Salas DCG3 preamp (line & headphone)

DCG3 with Toshiba 2SK170BL

Lovely sound!

First thing I noticed was how smooth yet detailed violins sounded. Significant improvement over the UPA's in this area. I have now listened for a few days with a variety of different material and the word that keeps coming to my head is balance. Nothing shouts out and yet it is all there. I thought that maybe the bass was slightly less but I was mistaken. It is just smoother. The entire presentation seems clearer and more resolved but not in a harsh or edgy way. It almost seems like the background quiet is quieter so instruments sound nicer and more correct.
Another important difference is in depth differentiation and resolution. When you are listening to an orchestra with certain instruments highlighted so these artists are in the front of the orchestra or for example tenors or sopranos in front of the chorus, the orchestra or chorus tend to get somewhat fuzzy in most systems. The 2SK170's do an excellent job here.
My DCG3 has the Diotec 327 mirrors and I thought it sounded great, but with the 2SK170's it sounds so much more refined. A bit more musical too. The overall improvement in sound quality is greater going from UPA68H to 2SK170 than switching from 560C to 327's in the mirror.
I experimented with 16.75v rails and 15v rails and in my present system I slightly prefer the 15v rails. The sound seems fuller and I seem to appreciate the newfound detail more. This is the DCBST with 18v transformers. Their heatsinks are now running at 55C as opposed to 45C at 17v. Could it be that the fuller sound I think I am experiencing at 15v is because the PSU is shunting more?
Thanks Salas for suggesting I try the 2SK170BL. I matched each pair under 0.1ma Idss with a spread of 0.35ma between the lowest and the highest. I also put the higher of each pair in the J1 position.
nash
 
diyAudio Chief Moderator
Joined 2002
Paid Member
K170 matched quad is an expensive thing these days and genuine is a hunt, hence the very decent monolithic uPA, but when you got them why not use them. The DCSTB PSU is not shunting because its series but more voltage across its pass elements makes them a tad more linear.
 
That's great nashbap, I have 30 original 2SK170s which were provided by a very polite american diyer, dshambala, 6 or more years back in the days Mouser used to store them. I hope I can get a quad matched.
What's the right margin for matching those, Nick ? (if I can't get it like nashbap down to 0.1mA)

Nashbap, one more thing, are you still using Takman resistors with DCG3 ? Unfortunately I can find the right values for Dale PTF so I've found 3 options for my relay attenuator update: CMF55, Takman (REY probably) and PRP. Any thoughts / experience with them ?
Hope I'm not hijacking the thread with this resistor talk.
 
That's great nashbap, I have 30 original 2SK170s which were provided by a very polite american diyer, dshambala, 6 or more years back in the days Mouser used to store them. I hope I can get a quad matched.
What's the right margin for matching those, Nick ? (if I can't get it like nashbap down to 0.1mA)

Nashbap, one more thing, are you still using Takman resistors with DCG3 ? Unfortunately I can find the right values for Dale PTF so I've found 3 options for my relay attenuator update: CMF55, Takman (REY probably) and PRP. Any thoughts / experience with them ?
Hope I'm not hijacking the thread with this resistor talk.

I was looking at a photo of the Ayre KX R preamp. It uses TC25 (I believe to be PRP) resistors. They wouldn't use them unless they sounded good.

Rush
 

Attachments

  • Arye resistors.jpg
    Arye resistors.jpg
    578.8 KB · Views: 348
Nash what are the Idss of your 2sk170’s that you used?


I used the following pairs by Idss:
7.39/7.37
7.47/7.41
7.53/7.49
7.63/7.61


I suspect the lowest distortion would be with matched pairs in each leg of the LTP. I would love to know which way the H2/H3 balance would go if there was a mismatch. For example, if I had two Jfets one 7.5 and another 7.2 Idss what result would I get with 7.5/7.2 in J1/J2 compared to 7.2/7.5 in J1/J2? Salas could you help explain?
Thanks. nash
 
Sorry to go back to this, but....

Better go for constant output resistance. You can fix a bandwidth and rise time that way. Steady for all volume positions. Zo+R1 make RC time constant with C1. The R1 C1 are there to ensure stability when pot is at max. In other words when reduced to simply a contact with no Zo. Or when there isn't any pot but a straight input.

I find 20k ordinary log pot's 5k Zo nice (see post#20). Makes 1MHz worst case or about there depending on wiring. But its not written in stone. In any case I wouldn't recommend high enough pot Zo capable of shortening the bandwidth below 200kHz.

LDR attenuators usually add very significant THD of their own. Its due to non linearity in the light elements.
The programmable one from here is one of the best to handle.

Does anyone have experience using an Autoformer Stepped Volume control (or Transformer Stepped Volume control)?

Is there sonic benefit to autoformer rather than say a high quality film type pot like a Penny & Giles or even the resistor stepped ones?

This site sells finished products as well as kits ($200 up):
intact audio

I have no experience with any of these yet.

Rush
 
Nashbap, one more thing, are you still using Takman resistors with DCG3 ? Unfortunately I can find the right values for Dale PTF so I've found 3 options for my relay attenuator update: CMF55, Takman (REY probably) and PRP. Any thoughts / experience with them ?
Hope I'm not hijacking the thread with this resistor talk.


Yes, I use the Takman REY 0.25w 50ppm version in the shunt positions and Shinkoh 1w in the series. As described in my earlier post I choose the combination values in my usable range that sound best. I have used Dale Rn, and KOA Speer in the shunt and Takman sounds the best. Also used low ppm KOA Speer, Vishay TX2575 in series and Shinkoh sounds best. I am considering the Caddock MK132 since Shinkoh's are now very scarce.
BTW if you do go with the Takman make sure you are getting the 50ppm ones. I get mine from soniccraft.com and they also have the PRP that Rush mentions.
nash
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I used the following pairs by Idss:
7.39/7.37
7.47/7.41
7.53/7.49
7.63/7.61


I suspect the lowest distortion would be with matched pairs in each leg of the LTP. I would love to know which way the H2/H3 balance would go if there was a mismatch. For example, if I had two Jfets one 7.5 and another 7.2 Idss what result would I get with 7.5/7.2 in J1/J2 compared to 7.2/7.5 in J1/J2? Salas could you help explain?
Thanks. nash

If I remember correctly from my DCG3 development experiments in 2016 where I tested K170s as well, you should now have less overall THD vs uPA68 due to higher yfs but third harmonic equal or more to second especially beyond 2kHz. Also less of a THD flat trend vs frequency. If there was a noise floor difference I don't remember but that would have stumbled on the mirror's contribution first I suppose.

I used a completely matched K170 pair in the tests not only in IDSS but for curves on the tracer. Thus I don't know what you may achieve in THD profile if you will deliberately differentiate. I don't expect much though due to corrective action of extra OLG converted to feedback with the more yfs.

There will be an Id difference between a pair due to offset trimming anyway.

You could run an FFT plot to see where you are now vs the uPA at least. By comparing past saved or published takes of yours.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
You got to experience it, its a very few active parts preamp and each different pick does change trends. Given the situation with NOS Toshiba fakes and matched quad pricing for rare genuine, if I was recommending the preamp on those only it would have been out of reach for many.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Does anyone have experience using an Autoformer Stepped Volume control (or Transformer Stepped Volume control)?

Is there sonic benefit to autoformer rather than say a high quality film type pot like a Penny & Giles or even the resistor stepped ones?

This site sells finished products as well as kits ($200 up):
intact audio

I have no experience with any of these yet.

Rush

Vgeorge I think in his #26 preamp. But he uses a ladder in DCG3. PM him if he will not not read this...
 
Attached please see two FFT plots. Both were done at 1v balanced in and 0.56v out using Focusrite/Arta The left is the UPA68H and the right is the 2SK170BL one. In my last posting #4041 I had shown an FFT but it was done at the default FFT8192. Now here as per Salas's suggestion they were done higher at FFT 32768.

To my untrained eye, except for the spikes under 1k(more on this later) the plots of the UPA68H and 2SK170BL look identical. Ear to tweeter test a slight reduction in hiss with the 2SK170.
Salas I isolated the spike at 320hz to the safety ground connection at the IEC. The PEM is a Schaffner with filter built in. When I pull out the wire to G at the PEM the spike goes down, reconnect and it goes back up. The left FFT with the UPA68H has it connected the right has it disconnected. I repeated the FFT for the 2SK170 several times, with and without the connection- without gave me THD+N of 0.0049% while connected gave me ).0050%.
Thanks.
nash
 

Attachments

  • DCG3balRch 15v UPA68H.png
    DCG3balRch 15v UPA68H.png
    72.2 KB · Views: 413
  • DCG3balRch2SK170BL15V rails.png
    DCG3balRch2SK170BL15V rails.png
    69.3 KB · Views: 407
Does anyone have experience using an Autoformer Stepped Volume control (or Transformer Stepped Volume control)?

Is there sonic benefit to autoformer rather than say a high quality film type pot like a Penny & Giles or even the resistor stepped ones?

This site sells finished products as well as kits ($200 up):
intact audio

I have no experience with any of these yet.

Rush


Yes, I have built it using two autoformers from Dave Slagle. I have only used it as a passive and this was after a Dale RN based passive. Compared to the resistor passive the sound was much fuller, stage bigger and a more enjoyable overall sound. Cant compare dynamics and detail retrieval to say the DCG3. Perhaps a DCB1 after it would really make it shine.
nash
 

Attachments

  • AVC Top (2).JPG
    AVC Top (2).JPG
    564.6 KB · Views: 379
diyAudio Chief Moderator
Joined 2002
Paid Member
Attached please see two FFT plots. Both were done at 1v balanced in and 0.56v out using Focusrite/Arta The left is the UPA68H and the right is the 2SK170BL one. In my last posting #4041 I had shown an FFT but it was done at the default FFT8192. Now here as per Salas's suggestion they were done higher at FFT 32768.

To my untrained eye, except for the spikes under 1k(more on this later) the plots of the UPA68H and 2SK170BL look identical. Ear to tweeter test a slight reduction in hiss with the 2SK170.
Salas I isolated the spike at 320hz to the safety ground connection at the IEC. The PEM is a Schaffner with filter built in. When I pull out the wire to G at the PEM the spike goes down, reconnect and it goes back up. The left FFT with the UPA68H has it connected the right has it disconnected. I repeated the FFT for the 2SK170 several times, with and without the connection- without gave me THD+N of 0.0049% while connected gave me ).0050%.
Thanks.
nash

Nice that you found the lower frequency spikes origin.

For THD looking the same between uPA & 2SK maybe the interface is at its limit and we just see its own?

Make a THD sweep with Arta Steps. There differences could arise.
 
Vgeorge I think in his #26 preamp. But he uses a ladder in DCG3. PM him if he will not not read this...

I used the autoformer at the output of the tube pre, in order to reduce noise and increase output impedance.
It worked really well in this position, but I do not think that DCG3 is the right application for an autoformer.
It is also, maybe preferred from a passive pre, due to the higher output impedance, at least when not in the least attenuation position. But of course you loose some bandwidth and resolution.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Cable tip

I didn't have any decent cable leftover lengths around for my deck but I had various male RCA terminals. While looking for Mogami local prices to get some bulk cable and terminate, I discovered that Roland offers decent quality ready made interconnects with narrow enough terminals to fit behind my cassette deck. Old gear almost always have small adjacent terminals clearance. My Amphenols don't fit, so I would have to use cheaper dubious terminals anyway.


Because the 1m Roland was below 15 Euro in a local music instruments store it wasn't costlier than making one myself with say Mogami W2524 Φ6mm guitar cable. That's how it looks inside on the package description. Like quality guitar cable. Other pro brands were far more expensive. So I thought to check one out. Roland - RCC-3-2R2R | Black Series Interconnect Cable

I measured 98pF/m at 1kHz on this terminated cable. Its very well made, soft, the bite on RCA barrels seems correct, not too hard but not loose. Its not completely separate runs, they attach together after 15cm from each end. Which is good for less interference by keeping a narrower loop area between channels. But they can also be pulled easily apart I believe. So I found it overall a handy purchase.

Maybe you don't want guitar cable with that extra anti-microphonic conductive PVC shield which is of no use in your home audio rig on the other hand. Mogami says its a necessary evil for noiseless stage handling not an enhancement.
 

Attachments

  • IMG_20191129_160451.jpg
    IMG_20191129_160451.jpg
    445.2 KB · Views: 316
  • IMG_20191129_160521.jpg
    IMG_20191129_160521.jpg
    490.1 KB · Views: 300