Do you mean you measure the output signal first, then attach a 10k load, and the amplitude drops by half?
Ouch. I think your MOSFET input is too heavy at 3k||10k||10k. Check your load lines!
An externally hosted image should be here but it was not working when we last tested it.
No, the correct theory is still fine. What goes wrong is that the naive theory which only considers output impedance can mislead if you try to draw too much current. If a CF has a standing current of, say, 15mA then you cannot draw more than 15mA peak signal current - and you will get significant distortion before that point.Billy60300 said:
Post the circuit of what you have actually built, with voltage and/or current readings. If it can't cope with a load of more than 7k then my guess is that either you don't have a CF or it has much smaller quiescent current than you think.
Indeed, it matters that output current doesn't alter the operating point, or more specifically the gm of the valve.
High gm valves make good CFs, having low Zo. if that is key for the application. High gm pentodes in pentode mode, such as EF184 have Zo ~ 75R, which is a nice number.
Requiring high Io in audio is rare, unless driving reactive loads as per the OP. But then small power pentodes are neat, also having reasonable gm. PL802 is a remarkable combination for this application, BTW.
High gm valves make good CFs, having low Zo. if that is key for the application. High gm pentodes in pentode mode, such as EF184 have Zo ~ 75R, which is a nice number.
Requiring high Io in audio is rare, unless driving reactive loads as per the OP. But then small power pentodes are neat, also having reasonable gm. PL802 is a remarkable combination for this application, BTW.
When it comes to driving capacitive load, it is slew rate which determines output current I = C dv/dt being the law and all. So in audio, much depends on programme material AND level when working out what is an OK load for a CF into a capacitive load. Its not as simple as hf response, because that is level dependent due to slew rate being level dependant for a given f.
Good evening dear experts
I made it fine with a ECC88 as the output valve and dropping the Rk to 11 k.
The signal keeps constant and very good down to a 1,5 k 2,2nf load, with a good looking signal up to 75 khz.
Now I will still try and replace the 2nd valve by a low power Mosfet.
Thanks a lot for your help.
I made it fine with a ECC88 as the output valve and dropping the Rk to 11 k.
The signal keeps constant and very good down to a 1,5 k 2,2nf load, with a good looking signal up to 75 khz.
Now I will still try and replace the 2nd valve by a low power Mosfet.
Thanks a lot for your help.
Great, phew - how large is the output signal in this test ? It's worth thinking about headroom, but in terms of slew rate - which is a hybrid of frequency and level when dealing with sinusoids.
Another way of looking at it is decide on an upper f limit, say 15kHz, and find at what level you have the onset of attenuation. This gives an indication of headroom at the audio bandwidth you're interested in.
PS: most audio programs roughly follow a 1/f law for level versus frequency, and roughly a +12dB peak to average level ratio.
Last edited:
Good morning, You will love it: at 75 khz I have NO attenuation which I can see on the scope. The signal remains at 23 Vpk. My most recent tests with a low power Zetex Mosfet show a still slightly better result, it goes as low as 1 Kohm load ! attenuation starts then. At -3dB (half the amplitude) it is about 850 ohms load. I adjusted the headroom by pushing the input signal and adjusted the V1Rk to get the best symmetrically clipping. The bottom part of the signal goes first, but we have there about twice what i need ! I think about testing different mosfets according to their transconductance to see which gives the best results.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.