Salas DCG3 preamp (line & headphone)

diyAudio Chief Moderator
Joined 2002
Paid Member
If I remember correctly, Q1 and Q2 came as matched pairs.
If you can't adjust below this half volt offset on both channels, it might be you mixed them.
If that's been the case, you could try to unsolder one of the right and one of the left channel and switch places.

Can you possibly check by signal generator and oscilloscope (or even AC multimeter) if basic operation works well besides the offset?

But there is a trimmer and t should have been able to force some convergence for low mV DC offset. Even with over-correction which is not a good thing but electrically doable. Normally with well matched Q1 Q2 only 4 Ohms or so is the range of compensation by VR1. A wrong or dead BJT or FET or another mistake looks more likely for 530mV offset and not further responding to trim.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I shorted the signal inputs (granted the alligator clips are lousy, but I am getting continuity between the pads) and measured DC offset of 545mA. This is nearly identical between channels. Twiddling VR1 gets me no lower than 530mA. I can't get anywhere close to the few mV around zero offset indicated in step 7 of the build guide.

So, I think I must have made some mistake. May I please have some guidance as to the most likely places to look for misplaced parts, wrong values, etc?

Thanks!

By the way after checking the tabs insulation and some circuit voltages as I described, take an offset reading without shorted inputs too.
 
Mige0 and Salas, thanks very much for your prompt replies.

If I remember correctly, Q1 and Q2 came as matched pairs.
If you can't adjust below this half volt offset on both channels, it might be you mixed them.

A good point - this was one of the first things I checked. The mosfets are in the correct positions. I recall that when I was testing them with the jig in step 2 of the guide they seemed very well matched. Thanks for the suggestion.

Basic things like the right MOSFET types order or the electrolytics orientations look correct. Check the insulation of each MOSFET's tab to the metal bar sink with the DMM's continuity beeper first. It must not beep.

:rolleyes: Got it in one. Yes, there was a cracked mica insulator. I must have overtightened it during installation. I replaced all of them (cheap insurance) to be sure and fired up the DCG3 again.

With the fix it was easy to adjust offset down to +/- 0.5 mV (without op-amp).
I let it cook a while and the offset seemed to rise 20-30mV as everything warmed up. Readjusted and it seems stable enough at less than 1mV variation.
With the op-amp in the offset is controlled at less than about 0.2mV. :)

So, all in a very good result except for the one problem.

Thanks very much for your help. I will come back with listening impressions when the casework is finished (I hope over the holidays :hphones:).
 
You're welcome. Little patience and some planned substitutions along more play time will surely lead you to most favorite and synergistic configuration for amping your modded AKG K712 cans. Keep us posted.

You are quite right.
After another day of break in the veil has almost gone completely and my strict dual mono setup of the DCG3 directly driven by Lundahl line transformers plus the 4-wire modification of the K712 already begins to pay out in a very agile and lively presentation.

On heavy bass lines the picture gets wobbly though. Will go straight to the planned stage one RC decoupling mod as BC237-40 are in backorder.
 
Thanks.
Order was already placed, but will take its time. Have to be patient

Maybe due to much less 1/F noise vs 560s there is characteristically good bass definition with the 327-40s in this circuit.
Definition is really ok for the time beeing, its the whole acoustic painting that gets wobbled.
From my experience its not a break in issue but most likely a PS issue.
Speaking of paintings, the colours of the DCG3 on the K712 are as saturated as been painted by Rembrandt himself. Like it!


It's good for THD when with below 80 Ohm cans anyway.
I have read about it, but wanted to try 100mA first, to get a grip on that instance as well.

soon to be replaced by an OPA 2209 anyway
sorry mixed it up, the OPA1622 was meant as headphone driver replacement of course.
The OPA2209 will replace the inferior NE5532's in the TEAC
 
Last edited:
I cranked the bias up to 150mA but it didn't yield any fundamental improvement.
The RC decoupling of the first stage did the trick, though.

If one happens to like Joni Mitchell's style for example, there are many songs that very much relay on heavy bass lines to establish an extreme wide and deep frame of acoustic space. If the bottom register gets wobbly, theses songs do not work at all.

Really looking forward to the split +voltage PS mod with the UltraBiBs to come!

I like tantals and I like them even more when paralleled with MKT (+100nF) and silver MICA (+100pF). The negative is wired straight to my close focus quasi star mass.
 

Attachments

  • DCG3_Sch_V1.03_BC237_RC_decoupling.jpg
    DCG3_Sch_V1.03_BC237_RC_decoupling.jpg
    193.2 KB · Views: 454
  • DCG3_Sch_V1.03_BC237_RC_decoupling_parts.JPG
    DCG3_Sch_V1.03_BC237_RC_decoupling_parts.JPG
    170.6 KB · Views: 458
Last edited:
"Love Puts On A New Face" on the "Taming the Tiger" album for example.
You can take almost any song from the 70 / 80ies recordings. Joni Mitchell seemed to have had a very strong and intimate almost tribal connection to such bass lines spreading out into eternity, which is rather unusual for a female singer IMO. It came to me when I recently rediscovered Joni for me.

Even better probably would be, to try this simple mod and have a look what it can do to the favourite recordings of yours you may know well. Despite your aversion to tantals I'd suggest you to use them here for once, as IMO no other lytic comes close regarding stiff and stringent bass.
Effect should be even more pronounced with a split PS, as RC decoupling has rather limited isolation at the very bottom end of course.

Will be fun once I get there.
 
Last edited:
Try your K701 as well, as this is very revealing regarding three-dimensionality due to its stock 4-wire cable. Though tantals need less break in time than other lytics from my experience, a day or two would be good to wait before final judgement
Sure low Z of the UBiB should be most beneficial, but split PS even for UBiB may pay out. Will see.

For those that don't want to step up to UltraBiB for one reason or another, the RC decoupling mod may be a mod well worth in my opinion. It also helps in sharpening the focus throughout the audio bandwidth, increasing separation and body of singers and instruments as far as I can tell by now.
 
Last edited: