Performance Issues Resolved
All of the original performance issues noted in the ratrig have now been resolved. As Don mention, we had a conversation in which he asked me to post all of my findings, not only for the learning possibilities they present, but also for the overall teachable moment the process provides as well. Here then are the areas I addressed to return the unit to its stock performance capabilities, in the general order in which they were addressed:
1. The 12AX7 phono preamp tubes and 5751 line/tone amp tubes were swapped to place the original specified tubes in the line/tone amp positions. This was done in an effort to address the significant sensitivity issues noted during initial testing. While the swap helped the sensitivity somewhat, it was still nowhere near specification.
2. Initial testing also showed a rising power output characteristic in the right channel only. Since it occurred at a thermal rate of climb, effort was then focused on the line/tone amp tube for that channel, where the TungSol 12AX7 was found to be very bad. It has been replaced with a temporary good replacement, which completely resolved the rising power output issue.
3. Initial testing also showed a variance in power output between the channels, which lead to testing of the 7591s. Of the four originally intended for installation, one was found to be quite weak, and of the two sent along as spares, one was bad. The four remaining four good tubes were then matched and reinstalled as decently matched pairs. This effort then produced nearly equal power output capabilities in both channels. However, it was also noted that the left channel developed a parasitic oscillation when both channels were driven. It was noted for later service efforts.
4. The Phase Inverter adjustments were then made, but it was noted that one control required rotation to one extreme to produce the lowest THD reading. It also let out quite a SNAP/POP when I did that. While no harm was done, it was also noted for later service.
5. An examination was then made underneath, where it was first determined that the AC heater circuit was still referenced directly to an output tube cathode, rather than the new cathode buss created when individual cathode sampling resistors were installed. While this generally doesn't cause any problems, it is still a discrepancy from ideal (relating to noise), and so was corrected.
6. Also noted was the manner in which the Screen Stability Resistors were installed. Rather than having each resistor power a screen terminal from a new buss, the resistors were installed daisy chain fashion, so that the first tube saw 100 ohms, the second 200 ohms, and so on through to the last tube. Correcting this installation then allowed both channels to produce identical maximum power output levels, and caused the Phase Inverter adjustments to be nearly centered now for optimum performance.
7. Further investigation showed that the resistors making up the NFB networks were reversed in each channel. This was causing not only the parasitic oscillation noted in the left channel, but the extreme loss of sensitivity since this reversal caused the NFB level to increase by a factor of about 5. Correcting this then returned normal sensitivity levels, and eliminated all traces of the oscillations noted earlier.
8. The last items noted were a cathode resistor in the line/tone stage of the left channel that was marked as one value, measured another, but in either case, was wrong. This was corrected to the proper value. Also noted was a B+ dropping resistor that was increased in an effort to achieve the operating voltages noted on the schematic.
This is really a practice that should be avoided. All of the small signal stages in all classic designs are cathode biased, which makes them largely insensitive to operating voltages with respect to gain produced. If operating voltages are of concern, then the best approach is to adjust the voltage to the entire unit -- not try to individually adjust each B+ distribution point. It is of limited value in the overall scheme of adapting a unit to today's higher line voltages, and in many instances, can actually hurt performance. The adjusted resistance was changed back to the original value.
Finally, the cantankerous phase inverter control was removed, disassembled, cleaned, reinstalled, and adjusted, with no further incident.
These efforts then allowed the unit to perform as originally designed. There are (obviously) some remaining modifications installed in the unit. But these are not such as to detract from stock measured performance. In total then, they include:
1. Installation of output tube cathode sensing resistors.
2. Installation of screen stability resistors.
3. The phase inverter coupling caps have been doubled in size, while the output tube grid return resistors have been reduced to 220k (from 330K).
4. The line stage coupling caps to the tone stacks has been doubled.
5. Phono preamp modifications have also been installed, which have not been measured or evaluated at this point.
In all then, this is a perfect example of how including measured performance with sonic evaluations can result in achieving the best overall performance from a given unit. Measurements and specifications do matter, as we don't always hear deteriorations in performance until it often achieves excessive levels. Don used this unit for close field low level listening use, so the defects were likely unnoticed -- but they were there none the less, and detracted significantly from the unit's original capability. Don has stated he wants to up his game in the performance measurement side of this hobby, so all of the preceding is offered as help in that vein.
For the moment then, listening tests will now be conducted on the repaired unit, while thought is given to best ways to proceed from here for improved performance, and interface into today's audio world. That project will be documented in a new thread devoted to that effort.
Pics include:
1. The heater connection error, and the daisy chained style Screen Stability Resistor installation can be seen in this before shot of the left channel output tubes.
2. The heater connection has been corrected, while the Screen Stability Resistors have been relocated on the tube sockets, using pin #1 as a buss terminal at each socket.
3. The original feedback network wiring, where close inspection will show that the 1000 ohm feedback resistor, and 220 ohm feedback load resistor have been inverted in each channel.
4. The feedback wiring has all been corrected in this shot, as has the resistor noted in the final pic. Individual channel power output with all the corrections in place is now 28.89 watts per channel in both channels, while 1 kHz distortion -- even with only 1/5th the feedback in place -- remains relatively the same at .75% average per channel at the onset of clipping. This is a perfect example of the fact that just like seasoning, some feedback is great, but too much spoils the pot. In this case, the instability issues caused by the excessive feedback canceled out any extra distortion reduction benefit it was providing. Clean power with both channels driven remains a rather timid 14 watts RMS, while distortion is over 1.5% in each.
5. What's up with this guy? Indications are that it's an 8.2K resistor, it measures 1.2K, and is supposed to be 1.5K. It can be seen as corrected at the very right in pic #4.
So, that's about that. Time for listening tests, and pondering. Output stage testing is complete now, with no doubt that its performance can be improved significantly, but that will be addressed in the new thread.
Dave
