Mullard 5-20

[alnewall]

Ok. so now is a good time to try reducing feedback.

But how is it done on this particular amp.

[Ianm2]

anyone make an easy guide to calculating feedback based upon resistor ratios?

al, its something to do with the feedback resistor, and the resistor to ground, I am not sure if its voltage or current, to the cathode, I have a feeling its current feedback.

In which case a bigger resistor will reduce current and reduce feedback? But I am wildly guessing..

anyone.?

[NealG]

Voltage Feedback.

You must know the gain of the amp without feedback.

Loop gain: is the gain of amp without FB x by FB fraction

Voltage gain (A):
5-20 for 20W o/p without FB needs 6.5mV input. V=sqrt of WxR or 12.65V. 12.65 / 0.0065 = 1946

FB ratio (B) is PD ratio of feedback network, shunt R / shunt R x feedback R

IE 100R / 100R + 5600R = 0.0175

Loop Gain (AB) therefore is:1946 x 0.0175 = Loop Gain of 34

Feedback factor is given as (1+AB) or 35

A Feedback factor of 35 is a 31db of feedback ((1/35) Log ) * 20


If you now changed the shunt R to 33R the FB (B) now becomes 33/33+5600= 0.00586

Feedback factor is therefore (1+1946 x 0.00586) = 12 Or 21.5db of feedback

Or

If you now changed the feedback R to 7K5 FB (B) now becomes 100/100+7500=0.0131

Feedback factor is therefore (1+1946 x 0.0131) = 26.5 Or 28.5db of feedback

Typically you would change the feedback R.

Note the HF compensation cap needs to be adjusted on test, with Paul’s favorite, a ‘scope and sig. generator. If they are not available, as a rule of thumb taking 6db out IE halving the feedback means halving the size of the HF cap.

HTH

[alnewall]

That's a comprehensive answer! Thanks for your time Neal.

Just to make sure i've got the basics right.
In a nutshell....
Feedback is reduced by either lowering the value of the shunt resistor or increasing the value of the feedback resistor. Right?

[NealG]

Yes, but typically you would adjust the feedback resistor.

Ooops! Just noticed a mistake!

[NealG]

Another thing! It's not so obvious to find out the gain of the amp without FB unless you can measure it. For K5881 MKII it stated to have 18db of feedback. and the FB ratio is 100R/100R+20000R = 0.004975

(((FBdb / 20)10^x)-1)

So, 18 / 20 = 0.9

10 to pwr of 0.9 = 7.943 take away 1 = 6.943

6.943 / 0.004975 = a Loop Gain (A) of 1396

(1+1396x0.004975) = FB factor of 7.945 or 18db

[david.rogers]

I have an unbuilt Maplin Mullard 5-20 kit I never got round to building. If anyone is interested I will put it up for sale - let me know. If so, will add it to the For Sale board.

[alnewall]

Another wet Sunday afternoon, still can't mow the lawn.

So I carried out the input mods, as suggested by Ian r in this thread.

Result...
Clearer, more natural mids and highs. Slightly less bass output but more tuneful. Overall a more laidback and maybe less dynamic sound. But better for it. Thanks Ian.

Also, earlier this weekend I tried a pair of the new World Audio Soniq caps at the Tripath amps input. After two days running.

Result...
Stunning clarity but seem to highlight the Tripaths weakness in the lower octaves, also seemed to reduce the height and depth of the soundstage. This may change with time.
Prefer the similarly priced Obbligatos in this application. They give a more balanced bigger sound in my setup.
The Soniqs are slightly less than half the size of the Obbs and are obviously easier to find room for. Maybe with a different amp and bigger bass drivers my impressions would have been different.

[Ianm2]

don't forget to try the srpp and mu stage later, Al, ....

[alnewall]

don't forget to try the srpp and mu stage later

I had forgotten, thanks for the reminder.

I can feel myself rapidly getting out of my depth.

There are more pressing matters however.
Having got rid of the horrible neon mains switch, I need to figure out a way of fitting a led to this thing. Is it ok to use the 6.3v heater tap as a supply???