Hi folks, having installed 6L6s in my TL10 with the leak recommended modifications, I have been looking at the on load frequency behaviour with my scope. The leak mod suggests omitting the cathode bypass capacitor (I've installed the single/common bias resistor config). Applying a sine wave and sweeping the frequency from about 100Hz to 18kHz, I find that the wave shape suffers visible distortion up to about 10kHz, then the shape improves until 18kHz. If I bypass the cathode resistor with a 25uF cap, the waveform improves below about 10kHz and worsens above. So I seem to have a choice of either high (with cap) or low (without cap) frequency distortion. Any observations or advice welcome. Cheers. Matt
Do you know if the 6L6s you have are fairly well matched? A bad mismatch could well contribute to ugly high frequency waveforms - there's less feedback up there to clean things up.
If you have a single bias resistor, any mismatch between the output valves could become exaggerated further - one might be drawing significantly more quiescent current. Not only does this cause a magnetic offset in the output transformer, the valve drawing more current will likely have more Gm at its operating point than the other - compounding the issue if it was the one with more gain in the first place!
If this is the issue or not I would still recommend independent cathode resistors / capacitors as per the original schematic, with a suitable value for 6L6s. A single resistor introduces more problems than it allegedly solves.
If you have a single bias resistor, any mismatch between the output valves could become exaggerated further - one might be drawing significantly more quiescent current. Not only does this cause a magnetic offset in the output transformer, the valve drawing more current will likely have more Gm at its operating point than the other - compounding the issue if it was the one with more gain in the first place!
If this is the issue or not I would still recommend independent cathode resistors / capacitors as per the original schematic, with a suitable value for 6L6s. A single resistor introduces more problems than it allegedly solves.
Last edited:
Thanks for your reply. The amp came as standard with the single cathode resistor. I have measured the anode currents and they are within a couple of milliamps- they were JJs bought as a matched pair. I have subsequently removed the load and looked at the open circuit waveform and the effect disappears. Maybe it's just the way the amp is when delivering power, maybe the o/p transformer is showing its age. I will try the separate cathode resistor config, just to see if things change. Was the single resistor method employed just for cheapness?
Obviously matched pairs are a necessity. That aside, when you run a PP output pair from a single cathode resistor without bypassing, you have created an LTP with a low impedance "tail". In my experience with my own PPP KT-120 mono-blocks, switching from a low impedance resistor to a high impedance CCS results in a significant improvement in SQ. As with all LTPs, a high impedance cathode "tail" acts to enforce AC balance. Doing this does restrict the output to class A operation. If you can live with the reduced power output, not a bad thing in my opinion.
SQ = sound quality.
How much current do you need to sink? In my case it's 300mA. I tried running various cascoded CCSs in parallel and had issues with oscillation. I do not recommend paralleling CCSs. If you current requirements are such that you can use a single cascoded depletion mode MOSFET current source then that should work. In my case I ended up using a TL-431 variant to control the current in a large power MOSFET. Kevin Carter at K&K Audio was kind enough to humor me by designing the PC board for this and supplying the boards. I believe he now will sell this type of current source as a kit. He also sells kits for cascode MOSFET CCSs. If you ever do use a CCS like this on the output stage of an amp be conservative with heat sinking and preferably have the fins external to the amp.
As an aside, I once had an output tube arc during a crescendo. There was no damage and the amp continued to play after the arc cleared.
How much current do you need to sink? In my case it's 300mA. I tried running various cascoded CCSs in parallel and had issues with oscillation. I do not recommend paralleling CCSs. If you current requirements are such that you can use a single cascoded depletion mode MOSFET current source then that should work. In my case I ended up using a TL-431 variant to control the current in a large power MOSFET. Kevin Carter at K&K Audio was kind enough to humor me by designing the PC board for this and supplying the boards. I believe he now will sell this type of current source as a kit. He also sells kits for cascode MOSFET CCSs. If you ever do use a CCS like this on the output stage of an amp be conservative with heat sinking and preferably have the fins external to the amp.
As an aside, I once had an output tube arc during a crescendo. There was no damage and the amp continued to play after the arc cleared.
Last edited:
Doh, SQ. I was thinking too technical!
I need about 70mA total. Now you've pointed me in this direction I'll do a bit of research. Thanks again. M
I need about 70mA total. Now you've pointed me in this direction I'll do a bit of research. Thanks again. M
At 70mA and I'm guessing about 50V or so you'd be looking at ~3.5W of dissipation. A simple cascode MOSFET should work fine. The IXYS MOSFETs that Kevin uses in his kits perform better in listening tests than the usual DN2540s.
Thanks. I found a good tube cad article on ccs and cathode bias. I'm gonna try using a 317 with their diode/capacitor bypass circuit first- mainly because I have those parts to hand. Best. M
- Status
- Not open for further replies.