I upgraded the DHT filament regulators to version 4. Rod has released a new kit which introduces temperature compensation for high current filaments. This is a key feature in my 4-65a design and I shall explain why:
The 46 drivers is working on filament bias. The whole amplifier as you know gets very warm. When temperature rises the regulator current sags a bit. This is mainly due to the CCS MOSFET drain current variation due to impact on VGS on temperature. If the filament current drops from 1.75A by 100mA this will be changing the bias point by 100mA*10Ω = 1V. Thanks to the anode gyrator load and its FET which will be maintaining the anode and mu output voltage to the expected level. Therefore the impact is on the anode current as we are varying Vgk by 1V then the anode current will change and anode voltage as well, but not the mu output which is fixed. Remember there is a mu resistor between the FET source and the anode. If the current changes then the anode voltage will change as well. The conclusion is that the driver is decently stable thanks to the gyrator voltage that is fixed (providing the voltage reference doesn’t drift with temperature as we will see later).
However, the output stage bias is dependent on the 4-65a filament current somehow. If the 3.5A regulator current sags then the Vgk will change with this. The grid voltage is set by the 46 gyrator load whilst the 4-65a filament current drops increasing the Vgk voltage (i.e. making it more negative). Therefore the anode current of the output stage drops slightly after 1-2 hours of playing. I was using plastic TO220 components which are not appropriate for these regulators if you are using +3A currents. You either need bigger heatsinks or use the right metal case TO220 devices.
Only the gyrator FET drift with temperature will impact the output voltage. From the IXYS curves for the 01N100D we can see that Vgs changes from -1.9V to -2.2V from 25C to 125C for a Id of around 25mA. This change of about 0.3V is also bringing slightly down the anode current of the output stage. Only solution here is a pretty big heat-sink for the gyrator to keep it cool enough. The bigger impact is on the gyrator CCS voltage reference. This is the LND150. and here is the temperature impact from the datasheet:
Well, simulating my CCS voltage reference in SPICE I found to get about 1V drift in 10C (from 20C to 30C). I could look at reducing the CCS current (down to 500uA) to minimise impact on Id versus temperature. This will be next step
Overall the Rod Coleman Regulators V4 are fantastic. Stability brought to the amp and sound impact is impressive. The tail CCS of the regulator has been optimised by Rod to perform really well in this scenario. I could notice a bit more clarity in the sound when testing the amp yesterday. A big step-up highly recommended!
Hi Ale,
over 10 years I am using my own DHT filament regulator. I never changed my starting setup: low current CCSed shunt (LND150) feds mosfet gyrator (IRF820) to f+. Ring-of two bjt CCS (TIP41C/BC547B) to f-.
I only once tried to change position combines Rk and Ck. For me, the best sound is Ck on f + and Rk to f-.
In the meantime I started using simulation and I saw that AC analysis looks much better with such a combination of, especially phase angle.
This is a contradiction to what Rod suggests, but I recommend you try this: Ck on f +, Rk to f- and listen few hours.
Best diying
Rajko
Hi Rajko
Thanks for the input. Interesting design for your DHT regulator. High current ones (+2A) are tricky.
Not sure if I follow your setup. Is this for output stage with cathode resistor and bypass cap correct? What you are suggesting is to move the Ck out from being in parallel with Rk and connect it to the F+? F- will stay on Rk, correct?
I’m about to try a 10Y linestage with this configuration so would be keen to get your simulation to play around if you have it at hand.
Thanks
Ale
>>>> Is this for output stage with cathode resistor and bypass cap correct?<<<<
Correct. It is classical 300B output with Rk and Ck.
I tried all variations of connections Rk and Ck to 300B filament/DHT regulator. As I said, in my system the best sound is Ck to the f+ and Rk to f-. In subsequent simulations, I saw why.
Rajko
BTW, my amplifier is C3g pentode mode driving 300B SET.