The GR-25

[GRollins]

What's this "Please complete the message field" nonsense? Haven't they ever heard that a picture is worth a thousand words?

Grey

[Nelson Pass]

Nice amplifier, Grey.

[c2cthomas]

GR - I LOVE IT!!!! This design would be a GREAT match up with my speakers!! Love the implementation - as only a true fellow "let's get 'er done" type could!!!!

[h_a]

Grey, thanks for reminding me that the damping factor is just another word for output impedance!

So around 20 per pair of vertical fets with 0.2R source resistors.

Nelson always points out that total bias is of interest, not the bias per part. So by paralleling some pairs (gets also dissipation per part down and reliability up), one could arrive at about 80 using say 4 pairs, even more with lower source resistors (but this increases distortion so maybe stick with the 0.2R). Really not bad I would say.

By the way, where's the master? While you're at it, a pic with the creator of his little creature would also be nice! Maybe with the fearless Cindi as well, if she doesn't mind a bit online popularity!

Have fun, Hannes

PS: looking at the X5, one is puzzled with the improvement NFB brings. The X5 sports a damping factor of 200, having 4 output pairs with 0.47R source resistors - without feedback it should be about 50.

[bogdan_borko]

VERY NICE GREY! javascript:smilie('')

[GRollins]

Damping factor is a little more complicated than just tallying up the Source resistors and dividing. The Source adds its own contribution and to get a good idea of that figure, you have to know the device type and how hard they're biased. An ideal device would have a Source impedance of 0 Ohms. If anyone has a box of ideal devices to spare, I'd appreciate it if you'd send them to me. Lacking ideal devices, the real ones have some impedance. I just pulled up the datasheet for the FQA19N20C. The Rdson is listed as .17 Ohms. On one hand, I'm not biasing the devices as hard as Fairchild is, so it'll be higher. On the other hand, their number could be a little optimistic. I probably overstated the case by saying that the Source resistors dominated the output impedance. They're probably more like a third to half. Nelson has told of some way to measure the Zout of an amp using another amp to drive it, but I don't remember the details and I don't have time to hunt it down at the moment.
The reason I don't get all twitterpated about the damping factor is that it's considerably overrated as a number. My old Conrad Johnson Premier One (a tube amp) had a fairly low damping factor, but it simply stomped most solid state amps in the low end. That plus the fact that damping factor is a diminishing returns sort of thing (Is a damping factor of 10000 really 100 times better than a damping factor of 100? I don't think so.) kinda puts it in the category of THD for me. Is it important? Yeah...up to a point...after which your time is better spent working on other problems. Like I said above--5% distortion in a subwoofer? Great time to employ feedback. That much distortion is a serious problem and feedback is the tool for the job. Damping factor of 1? Hmmm...not so good. Damping factor of, say, 20 or 50? Okay, let's take a look at something else and maybe we'll come back to the damping factor thing later if it seems like it needs more attention. If you match your devices well, you can use lower value Source resistors and get a lower Zout that way. Or you can parallel more devices. Or you can bias harder. Or you can add feedback. Or you can just sit back with some Highland Park 18 or a good Bordeaux and listen. I vote for the latter option.
What on earth do you want a picture of me for? Just go look at a picture of Shrek. Envision me as his older, uglier, and meaner brother. With even worse social skills. I sent Nelson a picture that happened to have me in it once upon a time. He commented that I 'didn't look as old as I write.'
Maybe I should figure out how to 'write younger.'

Grey

P.S.: According to the textbooks, an ideal device would have infinite input impedance, 0 Ohms Source impedance, and infinite gain (which actually proceeds naturally from the 0 Ohms if you work the math angle). Speaking solely for myself, I'd trade infinite gain for linear gain any day of the week. Finite gain suits me just fine. Personally, I think the guys who write the textbooks need their head examined. Yes, yes, I know...the assumptions for the ideal device include the use of infinite feedback to linearize the silly thing. But if we're wishing for an ideal device, let's take a look at the intended application before wishing for the wrong parameters. You know that old expression? The one that says: Be careful what you ask for, you just might get it. At least for audio purposes, I think the wish list should be modified.

[h_a]

Thanks Grey for replying, it was only meant as a rough estimate as you weren't to be pushed for numbers. Numbers have quantitative quality and have more meaning to me than when you simply say, damping is low. What means low? I guess everybody has a different measure for that.

If it's say 50, I'm perfectly happy. Same with THD, personally I want that odd number below 1% - where it is exactly I do not care that much. Same with IMD. But I want to know.

Why do we want to see a picture from you? Dunno, to assign a face to work - why not? You could always hold one of your books and push your sales

All the best, Hannes

[h_a]

By the way, I still haven't said it yet: I consider your GR25 a really mean amp. With THD below 1% at full output, damping at 50 it has very reasonable specs and that without any NFB. I haven't seen another feedback-free amp with such specs and I'm sure it sounds marvelous.

I'm just waiting that the NFB-crowd kicks in and makes some pcbs (so that I don't have to tear myself apart - again).

All the best, Hannes

PS: Grey, you told me to remind you to fetch your post-it with THD at full output. Give the beancounters their kick - thank you!

[KSTR]

Grey,

The notion that Rds_on has something to do with Zout is a misconception. Mr.Bob Cordell and others have written a few post on how the output impedance of a source/emitter follower is made up (pls read here and here)

Zout = (Rs + 1/gm_Nch) // (Rs + 1/gm_Pch) under class A conditions. When you enter class B one of the terms falls out of the equation (the one that belongs to the FET that goes into cutoff).

The 19N20 is quite like a IRFP240 and will have a gm of about 2..3S at the 1.3A you're running it at. This also is close to Bob's rule of thumb that the FET gm is roughly 1/10th of BJT gm at the same current (a BJT at 1.3A would weigh in at 50S). So your output stage might have about 0.1..0.2 Ohms (which seems to relate nicely to your DF=50@8R value). There is nothing wrong with that, I myself am not the high damping factor camp, of much greater concern is how constant it is (as any dynamic change of Zout IS plain distortion). So the trick is to find a bias current where the Zout is as constant as possible over the widest output current swing. See also the D2S discussion, where the goal is to have true constant gm at least in the most important low level region.


Measuring bulk output impedance is easy, drive the output of the amp (which has no signal applied) with another amp/channel via a bigger resistor, say 100Ohms or so, this practially is a current source with Iout=Vout/100. Measure the AC voltage at the DUT, say 10mV with 10V drive (=100mA), so Zout is 10mV/100mA=0.1Ohm. You can further drive the DUT amp with DC to have a standing output current underlying and can then get small signal Zout at varying output current and will probably find that Zout is lower at larger output currents, because gm increases greatly with current.

A counterchecking test is to drive to different loads and the obtain the Zout from the voltage and resistor ratios:
Say you have V1 with R1 and V2 with R2,
then set ratios v=V1:V2 and r=R1:R2,
and get
Rout = R1 * (1/v - 1) / (1 - r/v)
This one has less precision especially when the r/v ratio is close to unity. You must load the amp quite differently but not to lightly to get stable results.


Last not least I have to say that I like your design, especially the frontend.

- Klaus

[jacco vermeulen]

Mr GJ Rollins has been cloned, pretty unusual for a tube gent to even skip the impedance of the PS.

Does the South Carolina University have a vast supply of 2SJ313/2SK2013 ?
Toshiba stil manufactures the duo, but mere mortals will have a hard time buying them in <<1K numbers.