Ask Mark - Parallel Audio Output Tubes Explained

BG283 - We get the question a lot around "can I simply parallel output tubes to increase output power" and what all is involved.

[knifeswitch]

And this is a perfect example why I sub to Blueglow. My thanks Mark

[NickP]

Mark, this video was brilliantly explained. My grandfather was into HAM radio and had built a huge aprox 30’- 40’ ft. tall antenna attached to his garage as well as many huge tube amplifiers. He had a big vintage McIntosh system in his shop with huge Klipsch spkrs. This video just reminded me of when he used to explain and teach me some of this stuff as a kid, and it brought back some very fond memories. Thank you.

[rogerhowell]

I as well got my start in tube equipment via ham radio.
It's kind of the same concept but not. I started playing guitar again and my fender amp needed some work after setting up many years. Leaving in middle of nowhere West Texas finding a amp tech is near impossible. After a couple of conversations with so called amp techs and what I do know about tube RF amplification I could tell most were full of BS to put it mildly.
So out comes my old test equipment and did the job myself.
Subscribing to your channel as well as Uncle Doug and others have given me knowledge of understanding tube audio circuits.
Thank you very much for posting videos and renewing my love of all things tube.
By the way you look so much healthier now than your earlier videos.
Roger

[RichSad]

New subscriber here. As a teen in the 1970s I was a CB radio enthusiast and interested in electronics and the emerging home computer industry. I recall two projects both based on books from the town library. One was in my tween years I built an analog computer in a cigar box. It was simple and used little led lights, paper clips, wires, and potentiometers to add two single digit numbers together. While incredibly basic, it made the concepts of binary computing incredibly clear. The second project involved an "illegal" linear amplifier for RF. I tried two different things with it. One was putting it between the CB and the large home antenna. The second was turning it into a mobile device that could go in a car. I didn't fully understand the book or key concepts involved, but the simple act of doing it helped me to understand some of the practical applications of the concepts even if I didn't understand them at their core. FWIW I had to discontinue the use of this after a friendly visit from authorities. Apparently I was causing interference with neighbors and potentially other RF devices (TV, emergency communications). My career path took me in the direction of software systems and computing but I am still a guitar player who loves analog, tube amps. Lately as I approach 60 years old my interest in how all the stuff works is renewed. Thanks for the clear explanations.

[Berre]

Thanks Mark, your enthusiasm while explaining this is amazing.

[bucketlistbeginner]

Thanks a lot mark I really appreciate it. I’ve been looking for a long project to stay busy, I knew it wouldn’t be hi fidelity but thought it would work double duty with my video and music. Sounds like I’d have lots to try and tweak to make it right. Great explanation of everything, , thanks again, Rob

[williamvann]

This was very helpful to me, a guitarist, who loves tube amps and wants to learn and understand how they work. Currently, I have a Class A/B amp with 4 6V6GT power tubes, with a half power switch. The output transformer is wired to handle 16 ohm, 8 ohm, or 4 ohm speaker loads. If by using the half power setting, and cutting the output current in half, shouldn't I be concerned that the impedance the OT is feeling is not what it is set for? In other words, if I have an 8 ohm speaker, shouldn't I set the amp for 4 ohms when running it in half power mode? Thanks for all the great videos! I am learning a lot, but I am also learning how much I don't know!

[hugobloemers]

This is a great summary, specially the matching section was interesting to me.
I am currently in the initial phase of designing a headphone amplifier and one of the ideas I have is to use a single ended output stage driven by the two triodes of an ECC99 in parallel, per channel (obviously). So here we have two triodes in one envelope. In my naivety I was kind of assuming that if you have two triodes in one envelope there are no considerations to make for matching. The question is, how well are triodes in a single envelope matched in practice? (I have seen schematics with those in parallel).
Second question, I was planning on a common grid circuit, i.e. grids to ground and drive the cathodes with a single NPN transistor (probably a BC549 unless you can recommend a small signal audiophile transistor that is better). In such scenario, are there other matching considerations?
The alternative would be lower power and use a single triode, i.e. half an ECC99, which I calculated could deliver about 1, 5W per channel to the headphones. Would you recommend that instead to keep it clean and simple albeit with less headroom?

[elcocineroamericano]

Just day before yesterday I was digging through the RCA tube handbook trying to figure out what TV tubes would work best for a PPP amplifier, without really understanding the issues at hand. This video could not have been timed better!

A couple of questions:
1. Could the problem of impedance matching be solved by a custom interstage transformer?
2. Is there a formula or other way to calculate the necessary drive for output tubes in parallel? How is the input impedance of a circuit like that calculated?
3. Is it strictly necessary to connect the cathodes directly together? Could you not use adjustable cathode bias to balance each output tube?

Thanks for such a great and informative video.

[rexoliver]

Across from me in my office-looking at a 1950's era GE SW 250Kw AM broadcast transmitter whose 250Kw output stage is 2 7482 triode tubes in parallel.They are driven by a single 7482 tube.The single 7482 is driven by a 4CX5000 stage.The modulators are two 7482 tubes in PP.There is a balance control in the cathode circuit of the output stage allowing you to balance the output stage tube so they draw the same cathode current.Figure this could be done with the audio version as well.In the days those transmitters were designed and built tubes over 250Kw disappation were not available.And these stages have to be stable from 6Mhz to over 21Mhz.

[stephenhammonds]

Hey Mark, thank you!

*Side eye to my Peavey 5150 sitting in the corner*

[peteleoni]

Brass sets in dissimilar metal corrosion. Tubes are glass sealed. Dip in t
Tarn-x that will be quicker and probably safer.

[goodun]

Mark, why do your schematics often show a 10 mfd *non-polar* film cap as the 1st stage filter? Not that it's a bad idea compared to using an electrolytic cap, but most people who own or work on tube amps will never encounter a big film cap in the power supply. PS, you should probably show where you're taking the B+ off for the plates of the tubes ---- pre or post choke? --- especially when you are discussing a high current amplifier with multiple output tubes. It would require a massive choke if you wanted to choke-filter the plate supply as well as the screen supply and everything downstream.

[EngineeringVignettes]

Nice overview. Thanks Mark.
I'm doing a dual PP monoblock amp myself now, hope to get away with it as its designed as a sub-woofer (bass) driver for my stereo setup.. it's an experiment and it may end up in the scrap bin.. we will see. Uses 6AQ5 output tubes, closely matched for Gm but did not bother curve tracing them.
Take care now and Cheers,
(any news on the barn rebuild?)

[Piccodon]

impedance matching is great for min power loss in RF circuits.
Efficiency is not a common design goal is tube amps. For audio you don't care if your amp is under loaded, so a very low source impedance is OK, as you will not run out of drive power that easily.
A lower source impedance will make the job much easier for the output transformer design i.e. lower step ratio. Feed back from the load to the input will compensate for the losses in the output stage, and drive the amp harder where it fails, maybe not such a hot idea. Feed back only get you so far, and while looking good on paper with static signals your ears may think differently.
If you start with the speaker efficiency and impedance, which may be reactive, and demand more current that your tubes and power supply can gracefully muster, and work your way back across the output transformer to the power tubes you will have a better idea what tubes to use. How much SPL, what efficiency, what reactance, how much power, convert to peak to peak volts and then to typical tube voltages, with power in mind how may tubes, or what tube to use. As tube choices are limited, supply lower voltage is easier to deal with for many reasons. A 300B tube is an ideal tube from an ease of design point of view, 845 not so much, 1200V plate and very low gain so very high drive voltage is needed. The 6550 is OK and available works with 500 - 600V max, linearity is not stellar but OK with feedback, or ultralinear taps, you can easily buy transformers for these too. Matching for bias and gain balance with paralleling should be done, but individual bias for each tube is not hard, but gain balancing paralleled tubes is. Jadis used 3 parallel 6550 is push pull which sounded great. Push-pull amps cancel even order harmonics leaving the odd order ones, which may be a reason why some swear by single ended amps. Even order harmonics cannot be heard, your own ear produces them, and may well disguise odd order harmonics. Speaker choice and cross over freq to sub is key in amplifier design rationale.
In the glory days of tube amps very efficient speakers were used, by necessity. A 99dB/W/m is very different than 85dB in its power requirements. 10dB is 10 times the power, 20dB is 100 times etc, so speaker choice rests in the center.
As most of the music energy is concentrated in the bass, a powered sub makes the main amp design easier, and output transformer weight and price goes way down.
As a lot of music has high dynamic range, peak power is very important, a 90dB level at 1W average can be very loud, but the peaks of +15dB or more should not be clipped.
Most of this power happens in the bass, look elsewhere for power spectral distribution of music to find your design criteria. A sub up to 100Hz or more will take the brunt of the power, and I'm not afraid of class-D amps for this, the higher frequencies are filtered off and the bass speaker elements are not very good at the higher end either.

[laurencerogers]

Great video, I think ten watts is enough for me anyway!

[duanethompson]

Mr Carlson’s lab does in depth on a six pack mono block!

[davidnguyen]

Basically, if you want to change the output stage then you’ll have to upgrade the driver stage too.

[brianjuffs]

Plenty of 100w guitar amplifiers running parallel push pull. Not perfect device's.

[LrdGwyn]

One more complexity to add to the list, Mark, that I didn't see mentioned in your video specifically:

For the output transformer, if running a single-ended design, you are also doubling the quiescent current on the primary winding, meaning you have to have an OPT that is gapped to handle double the amount of current without saturating the core! And also maintain a decently high primary inductance for low frequency extension. Finding off-the-shelf transformers that meet those criteria is going to be tough, and for any reasonably sized output tube in parallel, the transformers will also be MASSIVE (which can also be read as expensive).