coppatone GM70 /2

I had no idea… jc sent me above image after I published part one, where I coined the coppatone name. I must admit that sunscreen lotion was not part of my joke.

Another reaction to part one was a comment by Pieter, questioning the interrelated gain/headroom/power complex I had set up in that post. With this I am thrown straight back to my daily praxis as interaction architect, where as a natural part of design processes critique, so-called fatal flaws, preconceptions and third-party tinkering have to be dealt with. The best way for designers to do this is: take the input serious, combine any new arguments with the dozens that were already known, think, decide again. So here we go.

First thing I realised is that there are conflicting requirements from the different classes of power amplifier design: a mass-market design; a specialised design; a completely bespoke design. Obviously I am not doing the former (good riddance), but conflict is rising from the fact that this is a bespoke design for one person with a variant for another, being published to a community of enthusiasts. When any of these pick it up to build it, that makes it a specialised design. And after just one round of design process one of the enthusiasts came knocking, with some tough questions.

The power of either designing your own audio gear, or having it custom designed for you, is that it can be so much more optimised. This is because a boatload of requirements—in compatibility, economy, size, manufacturing, parts sourcing—can be safely marked as being not valid in this situation. And hence do not have their generalising—if not degrading—influence on the design. Moving from bespoke to specialised design raises the number of aforementioned requirements, with resulting generalisation.

So how to do something inspired and escape the trap of generalisation? By running it as a bespoke design project for Thomas, with a variant for Daan. And offering the community pointers and methods to adapt the design to their own particular requirements, if these differ from the ones I am covering.

As a direct result of Pieter’s comment, I did a number of sanity checks/thought experiments to double check if I was really that far out (stop giggling at the back…). First, I thought: what if you do not need the 40 Watts? What if you just wanted to use the best part of the curves, no matter how little power you’d get out?

So I took the GM70 curves and set up a golden triangle of operating area by being just very, very picky: 100W max. plate dissipation (keeping that the same), not cross the zero volt grid line, curves below 80mA do not look straight enough:

Now let’s put in a load line. I spotted the -120V gridline crossing the right-hand corner of the triangle, set the operating point at where the -60V gridline crosses the 100W plate dissipation, and drew the line:

Operating point: 770V/130mA, max. power: 10.2W, on a 7k26 output transformer. So after letting the curves set the ‘ideal’ operating point and load line, out come values this close to what I calculated last time (840V/120mA, 7k OPT):

There’s something about my simple rules for operation points. More than with a usual class A1 operating point (higher plate volts, lower current), the first Watts (11, here) are brought forth in my set-up by the best, most linear part of the curves. So I see no quality penalty for using the operating point I chose. Designing the amplifier to put out all 40 Watts does create a luxurious amount of overall headroom, when a handful is all you need on a daily basis.

For the second thought experiment I thought about the ‘requirement’ that an amp must be driven into clipping by 1Vrms on the input. Why this input sensitivity? Well, maybe for minimal systems; source–amp–speaks. And in contrast to 1970, today the CD player sets the standard there. Thinking of how red book CD players put out at least 2Vrms—with a hard maximum—and our coppatone puts out 40W—and then clips hard: symmetrical—I connected the two and looked at full power, for 2V input. This asks for a gain of the input stage of 11.5. Compared to the gain I picked for the first stage (8, in practice), the difference seems to be almost trivial: 3.14dB. What are we actually arguing about?

Writing the previous blog entry I had a nagging suspicion and this time, while all this thinking was going on, it almost managed to trip me up: the input stage consists of a tube and an input transformer:

So my first community pointer is: if you are really set on getting a gain of 15–20 out of the first stage—and like me are not prepared to lose the glory of using something like a 10Y as input tube—then I would look into a step-up input transformer, 1:2 would do. If you are arguing for a gain of 20 because a 6SN7 type tube is the one at the top of your tube drawer, then I would carefully look at what you need (power) and what you have (voltage for the amp) and see if that all is not a bit too much...

What Pieter’s comment also made clear to me is that I had forgotten something: a good talk with the client about what the project is supposed to achieve. So I met with Thomas Schick a couple of weeks ago and we talked about why he is going to build the amplifier.

Predominantly, the answer to this is ‘because he can.’ Plenty of GM70s, let’s build a rack amplifier, like Klangfilm and Western Electric used to do for movie theaters. We agreed that this was not going to be the amp to deliver 0.1W to his horns. More something to hook up to his Celestion SL700 speakers. At 82dB/W/m, that is going to take 15 Watts to match the loudness of 3 Watts on my 89dB/W/m speakers. That a pre-amp will be needed to drive the coppatone GM70 to full power output is no problem for Thomas, he has been using one during all the years that I know him. He could not decide between 10Y or 205D input tubes (brilliance or romance? hmmm), so we are going to take both: let’s see how much fuzz it is to make a convertible input stage.

Next time, I will talk about a surprise requirement for the input tube in this circuit architecture and why the 76 may not be the best choice for an input tube.