Salas DCG3 preamp (line & headphone)

YMMV of course, those are very good op-amps. Used various music genres but on this track comparisons were easier (pun intended)

Happiness Is Easy (1997 Remaster) - YouTube

One of my favorite bands of the eighties!
I will make a comparison of the chips again, as it was mostly with vinyl and my speakers. Next time with digital and headphones too.
Also keep in mind that I use ultrabibs for regulation.
But you can' t go wrong with any of these chips and it' s nice to experiment.
 
Also keep in mind that I use ultrabibs for regulation.
But you can' t go wrong with any of these chips and it' s nice to experiment.

PSUs can be a factor in synergy and preference of course. I listen with a standard DCG3. Forgot to mention that DCG3 measurements did not change between BOM AD823 U1 and younger slicker SMD ADA4625-2. THD behavior is the same, standing on the edge of my sound card's own loop performance. As for noise floor I doubt that the interface could show any differences at this region even if the 4625-2 certainly has better noise spec. The 1/F ascend below 100Hz is card's own already. DC servo's noise is not directly impacting the whole of course. Its a servo not an input op-amp. Unless its one with much dated general spec to can give side effects. I had seen a TL-072 adding measurable noise and THD difference when used as U1 DC-Servo.
 

Attachments

  • DCG3-Loop.png
    DCG3-Loop.png
    87 KB · Views: 195
  • DCG3-ADA4625-2.jpg
    DCG3-ADA4625-2.jpg
    396.3 KB · Views: 181
  • EMU-Loop.png
    EMU-Loop.png
    89.9 KB · Views: 187
FWIW, twice now I've ordered pairs of DN2540s (from Mouser) and although Idss differed 30% between the pairs, the pairs themselves were very closely matched.

I received a Peak Atlas Pro DCA75 component tester today. I understand Salas uses one of these also. It turns out that I have ten DN2540s in my junkbox so I tested them. The bottom line is they were all very closely matched.

Again, I've seen significant (30%) differences between batches, but within a single order, very close tolerances.

Salas: Question. The DCA75 stubbornly identifies two of the eight DN2540s as voltage regulators. It also refuses to identify all one of four IRF9610s I tried. I suspect it doesn't have enough resolution to work reliably with such high gm devices. I wonder if you've seen similar misbehavior with your tester.

I sent an email to Peak technical support; we'll see what they have to say.

-Henry
 
Thanks. That's strange. I really hope they don't tell me I have to send it back. It's very odd that it works on 8/10 DN2540s, but always fails on the same two devices. I thought I had two bad transistors so I clip-leaded up a current source with a 10 Ohm resistor and both of the suspect transistors biased up as expected at about 120mA with a 9V battery power supply. I then put a second 10 Ohm resistor in parallel with the first and current increased to around 200mA. It seems the transistors are OK, at least in this simple test.

I accidentally left a word out of a sentence in my earlier post. I should have written, "It also refuses to identify AT all one of four IRF9610s." In other words, it identified three of the four.

I'll let you know what they say.

-Henry
 
Excellent idea, Salas.

I just found two more DN2540s and they both identified fine. I retested my four IRF9610s and confirmed that one of the four comes up as "No component detected." I also have some SiCFETs and get the same no-component result on all of them, but that's an exotic transistor type so I wouldn't necessarily expect it to identity.

On the IRF9610, two-wire tests identify two diode junctions between drain and source/gate, and a zener from gate to source. I don't know enough about MOSFET architecture, and nothing about the Atlas firmware, so I can't speculate at all about what's going on.

I also tested two or three TO3 bipolar output transistors successfully. On the other hand, I tried some TO92 DN2540s and got more "Voltage regulator" results on some of them.

I'm running the same firmware as you. IIRC, I saw somewhere the firmware is years old, so it's evidently not under active development.

I'm a little disappointed, but for the price I think it's still worth it. Mainly, I want to use it to match small-signal transistors. I have 36 genuine 2SK170s. Every matched quad I can come up with saves me $29 from the diyaudio store.

I also confirmed the measurements I made on my bag of BC327s late last year using my homemade test jig. I may have mentioned earlier that I gave up testing them because they all came in in the low 400 hfe range and seemed so well-matched that I stopped trying to test them. I was wondering if my jig was no good, but he results from the DCA75 are basically the same -- which is kind of cool, I think (but not surprising).

-Henry
 
No, this is me🙂
I will try with bc560 too and with a better offset controller.I have an smd opa2134 but i haven't pins for the adaptor.

My opinion on OPA2134 servo is it tonally blends more towards the newer high spec chips we tried in DCG3. I had one in DIP-8 package that I also swapped during the last U1 session. So its worth listening with it to see what suits your system and taste better. Try it before changing the mirror transistors so not to mix impressions.
 
connector pin purgatory

BTW prepared yet another high spec U1 chip alternative on an adapter to try later on as DC Servo

Does anybody have a connector pin part number that fits the precise dip 110-87-308-41-001101 8 pin dip socket. recommended pin size is .4 to .56mm.

Trying to do the SOIC to DIP8 thing

For example, the AD 823 has a pin width of .46mm.

My sockets are all installed so it is a real pain to change.

For the life of me I can't seem to find a method to search via the pin size.
 
the Quest

Thank you Salas - saves the day. 4625 in transit. Anxious to see what it does. Next up is the UBiB "upgrade". The sound is great but I don't know how high "up" is. Detail and crisp/clarity is always number one for me. Thank you for the DCG3 and FSP.

I brought in a listener and he was super impressed with the sound. Family is very musical including his dad playing in the symphony so I was very happy with the confirmation of overall sound quality.

This and a board change on Mx2 are in process. Then the UBiB, Then I think I need some speaker work/changes to get everything out of what I have.

Very appreciative of the circuits you have designed and the great followup service. Thank you.

R,
Don
 
Don, thanks.

What I searched in new generation servo chips for DCG3 is even more definition than with the AD823 but without some tightening of the rich tonal delivery and large soundstage of the standard BOM preamp.
Until now I haven't found that elusive golden section chip. It's more like one way or the other between the BOM chip and the newer ones. Each way may possibly suit one's system better or not of course.
I have in mind to also find out about C5 value tweaking just in case it could reach a better subjective equilibrium with some certain new chip. This one is tricky though because it changes the servo's response time in correcting offset move. It may work well with some chip or lose too much tracking. All aspects must be satisfactory for a firm U1 recommendation update.

Let us know how the ADA4625-2 worked for your system synergy and personal taste.
 
Report:

I have paired my DCG3 with Hifiman Ananda headphones. Sounds amazing :hbeat: You could say it's not that surprising, as those cans are the easiest planars to drive. Still quite some discoveries were made with this buy.

I'm not an expert in headphones. My previous cans were Beyerdynamics DT250, Sony Mdr-cd900st. Closed back, low to mid-fi - both very enjoyable with DCG3. And then a friend showed me his 1st gen Hifiman He400i, which were in a different soundstage, separation, top end presentation league, while at the same mid-fi bracket. This made me curious how the ever praised Sundaras performed. The local store had only the new He400i and Anandas for back to back listening in late autumn. Anandas were better, but arguably not by the that huge price margin (like 700$ more) better. So I left it at that, untill the beginning of this year, when my DT250 broke down. Soon enough Hifiman sale presented a significant discount for open box Anandas and I was faced with a choice between them and Sundaras. The reviews were raving both as very neutral, best in their price class planars, yadda, yadda, yadda, the frequency response measurements are close, both have their strengths and weaknesses, my head spins, I pull a trigger for Anandas, with a 30 days return pillow silencing the pain.

I was quite sure I was returning them after the first two days of listening. Curiously, in the first half of the day, it was ok. While in the evening, I was willing to send them back. Mainly, because out of the box, they were too bright and not that exiting for me. Maybe on some recordings it could be justified with the extra air and better resolution, but on others, especially heavily produced ones it was getting tiring soon. Luckily then, I have discovered the world of headphones EQ. To begin with I was not looking for headphones that would need EQing. I was looking to have a reference sound out of the box. But after some experimentation and trying out some recommended curves, I have found out what a huge difference EQ makes. Everything just sounds so right and goosebumpy without compromising other wonderful qualities these headphones have.

There might have been many different things involved – subjective and objective. Subjectively, I might be more leaning closer towards the Harman curve. On the objective side, this particular make could be simply be underperforming on the lower end. I have learned, Sean Olive bought the Sundara model almost at the same time. His measurements were showing quite a bad low end roll off making the community to question the pads and what not.

In conclusion (notes for myself):

1. DCG3 is amazing.
2. Evaluation length matters a lot.
3. Evaluation time matters: morning and evening, rested or tired can influences hearing.
4. Reviews are good only for general guidance.
5. Reminder: it's a hobby of combining both subjective and objective.
6. I might need to build another DCG3. It's a hassle to move between the main system and the work desktop.

🙏 Thank you again, Nick, for everything you provide.
 
With premium headphones beyond better construction quality we usually get a wider stage, more detail, more sub bass. If you got all those over the Sundara you chose wisely on the Ananda. Law of diminishing returns regarding price and value is always there though.