DHT Filament Supply rolling off HF

[Xpersephone2]

I am using a SMPS filament supply for the attached driver (for a single ended tetrode output stage). If I power the SMPS power supply with my low voltage lab power supply the HF frequency response is -3dB at about 350kHz.



If I power the SMPS with a 12.6VAV secondary from a heater transformer the HF frequency response is -3dB at 60kHz.



I assume this effect is caused by the capacitance to ground of the heater transformer... is this the case and can anyone suggest what I can do about it??

 

[euro21]

350kHz is unrealistic expectation.

In my practice the 10/VT25/801/VT-62 family upper -3dB at about 100-140kHz, depends of circuit (CCS or gyrator loaded VAS stage or cathode follower).

BTW in the second stage why not use cathode bypass capacitor?

 

[Xpersephone2]

Oops, My earlier figures were for -6dB - which is indeed still about 350kHz when I use the lab psu, driving the SMPS, as the filament supply.

Here are the scope traces showing -3dB is at 230kHz using the lab psu

BTW the cathode resistor on the VT25 is bypassed by 47u (tantalum)+1u (polypropylene) caps in the SMPS.

I have also posted a 10kHz square wave, using the lab psu, driving the SMPS, as the filament supply, and using the 12.6V heater transformer, driving the same SMPS.

 

[euro21]

BTW the cathode resistor on the VT25 is bypassed by 47u (tantalum)+1u (polypropylene) caps in the SMPS.

No.
The 47+1uF BETWEEN filament points is not cathode bypass, it's filament voltage smoothing.
Try 200..470uF in parallel with 4k3+4k3.

The correct measurement is to use "real" loading (operating power tube's grid, or it's simulation: R//C).
No way that type 10 tubes reaches in this situations several hundred kHz upper bandwidth.

 

[Mark Tillotson]

Your filament supply is the dominant part of the cathode circuit at HF, 8k6 resistor cannot compete with it.

 

[Xpersephone2]

I stand corrected, of course the 47u + 1u is across the filament.

I will try to bypass the 8K6 as suggested.

However, the frequency response of this circuit does reach 200+kHz, the scope traces tell that tale.. I cannot see how that is wrong?

 

[euro21]

However, the frequency response of this circuit does reach 200+kHz, the scope traces tell that tale.. I cannot see how that is wrong?

It's not wrong, but the load is not real, thus measurement is better than real, and DHT's cathode is "floating" (AC).

If you use cathode bypass capacitor, the DHT filament both side is on GND potential at AC, so filament supply parasitic capacitance to GND is irrelevant.

Agilent MSOX2024A is good 200MHz 8 bit digital oscilloscope, but has 1MOhm input.

Try 100kOhm // 25 ...100pF as load.

 

[Xpersephone2]

Euro21 - with a load of 100K / 47p in parallel, the HF is -3dB at 170kHz. In fact 33K makes little difference - cutting that back to 160kHz.


I have not tried bypassing the cathode resistor - but does this make any difference with a CCS anode load?.

 

[euro21]

-3dB at 170kHz. In fact 33K makes little difference - cutting that back to 160kHz.

It's looks better -closer to reality- than almost infinity load.

I have not tried bypassing the cathode resistor - but does this make any difference with a CCS anode load?.

A little higher gain and -especially- independence of heater supply parasitic components.

In this situation the earth independent lab supply and real PSU difference would negligible.

 

[Xpersephone2]

Euro21, you are correct. With the 8k6 resistor bypassed the transformer PSU and bench supply perform the same. The point about the CCS is subtle but in the end obvious - the AC impedance of the cathode resistors is infinite because of the CCS anode load.



Thanks for your help.