LEDs used as Cathode Bias

[acorn]

Hi Lads

What I am really concerned about is that nasty 1000uf in series with the cathode bias resister of the ECF80 cant it be soldered across the resistor.

Electrolitics of this size can be very detrimental to the overhall sound is there any modification I could implement to get around this pesky cap, over to you chaps.

Acorn

[NealG]

Unfortunately no, you'll need to redesign and recalculate the feedback cct to bypass in the way you want. I'm sure it can be done but I don't have the inclination! You see Andy Grove is a right clever fellow and this cct I think he first used in K6L6, then Kel34, Kit88, Kel84 etc. Cathode Bias is developed across the feedback resistor (R32/33 Kel84) and R8/9 forms the potential divider for the feed back. C6/7 is there for LF compensation increasing LF feedback by varying the potential divider according to frequency. You could change the value a little to experiment, smaller to reduce LF and subjectively speed the bass up or higher to go really deep and bloated!

[Richard]

Here you go. Working through the figures for 1Hz bypass cap for ECF80 pentode section gives 220uF for the 680R. The fb ratio is the same as the original and the cathode R value is similar as makes no difference. You could roll LF off earlier with 100uF or 47uF if you wanted. This way allows roll off to be set using cathode generated fb and global fb is kept linear. I suspect the fb cap C16/17 will end up much smaller when set with SW.

Kel84 method connects its global fb divider leg to ground via the cathode bypass cap and includes that leg in the cathode resistor bypass. A 220uF cap would bypass some of the cathode resistor generated fb but it starts to increase it's impedance below about 60 Hz and would progressively block the divider at LF so there would be progressively more global fb applied, rolling off response earlier. Hence the use of 1000uF which doesn't roll off till below 10Hz.



Rich

[acorn]

Richard

Thanks for that Richard, thats a good idea I will try it and see how it sounds on one channel and if it sounds better than the 1000uf I will implement the other Channel.

Acorn

[NealG]

Nice one Richard.

Did you really stay up until 1:40AM working it out!?

Seriously, I will be interested in Acorn's thoughts after trying this. Not sure if I have read the original cct correctly but with AG's layout it seems that Rk is not bypassed and so has local feedback. Bypassing Rk is going to remove it changing, among other parameters, the gain of the ECF80.

[Richard]

Originally Posted by NealG

Nice one Richard.

Did you really stay up until 1:40AM working it out!?

Seriously, I will be interested in Acorn's thoughts after trying this. Not sure if I have read the original cct correctly but with AG's layout it seems that Rk is not bypassed and so has local feedback. Bypassing Rk is going to remove it changing, among other parameters, the gain of the ECF80.

Hi Neal, I think I was awake he he... Circuit was easy but explaining it was much harder.

Neither bypass Rk completely, AG's bypasses more in fact, his is stood off 20R whilst the new version is stood off 50R. There will be a little less gain overall with the new version but it's the roll off which is interesting,

Have a look at these 2 impedance curves which show the ac impedance cathode to ground in each case including a figure of 161 factored in for rk.

This Kel84 one rolls off earlier and shallower but the 22R divider leg impedance increases as it does so, causing more global fb, so actaul roll off will be steeper,



The new one rolls off later and global fb will be linear as set by the divider,




Rich

[NealG]

Just theorizing, I'm not 100% sure those curves tell the whole story Rich. The cap seems to have a different function in each case, at least that’s the way I’m reading it. The classic way is to bypass Rk at all audio frequencies thereby reducing the degenerative feedback caused by Rk, we know this cap has a great bearing on sound quality. I would expect the gain to be a fair bit greater not less.

With AG’s cct the cap is acting more like a variable shunt in conjunction with the loop feedback and seems to me that it’s function is not to bypass Rk in the same way but provide LF feedback compensation. AG always said his cct gave great bass performance, I’ve not compared the two directly so I’m interested in Acorns thoughts when he tries it. I hope AG’s cct is not just a case of wanting it to be ‘different’ for the sake of it. Of course I’m most likely wrong and talking a load of hot air!

[Richard]

Hi Neal,

Updated the graphs and words as I'd forgotten to include the parallel rk value of 161.

The principle and operation is the same though. The bypass cap does the same function in both cases bypassing local feedback from the cathode resistor.

The main difference though is that with AGs version that cap also increases global fb as f falls.

Here's what happens to that divider leg as f falls,



As the divider leg impedance rises more global fb will be fed to the cathode. The benefit, I think, is that more global feedback gives lower OP impedance and so tighter bass?

Rich

[NealG]

I went and asked, there is more to this than meets the eye as I thought! AG was concerned with subsonic instability, his cct is one way of combating it. In a valve amp there are many LF time constants and when you apply loop feedback you can can get subsonic instability.

You can hear it with many amps, feedback or with none. There is a bloom to the Bass or a bloatedness indicating the possibility of subsonic instability in the amp. Many are just not aware of it's existence.

Acorn if he tries it may get more Bass, or at least he may think he is getting more but it could be that the amp is now unstable at subsonic F.

[Richard]

Hi Neal,

I explained how AG's works. It's clever.

Whether it actually gives better bass or not I don't know.

In Acorn's case the idea wasn't to get more bass. He will be able to vary roll off using values between 100uF and 1000uF to see what the effect is, and if the bass has any better, or different, quality.

Rich