Thai Tao

Hi, What's the current stable gerber file for CFA2? There is too many version in the google folder. Thanks!

Thanks, i will study the schematic more

Hi Kelvin, If i understand the schematic correctly, the CFA2 will have quite high open loop gain right? The CFA3 although output stage doesn't have gain like other buffer, it still wrap around the input stage which is also high loop gain so the zero feedback here only apply for the current buffer, right? My point is if really need that high current gain, then change the first mje15030 to pzta06 we will have current gain x5 in worst case ( 20 vs 100 ), and if we continue to the second mje15030, we will continue to get another x5 current gain and the pzta06 will still be able to handle 20ma bias. Or is the reason is we need thermal couple? Thanks!

hi @kevin gilmore, may i ask why there are 3 emitter follower at the output stage? I think it's just too waste silicon when only the outer most pairs do the heavy lifting. Instead can we just change to 2 emitter follower with the first one use small transistor like bc546 or mpsa06 so we can take advantage of high hfe to have high input impedance and the last one use mje15030 to do heavy work?

Hi all, I'm planning to order some board, do you guy think gold immersion and 2oz copper thickness make different in sound compare to HASL and 1oz?

Sparkoslabs make solid stuff with very detail document. Amb o78/o79 implement tps7a4700/3301, the sparkoslabs output noise is 3uV at 5v compare to 4uV of tps7a4701, 16uV of tps7a3301, PSSR wise, the sparkoslabs win completely at 120db compare to 80db of 4700/3301.

I found these picture of the Liquid Crimson https://www.head-fi.org/threads/new-cavalli-audio-liquid-crimson-amp.737694/page-14#post-14019045

so spice indicate at least 0.02% thd at 300ohm, real world measurement must be worse, can't understand their spec 0.0014% at 50ohm but at the same time i dont' believe some big company like cavalli do some marketing shit like that.

There's one thing i can't understand is that according to massdrop measurement, this amp has really low THD+N : only 0.0025% https://www.head-fi.org/threads/massdrop-x-alex-cavalli-tube-hybrid-amp-cth-dropping-monday.857673/. If it's a simple tube gain stage coupled to solid state buffer like the original CTH no way it could have such low THD. Even the schiit Vali2 with global feedback has higher distortion than the Massdrop CTH.

That is a tottally different design.

Yeah, i know it's liquid carbon, not liquid gold but as you stated that liquid carbon is mini liquid gold, i assume that the topology is the same, may be the liquid gold has better PS, transistor, ... etc. You seem to surf taobao a lot . I don't understand why you always want to reserve engineering cavalli product while you always call their product shitty? I'm also very curious about cavalli designs because i have heard the ehha, it sound very good and i really like reading detail his diy document. I guess the Liquid Crimson is like Schiit Jotunheim with input swap by tubes? I don't know other way then that you can make a balanced amp with only 2 tubes. Anyway, can't wait until you publish other cavalli design

So you said the LG is just a bigger LC, and now we can have a Massdrop LG for only 299$

I don't think the input jfet (2sk170bl) is fake as i buy it from a very high reputation local shop, the volume is 50k type, i will try it with 10k. Thanks for your suggestion.

Hi Kelvin! After some time, i have discovered that my amp have some serious problem: when i turn the volume up, after half volume the noise is increased to very audible level, so i measure the output with nothing connected, maximum volume and got this: Some serious 50hz harmonic at -55dbv lvl ! At min volume, it's fine: I thought it could be the power supply problem but i measure the PS and it's fine: So the problem must come from amp board, do you think what can cause this? Thanks!

Thanks you guys, actually that engraving and drill hole for audio jack (the volume pot hole come with the stock panel) cost me only 5$, incredible cheap!

Finally 3 or 4 months after finishing the krell, i have time to case it up. The wiring take me more time than populate all the components for the amp boards.

Yes, i have, input is 1Vpp sine wave or squarewave, output get from gnd-O+ and gnd-O-. The trace is a bit blur because steam get stuck on the screen, in reality, it look sharp. 1Khz, 1V/div 20Khz, 1V/div 1Khz/ 1V/div 20Khz, 1V/div

Yes, i know i should sound better than this, i left it cover in dust for months because it doens't sound so boring and i don't have a real reference (dynalo 1 or 2) to compare to, until now i have confident to say that it doesn't perform its true level so i finally ask here for advice. My power supply is +/-17v, ouput current for each transistor is about 13ma ( 0.27V across 20r). The power board kit i got from ebay, the kit is supplied with op27 and it sound awful, i swap for opa134, ad797 and lm6171 and it get much better, i find it sound best with lm6171. I also have the golden reference board, it sound slightly better than the Krell PS. I also modify another golden reference board back to walt jung super regulator with d44h11/d45h11 output transitor and this PS sound really good too. Right now i think i will make a power supply in perfboard with mc7812/mc7912 (which i have in hand) to try if it sound any different.

Yes, they're in parrallel so that's fine, i will post a better picture of entire board soon. I put 4 tl081 in, offset measure 0mv but i don't think it sound different from before, tried removing the connection between 2 boards input ground ( the ground at input+ and input-), it sound a bit different but when i switch to the original dynalo it's still a whole league better: it's sound smoother and clearer at the same time, the sound is more 3d, more open, imaging is better. Yes they share the same ground plane, but from my understanding, the output ground have high current flowing, and it should go back to the power source, where the current come from with it own path to reduce ground loop and interact with clean signal ground.

Yes, i use 10k pot. Here it is: Yes, i already checked that, i think if i mixed it up, it was already smoked.

May be i'm the only one who is not pleased with the Dynalo MK2? I have heard the original dynalo long time ago for brief time and it give me very strong impression, it sound very good. Then i decided to build the susy Dynalo but it doesn't sound good at all, i don't have totl head amp, just some entry level like Cayin C5, O2, Project Polaris and Schiit Vali and my Susy Dynalo doesn't sound better than any of these. Until today, finally i have borrowed the original dynalo and it just sound clearly better then my susy: it sound clearer, detail pickup is easier, better soundstage depth, everything sound more real! And this dynalo even using cheap transformer ( i can get it for less then 5$), the regulator is just basic 7815/7915. My build is pretty normal, the pzta56/06 and that340 is ordered from mouser, dale, vishay,bc resistor, panasonic fc , kemet mmk5 rail decoupling capacitor got from local store, the krell clone power supply board is a kit from ebay seller. My susy measure fine, 0.27V across 22ohm resistor, <5mv output offset, just can't understand how it sound like this.

Hi, how much current does this amp draw each rail?

I have finished my dynalo, everything measured ok, 0.3V across 20R resistor - 15ma for each output resistor at +/-18V rail, output offset for out+/- to ground is about <2mv without servo. Listen to some music, it sounds so so, not very good and gain seem to be low, i feed a 1Vpp sine wave at the input and get only 1.8Vpp sine wave at output, i use default resistor value with rin =5k, rf= 25k, does this seem strange?

Hi KG, I remember in 1 post you said that the default susy has gain of 6db in se mode, that mean 2, but i tried to simulate it in ltspice but it has gain of about 4, this got me confused as the feed back resistor is 25k and 5k, so the gain should be 25k/5k+1 = 6?

The P-channel is mmbf5460 then is the N-channel mmbf5457?