Improving the Fisher 400

[dcgillespie]

For those following my project 400 receiver, the path started with modifying it for EFB operation for extended tube life and lower distortion, in a thread that can be found here:

http://www.audiokarma.org/forums/showthread.php?t=509090

That effort had specific goals which were met, but also brought to light other areas where significant improvement can be made in this receiver as well. Additionally, there are also some rather simple, but logical changes that can also be made, to render the unit more useful in today's audio environment. In this post however, I'll start by laying out the problems with the phase inverter stage.

The previous work of installing EFB revealed that Fisher took precautions in this stage to protect the output tubes in these receivers. If you will access the circuit of either version from the data base, it will be helpful in following along.

Specifically, the design includes a 150K resistor connected between the B+ point supplying the phase inverter (PI) stage, and the cathode of this stage. This resistor is effectively an electrical noose around the PI stage, as will be shown.

From the schematic, the relevant information is:

1. The B+ level supplying this stage is 320 vdc.

2. The plate voltage of the PI stage is 292 vdc, which is the result of a 28 volt drop produced across the 47K plate load resistor for this stage.

3. The cathode voltage of the PI stage is 136 vdc.

From this information, since we know that the aforementioned 150K resistor is connected between two points representing 320 volts, and 136 volts, it has a 184 volt drop across it, which by Ohms law represents a current of 1.23 ma flowing through this resistor.

We also know that a 28 volt drop produced across the 47K plate load resistor represents a current flow of 0.6 ma through this resistor. Since the resistor is in series with the PI tube, it means that the PI tube itself is only passing .6 ma as well. This is a pitiful operating point for any output stage driver tube to operate at. With the 150K resistor flowing twice the current flow "around" the tube, versus that flowing through the tube, it represents a very big noose indeed. What is the net effect of this design then?

First, it reduces the effective supply voltage for the stage from 320 vdc, to 184 vdc (the voltage across the 150K resistor). With the majority of current flowing through the cathode resistor of the PI stage coming from the 150K resistor then (rather than the tube), it draws the cathode voltage up to a much higher level than would be produced by the current flow of the PI tube itself -- effectively lowering the available voltage for the tube to operate from, hence, the noose.

Secondly, with only a 28 volt drop produced across the plate load resistor, it means that even if the 12AX7 tube used in this stage is driven clear through the non-linear portion of its curves to cut-off, the stage only produced an output of 28 peak volts -- and a very distorted 28 volts at that since the non-linear portion of its curves would be used to develop this output level. Staying out of the non-linear portion of the operating curves for this tube means that the stage is really only capable of developing about 14 volts peak at each output. With the output tubes receiving a bias voltage of some -17 volts, it ultimately means that the PI stage runs out of gas before the output stage does. At higher frequencies, this becomes worse, since capacitance in the output tubes requires greater drive capability from the PI stage to develop full power. And if the 330K grid resistors of the output stage were reduced to 220K? It only ups the driving requirements of the PI stage even further, adding insult to injury.

Again, this was clearly all done to protect the screen grids of the output tubes, based on how they are operated in the stock design. But since EFB can properly deal with the output tube screen grid concerns, it means that the noose can be lifted from the PI stage, moving it to a much more linear operating point for the tube, allowing for a significantly improved drive signal to the output stage.

But the PI stage is directly coupled to the previous AF amplifier stage (the other half of the 12AX7 PI tube), meaning that correcting issues in the PI stage starts with a properly operating AF amplifier stage.

For starters however, the plate voltage of the AF amplifier stage rests at some 135 volts, which helps (in part, as discussed earlier) to set the cathode voltage of the PI stage at 136 volts. This is already a red flag, since the heater circuit for these tubes references ground in the design, meaning that a 136 volt potential at the cathode then exceeds the average 100 volt maximum heater/cathode voltage rating for these tubes. That places significant stress on the heater/cathode insulation in the tubes used in this position.

In looking to correct that issue however, that required a review of the feedback/stability circuits so that any changes made to the AF amplifier stage would not significantly alter the performance of these circuits -- but what a can of worms that opened! I'll lay out those concerns in the next installment, but stay tuned, because there is excellent resolve for all of these issues.

Dave

 

[larryderouin]

 

[sony6060]

I am all ears. Only reason I get away with a lot on my Fisher 400 is I only use 2 watts or so due to 98% efficient speakers. However, I should hear an improvement with more linear operating PIs. I also noticed the Fisher 400 PI sounds best with a very cool (not warm) sounding tube. I use 1951 era Tung Sol 12AX7 black plate that are even less warm than the Tung Sol square getter grey plate. The Brimar CV4035 is also a neutral, crystal clear tube, but too sterile (lifeless sounding) as a PI. The CV4035 are the best in the phono amp section. It is a fairly serious improvement placing the Fisher 400 on par with other phono preamps. Most interesting is Telefunken are not used anywhere & I usually like the Telefunken smooth plate sonics. One can only improve so much rolling tubes.

 

[Fisherdude]

This is just too cool.

 

[buglegirl]

Dave,

They could give you your own sub forum and I would hang out there like a smitten schoolgirl hanging on your every word.

Honestly you got it going ..

Thanks for inspiring us little folk.

Frannie

 

[dcgillespie]

Thanks all for the kind words!

I want to be very clear here though, in that I realize that such modifications are not for everyone -- and I surely appreciate that. Some might want to keep a pristine example of the original product, and I certainly recognize and endorse that line of thought. I have a number of items in my stable that I intentionally keep as absolutely stock as possible, not only for the historical element, but also as a basis to see if some grandiose modification is really "all that" -- mine or otherwise!!

What I am trying to do, is to make intelligent modifications that have a clear and purposeful result. I do this using the best of today's approaches, and apply them as the original Fisher engineers might have were they available to them. I am not trying to modify for the sake of modifying, until the unit is no longer what it was represented to be. We've all seen someone's handy work on a unit that ended up being so modified that it was no longer even recognizable as to what the product originally was -- to the point that in some cases, it simply became hideous. Fisher products are some of the best looking audio products ever produced, and certainly some of the most enjoyable to listen to. Still, today, improvements can be made, and I hope I can present them without offending those who feel that anything beyond required maintenance is sacrilege.

In this case, I hope those interested will stay tuned. The 400 already contains a superb FM tuner section (for those of you lucky enough to enjoy decent FM broadcasting in your area), and arguably, one of -- if not the -- finest switching multiplexer units of the day. Its got an excellent low impedance power supply, and wonderful transformers to boot. All the basic building blocks are there -- and with a little attention to the audio sections, tube life, sonic performance, and measured performance can all be significantly enhanced -- which is something I feel that Avery would endorse.

Back soon.

Dave

 

[bhamham]

Staying tuned, thanks Dave :thmbsp:

 

[eduarsan1]

Dave:

Once again many thanks for share with us your knowledge.

Like sony6060 says: where I have to put the funds??

Regards!!!

 

[dcgillespie]

Driver Circuit Revisions

Keep your schematics handy for the discussion.

With the problems of the phase inverter previously laid out, the fix really entails a complete rework of the driver stage for the 400.

The previous stage -- the AF amplifier stage -- works very well, but needs to have its plate voltage lowered if the cathode voltage on the inverter is going to be brought down to within the rated heater/cathode voltage limits for the tube. This was accomplished by simply sourcing the B+ supply for this stage (only) down to the next lower B+ source than was originally used. This is the same source that powers the line/tone stages as well. This simple move reduced the plate voltage on this stage down to ~ 90 vdc, which is just perfect.

The remaining fix for the phase inverter stage amounts to: (A) removing the previously discussed 150K resistor "noose", and (B) removing the old phase inverter AC Balance control and associated 120K resistor, and installing a new AC Balance control network, consisting of a 56K and 180K resistor in parallel, with this network then placed in series with a new 5K AC Balance control. This new network is then connected between the cathode of the phase inverter, and GROUND.

These moves cause the phase inverter stage to now conduct about 2 ma of current, causing the tube to operate in a very linear portion of its curve, while also providing plenty of reserve low impedance drive for the output stage.

The original design employed ~ 20 db of NFB, but the new inverter circuit reduces the plate voltage of the AF amplifier stage, and eliminates the positive feedback used between phase inverter and AF amplifier stages. This acts to reduce NFB as well unless the lost gain can be restored in some other way. Fortunately, it can, and very easily.

First, with the output stage now operating under the control of EFB, the grid bias voltage for this stage has been reduced from typically -18 vdc or more, down to ~ -15 volts. Less bias voltage equals less drive voltage for the same power output, meaning that the output stage has more gain. Then, the rest of the lost gain can be recovered by bypassing the 1200 ohm cathode resistor of the AF Amplifier stage with a 22 uF cap. These two measures then allow the revised circuit to display the same amount of open loop gain (OLG) as the original design did, allowing the same level of NFB and overall sensitivity to be achieved as well. Except that now, with the revised circuit, there is plenty of low distortion drive available for the output stage, that the original design was simply not capable of producing.

In verifying the results of the feedback systems applied to the power amplifiers in the 400, the results of the original design were frankly rather poor. The amplifiers were stable into most loads, but the outcome of achieving that with the methods used produced a 10 kHz square wave that showed that frequency response was significantly rolled off -- with the roll off extending well down into the audio range. The results were identical with the revised driver circuit as well. Tests were then made to see how much improvement could be had.

The output transformers were measured, and found to have an inherent response of +/- 1db to 60 kHz. Fisher specs say the power amplifier section will be down 2 db at 40 kHz, which like the power output specifications, were rather optimistic. My unit could do no better than -2db at 35 kHz. This is really very poor poor performance from a design with output transformers that indicated much better performance can be had.

Negative Feedback is one of those things that is very simple in concept, yet because it can affect so many performance parameters, the devil then is in knowing what the best mix of those parameters is to achieve with its application, and worse yet, then knowing how to best apply feedback to achieve the desired mix. It is something that most (but not all) manufacturers struggled with in their vacuum tube equipment, and Fisher was certainly one of them. The problem really only got resolved because the problem went away: Transistor amplifiers did not need an output transformer, making the job of applying NFB comparatively quite simple.

Applying effective NFB to the vintage designs is something I have made nearly a life study of. I can state here that far, far better results can be had with some simple changes to the 400's feedback and stability circuits, that together with improving the phase inverter circuit, and placing the output stage under EFB control, turns the power amplifiers in the 400 into world class units.

The old feedback and stability components are completely removed (2.7K FB and 47K input resistors, 560pF FB and 2pF and 18pF caps). Also, unless you are in an area of significant RFI, the 330pF cap at the 16 ohm tap can be removed as well. In their place a new 10K input resistor, 1.2K/390pF feedback network FROM THE 4 OHM TAP, and 27K/100pf step network are added as shown in the revised schematic.

The results of these changes are quite dramatic. The NFB level remains the same as that with the stock design, but now, frequency response of the power amplifier section is flat within +0/-.5 db to 60 kHz, with stability being absolute under all possible loading conditions. Also, high power 20 kHz sine waves no longer show distress before the onset of clipping as they did in the stock design, and the resulting 10 kHz square wave is near perfect.

Reconfigured in this manner, the 400's power amplifiers are now worthy of the finest associated equipment. To help facilitate that end, one of the practical changes that makes the 400 much more usable is to turn the Space Xpander jacks into Preamp Out / Power Amp in jacks. On the preamp side, there is work to be done there to help make this change produce the best possible results, but all of this will be discussed in a later post. For now, pics include:

1. A 10 kHz square wave, as produced by the original power amplifier section. They got all the squiggles out of it for sure, and it was stable for the most part into any load -- but frequency response paid a heavy price.

2. Working within Fisher's basic approach, frequency response could be improved (a squarer wave being produced), but then any modicum of stability went right out the window (squiggles). In this condition, even rather small amounts of capacitance as the only load would cause the amplifier to oscillate. Given that result, its easy to see why the stock frequency response is so poor, since such a heavy hand had to be used to facilitate stability (producing the first sq wave pic) -- which was a big deal to achieve at the time.

3. The results of the revised power amplifier circuit. The classic hump is produced by the inductance in the OPT, but notice that the wave is immediately and very well damped after the transient, and throughout the remainder of the wave top. This indicates a very high level of stability, and extended frequency response as well, producing a ruler flat response within the audio spectrum. This waveform is almost a carbon copy of that which my SA-100 clone amplifier produces. No amount of capacitance as the only load will cause the revised amplifier to oscillate.

4. The driver circuit of the left channel, as converted to the new design.

5. The revised schematic, as installed.

Sonically, the accumulative effects of these changes are one of adding lost detail, new found clarity of presentation, and effortless production of power. There is simply a new dynamic energy that is presented in a seamless fashion across the entire audio spectrum.

With the top to bottom makeover of the 400's power amplifiers complete, there is only one other area of significance to address -- next time.

Dave

 

[sony6060]

Sweet Again.

Nice work. My Fisher 400 awaits the regulation module when available & will get this PI + feedback mod installed.

For best sonics is the new 12AX7 cathode capacitor type critical? I have some Sprague polycarbonate 10uF @ 50 volts and could squeeze a pair in the space available per tube. Or can I get away with 10uF per tube with good flat response to 20Hz?

 

[dcgillespie]

Thanks Sony -- Going to a 10 uF bypass cap will still produce a flat response down to 20 Hz, but OLG will fall slightly, producing a slight increase in THD in the lowest octave. I suspect you would never know. However, keep the operating voltage for the cap you use as low as possible -- this so it will form correctly. The voltage across this cap is less than 1 volt in circuit.

Dave

 

[sony6060]

Comparing the two schematics, the 12AX7 cathodyne plate with new 47K moves up to the capacitor section that feeds the 7868 screens? I read schematics well (most in my head), just want to be sure.

 

[dcgillespie]

Hi Sony -- the phase inverter proper still has its B+ supplied from the same source as in the stock design. It is the plate resistor of the AF amplifier stage before the inverter (390K plate resistor) that moves to the next lower voltage supply point -- the point that feeds the line and tone stage tubes.

Dave

 

[larryderouin]

Dave; I got lost almost immediately on the schematic with the 10.2K @ 30W between the upper 7868 and the output transformer. What's this for??? The rest I'm pretty sure i can follow thru. Also what happens with a unit that has had the grid resistors dropped from 330K to 200K-220K and the .047uf cap raised to .1uf. would there be a change to the schematic or not?

Larry

 

[dcgillespie]

Hi Larry -- I thought about the clarity of that specification after I scanned it. The 10.2K figure is the primary impedance of the Fisher T1020-116-1/2 AX output transformer, and minimum rated power level it should represent.

For units that have had the grid resistors dropped to 220K with .1 coupling caps installed, no changes would be required. The revised phase inverter still has plenty of reserve drive capability to drive the output tubes even with the reduced value grid resistors in place.

With the lower value grid resistors, the revised phase inverter is still capable of 40 peak volts of low distortion drive per output tube, where as the original phase inverter was only capable of about 14 peak volts at a much higher distortion level, and only with the higher value grid resistors in place. Of course, each output tube only requires 15 volts peak to develop full power output in the revised circuit.

Dave

 

[larryderouin]

Thanks Dave. That clears it up from mud to water. I'm archiving this for future use o nany 400's that come this way. Waiting with baited breath on your version with a 800-C or 500-C. check the 800-C owners request i posted for. There may be a potential problem on the 800-C mainly with 10001-19999 series. If you need clarification, Email me.

Larry

 

[sony6060]

Hi Sony -- the phase inverter proper still has its B+ supplied from the same source as in the stock design. It is the plate resistor of the AF amplifier stage before the inverter (390K plate resistor) that moves to the next lower voltage supply point -- the point that feeds the line and tone stage tubes.

Dave

Got it. My 400 schematic varies from your posted voltages. Also, line voltages vary too. Your post is likely actual measured voltage. I will perform this tomorrow if parts are in my inventory or order as needed. Thanks again.

 

[sony6060]

Hi Larry -- I thought about the clarity of that specification after I scanned it. The 10.2K figure is the primary impedance of the Fisher T1020-116-1/2 AX output transformer, and minimum rated power level it should represent.

For units that have had the grid resistors dropped to 220K with .1 coupling caps installed, no changes would be required. The revised phase inverter still has plenty of reserve drive capability to drive the output tubes even with the reduced value grid resistors in place.

With the lower value grid resistors, the revised phase inverter is still capable of 40 peak volts of low distortion drive per output tube, where as the original phase inverter was only capable of about 14 peak volts at a much higher distortion level, and only with the higher value grid resistors in place. Of course, each output tube only requires 15 volts peak to develop full power output in the revised circuit.

Dave

10.2K ? I'll be darned. I like PP primaries unloaded (raised) from manufactures recommended specs. Building amps over the years I noticed more clear sonics. Other variables such as speaker impedance factor in, of course.

 

[dcgillespie]

It is unusually high for a fixed bias 7591/7868 class tube setting. That's what prompted my original thread regarding the potential differences in the output transformers as used in the two different versions of the 400. On the units that included the extra limiter in the FM section, they also changed the part number of the OPTs to include an "AX" suffix. There were circuit changes around the output transformers as well between the two versions, and of course there's the matter of at least the later units not agreeing with the original published specs as well.

Until somebody with an early version unit can measure their transformers and step forward with the results, we'll never know what the differences are -- although I suspect that the earlier transformers had a lower primary impedance.

Because of that, I warned early on that the EFB parameters set up for the output stage are based on the AX transformers, and indicated it on the schematic as well, since the feedback/stabilization networks are also based on the AX transformers.

Dave

 

[sony6060]

I have the later AX series transformers. I ordered new parts for PI mod. I spec'ed the 47K, 56K & 390K at 1 watt, rest at 1/2 watt. Also spec'ed silver mica type 100pF & 390pF.