For a headamp with high efficiency low ohm phones where no hum can be tolerated general rule is that one must buy the Tentlabs modules to use DHTs.
I know the standard solution are LM317 CCS's but these require capacitance on the output to avoid ascillations which defeats the purpose of a CCS.
However given the low .25A current requirement of the 71A, how would a regulated supply feeding 5 10M45's in paralell be for a DHT supply? Anyone tried it?
I know the standard solution are LM317 CCS's but these require capacitance on the output to avoid ascillations which defeats the purpose of a CCS.
However given the low .25A current requirement of the 71A, how would a regulated supply feeding 5 10M45's in paralell be for a DHT supply? Anyone tried it?
Not as complicated as it looks: http://www.diyaudio.com/forums/tubes-valves/38248-new-dht-heater-4.html#post446973
Not sure where you adjust the gyrator to deliver .25A vs the higher 300B current requirements.
Not sure what you mean about the LM317. The datasheet doesn't show any output caps for a CCS. A discrete CCS with a few transistors should handle 250mA easily.
You don't. R2 and R3 are a divider that set the voltage at DC. At AC, the gyrator looks like a high impedance source at the output. It acts like a choke, but it can't swing above the input voltage. In principle, you could use a choke in this position.
Sheldon
I know it doesn't, but I spoken to several who have tried and all found high freq oscillation without caps on the output rails. Headphones are very sensitive and pick up things speaker guys never notice.
I guess thats my issue with the gyrator. PMillet has a DHT fillament design basically an LM1085 voltage regulating into a choke. Yet in his commercial DHT headamp he had to use something better for the DHT filament supply, some sort of very high freq oscillator. So I'm not sure the gyrator would have the performance I need based on this limited info.
Thankyou dsavitsk, I imagine with a separate PS chassis its best to put the CCS portion in the tube chassis?
For lower current applications, a FET in the gyrator position works wonderfully. Choose a FET with the lowest (output) capacitance for the power rating. Choose a logic-level FET if you want the raw supply voltage to be low,, as LLFETs need only 2..2,5V across them (cp 4..6V for ordinary FETs).
In the gate circuit, use 470n..1uF film cap, [in C10 position of "New DHT heater"]- so you can set the filter resistance (R2, R3) into 500..910K range. A gyrator like this give huge noise & ripple rejection. If it's still not enough, run two in series. This should surpass any conceivable choke.
Yes, keep the trafo away from the DHT,, but the CCS near the filament, & twist the cables to the valve base.
For 250mA, set R4 to 2R2, and tweak on test.
In the gate circuit, use 470n..1uF film cap, [in C10 position of "New DHT heater"]- so you can set the filter resistance (R2, R3) into 500..910K range. A gyrator like this give huge noise & ripple rejection. If it's still not enough, run two in series. This should surpass any conceivable choke.
Yes, keep the trafo away from the DHT,, but the CCS near the filament, & twist the cables to the valve base.
For 250mA, set R4 to 2R2, and tweak on test.
I think parallel depletion mode MOSFETS would probably work OK but would like some headroom voltage. Probably no issue at all with 71A or 01A. I would use 3 or 4 DN2540s. With separate g and s resistors it may be a lot of parts. BTW, the DN2540 in any case style will not go much over 100 mA without positive gate voltage in my experience. The data sheet implies more but you should count on needing to parallel to get 250mA. You also will need at least 5V drain to source and maybe as much as 10V.
I think I'll build my DC filament supplies as per the attached sketch.
The transformer provides common mode isolation and allows the cathode to float on a very high impedance to "ground".
Since DC + ripple is basically a loop, i.e. only relative between the rectifier (+) and (-) outputs, a single MOSFET can act as ripple filter and bootstrapped gyrator.
I think making the cathode connection to either side is fine, but one could argue that the audio voltage across the filament is balanced with the resistor tap scheme.
There are low voltage MOSFETS that will have high gfs and decently low capacitance at small Vds.
You can also add your favorite tweaks, e.g. CMCs but...
Why would it need to be more complex than this?
I think I'll build my DC filament supplies as per the attached sketch.
The transformer provides common mode isolation and allows the cathode to float on a very high impedance to "ground".
Since DC + ripple is basically a loop, i.e. only relative between the rectifier (+) and (-) outputs, a single MOSFET can act as ripple filter and bootstrapped gyrator.
I think making the cathode connection to either side is fine, but one could argue that the audio voltage across the filament is balanced with the resistor tap scheme.
There are low voltage MOSFETS that will have high gfs and decently low capacitance at small Vds.
You can also add your favorite tweaks, e.g. CMCs but...
Why would it need to be more complex than this?
Attachments
- Status
- Not open for further replies.