decoupling capacitors in vacuum tube designs

[My3gger]

Both are decoupling caps, didn't catch that. RCA BA-1A has 1uF cap which looks quite small even for this circuit:
http://www225.pair.com/audio/waltzingbear/Schematics/RCA/BA-1A.htm

 

[moamps]

.. RCA BA-1A has 1uF cap which looks quite small even for this circuit..

Depends. Anode resistors for both stages have high resistance, so this cap does decoupling between stages very well. Besides, the power supply for this preamps looks very well engineered (unfortunately I don't have schematic), so I don't expect any significant noise from  that side. You should also consider a time when this preamp was built.

 

[DaveP]

The formula for this is in Morgan Jones page 301.

Vripple = half cycle time x amps/ C in Farads

His example with 50Hz power is:-

V ripple(pk-pk) = 0..01 x 0.12/0.000068F = 17.64V

In the US the half cycle time would be 60 x2  so 1/120 = 0.0083. not 0.01

Once you know the ripple and the voltage you can tolerate at any stage it is a simple voltage divider calculation.

For example:

A 10k feed resistor from  say 1V of ripple at 2mA and a 10uF cap would be:-

Where reactance of a 10uF cap at 120Hz is  132 ohms

1V x 132/ 10000 + 132  = 0.013V or 13mV.

Obviously a properly designed power supply would have several de-coupling stages before it got to the pre-amp stage described above.

I hope this is helpful

DaveP

 

[ruffrecords]

The attenuation due to an individual RC decoupling stage is:

2 x PI x f x R x C

So taking Dave's example of 10K and 10uF we get:

2 x PI x 100 X 10,000 x 10E-6 = 62.63 which will reduce the 1V ripple to about 16mV.

I prefer to convert this to dB which is 36dB. Three stages like this will get you over 100dB ripple reduction.

Not that a single RC section of the same total capacitance and resistance i.e 30K and 30uF gives an attenuation of only 55dB which is why it is better to have several smaller RC stages rather than one big one.

Cheers

Ian

 

[DaveP]

Thanks Ian,

Looks like I was in the right ballpark.

DaveP

 

[ruffrecords]

Thanks Ian,

Looks like I was in the right ballpark.

DaveP

It's capacitive reactance that always gets me. When the capacitive reactance equals the resistance you don't get 6dB loss as you would expect but only 3dB. It's the phasor thing.

The loss is (R + 1/2.PI.f.C)/(1/2.PI.f.C)

Multiplying top and bottom by 2.PI.f.C gives loss = (2.PI.f.R.C +1)/1 which approximately 2.PI.f.R.C

Cheers

Ian

 

[rmburrow]

There is no magic value for decoupling, only good engineering practices and verification.

Power supply decoupling is needed to reduce the HUM, RIPPLE & or SIGNAL from detreating the performance of the product. In the old days a scope was used to check for the problems and with today’s tools we have improved methods. Electrolytic capacitors have typical tolerances that are wide +20%, -50% is not uncommon. They sometime only work @ low frequencies and require additional HF bypassing.

I verify my designs or repairs with a scope and my Audio Precision analyzer to check for noise /signals on the power supply nodes. The Scope has high voltage input tolerance and has a 1 volt output that can connect to and analyzer for Analog or FFT measurements for debugging.

I also use a signal injection method to check the product for correct functions. Using a signal source I drive a current into the node and see what comes out of the product.
Duke

Electrolytic caps don't work well at RF frequencies (100 kHz and above)... It's a good idea to parallel electrolytic caps in critical decoupling stages with a 0.1 uF 400 v MYLAR capacitor which should get rid of RF "riding" on the power supply rails.  Also use a AC line filter (like one of those Corcom types) to minimize RF and other power line "rubbish"  "riding" on the AC power from the wall before the power transformer primary...

For stage decoupling (to minimize risk of "motor boating or other oscillation), 40 uF generally is adequate for most tube circuits, and the capacitor voltage rating must be greater than the power supply "surge voltage"...

 

[PRR]

All you need to know was in Radiotron 3rd, 1940, chapter 4.

http://www.tubebooks.org/Books/RDH3.pdf  {16MB PDF}

 

[pucho812]

All you need to know was in Radiotron 3rd, 1940, chapter 4.

http://www.tubebooks.org/Books/RDH3.pdf  {16MB PDF}

8)

 

[My3gger]

The formula for this is in Morgan Jones page 301.

Vripple = half cycle time x amps/ C in Farads

His example with 50Hz power is:-

V ripple(pk-pk) = 0..01 x 0.12/0.000068F = 17.64V

In the US the half cycle time would be 60 x2  so 1/120 = 0.0083. not 0.01

Once you know the ripple and the voltage you can tolerate at any stage it is a simple voltage divider calculation.

For example:

A 10k feed resistor from  say 1V of ripple at 2mA and a 10uF cap would be:-

Where reactance of a 10uF cap at 120Hz is  132 ohms

1V x 132/ 10000 + 132  = 0.013V or 13mV.

Obviously a properly designed power supply would have several de-coupling stages before it got to the pre-amp stage described above.

I hope this is helpful

DaveP

When calculating smoothing and decoupling i often read Scroggiefilters copy from Ian, continue with RDH beside others.
Not only for HT and anodes but screen grids, cathodes, etc. I just looked at V72 and don't understand well how 8uF cap between screen, choke and heater is doing:
http://www.trinaamp.altervista.org/Schemi/Siemens_V72_schematic.gif
My interest is also history of pro audio design in USA and Europe (Germany in particular, Danish...). For example,  V41 was advanced design already in 30's, while similar American OP-6 3 stage preamp was build in late 40's or early 50's. Can't say one school is better than another by looking at schematics. Without taking in account remote USA gear and lack of normal priced W.E. on the market, one can think their pro stuff was behind European. It seems Germans did big step in mechanical and electrical design soon after WWII, while USA had more than enough capacity to do great things.

 

[PRR]

> don't understand well how 8uF cap between screen, choke and heater is doing:

I wonder, on this tangled drawing, if you should RE-DRAW it to make the part of interest more straightforward.

I don't see any connection to heater (except that everything returns to common).

 

[scott2000]

Smoothing? And r6 is a humdinger...

Thanks PRR! That was mighty kind of you....