The first step in the building process of this preamp has been to get the active load boards ready. A reasonably quick assembly process this afternoon:
I’m aiming for a compact build here, so I mounted both boards with stand-off made of brass. The trimpots are at the top so I can adjust the operating point of the 01a.
After several careless tests, I blew some 2SK170 as I exceeded the VGS levels and also some DN2540 went pear-shaped with them so decided to add a protection 18V zener which I had at hand:
Also changed some of the component values based on what I have in my box currently. Nothing major but the frequency response is yet to be checked. I did a quick frequency sweep and the HF response was flat up to 60kHz or more. More than needed.
I added the cascoded LND150 to help improving the temperature response. Will see how this works in practice.
Now, looking at the FR response we can see that at very low signal level (e.g. 100mV) the output of this stage has H2 level at -75dB and the following harmonics at least 10dB down.
When looking at the response with a 1V drive (which is a very high input level for this stage in my system), the distortion increases from 0.02% to nearly 0.05% with the other harmonics 20dB below. H5 is higher than H3 surprisingly and even 2dB higher than H4:
So far, response has been as predicted and it mostly depends on the type of 01a valve used. I used a globe CX201a I have on the bench ready for abuse, not my best valve at this point for testing!
Will continue…
Cool.
What is the advantage of using the SK170 on the bottom of the cascode rather than another depletion fet? Is it “better” at low current?
cheers
tim
Hi Tim,
This is the first time I’m testing the normal FET (low-noise) in the lower position instead of a depletion one. An LND150 would have been ideal but given the current demands I have to use an DN2540 as the top device. The DN2540 has a typical input capacitance of 200-300pF whereas the 2sK170 is 30-50pF. Probably the latter given the low VDS value as the cascode bias voltages given top device bias method are not optimal (i.e. VDS is low). In theory the 2sK170 will provide far better HF response, which I confirmed briefly on the workbench last week.
I still need to listen to this design, so cannot fully answer this yet 🙂
cheers
Ale