A very interesting point was raised on the 4P1L DIYAudio thread around the gyrator circuit using CCS and whether a simple resistor divider was better than the CCS due to the LND150 temperature drift.
I’ve tried both options and I’d say I prefer the CCS despite the variation with temperature for the following reasons (which people may well disagree):
- It’s true the LND150 varies a lot with temperature (see attached), however if it’s operated at low current (e.g.<500μA) the variation is small. In a cascoded pair for this circuit the drift in the output voltage is small. Simulated in Spice I get about 6.35mV/°C. The resistor divider will be better of course but you need a smaller values to reduce impact of dR/dT. This creates another problem which is the reduced PSR. With a compromise divider to balance idle current and PSR you can get 5 times less variation with temperature in the circuit under discussion – see below (e.g. 1.4mV/°C)
- For a smaller value of resistor divider the PSR is impacted and significantly lower than the CCS. If you don’t have a well filtered supply, the PSR benefits of the gyrator will be reduced due to this. For example, I did some quick comparisons by simulating my 01a preamp. I used a 235KΩ/220KΩ and a 23K5Ω/22KΩ divider options with a typical film decoupling cap of 4.7μF.
The PSR of the CCS is above 100dB whilst the PSR of the resistor divider goes from 56dB (235KΩ/220KΩ divider) down to 37dB (23K5Ω/22KΩ divider).
In practice, I implemented two different circuits as I had a shunt regulator before when I had a resistor divider and now I don’t have any shunt regulator but I use the CCS version.
Looking at the output PSR as the gyrator provides additional rejection to noise. The resistor divider PSR is about 73dB and CCS is 30dB better anyway
The voltage variance is really small with temperature and this circuit in particular isn’t affected by such small drift in my view
Great work Ale,
Did you simulate a small capacitor (as small as 0.1uF) across R3 in your LND150 version?
Merry Christmas,
tim
Hi Tim
I’ve tried and doesn’t make any significant difference. What were you thinking there?
Happy Christmas!
Ale
Hi Ale,
I just noticed you used 4.7uF with the resister divider sim but nought with the LND150 set up.
With GP’s CCSs I figured the 0.1uF wasn’t just for coupling.
Using significantly less that 0.5mA also seems a good idea.
Merry Christmas
tim
But temperature drift of DN2540 in CCS seems much lower, right?
I have no practical observations.
After about 20 hours of listening:
Comparable to my old CCS (01N100D/DN2530), the gyrator sounds more like a tube sound.
CCS sounds a bit sharper with little faster bass. HF are identical on my supertweeters.
But with a gyrator sound there is a better live background. I like that very much.
Gyrator is made with 01N100D/BF862/LND150; CC gate resistors and NOS Beyschlag MF others. About 0.6 mA through LND150 and 15 mA through BF862. Measured on my mA-meter.
The gyrator capacitor is 1uF Ansar-CPA capacitor. Not tried another capacitors, I have great confidence in Ansar cap`s. Maybe here some Siemens polystyrene sounds the best, but it’s hard to find such a big value.
Rajko