F5 power amplifier

[jackinnj]

OK, problems solved -- I put a 1,000uF/35V cap on each of the rails immediately adjacent to the trimpots -- any lead length here adds to the stray capacitance. I haven't sought to optimize the value -- I bought jizillions of the capacitors from Tektronix 7 or 8 years ago.

Secondly -- I put 3.3n/100V polyester across R5 and R8. The ringing on the square wave test (200kHz, 1 Watt) is almost gone. Good enough for me. I tried values from 470pF to 47nF and these work best for me. YMMV.

THD% for 1W, 1kHz is now 0.0063% from 400Hz to 80kHz without any trimming of the source resistors on the JFETs or MOSFETs, and without any attempt to get rid of extraneous noise -- so without too much difficulty it can probably be brought down much lower still, i.e. to the levels seen in the article. I am using Fairchild devices, no thermal protection, etc., etc.

The THD% is still mostly 2nd order, however.

Funny thing about the Fairchild devices -- the Vgs for a batch of them are pretty tight, but the P's and N's are over 1.4V apart.

[Zen Mod]

Originally posted by udailey
Hmm. .....

is this make a sense ?

cap multiplier is same as series voltage reg ( where output voltage is set with zenner voltage , minus Uds of series mosfet) ;

there is just one tiny difference - lack of zenner

so - voltage on output will be always input voltage less Uds .

say that our start was series zenner reg ;
we took zenner from it , but we further augmented value of cap which was placed across zenner terminals ; there is point in name "cap multiplier" ...... just becasue mosfet have his own characteristic , known as "multiplication factor " , it will multiply value of that cap ......

some ppl like to say that cap multiplier isn't much different from plain choke ;

simplified it is , but in your case - exactly convenient , because Uds is always in ballpark of 4V , and that's exact amount of voltage you want to shave from your PSU .

further simplified - if you put adequate mosfet in line with each (and both) (positive and negative ) legs of PSU , you'll loose approx. 4V of voltage , in each leg.

[seventenths]

Originally posted by udailey
Hmm. Really does not make any sense to me. I wont try to pretend that I get it But I will try to understand it so I dont waste anyones time.
I tried reading up on voltage multipliers but for my situation it really didnt make sense. Plus that F3 is the amp not the power supply where I have the problem, but then again I dont understand the circuits anyway and I kind of need to be led by the hand at this point.

If I have a schematic I can replicate it and maybe even have it work but I dont actually understand the circuit. I do understand algebraic equations in circuits. So if you were to say I need to use a voltage divider circuit with Vout=Vin*C1/(C1+C2) and put that circuit on each secondary lead in between the transformer and the rest of the PSU circuit, then I would understand how to do that. But right now thats the limit of "getting it" for me.

In short, I am still left baffled. It seems like you guys think its possible for me to use my 27-29Vout transformer and it would probably save me a lot of money to be able to do that. Is there a simple way to do so? If you would be so kind as to show me I certainly would appreciate it.

I would guess that those in the know would suggest that you look for a different transformer. As it sits, you'd have nearly 40 VDC after rectifying and filtering.


Cheers,
7/10

[Tea-Bag]

Originally posted by jackinnj

THD% for 1W, 1kHz is now 0.0063% from 400Hz to 80kHz without any trimming of the source resistors on the JFETs or MOSFETs, and without any attempt to get rid of extraneous noise -- so without too much difficulty it can probably be brought down much lower still, i.e. to the levels seen in the article. I am using Fairchild devices, no thermal protection, etc., etc.

The THD% is still mostly 2nd order, however.

Funny thing about the Fairchild devices -- the Vgs for a batch of them are pretty tight, but the P's and N's are over 1.4V apart.

Jack, have you had a chance to see if you can 'hear' a difference here?
I dont have a scope, can probably borrow one from work....

I am wondering if this amp will be hard to keep sounding as expected without some serious DIY tools.
I look forward to understanding these tweaks that maybe needed as DIY F5's come to fruition.
Mike

[Nelson Pass]

Originally posted by jackinnj
I put a 1,000uF/35V cap on each of the rails immediately adjacent to the trimpots -- any lead length here adds to the stray capacitance.

Secondly -- I put 3.3n/100V polyester across R5 and R8. The ringing on the square wave test (200kHz, 1 Watt) is almost gone. Good enough for me. I tried values from 470pF to 47nF and these work best for me. YMMV.

THD% for 1W, 1kHz is now 0.0063% from 400Hz to 80kHz without any trimming of the source resistors on the JFETs or MOSFETs, and without any attempt to get rid of extraneous noise -- so without too much difficulty it can probably be brought down much lower still, i.e. to the levels seen in the article. I am using Fairchild devices, no thermal protection, etc., etc.

The THD% is still mostly 2nd order, however.

Funny thing about the Fairchild devices -- the Vgs for a batch of them are pretty tight, but the P's and N's are over 1.4V apart.

That's great. I haven't tried extra decoupling, but I'll play with it
next chance.

If you can still see 2nd harmonic at 1W, then some Source power
resistor trim can probably take that lower yet. If you see the
2nd harmonic positive lined up with the positve output, then
you want to slightly reduce the resistor on the N channel device,
and vice versa. Wit this I routinely get .001% to .002%.

I don't really hear a difference on that, but I don't see any
reason not to nail it if you have an analyzer.

And yes, my P's are over a volt higher Vgs also.

[jackinnj]

Originally posted by mithomas
Jack, have you had a chance to see if you can 'hear' a difference here?
I dont have a scope, can probably borrow one from work....

I am wondering if this amp will be hard to keep sounding as expected without some serious DIY tools.
I look forward to understanding these tweaks that maybe needed as DIY F5's come to fruition.
Mike

I think that this will work great just out of the box, not requiring exotic tools at all. Just be mindful of keeping the leads as short as possible on the input stage.

[fredlock]

Hi Jack,

See attached. Just want to verify if this is this your current setup?

[jackinnj]

For the moment I was replicating NP's setup -- thus I had ONE 1K resistor to both gates of the JFET's and 100K from gate to ground.

I am just using Dale 5W 50R resistors as I have a few hundred in stock. Thus 2 100R in parallel are unnecessary.

I have 1000u/35V at each rail. The potentiometer (trimpot) is part of the input so it really likes the power supply end at a very low impedance source. Elsewise it is having its immoral phase relationship with the C(gate-drain) of the JFET.

[defect9]

Originally posted by mithomas



If your talking about the owners manual then yes. I think there is a problem.
My personal feeling is that Nelson is still trying to get us to figure out the PS differences from the unregulated F4, if any, or how to sneak a few more volts in. The sixmoons article looked to be the same.

the users manual, which also contains the F5 article, works for me now:

http://www.firstwatt.com/downloads/F5-om_sm-080527.pdf

lots of tasty things in there, especially the last page

-Jared

[EUVL]

> Secondly -- I put 3.3n/100V polyester across R5 and R8. The ringing on the square wave test (200kHz, 1 Watt) is almost gone.

Would a small cap at the input JFET gates to ground not be a better alternative ?


Patrick