behind the project, this should put any speculation to rest:
SAM_3000.JPG

Dave
 
Dave,

I just noticed your hand drawn schematic pulls the global FB signal from the 8 ohm tap. I then saw that the original schematic does the same thing. I wonder why that is when 4 ohm and 16 ohm taps were available? I don't think I've seen a commercial amp pull the FB signal from the 8 ohm tap before, unless it was the only tap avaialble.

Those square wave shots are beautiful. It looks like the left side has more area inside it than the right side. Is that an optical illusion or is the duty cycle setting on your generator off a bit from center? No bother either way, just curious.
 
Al -- Scooter indeed! After passing his 1 year mark last month, he is now trying to upgrade his current rank of Junior Grade Engineer to Junior Grade Engineer 2nd Class. It's a slow process, but some of his suggestions have been absolutely purrrrrfect! :)

K -- Good eyes. The bench that the amps are on which is what so many of my pics are taken on, and has test equipment that primarily serves for basic design validation and troubleshooting. Most of my restoration work is done on that bench. As such the test equipment on it doesn't always get all the attention it deserves, but also because it gets the most use. So as you guess, the duty cycle in the Heath generator could use a tweak. My other bench is primarily used for alignment and performance documentation, and has a number of pieces with calibration traceable to the NIST.

As for the tap used for NFB, I've seen all manner of taps used, including dual taps (but not the way Fairchild did it), single taps, and balanced taps. Much depends on the supersonic performance of the individual taps, but in this case, the transformer is so carefully designed, it doesn't matter which tap is used to supply the NFB signal -- they all work equally well, so I just continued using the original tap that Fairchild used -- again, to minimize as much change as possible, but implementing enough change to make things right in the supersonic department. Dynaco transformers are the same way. You can use any tap you want to provide the NFB signal because like the Fairchild transformer, each tap displays the same supersonic characteristics. It's so great to work with transformers like that!

Dave
 
PROGRESS UPDATE

As with so many designs, the input stage primarily determines the finished supersonic response characteristics, since that is where the response controlling networks reside. With that stage all rebuilt -- and properly installed now -- it necessitated a slight adjustment of the phase advance cap in the NFB circuit (was 15 pF, now 27 pF) to produce the same results as before. Also, as a result of reworking the ground circuits, the unit now returns a rather remarkable -89 db S/N ratio -- this with no choke, and the bottom cover off, so it is one quiet amplifier indeed!

Dave

Below: New input stage wiring:
SAM_3003.JPG

Below: Fully operational again, only needing to replace remaining component parts on the tag board (except for the perfect condition military grade .2 uF cap near the end of the tag board, which is only in circuit when the test plug is full inserted into place -- otherwise, it just takes up space) to complete the basic restoration. Once that's finished, then the final phase of the of the project can begin - development and installation of EFB™. After that, copy it all into the other unit with the matching of component for component (as possible) along the way. Modifications installed at this point include removal of the damping circuits and installation of the new NFB loop and HF stability networks, rearranged ground scheme, and installation of output tube Screen Stability resistors:
SAM_3002.JPG
 
PROGRESS UPDATE

At this point, everything is now finished in the first amplifier, except for the installation of EFB™ (and a couple of mechanical details to attend to for the client). Next up then is to do another full performance analysis of the unit now that it is fully restored, to serve as a new base line for developing the EFB operating parameters from. That way, a true picture can be had as to how the unit is benefiting from the modification, versus an otherwise fully restored unit. The designated output tubes for normal use will also be used for this testing, instead of the tubes used for testing during the restoration work. In the mean time, a pic of the (temporarily) finished unit, ready to develop the new base line information from:

SAM_3004.JPG

SAM_3005.JPG

Dave
 
BASELINE EFB TESTING

The performance results presented here are for the fully restored 275 amplifier (before the application of EFB), that has been modified with:

1. New NFB and HF stability modifications with the Damping circuits removed,
2. Screen Stability resistors installed at both output tube sockets, and
3. Revised grounding scheme.

First however, a note about output tubes. The matched quad of coke bottle Tube Depot 6550 tubes were tried, with less than satisfactory results. Tests were then run on each tube in my power output tube tester, where one tube was found notably slow to warm up (and took a long time for the initial hum at turn on to balance out), and another was an eager beaver in terms of power output (the highest of all 8 tubes provided, and very near that of vintage American pieces). The other three however were dynamically close in terms of power output, but at a level some 8 watts below what an average new American 6550 can produce. The reduced power output capability shown in the tester explains why these tubes produced 7.5 less watts of power in the amplifier, and when the one eager beaver tube was used with any of the other three, clipping was unequal, and distortion appropriately elevated. These tubes also required almost 11% more bias voltage to reach a set current draw of 60 mA, or nearly 5 move volts of bias.

So, it was back to the Svetlana 6550C tubes (Wing C). They had been shipped in pairs, so originally I simply took one of the pairs, tested them, saw that they were reasonably matched and performing somewhat better than most Russian tubes of their day, and plugged them in. As is, the pair of Wing C tubes chosen were performing better than any combination of the Tube Depot 6550 tubes. But now that the unit was being base line tested for EFB, it is always best that the best output tubes be installed to see the widest variations produced in power supply voltages possible. With that then, the other pair of Wing C tubes were also tested and graded. It turns out that by picking one tube from each pair, a new pair of Wing C was made that had an extremely close dynamic match (basically identical) -- at a power output representing 94.12% of average new American pieces (down just 2.6 watts) -- and had a tight static match that is within 1.56%. These well matched and well performing tubes were installed to produce the restored but pre-EFB performance results provided from these tests. But first for comparison, a review of the amplifier's operation as it was before restoration (but with the three modifications listed above installed), with as many original (but still in tolerance) parts left in place:

POWER AND DISTORTION (distortion measured at 1 db below 60 watts output:
@20 Hz = 60.1 watts @ 1.55% THD
@1 kHz = 60.1 watts @ 0.32% THD
@20 kHz=54.4 watts @ 2.40% THD

IMD = 57.45 watts equivalent RMS @ 1.90%

FREQUENCY RESPONSE: 20 Hz - 20 kHz, +0/-0.1 db
@10 Hz = +/-0 db
@ 5 Hz = +/-0 db
@ 2 Hz = - 6.0 db

@ 40 kHz = -0.5 db
@ 50 kHz = -1.0 db

HUM AND NOISE:
-86 db below 60 watts.

SENSITIVITY:
0.45 vac for 60 watts

So the amplifier was hardly performing poorly, but the restored amplifier employs hand picked components that are not only within 1% of specified value, but where required, matched within 0.25% on either side of the push-pull circuit. At the time these components were chosen, components were also set aside for the other amplifier, so that it too will employ components within 1% of specified value, and within a 0.25% match on either side of the push-pull circuit. This will effectively match the performance of the two amplifiers very closely. The results for the first amplifier are presented now:

POWER AND DISTORTION (distortion measured at 1 db below 64 watts output)
@20 Hz = 64.0 watts @ 0.85% THD
@1 kHz = 64.0 watts @ 0.28% THD
@20 kHz=60.1 watts @ 2.70% THD

IMD = 62.1 watts equivalent RMS @ 1.45%

FREQUENCY RESPONSE: Essentially unchanged from data above.

HUM AND NOISE:
-88 db below 64 watts

SENSITIVITY:
0.47 vac for 64 watts

OUTPUT TUBE B+ SUPPLY:
With all chassis controls properly set, B+ dropped from 450 vdc under quiescent conditions, to 410 vdc at the onset of clipping at 1 khz. Under the same conditions, bias voltage dropped < 0.50vdc, causing a significant shift in the operating conditions. This shift is even greater at 20 Hz and 20 kHz.

So the restoration and change in tubes added a few more watts, producing less distortion than before, and at a greater power output level. At 20 kHz, power and distortion are OPT limited at that frequency largely due to the sheer size of the output transformer. Distortion has risen incrementally at this frequency, as has power output as well, with the higher power output produced being responsible for the increase in THD noted at this frequency. This is quite insignificant. In any event, the measurements show an amplifier that is quite healthy in it's performance, but also one that just may respond very nicely to the addition of EFB, so work will begin on that next. As always, a few pics are provided.

Dave

Warming up for the testing process:
SAM_3011.JPG

Everything nearly finished up:
SAM_3010.JPG

Final HF transient performance: 10 kHz square wave with standard resistive load. I see Scooter just had to leave a calling card almost dead middle of the screen.......
SAM_3009.JPG



 
They are Tung-Sol reissue tubes (clearly marked as such) with the typical Tube Depot sticker applied that is marked with a quality number of some kind for matching (56 in this case). I must admit that I was somewhat surprised, and had thought they would have a better showing (I've never tested them before). The best modern manufacture tubes I have tested of this tube family is the reissued KT88 under the Genalex label. I'm doing extended life testing on them now, but out of the gate, those things test every bit as strong as the Real McCoy does.

Dave
 
problem with single operating point matching is that unless the amp happens to run at that same point the tubes may not play well together in the amp.

same reason why "22" and "56" don't mean anything without context. Power output? Idle current at some test condition? The number chanted when the magic 8 ball said "conditions are favorable" ?
 
6 -- Don't know -- they were provided loose by the client, so I don't know what their history is, although they were supposedly either very close to or in fact, brand new.

Dave
 
AMPLIFIER #1 FINISHED

The EFB™ circuit has been designed, proven out, and installed in the first unit now. From the (now) extensive experience accumulated from the concept over the years, the resulting improvements made were right in the ballpark of expectations for this type of design. The point of "this type of design" primarily refers to the power supply, which in this case means a classic vacuum tube full wave rectifier design (albeit a heavy duty one) with capacitor input filtering, with no further filtering employed for the output stage. The Fairchild 275 therefore depends heavily on a current-balanced output stage so that the ripple voltage produced at the rectifier output is canceled out when summed in the OPT secondary winding. The particular ripple voltage produced from this type of rectifier is somewhat more difficult to filter out than that produced from more modern SS voltage doubler designs, so balance is of significant importance. Fortunately, once set, the proper balance setting has held very well in this amplifier.

More importantly however, is that it is my experience and observation, that designs from the mid 50s or so -- right where the 275 came from -- suffered most from a distortion standpoint in Intermodulation Distortion (IMD). This speaks squarely as to why Fairchild didn't even bother to provide the more customary (today) Total Harmonic Distortion (THD) specifications for the amplifier. In deed, the restored/stock/but modified with new NFB loop design is quite credibly low in THD, as the information in post #72 shows. The combination of tapped screen operation of 6550 tubes, a superior OPT, and about 15 db of global NFB will net you that result. But when powered by even a heavy duty version of the power supply installed, IMD invariably suffers. Since EFB produces the most results from the weakest performance characteristic, it is the IMD that received the biggest benefit in this case. By comparison, amplifiers that employ more modern SS power supplies invariably have much lower IMD from the get-go, so that EFB primarily boosts THD performance in those cases. This was the expectation going into this phase of the project, and once again, what was delivered from "this type of design".

The measured performance for power output, frequency response, transient response, hum and noise, and sensitivity are essentially unchanged, and so won't be re-posted here. Post #72 provides all that information. The focus here then is on the benefits produced from the addition of EFB in this application, with respect to distortion reduction. The test conditions and power levels the tests were executed at are exactly as before when a base line of information was generated from the restored non-EFB amplifier. The results achieved then spell out as follows:

1. At 20 Hz: THD = 0.63% (25.9% drop in distortion).
2. At 1 kHz: THD = 0.29% (a rise of .01%, within measurement error)
3. At 20 kHz: THD = 1.90% (a 29.6% drop in distortion)

4. IMD = 0.75% (a 48.3% drop in distortion)

There is one other thing that the application of EFB allowed for. It will be found that for typical 6550 tapped screen designs (of which the 275 is no different), the distortion (IMD or THD) continues to drop as the output stage quiescent current level is increased. This continues to the point that the low distortion operating point is most often produced at a point that would well exceed output tube dissipation limits under quiescent conditions. The 275 is no different in this regard, either. The low distortion operating point for this design actually occurs with a quiescent current of about 100 mA per tube, resulting in an output tube Pd level of 42.6 watts which is rather tough on tubes with a 35 watt (Design Center) rating. All 6550 tubes carry a 42 watt Design Maximum rating, but operation at that level brings in the scary nightmares that Citation II owners know all too well. Even Fairchild backed off on the quiescent current setting for the tubes, specifying 170 mA (85 mA per tube) in the original design, but 152 mA in the updated design. It is this area between the true low distortion operating point of either Fairchild version (100 mA per tube), the the specified operating point (76 mA per tube) they ultimately established, that EFB operates in to advantage. In fact, with EFB installed, the true low distortion operating point is now shifted to a much more reasonable 80 mA per tube, which results in a very satisfactory 33.6 watts of plate dissipation under quiescent conditions. As a percent of Design Maximum rated plate dissipation for 6550 tubes (42 watts), this represents operation at exactly 80% of rated dissipation, a percent that is so often used as a general target point for setting output tube bias. It is important to note that the 80% "rule" (if you will) is based on Pd as determined under the Design Maximum rating system. The performance results produced above are with the quiescent current level adjusted for 80 mA per tube, operating under the control of the EFB regulator.

So this project is starting to see the beginning of the end. My client is sending some alternate output tubes to test, which I'll be happy to post the results of, but otherwise, the fat lady has sung on this unit. It performs superbly now in all the performance characteristics that are a target goal for an amplifier of this quality level. A few pics to post, a slight break to clean up the lab (and a few days break anyhoo), and then I will build the same exact work into the second unit, which of course will go much quicker. Testing will be done to ensure matching performance between the units. When the second one is done, then listening tests and dependability proving will begin for a period of 3 or 4 days. When they are both ready for return shipment, I'll post some pics of the shipping container my client had specially made for shipment of these amplifiers. Heavy as lead of course when the amplifiers are placed in it, but it doesn't get any better than what he had made for a pair of rare gems like these!

Dave

Below: The underside of the finished amplifier. The lack of clutter compared to the as received condition is notable.
SAM_3012.JPG

Below: A closeup of the EFB regulator installation.
SAM_3013.JPG

Below: The rear panel. The damping control has been removed and a hole cover cap installed in its place. A new gold input jack has been installed, and the original power cord -- still in excellent shape -- cleaned right up with a little Armorall applied to it.
SAM_3015.JPG
 
Last edited:
Awesome job Dave. Guess It’s time to look at adding EFB to my 6AV5 triode strapped PP amp. You’ve given me some ideas of how to determine if it will improve performance or not.
 
You must log in or register to reply here.
Back
Top Bottom