Cathode bypass Cap vs. No Bypass Cap

[rcollege]

Hi all,

I was just wondering if there are sonic differences with bypassing the cathode resistor on a preamp. vs. no bypass? Or is it just amplification?

I'm currently running a DIY SE preamp using a single tube per channel.
My source puts out 200mA tops...so I need about 5x amplification
Currently driving a power section of an SS Marantz.
It will be driving a Dynaco ST-35 soon.
Any info or comments are appreciated.

Thanks
Roger

 

[Shoog]

No bypass means higher output impedance (cathode resistor x mu + rp) which may or may not be an issue depending on the following stage.
The reason gain is lower is because the unbypassed cathode resistor introduces degenerative feedback which like all feedback reduces gain, but also reduces stage distortion. Again this may or may not be an issue depending on the linearity of the gainstage been used. That is to say a linear triode with bypass will have lower or equal distortion than a unlinear gainstage with dengenerative feedback.

Its about the design and the choices which are appropriate to the design. As an example I would use a no bypassed input stage if the loading of the preceding stage was light - such as a cathode follower. This would give lower overall distortion at the cost of a slight loss of gain.

Shoog

 

[M Gregg]

The Sound of the Machine

Have a read..🙂

Regards
M. Gregg

 

[Miles Prower]

Hi all,

I was just wondering if there are sonic differences with bypassing the cathode resistor on a preamp. vs. no bypass? Or is it just amplification?

It depends. Emitter/source/cathode degeneration is just amplification. For transistors and pentodes, this degeneration reduces g_m and increases linearity by the added NFB. The same thing happens with triodes, but with an added complication. For just triodes, what you hang on the cathode/plate has an effect on the impedance seen at the plate/cathode, very much like the impedance transformation seen at the base/emitter of a BJT.

This can have an impact on sonic performance since an unbypassed cathode resistor elevates the effective plate resistance. Since triodes like light plate loading, this increased effective plate resistance can mean your passive plate loads are no longer big enough. You still get the linearizing effect of NFB, but are at the same time fighting it as well.

If you need to reduce the gain of a triode gain stage, it's better to tap off the plate load to reduce the output swing rather than degenerating the cathode.

 

[Bigun]

The Sound of the MachineHave a read..🙂

I've seen this before and I find it difficult to interpret the data when it's split up into even and odd harmonics. I prefer to see all the harmonics on the same chart. And I think this is important because of the psychoacoustic affects that lower order harmonics have in masking higher order harmonics.

I've recast the data to match my preference (attached). All these curves look very good by the way. But what I notice is that the differences show up when we get to the 5th harmonic - generally a bad sounding one. The Xfmr coupled result (without bypass) looks bad, it has a high 7th too which is bad again. The SRPP Plate looks very good - really low 5th and the only harmonic pushing above the -100dB level is the 6th - and only just. The SRPP Cathode is also very good. The SRPP would be my pick from the data. And that would be in conflict with what the article says.

 

[Mescalero]

An interesting point to compare would be the anti-bypassed (new phrase?) cathode behavior of the PP class A designs which have a current regulator on the cathodes (Oddwatt, etc..).

 

[Johan Potgieter]

Bigun,

The main reason for showing odd harmonics separately is that they are the troublesome ones; as I understand, particularly 7th and 11th, but also others. Even harmonics are for the most octaves above the fundamental and not disharmoneous. That is not to say thay should be ignored, but if one prefers to look at odd harmonics the figures for even ones will unnecessarily 'clog up' a graph or table.

 

[Bigun]

The issue is that odd harmonics can be masked by nearby even harmonics - it's a psychoacoustic effect. This is why it's handy to see all the harmonics together.

 

[Johan Potgieter]

I read what you say, Bigun - need to brush up my psycho-acoustics.

I once saw data where higher odd-harmonics and some I.M. products of the same could be noticed even when as low as the threshold of hearing. But that was a long time ago. As said, I will Google psycho-acoustics again and brush up. Thanks for the heads-up.

 

[gingertube]

Harmonic Distortion and Intermodulation Distortion is produced (simultaneously) by the same mechanism, namely mixing of signals in a non-linear system.

Higher order harmonic distortions are accompanied by many more Intermodulation Sidebands
Even harmonic distortions are accompanied by IM Sidebands remote from the original frequencies
Odd harmonic distortion are accompanied by IM Sidebands close to the original frequencies (and therefore more objectionable).

As Johan states the 7th and 11th Harmonics are musically "quint" and are therefore very objectionable, even levels 0f 0.001% of these are fingernails down the chalkboard stuff, it just grates.

Cheers,
Ian

 

[fdegrove]

Hi,

I was just wondering if there are sonic differences with bypassing the cathode resistor on a preamp. vs. no bypass?

To my ears the bypass caps themselves are a major source of colouration.
If you want to retain the advantages of bypassing the cathode resistor without having to deal with the disadvantages of a large value cap then LED bias is a good idea.
It does away with the cathode resistor and there's no need for bypass caps.

Alternatively battery bias can be used as well.

Ciao, 😉

 

[Bigun]

I can't remember why but Thorsten didn't like the sound of LED bias, I don't know if that was an old opinion or if he's now a fan. I do remember that there is a need to keep enough current flow through the LED and the 'trick' I saw posted once is to run a resistor from the top of the LED to a quiet power rail so that the LED has 10mA even if the tube has only 1.5mA (e.g. 6SL7).

Mind you, try using LEDs on the cathode of a single ended 6AS7 😱😱

 

[fdegrove]

Hi,

Mind you, try using LEDs on the cathode of a single ended 6AS7

You'd need a X-mass tree full of them....😱😀
Joke aside, tubes such as 6AS7s and consorts are best bias from a dedicated negative rail.

Ciao, 😉

 

[Shoog]

Unfortunately cathode bias is the only safe way to bias up a 6AS7, or use a CCS.

I messed with LED bias for a time - but ultimately I prefer unbypassed or bypassed. If I remember correctly LED sounds to clean - to solid state for my tastes.

Shoog

 

[fdegrove]

Hi,

Unfortunately cathode bias is the only safe way to bias up a 6AS7

There plenty out there though that do not use cathode bias. OTL amps for instance.
I run mine that way (10 bottle per channel) with no problem whatsoever.

I'm well aware that the manufacturer warns against it though.

Ciao, 😉

 

[DavesNotHere]

Unfortunately cathode bias is the only safe way to bias up a 6AS7, or use a CCS.

I messed with LED bias for a time - but ultimately I prefer unbypassed or bypassed. If I remember correctly LED sounds to clean - to solid state for my tastes.

Shoog

you're right, you should be a skeptic about using solid state ccs in the cathode. because its a trade-off. and the signal does interact with the ccs. and yes you can use tube for it, but in some cases, those even effect the response of the signal.

the simple resistor is the best linear response.
the bypass cap should be only used if you need to boost the gain.

as far as the 6as7, yes cathode bias is the safest way for class A amplification.

And yes, you can run it in class B self biased, because class B Push pull amp runs efficient in the non linear regions of the tube.

 

[rcollege]

Thank you all for your comments...I hope they keep coming 🙂

I did use a 27K resistor bypassed by a 2.2uF non polarized cap- I believe it is Mylar.
The bass response is good if not too much.
There is a much fuller sound in the mids...but still clear.
However now I have too much gain. Thus I have changed to the higher value cathode resistor. I also added a 7.5K resistor after the pot on the input. I have a 220K pot that I may change to in order to lower gain.


My circuit is very, very simple -

Tube - NOS 6J5GT it has a flat plate ...I also have NOS 6C5GT round plates Tungsol.
Plate resistor - 100k
Cathode resistor - 27k
Bypass Cap - 2.2uF
input coupling Cap - 100n Oil Paper
Output Decoupling Cap - .47uF 400V Orange Drop
Resistor after Input Cap - 220k
Resistor after Output cap - 1meg
50k pot followed by 7.5k resistor

Plate voltage approx. 130V
Cathode current approx 1ma- I fear it is too low.
B+ is approx 176v

Rectification is two 1500v ss diodes followed by 470uF filter caps. 235uF - 1.2k -235uF -7k (Caps are 2qty 470uF 385v in Series 100k balancing resistors)
Power Transformer is a 220-0-220 50VA - (I also have a 330-0-330 50mA I can use.)
Medical grade switcher provides heater 6.3v DC. Thinking of switching to AC transformer rather than switcher.

I remember 1H4G giving me better imaging and clarity (even though it is a battery tube)

Any recommendations are appreciated.

 

[fdegrove]

Hi,

I did use a 27K resistor bypassed by a 2.2uF non polarized cap- I believe it is Mylar.

That value is quite likely way too small to get a decent LF response. You probably need something in the order of a 220 microF or even 470 microF.

Ciao, 😉

 

[painted]

... the bypass cap should be only used if you need to boost the gain.

[FONT=Verdana, Helvetica, Arial, sans-serif]Leaving out the cathode bypass cap has impact on the proportion of upper harmonics[/FONT]; the triode without bypass has [FONT=Verdana, Helvetica, Arial, sans-serif]more higher upper harmonic content without adjustment of the load.[/FONT]

 

[Merlinb]

Hi,

I did use a 27K resistor bypassed by a 2.2uF non polarized cap- I believe it is Mylar.

That value is quite likely way too small to get a decent LF response. You probably need something in the order of a 220 microF or even 470 microF.

2.2uF is just about tolerable if he really does have a 27k bias resistor. Is that correct?