Low voltage-drop CCS for LTP input stage?

[Svein_B]

I am considering conversion of the input stage of a small EL84 amp to use differential LTP with CCS.

The challenge: I do not have any negative voltage available, and space is very limited so it is difficult to add another tranny. I would like to avoid adding input caps if I can.

Here is the idea: Generate negative voltage from the 6.3V filament AC. This could probably get me -7 to -8 volt and would give around 9V of room for the CCS. Would this be enough?. Are there any CCS solutions that would be comfortable with such a small "headroom" ?

The input tube is presently 5670, but this could be changed if there are other suitable driver tubes that would get more voltage up to the cathode.

My approach will require grounding one side of the 6.3V AC instead of the center, but for a line level/output stage I would assume that this would be acceptable (I have seen some highly recognized commercial amps doing the same thing.)


The attached diagram illustrates my idea.

Svein

 

[reinhard]

Will work, try a 2SK170 Fet with resistor to set the current.

Reinhard

 

[panos29]

You can also try doubling on the filanment vontage to get around 12V. I tried LTP cascoded but altough good on distortion measurements and swing, the sound seems kind of transistorized. There is no warmth and not that analytical. On the other hand you can also try normal LTP with a mosfet CCS with 5670, that will drive the el84s o.k, and in my ears is better than the cascoded option. In orer to adjust the operating point of cascoded you need a spectrum analyzer in my opinion (i use spectrum lad with excellent results).

 

[darkfenriz]

why do you need negative voltage?

 

[Svein_B]

You can also try doubling on the filanment vontage to get around 12V.

How would you go about doubling the filament voltage for -12V, while still maintaining a stable ground reference for the filaments to avoid hum ?

BTW: My circuit is non cascoded. the EL84 is sooo easy to drive

Svein.

 

[SY]

9V of room should be adequate, though barely so. For an input sensitivity of 2V, the max signal at the cathode will be 1V (or 1.4V peak). So you've got 7.6V of margin, certainly enough for something like a ring-of-two or a chip CCS.

 

[Svein_B]

why do you need negative voltage?

Well, I thought that this would be necessary in order to be able to keep the grid at zero level and avoid an input cap.

Then I found a reference to the Coldwar amp in this thread. It also uses the 5670 tube in LTP with CCS, but no negative voltage. I do not understand how it works.

Svein

 

[SY]

Redraw it like this and the operation becomes a bit clearer. The voltage across the resistor always has to equal one diode drop, thus the current is held constant.

 

[Svein_B]

Aha!. So the mystery is solved. Thanks SY!

Although the Coldwar referred to is running at 300V/8mA, I guess it must still be a "tight fit" for the Widlar CCS under the cathodes.

I will try with the -8V and a "ring-of-two" CCS. Then the next challenge will be how clean power it can provide, and if the slightly unbalanced filaments will cause any hum.

Svein.