Scott 299-C Modifications
This is a verbose description of the process of restoring and modifying a Scott 299-C integrated tube amp. I am not an engineer and do not pretend to have in depth understanding of circuit design. However, over the years I learned a lot about optimizing existing designs to squeeze out the best possible sound. The goal in this endeavor was like unto attempting to transform a Chevy Impala into a Mercedes.
I brought up the amp slowly on a variac and did a quick check of the DC voltages before doing a initial listening evaluation. At the beginning of the listening session one section of the filter capacitor shorted out. I disconnected the shorted section and continued listening but with only half of the designed power supply capacitance.
In it’s stock form the amp sounded quite respectable. The treble was a bit rolled off on the high end and there was some coloration in the lower midrange. The sound stage and imaging were only fair and there was not a lot of detail. However, in spite of it’s flaws the amp was very musical. The Scott had warm and soft character that was quite appealing.
The initial set of measurements (figures 1 and 2) show that the frequency response was not very flat even with the tone controls set straight up. There was significant low frequency roll off and a hump in the upper bass. Apparently the low frequency roll off is essentially a rumble filter that is build into all Scott amps. The 1khz , 1 watt distortion figures in figure 2 were typical for vintage tube amp.
The first modification was to bypass the pre-amp and tone control circuitry. To make it possible to still use the pre-amp if desired I disconnected the phase switch and placed it between the volume control and the pre-amp output. The pre-amp is connected only when the switch is pushed to the right. To avoid putting a switch contact in the signal path a length of high quality Cardas twin-ax cable was connected directly from what was the extra input jack to the volume control. When the pre-amp is being used nothing should be connected to the "extra" inputs. Likewise, when the direct wired "extra" input is used the phase switch should be in the left position.
With the pre-amp bypassed the lower midrange coloration was gone and there was just a bit more clarity. The measured frequency response shown in figure 3 shows a much flatter frequency response. In figure 4 we see that there is also a small decrease in distortion.
With the pre-amp bypassed the next priority was to assess the potential of this amp by measuring the performance without negative feedback. The results are shown in figures 5 and 6. The performance was about as expected, reasonable but not exceptional. The biggest concern was the significant roll off of the high frequencies.
The next change was to convert the output tubes to run as triodes rather than pentodes. This change reduced the power output from 25 to only 11.5 watts. Subjectively, the sound quality with triode outputs was considerably better.
Distortion measurements in figure 7 show a small reduction in distortion with triode outputs. Figure 8 shows little change in frequency response. The most significant change is the reduction of negative feedback shown in figure 9.
Since the 299-C uses a pentode input stage, converting it to run as a triode was the next target for improvement. This change had a significant effect on performance. The gain was greatly reduced and correspondingly there was a big reduction in negative feedback. Because of the lack of feedback the distortion in figure 7 increased. Again due to the limited feedback there is much more high frequency roll off (figure 8). More important there was reduction in overall gain. With this reduction in gain it was not possible to drive the amp to full power from my CD player. Because of the reduced gain and more limited high frequency this change was implemented such that it could easily be changed at any time. This was done by adding a pair of SPDT switches (one for each channel) near the 6U8 tubes to select triode or pentode mode for the input stage.
In spite of the measured problems the subjective sound quality with the input tube wired as a triode was much better. There was much more life to the music and an improvement in both detail and soundstage. Figures 9 and 10 give clues as to why. In figure 10 we see that the open loop distortion is drastically reduced. This confirms the common notion that an amplifier’s linearity before feedback is applied is a significant factor in it’s sound quality. In addition it has been my experience that reducing negative feedback improves an amplifier’s subjective quality. In figure 9 it is clear that by using triodes throughout that there is a large reduction in negative feedback. However, reducing negative feedback also reduces the amplifiers damping factor. With some speakers (particularly those with large impedance variations) lower damping factor can introduce frequency response and distortion problems. With my speakers (Ariels, designed by Lynn Olson) this was not a problem.
Two pairs of the corroded tin input jacks were replaced with high quality gold plated teflon insulated jacks. On the output side the old fashioned screw strips for speaker connections were replaced with a pair of Vampire gold plated binding posts. Only the 4 ohm tap is connected to the binding posts. If 8 or 16 ohm speakers are used, the binding posts must be re-wired.
During listening sessions there was some hardness and grit in the treble that became more apparent as upgrades were done. The new connectors eliminated almost all of the grit. It’s interesting how much difference "little" things can make.
During the initial listening session one section in the dual section 30uf power supply capacitor shorted out so all listening so far had been done with only 30uf capacitance in the power supply. The defective capacitor was replaced with a dual 47uf Elna Cerafine capacitor. The effect of the capacitor change was surprising. The sound improved across the board. The magnitude of the change was second only to changing the input tube to run as a triode. I don’t know if the change can be attributed to the increase in capacitance or if it is due to the superior performance of the Cerafine capacitor, though it is most likely some of both.
Replacement of the often unreliable selenium rectifier in bias supply was the next task. As noted earlier the bias supply had been modified. This modification had evidently been done due to low output from the selenium rectifier. With the rectifier replaced the original bias supply filter capacitors and resistors were reconnected and functioned properly.
For optimum performance and to eliminate the need for matched tubes the amp was modified to provide individual bias adjustments for each of the output tubes. The original circuit had no bias adjustment but did include a "DC balance" control. The DC balance control adjusts the relative bias between the output tubes but does not affect the overall bias level. Two 50k potentiometers were added (one for each channel) to facilitate setting the bias level. The existing DC balance controls were retained. This results in a slightly more cumbersome method for adjusting the bias but was easier to fit into the existing design. To measure the bias, 10 ohm resistors were inserted between the output tube cathodes and ground. In addition, four test point jacks were added to the left side of the chassis and connected to each of the output tube cathodes for measuring the bias levels. To provide enough adjustment range the grid resistors for the output tubes were adjusted slightly.
The final change was to replace the aged, often out of tolerance passive components in the power amp. The coupling capacitors were replaced with Audio Note paper and oil caps. The Audio Note caps were probably a bit of overkill, but I like their mellow sound. All of the resistors were replaced with precision Holco metal film parts. Several lengths of coax cable in the signal path were also replaced with Cardas twin-ax cable.
Upgrading the passive components removed the last of the grit and hardness. There was also a notable improvement in clarity and low level detail. However, the music was not very involving. The amp sounded "nice" and I could not hear any obvious flaws, but yet something was not right.
Frequency response measurements indicated 6db of roll off at 20khz. In my opinion a 1 to 2 db roll off at 20khz is desirable to soften the high end a little. However, 6db is far too much. It’s interesting that during listening tests it was not obvious that the high frequencies were so severely rolled off. Investigation showed that the high frequency loss was due to high impedance in the grid circuit of the input tube. The input tube grid circuit consisted of a 500k volume control potentiometer connected to the grid via a 150k resistor. The 500k volume control was replaced with a high quality 20k Noble conductive plastic potentiometer connected to the input tube grid via a 4.75k precision resistor. This resulted in the on target frequency response shown in figure 11.
With solid state circuits I am accustomed to trying to keep impedance’s low but it is not clear why this was a problem given tubes inherently high input impedance. Perhaps the same principal of keeping impedance’s low applies to tube equipment as well.
In figure 12 we see that the final distortion is has a different spectrum but the overall level is about the same as stock. However, this is obtained with much less negative feedback. Note that the final distortion for both input and output tubes in triode mode is much less than that shown in figure 7. During the resistor upgrades I discovered that the original plate resistors for the input stage were the wrong value (220k rather than 150k). This caused the input stage to be seriously under biased. With the proper plate resistors distortion was reduced at all power levels but most significantly at higher levels. There was most likely some additional distortion reduction due to better matching of the plate and cathode resistors for the phase splitter.
Figure 13 shows harmonic distortion at various power levels. These are excellent results for a tube amp, particularly one with so little negative feedback.
The initial open loop measurements were not particularly good. However, the high levels of distortion and high frequency roll off were both due to design and build flaws. By correcting these flaws the open loop performance is excellent. In addition to fouling up the sound quality, negative feedback tends to hide serious flaws that otherwise would be easy to identify and correct.
As noted in the previous section, this amp sounded good but something seemed to be missing. Well after upgrading the volume control and consequently extending the high end, this amp was transformed. The difference was simply remarkable. There was lots more depth, detail and most important magic. Often it is difficult to be objective with your own work so I asked my "golden ear’d" son have a listen. His jaw dropped and he exclaimed "what did you do!". I suspect that some of the improvement was the better quality potentiometer, but probably the extended high end was more significant. Either way I am at a loss to explain way the change was so dramatic.
The difference between the stock amp and it’s current form is quite dramatic. On the negative side there was a small loss in warmth, although there are big gains in every other aspect. It’s difficult to gage how this amp would compare with modern high end tube amplifiers. However, I expect that it would stack up well next to the best of them. I continue to be impressed with the value of vintage tube amps when properly restored and modified.