• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

GB for DC coupled B1 buffer with shunt PSUs

Got the new Gerbers.
I can't tell too much more about the relays, other than BillyK's gift of BC550s did not stop the chatter either, so I think placing different bipolar there does not help, and that different relay's seem to be cause of problem.

Maybe more builders will come into the mix soon..If the international builders ever get there boards. Maybe the bom should make a preference to Omrom 12v, if they are easily available.

I'll be happy to send off one of my relays... It's 5v though. LMK
 
OK bad news for me, one of my beloved Mullards in the matched pair in mt cj PV7a has died. :( I cannot really use it for comparison now. I have some JANs in there now burning in.
Against the Diamante, no comparison really and I do think the Diamante sounds good, really good. Perhaps it's the Lightspeed, however it's just not in the same class.
I do not have what folks would consider high-end stuff. At the office where I started this journey I have a Mauro Pensa MyRefC set of mono blocks and NHT Super Zero speakers. At home here I have a Parasound amp and a some older Polks. Does not seem to matter, the DC-B1 withe a Lightspeed is a revelation to this poor soul. I am an emotional person and my wife kids me that I cry during movies and powerful musical passages, well I have been through a box of tissues today!

This is a special device, Salas, Thank you so very much! I did not think I could have this sound, and now I do.

Pink Floyd Dark Side of The Moon, track 4 you can hear the different ohh ahh, hii heyy and ho as well as the inflection and almost pain in her voice, could not hear that with the Diamante. Was disappointed in the inconsistancy of Snoopy's Christmas, the tracks varied wildly regarding image placement and quality. This was noticable on the Diamante, but was unpleasantly present on the DC-B1, however what I did notice clearly was the use of piano pedals you could really hear the cut off of the sustain or the increase depending on the pedal. I spent a good deal of time with some of my favorite live recordings, wow, wow, wow. Hot Tuna from the Ram's Head Tavern, they were in my room, I could hear Jorma's fingers on the strings, there was also a lot of emotion brought on by the dynamic range. My Grateful Dead was another revelation. There are some newly available "Matrixed" recordings, a skillful blend of soundboard and audience recordings that for the most part give the best of both and present a nice experience. Needless to say the DC-B1 add a new dimension to these too, in comparison to the Diamante.
Enough of my rambling, I am going back to my listening chair.
 
Sorry if ya'll tired of me.
I think I have a good feeling for what it is about the DC-B1 that is different, I think it is the ability to provide the "emotional" character of the music. When I listen to live music through the DC-B1 as compared to my other gear I get the emotional feelings I had at the live shows I attended. I think that sums it up nicely.:D
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I am an emotional person and my wife kids me that I cry during movies and powerful musical passages, well I have been through a box of tissues today!

That is the point. If music can involve, and transcend, it makes us limp for a fleeting glimpse to priceless truths, and lifts the weight of this mortal coil. Best education of them all.

P.S. Did the 12AU7 die in the CJ?. Best English was CV4003 IMO. And I liked the French Radio Cifte too. Had a couple with the crossed banners and 5 stars above them.
 
I've been patiently waiting for a "tester PCB" to arrive while keeping up with this thread. Tea-Bag sent the PCB's out promptly, but it takes a while to get to international destinations...10 working days to Australia is the norm. It should arrive this week. Thoughts about how to speed up the build occupied my mind...

I saw some recent posts about matching the 2SK170's in the buffer, starting around post #902. So with nothing better to do I tried that on 18 unmatched 2SK170's, using 10V as suggested by AndrewT, with a DMM reading milliamps, and the gate tied to source as in Tea-Bag's link here.

The current I measured was (in mA) 6.74 6.82 7.10 7.24 7.50 7.72 7.73 7.85 7.90 8.32 8.38 8.53 8.67 8.68 9.64 9.65 9.90 10.95
I tested 3 of them at 9V and the result was approx 0.1mA lower. Looks like a few good matches there.

Still feeling restless, I tried matching some LED strings of 3 and 5 to get 5.4V and 9V as per Salas' post #642. The forward voltage drop seems to depend on the current through the string, so I breadboarded the string of 3 with the 2SK170 and 100 ohm resistor, which gave a current around 7mA. On the string of 5 LED's, I breadboarded them with the 2SK170 and 10 ohm resistor and the current draw was 16mA. So I managed 5.40V for the string of 3 LED's, and 9.39V for each string of 5 was the best I could do, should be OK. Be interesting to see if the matched strings of 5 actually result in equal pos and neg rails...being anal I included the 2SK170 with each string that I matched in the bag with the LED's, ready to go into the PCB. The Idss of the 2SK170 can be used to fine tune each string, but who knows what happens when it all burns in. If the currents through each string really are different when the cct is built (7mA vs 16mA), I don't understand how we can get the ideal 1.8V voltage drop across each LED by using the same part number LED in both strings - maybe it doesn't matter that much.

AndrewT said in post #910
"Within 1% is so tight, many of us cannot measure Idss to this accuracy AND hold all the other parameters at the correct values. I certainly can't. I doubt whether I can measure an absolute 5% tolerance for Idss."

From this I gather that the Idss measuring method in the link is very basic, or is not taking into account many other factors, or maybe it just doesn't test the true definition of Idss, I don't know, I don't have the experience that others do here on this forum. It seems that we are aiming for a minimum DC output offset by having the 2 x 2SK170's matched. So I breadboarded a simple cct to try to simulate what might happen as far as DC offset goes. It might be a bad idea this cct, but I was curious. Cct shown below.

I put in my "matched pairs" of 2SK170, set the power supply to 20.00V, set the pot wiper to 10.00V, and measured between points A and B to try to measure the DC offset. The result of this little experiment was, (whether it be a good idea or completely stupid) that the "matched pairs" I came up with by using the test method from the link above, returned small DC offsets from 0.2mV to 2.4mV.

Pair Idss 7.72/7.73 offset 0.2mV
Pair Idss 9.64/9.65 offset 0.3mV
Pair Idss 8.67/8.68 offset 0.7mV
Pair Idss 7.85/7.90 offset 1.4mV
Pair Idss 8.32/8.38 offset 1.4mV
Pair Idss 6.74/6.82 offset 2.4mV

Swapping the positions of the upper and lower 2SK170 often changed the DC offset by about 0.2mV, but there is drift going on and final values after burn in could be different. Will be interesting to see what happens when I put the bits in the PCB. Thanks for sharing you guys, Salas, NP, Tea-Bag, crt and all the others I've missed.

Ian.
 

Attachments

  • test1.GIF
    test1.GIF
    2.4 KB · Views: 527
diyAudio Chief Moderator
Joined 2002
Paid Member
If the currents through each string really are different when the cct is built (7mA vs 16mA), I don't understand how we can get the ideal 1.8V voltage drop across each LED by using the same part number LED in both strings - maybe it doesn't matter that much.

Your 2SKs that are going to run the Vref 5 Led strings are going to work under a BJT's Vbe (BC550). The Idss you measure at 9 or 10V is going to fall at roughly 70% of that value. So match your Vref strings by testing your leds with 5-6 mA. Try to get elements that add up to about 9.4V for a row of 5. Nice that you do groundwork, its educating. But no real worries, as long as you gonna be in the ballpark it will play good. There is elasticity built in to it right from the start. So it could offset matching if made opposite. Reg Vout differences will only manifest by same voltage amount at clip. Nothing else, if your DC offset is at bay.

Bad scenario:

Imagine the catastrophy of clipping at plus 21.5V and minus 20.5V. Dreadful consequences will apply! Your 25V input sensitivity amp will show the assymetry right away on your 60dB sens 3kW speakers while your 30Vrms nominal output CDp is going to push the envelope even further...:Pinoc:
 
I came up with by using the test method from the link above, returned small DC offsets from 0.2mV to 2.4mV.

Pair Idss 7.72/7.73 offset 0.2mV
Pair Idss 9.64/9.65 offset 0.3mV
Pair Idss 8.67/8.68 offset 0.7mV
Pair Idss 7.85/7.90 offset 1.4mV
Pair Idss 8.32/8.38 offset 1.4mV
Pair Idss 6.74/6.82 offset 2.4mV

Swapping the positions of the upper and lower 2SK170 often changed the DC offset by about 0.2mV,
you are measuring Vgs. That is not quite the same as output offset.
But, it does allow a comparison.

Take the first three pairs.
Your measuring of Idss indicate that each pair have a mismatch of 0.01mA. That is 1part in many hundreds. i.e. a lot closer than <=0.3%.
Look at the Vgs variations. 0.2mV to 0.7mV. There is little correlation between the <0.3% selection and the output offset. That tells me the Idss was not measured to better than 0.3% it was more likley that it exceeded 2%. Now take these three pairs and set them up as an LTP, with the sources connected and the gates connected and now measure the volts drop across the 1k0+-0.05% load resistors. Connect gates to sources so that each half of the LTP is passing Idss. I guarantee that your load currents for all three pairs will not match to better than <=1%. i.e. your initial selection of 0.01mA is only good for batching into similar groups. Our DIY methods of measuring Idss and other parameters are flawed when attempting absolute measurements. That's why I compare one (DUT) against another (REF), I cannot achieve accuracy for absolute measurements and recognising this means I never fall into the trap of claiming absolute accuracy of my measurements.

Now look at the gross range of your six pairs. Here we can se the trend of rising Vgs as Idss difference increases. That conforms that the method has some reproducability. It is not worthless, just don't rely on it for absolute measurement.
The biggest problem as I see it, is holding Tj=25degC. I can never achieve that.

A final reminder.
The B1 does not need matched pairs. It simply requires selection of similar Idss pairs.
Any amateur measurement procedure with a bit of care can achieve that.
 
Last edited:
diyAudio Chief Moderator
Joined 2002
Paid Member
Be interesting to see if the matched strings of 5 actually result in equal pos and neg rails...

There are foes around...critters that sport Vgs, Vbe, TC, and other non matched or even batched vices.:D

If someone wants to see two same Vout regs will need either to swap the couple of X Vf leds AFTER assembly and test, or use 500R trimmers instead of the 5th Leds and dial in final Vout...which will be tempted to make somehow assymetric so to offset the audio part NJFET minimal offset!:headbash:

A guy did it. He is in the original building thread, he played with Leds and got em regs down to 0.05V dif.

His words:

''Here it is....

It took a bit long time to find closest voltage of LED (finally i found around 1,8V LED, but then i realize that even i have all same voltage on LED, it won't give the same output voltage to supply the buffer because we dont match the transistor and other components, so we should compromise this by adjusting our LEDs. Therefore the LED voltage between positive and negative side isnt the same but resulting the same output voltage). The output voltage is about 10,7V - 0V - (-10,7V) with +/-0.05V ripple. Offset is about 0.7mV at positive voltage and 1.7mV at negative voltage.

I can feel more detail in sound.''

Then Kannan matched nothing else, bought 'em 2SK pairs 2% from Spencer and got 0mV offset! But got a chattering relay if without a comp cap across C,B, on its driving NPN! Him and Tea-Bag only for now. Ain't life a bitch?:D

Beyond humor, one thing I know for sure. You will all ''feel more detail in sound''.:hphones::cloud9:
 
A final reminder.
The B1 does not need matched pairs. It simply requires selection of similar Idss pairs. Any amateur measurement procedure with a bit of care can achieve that.

Thanks Andrew, a lot of readers of this thread will be happy with that, considering there may be another 200 or so builders. When testing the Idss, just breathing/blowing on the unit under test would change the reading, you definitely want the windows and doors closed. With the data on the pairs I measured, another digit of accuracy would have helped, I posted what I found, no point to prove, just posted it to give builders a fairly rough idea of how close might be "close enough".
 
There are foes around...critters that sport Vgs, Vbe, TC, and other non matched or even batched vices.:D

If someone wants to see two same Vout regs will need either to swap the couple of X Vf leds AFTER assembly and test, or use 500R trimmers instead of the 5th Leds and dial in final Vout...which will be tempted to make somehow assymetric so to offset the audio part NJFET minimal offset!:headbash:

A guy did it. He is in the original building thread, he played with Leds and got em regs down to 0.05V dif.

His words:

''Here it is....

It took a bit long time to find closest voltage of LED (finally i found around 1,8V LED, but then i realize that even i have all same voltage on LED, it won't give the same output voltage to supply the buffer because we dont match the transistor and other components, so we should compromise this by adjusting our LEDs. Therefore the LED voltage between positive and negative side isnt the same but resulting the same output voltage). The output voltage is about 10,7V - 0V - (-10,7V) with +/-0.05V ripple. Offset is about 0.7mV at positive voltage and 1.7mV at negative voltage.

I can feel more detail in sound.''

Then Kannan matched nothing else, bought 'em 2SK pairs 2% from Spencer and got 0mV offset! But got a chattering relay if without a comp cap across C,B, on its driving NPN! Him and Tea-Bag only for now. Ain't life a bitch?:D

Beyond humor, one thing I know for sure. You will all ''feel more detail in sound''.:hphones::cloud9:

It is nice to know that people experiment and explore the fundamental electronic design as well test procedures in these threads. This makes interesting learning. That is what I like about this site. I read this along with NJfet RIAA and shunt regulated PSU threads many times and felt comfortable in using whatever LED ( I bought them at Radio Shack in the last minute thinking that I have not ordered earlier! ) that were available. After completion i realized that my large bunch of green LEDs from muser is sitting in a BAG!. I even lost three leds while trying to measure some voltages. It was fun . If you look at my photos I had connected the ALPS POT wrong. Now I corrected and it is playing it on my Nakamichi Receiever ( which is not great but still a very decent Power Amp- Nelson' Statsis topology) I connected this after the Volume control by connecting a seperate line inside the receiver to by pass all including its Volume control to check the quality of this as a preamp.
This one is extremely good as a preamp ( due both to PSU and buffer) very well detailed. I am not a good reviewer but I would recommend this to new kit builder. Easy to build and contains everything you need to build a preamp including line selection through remote relay/switch on delay. Only care you need is input side as the buffer impedance is high. And if all the information can be placed together in one kit help doc, it will be very easy for the builders. I am even planning on checking different relay to find the reason for chatter ( still mysterious why it happens to my kit) Nice to be in the testers list as you get the first hand chance to build , nice time I had - thanks Salas / CRT and ofcourse Tea Bag for their ivaluable contributions. I am waiting for the final PCB to use this for making only shunt ( for my experiments on other applications!) as this one will remain without any change as proto for this GB. I forgot to mention that i used 10K Alps Volume pot ( do I have to change it to 20K as originally recommended?)
kannan
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I forgot to mention that i used 10K Alps Volume pot ( do I have to change it to 20K as originally recommended?)
kannan

20k will be easier on your sources. Sweet spot value for being drivable enough and good on what DCB1 likes to see.

Is there any el cheapo stepped 20-25k on epay with fake Dale pre assembled for you to try?:D Must do well for targeting tight image focus at least.:bullseye:
 
I like my Power amp filters to be set to ~90ms and 0.7us.
I recommend that any source feeding the power amp should have the filters set ~ one octave wider at both extremes.
This requires the RC time constants to be (for me) ~ 190ms and 0.4us

A 10k volume pot would require 18uF of DC blocking cap to achieve that 180ms time constant.
A 22k volume pot could live with 8u2F for DC blocking.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Now I corrected and it is playing it on my Nakamichi Receiever ( which is not great but still a very decent Power Amp- Nelson' Statsis topology) I connected this after the Volume control by connecting a seperate line inside the receiver to by pass all including its Volume control to check the quality of this as a preamp.
This one is extremely good as a preamp ( due both to PSU and buffer) very well detailed. ... I am even planning on checking different relay to find the reason for chatter ( still mysterious why it happens to my kit) Nice to be in the testers list as you get the first hand chance to build , nice time I had - thanks Salas / CRT and ofcourse Tea Bag for their ivaluable contributions.
kannan

The chatter is your Zen for balancing out you got spot on lucky with offset. Your missmatches worked right on the money.:$:

So just bcs the DCB1 has got Pass DNA and your Nak too, Zen goes with the territory.:hypno2:

On a serious note, its nice to see if another relay is silent without the comp cap. Happened in two cases, one was Tea-Bag's Nais 12V, the other was your Omron 12V, yet the other guys got no problemo with exactly those types and the 5V Nais too. I would change a 7812 before, which is easier, just out of curiosity. The FORTUNATE thing is that just a ceramic cap KILLS it if it occurs and nobody will have to cripple his PCB bcs of the odd relay coil interaction.:happy1:
 
The chatter is your Zen for balancing out you got spot on lucky with offset. Your missmatches worked right on the money.:$:

So just bcs the DCB1 has got Pass DNA and your Nak too, Zen goes with the territory.:hypno2:

On a serious note, its nice to see if another relay is silent without the comp cap. Happened in two cases, one was Tea-Bag's Nais 12V, the other was your Omron 12V, yet the other guys got no problemo with exactly those types and the 5V Nais too. I would change a 7812 before, which is easier, just out of curiosity. The FORTUNATE thing is that just a ceramic cap KILLS it if it occurs and nobody will have to cripple his PCB bcs of the odd relay coil interaction.:happy1:

Yes - I have also a feeling about 7812 behaving badly ? When I get time this week try out 317T or another 7812 just to check out. Cap solution is the easiest and quick rather than trying out anything else at this point in this GB effort. Nice to see the total tally of requirement is exceeding 225. I expect this to cross 300 the least.
kannan
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I like my Power amp filters to be set to ~90ms and 0.7us.
I recommend that any source feeding the power amp should have the filters set ~ one octave wider at both extremes.
This requires the RC time constants to be (for me) ~ 190ms and 0.4us

A 10k volume pot would require 18uF of DC blocking cap to achieve that 180ms time constant.
A 22k volume pot could live with 8u2F for DC blocking.

I would add that if a source is a phono, better go 3u3 to 4u7 in it for 22k because considerations of deliberate mild sub LF filtering come in to play. If with largish speakers, the small for such wavelengths living rooms tend to behave as compressed chambers with lots of reverberation build up even with minimal excitement by the largely subdued output of almost any speaker so low. That can find its way back to the TT's tunings.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Yes - I have also a feeling about 7812 behaving badly ? When I get time this week try out 317T or another 7812 just to check out. Cap solution is the easiest and quick rather than trying out anything else at this point in this GB effort. Nice to see the total tally of requirement is exceeding 225. I expect this to cross 300 the least.
kannan

You would have to do creative dead bug maneuvering if with a 317T. Different pins order and a voltage set divider needed. Maybe start with a randomly different batch nobody knows where made 7812.
 
Disabled Account
Joined 2002
Got the PCB's yesterday. I am still gathering parts but I will try to do the building faster than normal.

Very nice PCB but I discovered the Deluxe is "mirrored" as the power section is on the right. Not a big problem but most gear has the power supply at the left. Minor inconvenience.

I miss a decoupling cap on the 7812 but I guess that's what you spoke about. No problem as a small cap can be soldered straight at the 7812 output legs.