Mcandmars ramblings \ build thread...

[Paul Joppa]

The current production C-3X may actually be made by Hammond.

[i luvmusic 2]

THANK YOU!

[mcandmar]

Latest experiment (eXperiment is in the name after all) was to replace the bias resistor in the driver stage with a resistor/LED string.  I had a couple of nice precision Dale 499r resistors that i used in the C4S, and a couple of HLMP6000's in stock so there really was no excuse not to try it.

If i worked it out correctly the manual states ~2.5v bias which implys to me the operating point is around 2ma.  On that basis a 499r resistor would give a bias of .98v, plus the 1.55v of the HLMP should give ~2.5v combined.  In practice i was seeing ~2.2-2.3v before the mod, and approx .2v higher on each side after the mod. Good enough.

Now for the hard bit, does it sound any different?  I could convince myself it sounds cleaner and more in focus. But that's probably because that is what i am expecting it to do.  Kind of kicking myself i didn't take some frequency response and distortion measurements with RMAA before hand.  Either way it cost peanuts, and i dont fancy playing a game of operation trying to get the soldering iron back in there without melting something so they will just have to stay

[Grainger49]

I feel the same way about a number of minor mods.  Taking them out is harder than putting them in.  If it didn't harm the sound they stay.

[Paul Birkeland]

You can use a TL431 or LM431 to get 2.5V of bias, much like an LED.  The middle leg goes to ground, and the two outside legs go to the tube's cathode. (The Stereomour uses this)

-PB

[mcandmar]

Interesting, i see in the Fairchild datasheet they have a graph for dynamic impedance showing a flat line of .2ohm up past 40Khz.  I assume it outperforms the HLMP if you are using it in the higher end kits?

Also have you ever taken any measurements of the S.E.X. amp?  I've started using RMAA and i'm not 100% sure of the accuracy of my plots. Its telling me around .3 to .4% THD, but the frequency response seems a little odd, large hump at the very low end and starts to roll off the high end sooner than i expected.

Cheers,

Mark

[Paul Joppa]

That looks about right to me - my design target for the output transformer was 40kHz at -3dB, which is -1dB at 20kHz. I can get much better numbers but it would not sound as good for other reasons.

A 1dB peak at 20Hz is reasonable for a parafeed output. It will change depending on the spaeker mpedance - did you load the output with a resistor?

Yes, I'm still on vacation in France, but nursing a cold and bored in the apartment so I'm checking the forum ...

[mcandmar]

Thats good to know, thanks Paul. The testing was done through a Pete Millett sound card interface so i'm not sure what the input impedance is, i think its around 600ohms.  I re ran the tests into 300r and 32r loads and the 32r load does indeed flatten the low end spike, but also doubles the distortion figure. I assume that is to be expected being a more difficult load to drive.  The 300r load is pretty much identical to going straight into the interface, which implies the interface is 300r or higher.

I switched out the HLMP/500r string with LM431's as per Paul B's suggestion and the high end frequency response jumped up to pretty much exactly what you specified, -1db at 20khz on the money.

Initially i only changed the left side of the amp and noticed the difference when i compared the left/right channel response. I then changed the right side to an LM431 and the frequency response jumped up there too so its 100% the difference between the two bias setups and not some random change in my testing setup.  I'm tempted to solder the stock resistor back in now and measure it out of curiosity

I debated about where to install the 431's as the free positions on the C4S board were very tempting, but in the end the added lengths of wiring from the socket to board swayed me into soldering directly onto the tube socket.  Lead lengths were a perfect fit too.

Thanks again guys!

P.S. Get well soon PJ, such a shame to have the sniffles when you are surrounded with nice wines.  ...assuming you drink.

[mcandmar]

Bonus Content:  Tung-Sol vs Sylvania Black Plates

Found this rather interesting, Sylvanias seem to have a slightly flatter low end response, but roll off the high end slightly sooner.  They also measure lower distortion figures too. I have attached both 32ohm and 300ohm load plots.

Obviously take all the results with a pinch of salt as results may vary from tube to tube. I dont know how many hours of use the Tung-Sols have, its probably close to six months since i got them. The Sylvanias were fresh out of the box, pretty sure i have never used them.  Its still interesting all the same...

[mcandmar]

Found a problem, having just plugged in some headphones i now have a hiss on both channels, i can only assume the LM431's are creating the noise.  Have you experienced this Paul, or is there a specific model of the 431 i should be using?  Failing that i may have to go back to the LED/Resistor string.

Cheers,

Mark

[Paul Joppa]

We did once get a bad batch of "431" chips - they may have been fakes, or factory rejects - that were very noisy. Hard choices there - resistor/capacitor is always quietbut you can hear the cap, resistor alone risks high frequency drive to the power triode, LEDs can be noisy (the predecessor of theHLMP-6000 was measured to be very quiet - but long ago), chip regulators such as the 431 can be noisy (the specs on the major brands say they should be relatively quiet, but fakes and rejects abound) ..

Thanks for the kind words guys, feeling much better now - I'll probably be pretty occasional on the forum over the next few weeks!

[mcandmar]

I did a little research and found a few sources who claim the On Semi LM431A measures the quietest so i ordered a few different brands from Mouser to see. Using the highly scientific testing methodology of a pair of Grado RS1 headphones on my head i found the TL431's were actually considerably quieter than the On Semi, Fairchild, and Ti LM431's.  Unfortunately it still wasn't good enough with a constant white noise in the background.

I found some test measurements which noted the LM431 performed much quieter when set for higher voltages, i assume that is due to the resistor added between the reference and cathode. The test report measured the LM431 noise in the 20uv range, when configured to 5v it measured down around 3uv, vs a selection of red LED's that all measured under 1uv.   It seems to me that adding the two resistors across the 431 would defeat the purpose of using the 431 in the first place, and similarly adding bypass caps would do the same so i came to the conclusion it just wasn't going work and threw in the towel.  Frustrating really as there was clearly better high frequency response and lower distortion while using it.

So back to the HLMP/500r sting, i did one final frequency response plot with the stock 1.27k resistor and sure enough the HLMP helps the high end response.  Its kind of a half way point between the stock resistor and LM431, but more importantly its completely silent.

Before i close the door on the LM431 idea i should point out i am using very sensitive Grado headphones, the noise may not be audible with less sensitive headphones or speakers so it may work perfectly fine in other setups.  As always, please report your findings

[ZacharyP]

I have been following this thread like trial and error gospel ever since I submitted my order for the sex 2.1, great job to all who contribute and especially mcandmar for sharing all the fun with us!

My kit and c4s kit should be arriving within a couple of weeks and I've already gathered some mundorf Supremes (2.2uf, and 0.1uf for parafeed and interstate caps respectively), two triad chokes, some 100r resistors for heater supply, and all I'm waiting on now is for my mind to finish an internal debate over a goldpoint or this; http://www.diyaudioblog.com/2011/03/elma-stepped-attenuator-populated-with.html?m=1

Anyway I plan to build the kit entirely stock (except for the attenuator just to make wiring it up easier).  Almost all of my listening will be with Alpha Dogs headphones and a set of speakers I'm building with my father's friend (the speakers are rated 8 ohm too I believe, maybe 4 though). 


First question, if anyone has experience with any of the Fostex mods, ZMF or MrSpeakers Mad Dogs, etc. should I wire for 8 or 4 ohms?  Any help would be greatly appreciated.  I'm leaning towards 8.

Second question:

In what order should I implement the upgrades and modifications if I wish to enjoy each individual change as much as possible?  Would putting the c4s in first only make all the changes brought by capacitors that much more transparent, or does it have an opposite effect?

Thanks for reading guys!

[mcandmar]

Glad to be of help, sorry about your wallet

I would start with 8ohms, i've stuck with that for a variety of different headphones, and since your planning on driving speakers that would make 8ohms an ideal setting.

As always, build it stock first, then move onto the Bottlehead options, impedance switch boards, then the C4S.   Basically you want to get all that up and running before you start adding in custom parts to the mix.

As for the upgrades, the capacitors will give you the most immediate change, but i would would fit them last as they limit access to the rest of the circuit around them and its all too easy to ruin a nice cap by hitting a hot soldering iron against it.

Also resist the urge to chuck all the parts in at once, best to use it for a while so you can fully appreciate the differences each change will make.  Enjoy, and keep us posted!

[ZacharyP]

Glad to be of help, sorry about your wallet

I would start with 8ohms, i've stuck with that for a variety of different headphones, and since your planning on driving speakers that would make 8ohms an ideal setting.

As always, build it stock first, then move onto the Bottlehead options, impedance switch boards, then the C4S.   Basically you want to get all that up and running before you start adding in custom parts to the mix.

As for the upgrades, the capacitors will give you the most immediate change, but i would would fit them last as they limit access to the rest of the circuit around them and its all too easy to ruin a nice cap by hitting a hot soldering iron against it.

Also resist the urge to chuck all the parts in at once, best to use it for a while so you can fully appreciate the differences each change will make.  Enjoy, and keep us posted!

I never really shared any pics of my old Crack's internals, I can definitely work with tight spaces after stuffing that headphone amp!

I attached some different shots of it between mods, without the capacitors mounted with zip ties of course, much neater in the box.  Mounting the choke at an angle to the transformer resulted in absolutely zero hum.  Dunno, I tried to follow my Morgan Jones amplifier design text to lead me.  For my first time with a soldering iron since my freshman HS electronics course (22 now) I think I did a decent job with the crack.  Now time to move on and keep improving.

Planning a much cleaner build this time; starting with the wooden base.  An extremely light burgundy stain followed by a linseed oil finish, cream white top plate (if I can find a powder coated to do it) with claret/red wine accents to match my Alpha Dogs.  Will definitely have to try to make a thread 1/100th as useful as yours!