CALRAD SA 30:
This caught my interest on Ebay. It looked as though it might have decent output transformers which proved to be true, however the circuit design was simply awful. Since it powered up OK I had a listen - TERRIBLE. And so I traced out the circuit. I could not believe my eyes. 6BM8s in pentode connection into 3.6k p-p with 6.5dB negative feedback. The plate resistance of a 6BM8 pentode is around 20k so loading it with 900 ohms is not a good idea! I measured the output and it barely made 1W before the distortion shot off the scale. Given the loading this result was not surprising. However I checked the tubes, they were not great but within acceptable limits. I also checked the voltages around the circuit and was able to reconcile them with the design. The unit includes a phono stage. DC current for the first two stages is provided from the B+ supply; the heaters of the first two tubes are in series and both output stages are 'stood' on them! A resistor is in parallel with the heaters because the combined current of the four output tubes would be too great. Looking at frequency response I could not get a flat response from it, no matter what I did with the tone controls. The culprit was the balance control arrangement which effectively put 470k in series with the Miller capacitance of the next stage. There was a cap shunting the control to the grid of the next stage in an ineffectual attempt to correct this but the result was a huge variation in HF response as the balance control was rotated. Such a shame, the amp is well made and the output transformers turned out to be excellent. I found myself wondering how these things sold in the first place.
Click for the original schematic:
I decided to rework the design addressing the following:
1/ Reconfigure the output stage as triode PP. To avoid the need for negative feedback, I reconnected the output transformer taps so that 4ohm tap goes to the 8ohm terminal and 8ohm tap goes to the 16ohm terminal so as to obtain a nominal reflected impedance of 7.2k P-P. (The 4ohm terminal and 16ohm tap transformer are no longer used.) To do this properly, I needed triode curves for the 6BM8. A web search came up blank and so I plotted them:
Click for 6BM8 Triode Curves:
In addition to the triode conversion, I decided to use DN2540 depletion mode mosfets (available from K&K Audio) as constant current sources (CCS) to configure the output stages as power long-tailed-pairs. No matter how good the balance of the PP signals driving a PP output stage, there will always be some imbalance of the output transformer phases due to imperfectly matched tubes, transformer windings and coupling. The use of a CCS tail enables the output stage to self-balance resulting in near perfect even harmonic cancellation and improved low level resolution. I did this modification after first using standard cathode bias and I must say that the sound is somehow more articulate and effortless. Additionally, the static damping factor into 8ohms has improved from around 3.5 to around 5.
2/ I needed to remove the weird DC heater arrangement for the phono stage to liberate a higher value of B+ voltage for the new output stage. However, I also needed to leave a resource for a front end DC heater supply. I replaced the rectifier with 'fast' silicon diodes and installed a mosfet on a small board mounted over the rectifier socket. The mosfet is controlled by a long time constant zener arrangement so as to preserve the desirable slow turn on characteristics of the rectifier. Now, I can use the liberated 5V winding to run a voltage doubler and 6.3V regulator for the front end heaters.
3/ I left the classic DC coupled cathodyne 'phase splitter' alone. The line input stage used a 12AX7 and the gain was too high. I re-designed this stage to use a 6N1P. The 6N1P has a decently low Rp and I took advantage of that to re-design the balance control which was nominally a 1Meg pot in each channel. Actually, the pots measured around 1.8Meg. Thus, the Thevenin resistance presented to the Miller capacitance of the cathodyne input stage was around 450k, far too high. I took advantage of the low Rp of the 6N1P to re-configure the balance control as a variable shunt attenuator, using a 22k resistor in series with the 6N1P output. I replaced the pot with a 250k unit (what I had on hand) which is configured as a variable shunt to ground. Now, there is very little interaction with the Miller capacitance at audio frequencies.
4/ Now I was left with the question of what to do with the tone controls. I had noticed that the tone controls were badly off centre. On investigation, I found that the 1Meg pots measured around 1.8Meg! Furthermore, passive tone controls work best if there is an input and output buffer. I could use the 6N1P line input stage as an input buffer (of sorts) but connecting the output of a passive control stage to the input of the cathodyne stage was not a good idea since, once again, I would end up with the tone stage impedance interacting with the Miller capacitance of the cathodyne input stage. Then it hit me! I still had more gain then I needed for a nominal line level of 2V and so I could use place an active Baxandall tone stage around cathodyne input stage.
These modifications work really well. Since I am operating the amp in Class A the output is a modest 3W PC but it sounds really nice. Also, the tone controls make it fun to play with.
Click for the modified schematic:
Phono Stage Rebuild:
I replaced the existing phono-stage with a one featuring a hybrid cascode front-end. In the past, I have found myself wondering about the effect of Miller capacitance on a high-output MC or MM cartridge. (My second TT is fitted with a Sumiko BPS). This curiosity was stimulated by an article by Raymond Futrell (Audio Xpress, Jan 03) in which he described his analysis and listening tests pertaining to the shunt capacitance element (both due to the cable and due to the Miller effect) of the cartridge termination . His comment was that "the improvement in the sound was dramatic, not subtle. The dull midrange and shrill violins went away." I recalled similar a similar experience when experimenting with an all tube cascode and an ancient AT cartridge (type unknown...). So, I wanted to take another look (or "hear") at this. I'm glad I did. Low C termination most certainly helps the sound of my Sumiko. I quite liked this cartridge already but with the cascode, it is a great second to my Benz Micro LP2.
Click for the phono stage schematic: