Cathode Follower Output Stage

[RicharD]

Here's another interesting article:
http://www.tubecad.com/2005/July/blog0050.htm

Some years ago, I sorta accidentally purchased the remaining gear from an old recording studio.  One of the items was a home brew monitor amp.  I had to drive 1/4 way across the country to pick this stuff up and one of the thing the seller was most proud of was this monitor amp.  I distinctly remember him telling me it was a cathode follower amp.  At the time I was farr too uneducated to have a clue what he was taking about.  For some reason, this dawned upon me this evening so i did a little research and read the link posted above.

I'm guessing the reason he was so proud of this circuit is because cathode followers are very low distortion.  (guitar amp guys click back button here.... maybe)     Every single other amp I've seen is a grounded cathode amp, most of which are push pull.  So if cathode followers are lower distortion, then why aren't all the hifi tube amps cathode followers?  Here's the deal.  A cathode follower has a gain of less than unity, let's call it .9 for the sake of discussion.  This means that the PI or proceeding stage has to make greater voltage (not current or wattage) than the output stage.  Ah.... so here's the paradox.  The output stage may be capable of doing it's job with less distortion but this means the preceding stages require more gain and therefore will distort more.  Gain stages multiply so in a hifi situation you want your front end to be as clean as possible.  If you're gonna hafta allow for distortion, you want it to be on the back end, but in geetar amps we like distortion.  Wrong kicks ass.  This appears to be wrong.  Have any of you seen a cathode follower output stage?  We do a lot of reinvention of the wheel around here.  I try to stay at the edge if not entirely outside of the box, so I find myself here wondering about cathode follower output stage.  The Broskie link above stars with a picture of a wrong way sign which of course peaks my curiosity.  In my mind I'm still a child,  Telling me "NO" will almost ensure I will at least try it.  I'd love to hear some feedback.

-Richard

[PRR]

If the driver for a plate-loaded stage needs a 300V B+, the driver for a cathode-loaded stage "can" give the same distortion with a 3,000 B+.

The McIntosh is halfway cathode loaded.

[darryl]

Tubelab has investigated cathode follower amps, and provided further links here:

http://www.tubelab.com/CathodeFollowerAmp.htm

[RicharD]

Interesting read.  I can't find any specs on the tube he's using but I don't see why he's using an opt with a 600 ohm primary unless this tube has a really freakishly low Rp.

[PRR]

> I don't see why he's using ...600 ohm

Nor do I; if he's hitting 80W Pdiss at 550V then it's 0.15A and perhaps 3K load for an ideal device, ~~2K for a real bottle. Oh, but half of that because his driver can not swing 1,100V peak to peak, but less than the 550V (may be 530V, very good).

Ah, but if driver can't swing the big bottle half-way, then how is he reading 35% efficiency, 28W (sine?) out for 80W Pdiss? Or did the driver change to choke-loaded and I missed it?

IAC, this is all wrong for rock and roll.

And even for super-clean, you can usually do better, with less heroics, by plate-loading the power stage and applying the extra gain toward NFB. Cathode-loaded power stages are VERY VERY rare in professionally designed gear.

LA2a is one exception, and it could be argued that the plate-loaded half of the WCF is vital (also that it isn't so well designed, just a platform for the patented opto-attenuator). There's a rack mike preamp (perhapsd many) in recent production with CF out; but you need 3 triodes to do the job and using popular twin-triodes you have one left over, which allows a simpler parallel-CF output rather than plate-loaded.

[jojokeo]

NO!  >:) 

Telling me "NO" will almost ensure I will at least try it.  I'd love to hear some feedback.

[DummyLoad]

below is an excerpt from tube CAD on the transformer coupled P-P cathode follower output stage. the lower distortion can be alluring, however, with such high grid swings needed, you'd need a PS with higher V+ to achieve the swing needed for the driver stage to drive the output stage to full power - or you live with reduced output power. even though the output impedance is lower, you use the similar values for OT since the cathode currents and voltage swings are similar to that of the plate loaded design.

when you were designing your tube mic-pre (the first one), i worked up my own model using a P-P cathode coupled output stage up to the point that cine-mag gave me quote for the OT. i pulled the plug on the whole mess when the light bulb went off about the grid V swing requirements and tilted PS requirements to drive to full power. i suppose i could have used a V doubler of sorts for the PS, but that's another story of lost interest.


This reversal greatly reduces the distortion and output impedance, but at the cost of no gain and, thus, the need for huge grid voltage swings. Input voltage swings of 300 to 400 volts peak are commonly needed, as the cathodes must experience the same voltage swings that the plates would experience in a grounded-cathode transformer-coupled amplifier. Furthermore, because this amplifier will undergo the same voltage and current swings as it more conventional cousin, the grounded-cathode push pull amplifier, similar plate-to-plate load impedances are used (in spite of the cathode follower's much lower output impedance).

CF operation does not change the IV dynamics that the tube must undergo. For example, when using a single cathode follower circuit, we do not load a 6DJ8 with a 200-ohm cathode resistor (twice the effective output impedance), even though that value would adhere the 2rp rule; rather we use something like 10-20k.

Here's a quick thought experiment. Take sum of half of the output tube's rp placed in parallel and half of the primary winding's DCR value and one fourth of the primary winding's reflected impedance and divide this sum into the B+ voltage. For example, given a total of four 2A3s, a B+ voltage of 250 volts, a plate-to-plate reflected impedance of 2400 ohms, and a primary DCR of 200 ohms, the result would be 227mA (250 / [800/2 + 2400/4 + 200/2]). The result is (roughly) the peak current that the tube will undergo when the cathode and grid voltages are equal. This current flow against the half the primary winding's DCR value and one fourth of the primary winding's reflected impedance will give you half of the peak voltage that will develop across the primary. For example, given the previous values, the result is 159 volts. If you play with various values, you will see that something close to 2rp works best.

A common-cathode resistor can be used in place of the fixed bias voltage by placing the resistor in series with the output transformer's center tap and ground. The primary's DCR (DC resistance) becomes part of the cathode bias circuit.