Rp = 2Ri

hi!

when finding out a working loadline and bias spot and aiming for the criteria:

plate resistor is twice the internal resistance of the triode at the working point. Rp = 2Ri

is Ri the same for triodes with unbypassed and bypassed cathode resistors?

or do we look on the plot for bypassed
and use the formula Ri = ra + (mu + 1)*Rk for unbypassed ?

where ra is the internal plate resistance at the bias point.

Bypassing the cathode of a tube stage will reduce the output resistance ,it will also increase the gain. As in the case of unbypassed, the signal voltage appearing at the cathode causes a loss of gain .the value of the bypass cap determines the low frequency performance of your amplifier. Using a plate load of x2 the output resistance of the tube is generally regarded as the minimum,you certainly can go higher ,you'll get a little more gain but loose some bandwidth in the high end. On the output side the following  grid resistance is generally made as high as practical,  as the following grid resistor value drops  a larger coupling cap is required for the same low frequency response  and the lower the load is presented to your circuit.

hope thats some help to you

5v333 said:

hi!

when finding out a working loadline and bias spot and aiming for the criteria:

plate resistor is twice the internal resistance of the triode at the working point. Rp = 2Ri

is Ri the same for triodes with unbypassed and bypassed cathode resistors?

Click to expand...

No

or do we look on the plot for bypassed
and use the formula Ri = ra + (mu + 1)*Rk for unbypassed ?

where ra is the internal plate resistance at the bias point.

Click to expand...

Sort of yes. The plots show voltages relative to the cathode. With unbypassed, the the cathode to plate and cathode to grid voltages vary  due to the negative feedback so you cannot use the plots to predict overall gain or distortion - they just show what the tube is doing.

Cheers

Ian

cool!

ive made some cathode lines, tried different loads and using the formula until i fell into the zone Rp = 2Ri

with the rather small B+ i have for now i get quite low idle Vdc and Adc.

but its worth trying out i guess. if it sounds like it should i might have to source a bigger mains transformer.

further checking the plots, that same Rp and Rk combination seems to work just as good with any  higher B+


got some resistors from the shop before i went to work. looking forward to get home and warm the iron up!


by the way.

since the unbypassed cathode yields a much higher Ri, it seems to get harder to make the amp however you want it.

is there any clever way to reduce the resistance? like applying global negative feedback..?

5v333 said:

is there any clever way to reduce the resistance? like applying global negative feedback..?

Click to expand...

The unbypassed cathode resistor provides series derived. series applied negative feedback (NFB). Because it is series derived it raises the output impedance. Because it is series applied it raises the input impedance.

To lower output impedance you need to use shunt derived NFB. In audiophool circles, Shunt derived, shunt applied NFB around a  triode is often called an anode follower.

The simplest way to reduce the output impedance is to bypass the cathode resistor thereby removing the NFB. If this gives you too much gain the either try a tube with a lower mu or use a split anode load resistor.

Cheers

Ian

The unbypassed cathode resistor provides series derived. series applied negative feedback (NFB). Because it is series derived it raises the output impedance. Because it is series applied it raises the input impedance.

Click to expand...

this i do not understand...

To lower output impedance you need to use shunt derived NFB. In audiophool circles, Shunt derived, shunt applied NFB around a  triode is often called an anode follower.

Click to expand...

would this be the same as local feedback? plate to grid?

The simplest way to reduce the output impedance is to bypass the cathode resistor thereby removing the NFB. If this gives you too much gain the either try a tube with a lower mu or use a split anode load resistor.

Click to expand...

my eyes just fell on 6n30pi.

the concept of feedback can be a hard one to grasp alright ,
heres an article from way way back written by Cathode Ray, real name  MG Scroggie ,he has a great abillity to put complicated things into easy to understand language .

https://www.scribd.com/document/312546624/When-Negative-Feedback-Isn-t-Negative-the-Cause-and-Prevention-of-Oscillation-Distortion-Cathode-Ray-Wireless-World-May-1949

do you think my approach is a sound one, in finding a tube,  a loadline and a bias that  fullfills the criteria?

what is your approach when setting up a common cathode (unbypassed) resistive coupled stage?



thanks for the link by the way!!

Well Im very much trial and error in my design technique ,I dont tend to use a whole lot of maths to work things out ,tubecad/Se ampcad  does come in very handy though ,does all the number crunching for you . Another very usefull program is Psud from Duncan designs . I find for high quality tube mics I dont often need much more than 20db or so of gain ,keeping the input transformer seperate from the pre works for me ,then you can select the most appropriate transformer  as the need arrises . I did make a pre recently using an E88c ,low impedence triode ,but with quite high gain ,10k anode load ,cap coupled into a step down transformer ,so I can drive mic inputs if nessesary ,I added feedback from the transformer secondary through a pot into the lower end of the grid resistor, input impedence does vary a bit depending on source and the amount of feedback ,but does allow me to trim my input levels .I guess the best way to go is to have some idea of the amount of gain you need before starting on a design . Like with all tube gear its involves high voltages ,putting a high value resistor across the low end of the Ht supply to ground (220K ohm 2 watt ) is a good plan ,as this will bleed off any residual voltage on your smoothing caps within a few minutes of powering the unit down  . Of course at some point you will need to test the voltages with the circuit powered ,make sure the multimeter has a good ground connection on the equipment ,a croc clip works best here. I also usually insulate the probe appart from the very tip ,with a piece of wire insulation ,just to lessen the chances of shorting something  and sending sparks flying!
Old record players and reel to reel tape machines from the tube era make a great place to find cheap transformers and chokes  for the experimenter , surplus pc powersupplies  contain high voltage Electrolytic caps which are perfect for your power supply also. Many great books and articles from the old days are up online free to download nowadays ,The radio designers handbook is one you can find easily ,If you want something a bit more up to date any of the titles by Morgan Jones are full of great stuff..... have fun.

5v333 said:

do you think my approach is a sound one, in finding a tube,  a loadline and a bias that  fullfills the criteria?

what is your approach when setting up a common cathode (unbypassed) resistive coupled stage?



thanks for the link by the way!!

Click to expand...

Well, so far you have not actually said what you want the circuit to do. What gain do you want, what input impedance, what output impedance, what drive capability, what distortion, frequency response etc. That is the first step.

Cheers

Ian

> aiming for the criteria: plate resistor is twice the internal resistance of the triode

Why?

That's a high distortion condition.

It ignores the *load to be driven*.

It picks a triode before the job is defined.

What is your *load*? For much small in-box audio, it may be near 200K+100pFd. This is really 200K to midband, falling to 20K at 80KHz. That's ample for AM radio, dubious for hi-fi. So you need to ask: how much supersonic response do you need. How much capacitance?

Be realistic. Tubes are high-resistance devices. You won't drive heavy low-Z loads without heroic effort (hot chassis).

Pick RP (plate resistOR) 1/2 to 1/5 of the worst-case load.

Find an Rp (plate resistANCE) 1/2 to 1/5th of RP.

Generally you can't find an Rp (1/5)*(1/5)= 1/25 of the load. Even taking 200K, that's 8K, and leads to low-Mu triodes (or medium-Mu with very fat cathode and high cathode current).

it will drive  a 6sn7 which drives the output transformer. (about 80pf..?)

load resistor will prob be 100k.

i would like low distortion and high bandwidth for pleasureable listening and some what transparency.

the whole amp  is push pull, transformer in and out, and will do about 20db gain for mic amplification and passive eq gain.

the output tube 6sn7 is doing almost all the gain so the first stage is for dbs that will be sacrificed in global feedback.

ive been playing with ecc802 and e88cc. and had a hard time reaching Rp = 1/2RP without ending up far right and and quite low on the curve plot.

Ok ,thats a bit more to go on .
You will need to provide balanced drive to the grids of the 6sn7,
You could use a split load arrangement with a single section of an e88cc or seeing as your using an input transformer ,you could use both sides of the e88cc in  push pull and avoid any phase splitting  . The benefits of using push pull right through the amp will be mains hum tends to cancel ,you'll also have a higher drive voltage available at the grids of the 6sn7,and incidental capacitance will tend to cancel ,giving wider bandwidth. Adding feedback to a fully balanced circuit requires a bit more thought than in the case of single ended ,but it is do-able .Feedback across two transformers and two gain stages is a bit more complicated also, as at the frequency extremes phase shift will mean your negative feedback may  end up being positive which tends to be unstable .Its suprisingly easy for an amplifier with feedback over multiple stages to become an ultrasonic oscillator!
Keep feeding in more details about input/output impedences and transformers you plan to use
E88CC gain stage and 6SN7 output seem like a reasonable choice  for what your doing though .
Ill post a new thread on one of my preamps  later ,and hopefully get Ian(ruffr) and Prr  to critique it and help me iron out a few bugs.

Love you guys on this forum!

Im working with live gigs today n tomorow.

I can post a precent schematic when i have the time.

I havnt thought  about capacitance before. Thanks for the headsup on this!!

> it will drive  a 6sn7

Then don't bust brain cells. You don't need a ton of output. You don't have a ton of capacitance. Use the values in the Resistance Coupled Amplifier table for that tube. Some white-shirted junior engineer worked all morning to tabulate those values, and they are usually as good as it gets.

back from workin some concerts..

attached my circuit as it is right now.

the mu of the first tube dont need to be higher than 20 actually.

Then don't bust brain cells. You don't need a ton of output. You don't have a ton of capacitance. Use the values in the Resistance Coupled Amplifier table for that tube. Some white-shirted junior engineer worked all morning to tabulate those values, and they are usually as good as it gets.

Click to expand...

didnt know about theese tables! thanks for the info!
however most tables seem to adress bypassed cathode stages..? my main wondering in this thread is about the RP=2Rp criteria of an unbypassed stage. should i care ? using thoose numbers i get for ex Rp = 56700r for RP = 47000r with 250V B+ and 870Rk.


i will see if i can make the first stage bypassed and loose some gain somewhere, perhaps connect the input transformer 1:2 instead.

for now i get a bandwidth of about 30-35Khz but theres some instability that needs to be adress and this brings the bandwidth down even more. inside the circuit the bandwidth is more around 45Khz...

Layout /wireing can be very tricky when it comes to feedback amplifiers ,incidental picofarrads between the wrong points in the circuit can make the difference between oscillations and not .

Your main problem right now is the two poles due to C1 and C3 (and the corresponding pair in the other half).  Without doing any calculations it is hard to know how close together they are but with similar values there is a good chance they are not far apart. For stability you really need these poles to be a decade apart otherwise you will get low frequency instability.

At a rough in the head guess, one pole is at about 250Hz and the other is at about 40Hz. You could change the 40Hz one to4Hz just by changing C3 and c4 to 300nF. Increasing R8 and R9 to 470k should take close to 1Hz.

Cheers

Ian

yeah layout seems crucial ive noticed. tubes are touchy fellas!

the output transformer has a lowend bandwidth of 10hz.

c3/r8 and c4/r9 is set to about 50hz. any higher and i get a bump just below 10hz.

the feedback components are set to about 2,5hz but i might done it the wrong way... this freq point corresponds to the lowend bandwidth of the input trasnsformer which is some where around 2 - 2,5hz.


sun is shining! think i head out for icecream!

cheers!!

Sort of yes. The plots show voltages relative to the cathode. With unbypassed, the the cathode to plate and cathode to grid voltages vary  due to the negative feedback so you cannot use the plots to predict overall gain or distortion - they just show what the tube is doing.

Click to expand...

Ian, would you like to explain more on this, how to go about when setting up an unbypassed stage?

im starting to think that my mind has stuck on this criteria for no use, and that it is just a ruff guide to the area where good symmetry is found..

and when the bypass cap is removed, impedance and gain is different, and thats it..!

cheers!