diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Pass Labs (http://www.diyaudio.com/forums/pass-labs/)
-   -   The Valorous Amp - A modest aggregation of pass inspired ideas (http://www.diyaudio.com/forums/pass-labs/229302-valorous-amp-modest-aggregation-pass-inspired-ideas.html)

kasey197 4th February 2013 03:25 AM

The Valorous Amp - A modest circlotronic aggregation of pass inspired ideas
 
1 Attachment(s)
EDIT>>
Latest schematic is here:

http://www.diyaudio.com/forums/pass-...ml#post3594179

Hullo,

Here is the little result from a *LOT* of work these last 6 mths. Most of my time was spent finding out how NOT to make a hi-end amplifier with few parts that works well.. Until now :)

This is a simple circlotron with ideas culled from some of the excellent work by the august authors in this forum (with some small contributions from myself.) You can see shades of Rotharcher's AFC, NP's Aleph input diff stage & F6 thots plus some ideas for the CCS from Horowitz & Hill too for good measure.

As you can see, I am making no claims for originality :D

One of the nice things (to me)of this cct is that it takes unbalanced inputs directly & the output offset is well behaved. There is also very good symmetry if that keeps things on balance for you lols ...

The circuit works well and you can use hitachi laterals as well if you prefer instead of the semisouths. Jfet inputs will be played with another time as I'm quite happy with it as is.

The bias set and diff pots allow you to set up bias and harmonics to your liking. [edit- not quite true with this schematic - see post #9] I chose not to set the bias by varying R10 as the open loop distortion of the LTP ip stage is quite sensitive to the voltage across R6 and R7.

BTW., if you don't bypass the power rails, the amp will oscillate. For sure. I mean it will - really. Anyway....

V1 and V2 are the beefy rails. For V3 you can use pretty much anything as the current drawn is minimal.

Gain is double that implied by a cursory look at the FB resistors as the FB network sees only half the output amplitude.

Any questions/brickbats pls fire away! I'll be travelling soon so bear with me for any short delays in responding.

ps. the schematic is not simmable directly as I am quite unskilled at LT spice so don't know how to insert pots properly so these were hard-drawn :o

flg 4th February 2013 05:02 PM

Have you tried compensating the diff pair vs. the output for that osc problem? Or bigger gate stoppers or miller at the output?
You have no temp comp? Is it that the R100 and your input pair do well enough without any?
Alas it is not a follower? I have Circlotrons in my head as followers? I personally think we should dream up another name...

AudioLapDance 4th February 2013 05:20 PM

I hope it sounds as good as the title! :D

kasey197 5th February 2013 12:08 AM

Quote:

Originally Posted by flg (Post 3355719)
Have you tried compensating the diff pair vs. the output for that osc problem? Or bigger gate stoppers or miller at the output?
You have no temp comp? Is it that the R100 and your input pair do well enough without any?
Alas it is not a follower? I have Circlotrons in my head as followers? I personally think we should dream up another name...

I did not end up with any compensation caps since the source of the oscillation seemed to be poorly bypassed power rails. I added some caps as shown in the little box in the schematic and then all was well. I used 0.47uf caps as that was what I had in my box. Fingers crossed :) There's not much OLG hence fb is quite low so there's a decent chance we can do without 'em.

And you're right - it's not a follower - there are a few examples of circlotron output stages like this - eg the excellent circlotron by Mike Rothacher which is on the passdiy website - recommended reading! The simplest follower type circlotron using enhancement Fets is probably the one on audiodesignguide.com but that approach leaves us with the voltage gain stage to deal with (well, we can always use tubes as he has)

As regards temperature compensation - I did use the R100s for the outputs and suggested the hitachi laterals (2sk1058) as an alternative. The F6 thread has some very good discussion of this point and these are equally applicable here.

Zen Mod 5th February 2013 12:18 AM

put C1 between V3+ and R9/R14 node

kasey197 5th February 2013 12:23 AM

zm - that will reduce the noise from the current source ?

Zen Mod 5th February 2013 11:20 AM

at least in theory

;)

kasey197 5th February 2013 02:04 PM

Cool - will try ... Running it thru some torture tests now to see if anything breaks ;)

kasey197 7th February 2013 12:09 AM

2 Attachment(s)
Turns out that its not that easy after all to play with the harmonics balance and nulling in my design - the F6 really rules here :) I think the only way to do it properly is to have separate biasing and capacitor coupled ac drive - maybe I will add that to a to-do enhancement list after i get the basic design properly nailed down.

For now, things are progressing well. Sine attached and picture of the amp clipping at 47V attached as well. Can't see any signs of sticking or wobbelies during clipping so this should sound quite clean on transients.

At this point the circuit is exactly per the schematic. The pot on the diff pair source was left at very close to center and later when I get matched fets for the input I will ditch the pot altogether.

One thing I notice is that HF response is *NOT* great - its down 1.4db at 20khz. I know that would not be audible but still its annoying so I'll try to spend some time figuring out why. I think it could be down to either the input capacitance of the LTP mosfets and/or the bypass caps. These caps see audio signals so that may certainly be a factor but if I get rid of them the amp has a tendency to oscillate. Am trying smaller values & also different locations (ala FLG's suggestions) to see if I can do better - any ideas here would be welcome! Very large gate stoppers would prob work but the performance expense would be be considerable imo.

[unproven speculation alert] I think that circs have more tendency towards this as a spike in one half induces an opposite going spike in the other. With complementary op stage - this snuffs out the oscillation as both halves need to see the same phase drive to amplify. With circs, the opposite phase spikes get amplified and then that sets things off. Thats my theory anyway! So if that is true, then some sort of bypassing will be essential - the trick is finding out the most optimal way to do it...

flg 7th February 2013 11:24 AM

Some gate stoppers on the input MOSFETs? 200ohm maybe? Bypass your R12 with a tiny cap? 10p maybe? :D
The Power Supply could also be causing issues. I beleive N.P. said the F5 was not tolorent of caps right on the output transistors. It caused osc.


All times are GMT. The time now is 08:30 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2