The 5687 is a nice, low distortion, 9-pin twin triode, developed by Tung-Sol in the 1940s for industrial-military use.
In audio, the 5687 was chosen by Hiroyasu Kondo for his legendary Audio Note Ongaku amplifier (and other Kondo designs). Pretty good credentials then.
The 5687 does not deliver high gain (mu of ~17) but a Linestage does not need much Gain, only around 20dB (10x) will do. The 5687 was reportedly clean sounding and low in microphonics. The signal schematic – it doesn’t get much simpler than this – just one tube + an output transformer and volume control (originally a DIY resistor ladder step attenuator, later changed to a DACT SMD resistor type, then a TVC Transformer Volume Control – more later). However, Diego’s design had some special PSU features:
- Huge, high energy storage 1800uF/400v cap for each channel, to emulate ‘quasi-battery’ supply for the 6X5 tube-rectified high voltage.
- A dual symmetrical SS regulated +/-6.3v heater supply to minimise cathode-heater leakage on the 5687.
The inputs were selected via 5v relays, minimising input signal path. I found that the relays (Omron G5V-2, silver/gold clad contacts) dulled the ‘dynamics’ and later reverted to a simple silver contact input selector switch.
Note that a single tube stage (except for cathode followers) will reverse phase and the OPT needs to be connected to correct phase.
This preamp sounded uncoloured and ‘invisible’ but not a lot of ‘magic’. However I had some serious problems:
- Noise problems with my attenuator switch (popping) and system hum.
- Frequency Response was not excellent, especially the LF and the HF response on the scope also showed some square wave ringing, indicating OPT poor performance. Loading down the secondary helped, a little.
I decided to re-house the PSU into a separate outboard box – but this did not solve the noise problems.
I tried box-encasing the OPTs, shielding them, different orientation – no improvement. I then contemplated springing for costly, new, higher performance Line OPTs. But finally, I threw in the towel, I gave up!
I decided to build….a JE Labs Linestage (story to follow).
Postscript: With a TVC at the input end and an OPT at the output, I was eventually able to kill the hum by breaking the -ve Grounds at input & output – ie. eliminating a couple of Ground/Earth ‘loops’ by using the TVC and OPT as isolating transformers.